• Submit Earth, Atmospheric, and Planetary Sciences Papers
  • Science Sessions: The PNAS Podcast Program

Properties of developmental gene regulatory networks

  1. Eric H. Davidsona,1 and
  2. Michael S. Levineb,1
  1. aDivision of Biology, California Institute of Technology, Pasadena, CA 91125; and
  2. bDepartment of Molecular and Cell Biology, 401 Barker Hall, University of California, Berkeley, CA 94720
  1. Edited by Gertrud M. Schüpbach, Princeton University, Princeton, NJ, and approved September 11, 2008 (received for review July 24, 2008)


The modular components, or subcircuits, of developmental gene regulatory networks (GRNs) execute specific developmental functions, such as the specification of cell identity. We survey examples of such subcircuits and relate their structures to corresponding developmental functions. These relations transcend organisms and genes, as illustrated by the similar structures of the subcircuits controlling the specification of the mesectoderm in the Drosophila embryo and the endomesoderm in the sea urchin, even though the respective subcircuits are composed of nonorthologous regulatory genes.


Developmental gene regulatory networks (GRNs) provide the specific causal links between genomic regulatory sequences and the processes of development (1,3). They consist of the regulatory and signaling genes that drive any given process of development and the functional interactions among them. The design features of the GRN directly explain why the events of a given process of development occur; for example, why a given set of cells becomes specified to a given fate, why it emits particular signals to adjacent cells, and why it differentiates in a given direction. The architecture of a GRN is mandated by the cis-regulatory sequences of the enhancers that control each gene of the network. These sequences determine what inputs affect expression of each gene, and how these inputs operate in a combinatorial fashion.

The individual components of a complex developmental process are in general controlled by GRN subcircuits, and it is their architecture that illuminates the basic logic of development. In this Perspective, we summarize several types of subcircuits that are used in development and focus on those used for similar functions. The structural features, or architectures, of subcircuits provide new insights into the basic developmental processes they control. Table 1 lists several examples.

Table 1.

Network logic: Commonly encountered subcircuits and the complex developmental jobs they do

Double-Negative Gate

We first consider the double-negative gate (see Fig. 1A). This is a counterintuitive feature of network design that is a critical component of both the sea urchin and Drosophila GRNs controlling early embryogenesis. It is used to install a specific regulatory state in a specific region of the developing embryo. Instead of employing a developmental control gene as principal activator, it operates by relieving an otherwise global inhibitor through localized transcriptional repression of the inhibitor in a specified domain. Here are some examples:

Fig. 1.

GRN subcircuits used in both Drosophila and sea urchin development. Horizontal lines indicate the relevant cis-regulatory modules of the indicated genes, and the bent arrows attached represent the transcription unit; incident vertical arrows indicate positive transcription factor inputs into these cis-regulatory systems, and barred stems indicate negative inputs; double sideways “v” indicates the Notch receptor for the Delta ligand; “Regulatory State” represents the sum of expressed transcription factors at the given time and place. (A) The double negative gate. (B) Notch signaling circuitry. In Drosophila, the delta ligand is expressed in mesoderm (Meso), and Notch signaling occurs in adjacent neurogenic ectoderm causing expression of the midline gene sim; in the sea urchin, the Delta ligand is expressed in the skeletogenic mesoderm (SM), and Notch signaling occurs in adjacent nonskeletogenic mesoderm, causing expression of the pigment cell regulator gcm.

In the specification of the skeletogenic mesoderm of the sea urchin embryo the double negative gate consists of activation of genes encoding a lineage-specific repressor, Pmar-1, which prevents expression of a global transcriptional repressor of the genes that establish the skeletogenic regulatory state (3, ,4). Interestingly, the same basic regulatory logic is used in the oral ectoderm of this embryo. Here, the domain-specific repressor Goosecoid is used to repress expression of an otherwise pan-ectodermal transcriptional repressor, thereby permitting a set of regulatory genes to be activated only in the oral ectoderm.

