Genes & Development: Molecular and Logical Themes
7 – 12 May 1990
Organisers : Fotis Kafatos and Spyros Artavanis-Tsakonas
Spyros Artavanis-Tsakonas (Yale University), Seymour Benzer (California Institute of Technology, Pasadena, USA), Sydney Brenner (Medical Research Council, Cambridge, England) Thomas W. Cline (Princeton University, USA), William F. Dove, organiser, (University of Wisconsin, Madison), Gerald R. Fink (Whitehead Institute for Biomedical Research, Cambridge, USA), Antonio Garcia-Bellido (Universidad Autonoma de Madrid, Gyorgy P. Georgiev (Académie des sciences de l’URSS, Moscou), Peter Goodfellow (Imperial Cancer Research Fund, London, UK), Ira Herskowitz (University of California School of Medicine), Jonathan Hodgkin (University of Wisconsin, Madison, USA), David S. Hogness (Stanford University School of Medicine, USA), H. Robert Horvitz (Massachusetts Institute of Technology, Cambridge, USA), François Jacob (Institut Pasteur, Paris, France), Fotis Kafatos, organiser, (Harvard University, Cambridge, USA), A. Dale Kaiser (Stanford University School of Medicine, USA), Nicole Le Douarin (CNRS, Nogent-sur-Marne, France), Ruth Lehmann (Whitehead Institute for Biomedical Research, Cambridge, USA), Anne McLaren (MRC Mammalian Development Unit, London, UK), Barbara J. Meyer (Massachusetts Institute of Technology, Cambridge, USA), Mark Ptashne (Harvard University, Cambridge, USA), Janet Rossant (Mount Sinai Hospital Research Institute, Toronto, Canada), Lucy Shapiro (Stanford University School of Medicine, USA), Patrick Stragier (Institut de biologie physico-chimique, Paris, France), René Thomas (Université libre de Bruxelles, Belgique), William B. Wood (University of Colorado, USA)
A number of investigators accustomed to the study of gene regulation in microbes have been involved in studying development directly with metazoons. A group of such biologists joined together, autour de François Jacob, in May, 1990, at the Fondation Les Treilles in Provence, France, to discuss their experiences and prospects. This essay explores four themes on which there was extensive conversation: the strengths and limitations of the operon paradigm; the dialogue between gene action and morphogenesis; the formal logical elements of complex biological systems; and “back to the bench”, challenges for developmental genetics in the 1990s.
The operon paradigm and its limitations
Mark Ptashne presented a scheme, supported by considerable experimental evidence, in which the molecular mechanism of action of regulatory proteins in eukaryotes, including higher organisms, could constitute a reenactment of the same molecular principles discovered initially in the regulation of bacteriophage λ.
In outline, the following molecular modules could be defined as sufficient and necessary:
– A DNA-binding domain that recognizes a short sequence of DNA, usually about 5-8 nucleotides, which provides the address.
– A dimerizing domain that allows the protein to form homodimers with itself or heterodimers with another recognition protein to enhance both selectivity and affinity.
– A patch for direct or indirect interaction with another protein that itself has a patch interacting with the transcriptional machinery.
– Patches for interaction with other sets of proteins of the same type. This feature allows action at a distance, from several addresses, which in higher organisms are not confined to a few dozen base pairs at the 5′ ends of the genes. There are even effects from paired homologs. The network of interactions permits the assembly of a complex regulatory agglomerate (“regglomerate”) with many different components. The specificity rules for interaction may be quite relaxed (LIN et al. 1990). All of this can be, and has been, tested by experiments of the kind done most extensively with the GAL4 regulator of yeast, by artificially joining different domains (Brent and Ptashne 1985).
A good example of formation of heterodimers to create a new regulatory protein comes from yeast, where the regulator a1/α2 is formed by association of a1 and α2 polypeptides (Goutte and Johnson 1988).
A spectacular example of this combinatorial association has been described by Schulz et al. (1990).
The howls of protest were generally of two types: The molecular mechanisms and How might such regulation occur in cellular development?
Herskowitz summarized the yeast situation (see Herskowitz 1989a). Haploid yeast cells (either a or α) are partially differentiated: they have receptors on their surface and they produce the mating factors. When prospective mating partners approach each other they signal their presence with these factors, thereby inducing the final differentiation into cells competent for mating. The mating factors arc differentiation signals that induce the expression of various genes involved in mating (such as for cell fusion per se). There is a good argument that the mating pathway in yeast culminates by regulating the activity of a transcriptional activator protein, STE12, which might well be activated by phosphorylation.