In the Drosophila embryo, the mesodermal Snail repressor prevents transcription of Tom, which inhibits processing of the Notch signaling ligand, Delta (5). This results in mesoderm-specific expression of the Delta ligand, with the consequence that Notch signaling is constrained to the adjacent cells where it activates sim and other genes required for the specification of the ventral midline of the neurogenic ectoderm (2, ,5). Although not explicitly described, the entire anterior–posterior patterning network of the Drosophila embryo is also composed of a series of double-negative gates, in which broadly distributed gap gene repressors establish domains of pair-rule gene expression by cross-repressive interactions among themselves (6).

What is the functional advantage of this unexpected design feature compared with simple regional transcriptional activation? The double-negative gate appears to be an effective mechanism for ensuring spatially restricted patterns of gene expression, and for actively preventing these genes from functioning elsewhere in the embryo by sequence-specific transcriptional repression. In addition, this mechanism might relieve the regulatory complexity of highly pleiotropic genes such as those encoding Delta and other cell signaling molecules. For example, Delta is used in many different cell types at numerous stages in the development of virtually all animal embryos. It might not be feasible to dedicate a separate enhancer for every aspect of Delta activity. The double-negative gate provides a means for producing cell-specific regulation without the need for a distinct and dedicated enhancer.

Transcriptional Exclusion of Alternative States

A second design feature of GRN subcircuits is the cryptic, intraterritorial repression of other possible regulatory states during the process of specification (Table 1). We say “cryptic” because such functions are not revealed until they are experimentally interrupted, with the consequence that the state of the specification switches because of the absence of the repressor. Often, the function is to prevent a set of cells from responding to a signal that they are exposed to. For example, in the sea urchin embryo the skeletogenic mesoderm emits a Delta ligand, which in consequence of Notch signaling in the adjacent mesoderm cells activates transcription of the gcm regulatory gene (Fig. 1B). The skeletogenic cells are prevented from responding to Notch signaling themselves by the intraterritorial repression of gcm.

An exactly parallel mechanism is used for the localized expression of sim and the specification of the ventral midline of the neurogenic ectoderm. As discussed earlier, the Snail repressor leads to localized synthesis of the Delta ligand in the ventral mesoderm of the early Drosophila embryo via a double negative gate. This triggers Notch signaling in adjacent cells. In principle, the ventral mesoderm could respond to the Delta signal as well, but the target gene of Notch signaling in the neurogenic ectoderm, sim, is specifically repressed by the intraterritorial Snail repressor (Fig. 1B) (2). As in the case of the double-negative gate, we see an example of a regulatory design that is used in very different embryos to implement a similar job.

Such exclusion functions are a widespread feature of GRNs, and examples have been noted in Xenopus, Caenorhabditis elegans, and mouse, as well as Drosophila and sea urchin (7). This is one of several GRN design features that contribute to the remarkable stability and reproducibility of animal development.

There is a further exact parallel between the wiring leading to mesodermal gcm expression in the sea urchin and to neuronal sim expression in Drosophila. After their initial activation by a double-negative gate that induces the mediator of Notch signaling (SuH), both genes lock down their respective states of expression by means of positive autoregulation (Fig. 1B) (8, ,9).

Other General Subcircuit Design Features

Table 1 includes additional GRN subcircuit design features that are responsible for the execution of specific developmental functions.

The “community effect” (1) is an intraterritorial signaling function, in which all cells of the territory produce and respond to the same signal ligand. The underlying circuit design feature is that the gene encoding the signal ligand is activated by the same signal transduction system that it triggers in adjacent cells. The general consequences of community effect circuitry are: (i) establishment of a regulatory state through the activation of a set of regulatory genes downstream of the signal transduction system in all of the cells of the territory; (ii) averaging, that results in installation of approximately the same levels of expression in all of the cells within the territory; (iii) maintenance of the regulatory state so that, if signaling is interrupted, the regulatory state is lost.