The dialogue between gene action and morphogenesis
Embryogenesis involves processes that are visibly complex in both space and time. Gastrulation reaches levels of complexity that baffle the imagination of those comfortable with the one-dimensional character of chromosomes, genes, and polypeptides. Are developmental genes qualitatively distinct in their complexity (Waelsch 1989)?
A reduction of this issue of developmental complexity is to ask how one-dimensional genes and their polypeptide products can give rise to three-dimensional patterns. In Caulobacter crescentus, the products of certain flagellar fla and chemotaxis (che) genes are specifically directed to the swarmer cell in the polarized division process. Two mechanisms were discussed by Lucille Shapiro.
The methylated chemotactic protein encoded by the mcp gene carries a carboxy-terminal segment that is necessary and sufficient for segregation to the swarmer cell. By contrast, the gene (flaK) encoding the flagellar hook protein directs its transcript selectively to the swarmer cell. Fusions of the flaK promoter alone to the marker polypeptide neomycin phosphotransferase can direct the marker antigen to the swarmer cell. In this case, the problem of gene product segregation is replaced by a problem in selective transcription. The issue becomes whether the swarmer cell contains a distinct transcription factor for the flaK promoter, or a distinct chromosomal template in which the flaK gene is active, or both.
Herskowitz summarized a mutational analysis indicating that the axial pattern requires the action of five genes, two of which are unnecessary for the polar pattern. For at least three of these pattern-determining genes, null alleles are fully viable. Thus, the cell division patterns of S. cerevisiae seem to be completely dispensable in the Iaboratory; the prospects for an extensive genetic analysis are correspondingly large!
The nematode C. elegans illustrates the importance of cell division patterns in metazoan development (see Horvitz and Sulston 1990). Bill Wood summarized the three early cleavage divisions by which the anterior/posterior, dorsal/ventral and left/right asymmetries are established in this species. He has begun to describe mutants affected in left-right asymmetry.
Spatial morphogenesis is but one aspect of embryogenesis. How does the morphology of a particular stage lead to the patterns of gene expression that generate the next stage? Patrick Stragier gave several illustrations, from studies of the sporulation process in B. subtilis, of particular morphogenetic steps that activate new patterns of gene expression (Stragier and Losick 1990).
A mutant described by Stragier exhibits a perforated forespore membrane and fails to synthesize σ even though the DNA-binding regulators of σ have been synthesized. Thus, the cascade of transcription factors that drives the sporulation process in B.subtilis is thoroughly coupled to morphogenesis in the developing system.
Dale Kaiser gave examples of factors that act between cells to promote development of the microbe Myxococcus xanthus. The gene product of the csgC gene that acts at the final stage, C, is a 17-kD protein tightly associated with cell membrane.
Interestingly, the successful transfer of C factor requires intimate cell alignment, which normally depends on the motile behavior of the bacterium. The notion of “microenvironment” is appropriate to the C-factor interaction in M. xanthus, just as it is to the action of the Steel gene in murine hematopoiesis.
The richness of analysis of these issues in the development of Drosophila was convcyed by Spyros Artavanis-Tsakonas, Seymour Benzer, Tom Gline, Antonio Garcia-Bellido, Dave Hogness, Fotis Kafatos and Ruth Lchmann, but we lack the space to describe their observations.
Formal logical circuits
The concept of operon is generative; not only is it sufficient to cover the diverse situations that prompted its discovery, but it helps in imagining new situations that it may also cover. An amplification of the operon paradigm prolongs and extends it rather than replacing it; the paradigm uses the elements of the operon and similar elements as building blocks for ever more elaborate networks. As developed by René Thomas, the individual wheels of these nets are feedback loops (Thomas and D’Ari, 1990).
A negative feedback loop can abolish gene dosage effects. Thus, Shapiro observed that when the Hook operon of Caulobacter is cloned in a multicopy plasmid, there is no gene dosage effect. And the negative loop involving the λ cro gene presumably results in the levels of products of cII, O and P being relatively insensitive to the number of gene copies. In the absence of such regulation» the rate of synthesis of these products (which is already sufficient for replication when there is a single copy of the λ chromosome) would be over 100 times higher after 20mn. Such levels are probably unnecessary for O and P and toxic with regard to cII.
A positive feedback loop may associate with a choice. The detailed mechanisms vary from case to case but the principle is the same: positive autocontrol by the λ cl gene (Ptashne); positive autocontrol by various σ factors in B. subtilis (Stragier); induced synthesis of receptor and mating factor by the yeast mating factors so that, by positive reinforcement, a and α cells are indeed making a commitment to differentiation and a commitment to each other (Herskowitz); a positive control loop involving the Sxl gene in sex determination in Drosophila (Cline); and a positive control loop of the C-factor gene in Myxococcus (Kaiser). Thomas suggested that a fruitful avenue for cloning developmental regulatory genes is to screen for activities that have a positive effect on their own expression level, directly or indirectly.
Chains of control units are also an important feature of the networks involved in development. Primary sex determination in Drosophila and Caenorhabditis involves multiple regulators acting in series, as summarized by Hodgkin (sec Hodgkin 1990). In contrast, Anne McLaren noted the apparent simplicity of primary male determination in mammals by a positive regulator (see McLaren 1990 and references cited therein). But λ regulation was also simple at first glance.
The genetic analysis of development:
Three topics were intensively discussed.
The saturation genetics approach: One of the fundamental genetic strategies for identifying the components of a process, whether it be biosynthesis of histidine or mating. ability of yeast, is to isolate mutants defective in the process and then figure out what the wild-type genes do. Can this “saturation genetics approach” identify all of the important molecular protagonists?
The simple answer is that many important genes and proteins have been identified in this way, including several that fill the bill of being regulatory proteins of the type conceived by Jacob and Monod. The discovery of these proteins by genetic methods represents a tremendous advance in understanding the programming of development, but is it complete? The tools that we use to study the process may introduce a bias in our view of the process.
Essentiality and redundancy: There are many genes identified in fruit flies, nematodes, mice and other organisms in which the canonical mutation is leaky and in which null alleles cause inviability. In fact, MCM1 of yeast was originally identified genetically in this way. Although genetic analysis can identify important genes such as MCM1 of yeast and daughterless of Drosophila (Cline 1983), special mutations of these genes that allow viability but confer a mutant phenotype are obviously much rarer than mutations that simply inactivate genes. When Benzer discussed some of the more than 100 genes known to affect eye development in Drosophila, Garcia-Bellido pointed out that, for the vast majority of these genes, null alleles are lethal. Barbara Meyer developed this theme in depth for sex determination in C. elegans. The control gene xol-1 (XO lethal) is vital for dosage compensation (for review see Hodgkin 1990).
Redundancy strikes fear in the hearts of geneticists. Bob Horvitz described at least three different types of functional redundancy, and lie and Gerry Fink gave examples of these from nematodes and Arabidopsis. He characterized them as redundancies of genes, of pathways and of cells. McLaren noted the analogy to the principle of “double assurance” from experimental embryology. There are even examples in nematodes and yeast in which genes are triply redundant: three ACE genes in nematodes (Rand and Russel 1984) and three CLN genes in yeast (Richardson et al. 1989). In the cases, certain mutant phenotypes are observed only in triply mutant strains!
Arabidopsis provides a good example of redundant pathways. Fink described recent studies indicating that it has two independent pathways for tryptophan biosynthesis (Berlyn, Last and Fink 1989). Inactivation of any of three genes in one of the pathways causes a leaky tryptophan requirement, apparently because the other pathway can provide some tryptophan. Nematodes provide an example of redundancy at the cellular level. Horvitz described studies of the excitation of certain intestinal cells by neighboring neurons called AVL and DVB. Although ablation of cither neuron has no phenotype, ablation of both causes a mutant phenotype, defective defecation. Thus, genetic analysis of either the AVL or DVB cell type must be performed in a strain defective in the reciprocal neuron.
Given functional redundancy, is it ever possible to obtain mutants defective in such a function? In cases where one gene of a multigene family has been cloned, dominant negative versions can be designed in vitro and then introduced into the genome, where overproduction is likely to cause a mutant phenotype.
Janet Rossant described the new era of gene knock-out in mouse genetics; it is now possible to use a mouse gene mutated in vitro to inactivate the wild-type gene, by first modifying cells in culture and then deriving a mouse with the mutation in its germ line. That is the good news. Now the bad news: sometimes the mutant has no overt phenotype! One explanation for this disappointing result is functional redundancy. A strategy to contend with this difficulty is to begin another mutant hunt using the silent mutant as the starting strain. This approach has been used successfully in yeast, nematodes, and fruit flies, but of course it has no guarantee of success and can be quite cumbersome.