There are many examples. In the sea urchin embryo, specification of the oral ectoderm depends on a community effect where the ligand gene is nodal (10, ,11). Similarly in the specification of the endomesoderm a community effect obtains where the ligand gene is wnt (12, ,13). Within the ventrolateral mesodermal territory of the Xenopus embryo the gene encoding eFGF similarly responds to a Brachyury input (14), but brachyury gene is activated in consequence of the FGF signaling pathway (15).

As extensively discussed in refs. 1,3, one of the most widespread and important equivalences between GRN topology and developmental function noted in ,Table 1 is the use of intra and intergenic feedback loops for stabilizing transcriptional regulatory states. We have already seen examples of autoregulatory feedback (,Fig. 1B), and many cases of cross-regulatory intergenic feedback have come to light in GRNs from sea urchin, mouse, Drosophila, and other developmental systems. These typically appear in the network architecture just downstream of the transient signals or other initial states that trigger the first transcriptional events in a specification process. They consist of sets of 2 or 3 coexpressed regulatory genes, the cis-regulatory systems of which use as positive inputs the transcription factors encoded by other gene(s) of the interacting set. Thus, these transcriptional feedback subcircuits function as “dynamic state lockdown” motifs; once the participant genes are under each other's transcriptional influence, the regulatory state cannot turn off, and these genes together with others they control represent the regulatory state defining that specification event. Thus, paradoxically, a stabilization function depends on continuing dynamic transcription. Other mechanisms of transcriptional state lockdown, such as those operating at the chromatin level, lie downstream of these GRN subcircuits.

Dynamic Aspects of Regulatory State Change in Space and Time

Just as dynamically functioning transcriptional circuits can produce stable regulatory states, the static genome can produce dynamic spatial and/or temporal transcriptional patterns. Such processes are also mandated by the design features of specific GRN subcircuits. One well-defined example is the generation of pulses of gene expression, which rise and then fall because of regulatory autorepression. The same regulatory design produces either peaks of gene expression or successive oscillations mediated by cis-regulatory autorepression (Table 1). Several examples have been discovered in the sea urchin and sea star embryo GRNs (,16).

The blimp1 gene presents an interesting autorepression system of this kind in the sea urchin embryo. blimp1 turns itself off after some hours of expression when its product accumulates to high levels, and then acts as a repressor. However, blimp1 also provides a required input into the gene encoding the diffusible signaling ligand, Wnt8, which in turn is locked into a feedback relationship with blimp1: The Wnt8 ligand induces Tcf to activate blimp1 expression. The consequence of this unexpected subcircuit design is that an expanding torus of coordinately expressed and coordinately silenced wnt8 and blimp1 expression sweeps across the endomesoderm of the embryo. It begins in the centrally located skeletogenic mesoderm, then as Wnt8 ligand diffuses outward, it extends to the surrounding nonskeletogenic mesoderm and finally to the peripheral endoderm, while, in each previously active domain, the expression of both is extinguished due to blimp1 autorepression (13).

Additional regulatory genes are also entrained in the expanding torus subcircuit (17). In an accompanying article in this Special Feature, this GRN subcircuit is shown to control both Wnt and Notch signaling across the entire endomesoderm (,18). The subcircuit not only includes autorepression but also transrepression of notch by high levels of Blimp1 and of another gene that repressively controls delta expression. It combines in one almost all of the individual subcircuit functions/architectures shown in Table 1. Its overall developmental role is to ensure a dynamic, Boolean march of successive and exclusive domains of Wnt8 and Notch signal transduction that moves peripherally from one fate domain to the next as development proceeds.

GRNs in Metazoan Development and Evolution

Here, we have focused on examples of the ways in which GRN subcircuits mandate developmental logic. The design features we have considered are devices used to drive the development of all animal embryos, and as the parallelism illustrated in Fig. 1 shows, disparate organisms, in different tissues, using different genes, nonetheless execute similar developmental decisions with the same circuit designs. We believe that in the near future a repertoire of such GRN subcircuits will be revealed, a repertoire that has been assembled in countless combinations throughout the evolution of diverse body plans among the Metazoa.