For the mouse, the tools for genetic analysis are under active development. Jacob raised the conundrum that patterns in the expression of lacZ insertions do not match classical embryological lineages. Peter Goodfellow noted the difficulties of identifying sex-determining elements by analyzing translocation chromosomes. Gyorgy Georgiev explored the development of transfection assays for genes controlling metastatic cell behavior. The new genetic tools in mice may soon participate as partners comparable to transplantation and cell culture analysis of the neural crest as summarized by Nicole Le Douarin.
Identifying interacting components:
Along with saturation genetics and gene knock-outs using cloned genes, another important genetic strategy for the nineties is the identification of interacting components. At least four different schemes were discussed in the formal sessions and during the unforgettable Provençal meals. The central idea is to use one mutant strain as a springboard to identify another gene, one that is functionally related in some way to the original gene.
The prototype has recently been described for a potentially powerful new method for identifying interacting components. Dubbed by the group the protein interaction trap, this method is based on the two-domain structure of the yeast transcriptional activator protein GAL4-one (D) the DNA-binding domain and the other (A) necessary for transcriptional activation (Brent and Ptashne 1985). Fields and Song (1989) have exploited the fact that these two domains must be physically associated in order to function. They have shown that the two domains can be brought together with two hybrid proteins, D/P1 and A/P2, where P1 and P2 are protein segments that associate with each other. Fields and Song suggest that it is in principle possible to use this as a screening method. For example, a protein segment of interest (P3) could be attached to the GAL4-DNA-binding domain to form D/P3. Then a library of hybrid proteins (A/Pi), formed by joining the GAL4 activation domain to random coding segments, could be examined for one that can associate with the DNA-binding domain. The regglomerate may be dissected in this way.
Sydney Brenner gave the following overview in the opening conversation.
Biology is concerned just as much with particular implementations as with general principles. Gunther Stent, long ago, put forward the idea that development was a trivial problem because, given the Jacob-Monod model of gene regulation, all development could be reduced to that paradigm. It was simply a matter of turning on the right genes in the right places at the right times. Of course, while absolutely true this is also absolutely vacuous. The paradigm dues not tell us how to make a mouse but only how to make a switch. The real answers must surely be in the detail.
However, there is a global constraint, and that is the connectedness by descent of all present living systems and the impossibility in evolution of going back to the drawing board once a certain level of complexity has been reached. Thus, anything successful that appears in evolution will be retained and elaborated as organisms advance in complexity and will be found over and over not as the instantiation of a general principle, but only thorough the continuity of utility.
Excerpted from S. Brenner, W Dove, I. Herskowitz, and R. Thomas.
Genes and Development : Molecular and Logical Thèmes.
Genetics 126:479-486 (1990)
Spyros Artavanis-Tsakonas: Cell interactions in neurogenesis
Seymour Benzer: Flies’eyes
Thomas W. Cline: Drosophila sex determination
William F. Dove: Developmental genetics of vertebrates
Gerald R. Fink: Duplicated functions in yeast and arabidopsis
Antonio Garcia-Bellido: The Drosophila wing
Gyorgy P. Georgiev: Metaslasin 1, the gene related to control of melastatin behaviour of tumors.
Peter Goodfellow: Sex determination in mammals
Ira Herskowitz: Cell specialization in yeast
Jonathan Hodgkin: Nematode sex determination and sexual dimorphism
David S. Hogness: Genetic control of Drosophila segmentation and of coordination of tissue development
H. Robert Horvitz: Cell lineage, cell death and neuronal differentiation in nematode development
François Jacob: Patterns of gene expressions in transgenic mice
Fotis Kafatos: Functional and regulation of insect chorion genes
A. Dale Kaiser: Signalling between cells regulates fruiting body development in Myxococcus
Nicole Le Douarin: Cell lineage in neural crest ontogeny
Ruth Lehmann: Genetic analysis of development
Anne McLaren: Mammalian genetics / sex determination
Barbara J. Meyer: Sex determination and dosage compensation in the nematode C. elegans
Mark Ptashne: Mechanism of control of gene transcription
Janet Rossant: Mammalian developmental genetics
Lucy Shapiro: Control of temporal and spatial organisation during cellular differentiation
Patrick Stragier: Cascades of sigma factors
René Thomas: Formal description of Systems comprising feedback loops
William B. Wood: Axis formation, Ieft-right asymmetries and cell interactions in Caenorhabditis embryos.