This PNAS Special Feature contains 10 articles covering a variety of contemporary topics relevant to the role of gene regulatory networks in animal development and evolution. The first 2 articles, by Hobert (19) and Hong et al. (20), respectively, provide Perspectives on two long-standing problems in metazoan development. Hobert discusses recent advances in our understanding of the gene regulatory networks responsible for the specification of individual neuronal cell types in C. elegans (19). Hong et al. (20) summarize the use of postgenome technologies in determining how different concentrations of the Dorsal transcription factor produce a variety of gene expression patterns in the early Drosophila embryo.

The next 4 articles are original research papers that present new insights into our understanding of how gene regulatory networks control different aspects of embryonic and postembryonic development, as well as changes in body patterning during animal evolution. The articles from Tumpel et al. (21), Nikitina et al. (22), Ochoa-Espinosa et al. (23), and Smith and Davidson (,18) describe advances in basic embryonic patterning processes, including the specification of the posterior hindbrain in vertebrates, the specification of neural crest progenitors in lampreys, the combinatorial control of A/P patterning of the Drosophila embryo, and the specification of the endomesoderm territory in the sea urchin embryo.

Two more articles are devoted to one of the major challenges in developmental biology, namely, unraveling the complex regulatory networks underlying the formation of postembryonic tissues and organs. The article by Ririe et al. (24) examines vulva development in C. elegans, with an emphasis on how gene networks coordinate individual cells to produce a complex organ. Georgescu et al. (25) examine the fascinating problem of T cell specification and diversification in the mammalian immune system. Evidence is presented for dynamic networks that are generally more plastic and reversible than those seen in hard-wired developmental processes such as endomesoderm specification in the sea urchin.

The final 2 research articles address problems in the evolutionary diversity of animal morphology. Gross et al. (26) explore the genome organization of the Mexican cavefish, Astyanax mexicanus, in an effort to understand the basis for its peculiar mode of adaptation, including the loss of eyes. Finally, Usui et al. (27) investigate the large sensory bristles (macrochaetae) of the adult fruitfly as a paradigm for understanding the evolution of morphological diversity.


We thank Susan Marty for superb administrative assistance and David Stopak for his editorial oversight. This work was supported by the National Academy of Sciences Sackler Colloquium program and National Institutes of Health Grants HD37105 (to E.H.D.) and GM34431 (to M.L.).


  • 1To whom correspondence may be addressed. E-mail: davidson{at}caltech.edu or mlevine{at}berkeley.edu
  • Author contributions: E.H.D. and M.L. wrote the paper.

  • This paper results from the Arthur M. Sackler Colloquium of the National Academy of Sciences, “Gene Networks in Animal Development and Evolution,” held February 15–16, 2008, at the Arnold and Mabel Beckman Center of the National Academies of Sciences and Engineering in Irvine, CA. The complete program and audio files of most presentations are available on the NAS web site at http://www.danielhellerman.com/SACKLER_Gene_Networks.

  • The authors declare no conflict of interest.

  • This article is a PNAS Direct Submission.


  1. ?
  2. ?
  3. ?
  4. ?
  5. ?
  6. ?
  7. ?
  8. ?
  9. ?
  10. ?
  11. ?
  12. ?
  13. ?
  14. ?
  15. ?
  16. ?
  17. ?
  18. ?
  19. ?
  20. ?
  21. ?
  22. ?
  23. ?
  24. ?
  25. ?
  26. ?
  27. ?

Online Impact

  • 667601844 2018-01-21
  • 385442843 2018-01-21
  • 496686842 2018-01-21
  • 915288841 2018-01-21
  • 885256840 2018-01-21
  • 726268839 2018-01-21
  • 25737838 2018-01-21
  • 143837 2018-01-20
  • 725372836 2018-01-20
  • 469379835 2018-01-20
  • 935424834 2018-01-20
  • 37813833 2018-01-20
  • 904770832 2018-01-20
  • 626160831 2018-01-20
  • 447659830 2018-01-20
  • 133161829 2018-01-20
  • 190427828 2018-01-20
  • 734726827 2018-01-20
  • 956922826 2018-01-20
  • 93519825 2018-01-20