Computational Models for Neurogenic Gene Expression in the Drosophila Embryo

Current Biology 16, 1358–1365, July 11, 2006 ª2006 Elsevier Ltd All rights reserved

DOI 10.1016/j.cub.2006.05.044

Report Computational Models for Neurogenic Gene Expression in the Drosophila Embryo Robert P. Zinzen,1 Kate Senger,1 Mike Levine,1,* and Dmitri Papatsenko1 1 Center for Integrative Genomics Department of Molecular and Cell Biology University of California, Berkeley Berkeley, California 94720-3204

Summary The early Drosophila embryo is emerging as a premiere model system for the computational analysis of gene regulation in development because most of the genes, and many of the associated regulatory DNAs, that control segmentation and gastrulation are known [1–5]. The comprehensive elucidation of Drosophila gene networks provides an unprecedented opportunity to apply quantitative models to metazoan enhancers that govern complex patterns of gene expression during development [6–11]. Models based on the fractional occupancy of defined DNA binding sites have been used to describe the regulation of the lac operon in E. coli and the lysis/lysogeny switch of phage lambda [12–15]. Here, we apply similar models to enhancers regulated by the Dorsal gradient in the ventral neurogenic ectoderm (vNE) of the early Drosophila embryo. Quantitative models based on the fractional occupancy of Dorsal, Twist, and Snail binding sites raise the possibility that cooperative interactions among these regulatory proteins mediate subtle differences in the vNE expression patterns. Variations in cooperativity may be attributed to differences in the detailed linkage of Dorsal, Twist, and Snail binding sites in vNE enhancers. We propose that binding site occupancy is the key rate-limiting step for establishing localized patterns of gene expression in the early Drosophila embryo. Results and Discussion Arrangement of Dorsal and Twist Sites Is Critical for vNE Enhancer Function Enhancers associated with the rho, vn, and vnd genes (Figure 1D) direct localized expression in ventral regions of the neurogenic ectoderm (vNE) [16, 17]. vNE enhancers are activated by the combination of Dorsal and Twist, but repressed by Snail in the ventral mesoderm. All three vNE enhancers contain conserved arrangements of Dorsal (Dl) and overlapping Twist (Twi)/ Snail (Sna) binding sites (‘‘DTS’’ composite elements, see Tables S1 and S3, as well as Figure S2, in the Supplemental Data available online). The composite Twi/ Sna site ‘‘points toward’’ the closely linked Dl site. These sites are essential for gene regulation [e.g., 17]; for example, point mutations that eliminate the core Dl

*Correspondence: [email protected]

consensus sequence in the rho DTS cause a severe loss in vNE activity [18]. However, it is unclear whether the exact arrangement of sites is also important for gene expression. To assess the importance of DTS organization for vNE activity, we inverted the orientation of the composite Twi/Sna site within the vn DTS (Figure 1C). According to the results of SELEX assays (see Table S3), this inversion should not alter the binding affinity because the modified sequence still conforms to an optimal binding site, though on the opposite strand. However, the modified vn enhancer is strongly impaired. There is a severalfold reduction in the levels of reporter gene expression driven by the mutant vn enhancer as compared with the wild-type enhancer in transgenic embryos (Figure 1B, compare with Figure 1A); lateral stripes are observed only after prolonged enzymatic staining. Previous studies have shown that the order of binding sites may be important for short-range repression in vivo [19, 20], and that orientation is essential for the activities of enhanceosomes in cultured cells [21–23]. The arrangement of REL and GATA sites has recently been shown to be important for the activation of immunity genes in the Drosophila fat body [24]. The current analysis of the vn enhancer provides evidence that the orientation of linked activators is also essential for proper gene expression in the Drosophila embryo.

Quantitative Measurements of the Dl, Twi, and Sna Gradients Site-occupancy models were used to predict vNE activity in early embryos (see below). These modeling efforts required detailed measurements of the relative concentrations of the Dl, Twi, and Sna protein gradients along the dorsal-ventral (DV) axis of precellular embryos. Particular efforts were focused on a specific time point, early to mid-nuclear cleavage cycle 14, when vNE enhancers first direct discrete lateral stripes of gene expression in ventral regions of the future neurogenic ectoderm. Quantitative confocal imaging was done with fixed embryos stained with antibodies directed against Dl, Twi, and Sna. All three proteins were simultaneously visualized in individual embryos in order to obtain accurate measurements of their relative concentrations (Figures 2A and 2C; Figure S3). The Sna repressor displays a steep distribution profile, with an abrupt decline over a span of just two nuclei (Figures 2A and 2C) at the future boundary between the mesoderm and mesectoderm (the ventral-most row of cells in the neurogenic ectoderm). There is also a reversal in the relative levels of Dl and Twi at this boundary: There are higher relative concentrations of Twi than Dl in ventral nuclei, but higher relative levels of Dl than Twi in lateral nuclei (Figure 2C). The steeper Twi gradient appears to be more critical for setting the limits of vNE gene expression than the gradual Dl gradient (see Supplemental Data).

Quantitative Modeling of Enhancer Response 1359

Figure 1. Structural and Functional Features of vNE Enhancers (A–C) Orientation of binding sites in the DTS element is crucial for vn enhancer activity. (A) The wild-type vn enhancer drives considerably stronger expression than a mutated vn enhancer (B) that contains a flipped Twist site, as indicated in (C). Staining in (A) and (B) were done in parallel. (D) Summary of vNE enhancer structures observed for vnd, vn, and rho; the DTS elements are enclosed by dotted lines, green arrows indicate the orientation of optimal Twist sites, and solid boxes indicate high-scoring binding site motifs. The two-toned box indicates a lower-scoring motif.

In situ hybridization assays were done by fluorescent RNA detection [25–27] to visualize the vn, rho, and vnd expression patterns (Figure 2B). Individual embryos were hybridized with all three probes, which were simultaneously visualized along with DV positional markers (see Supplemental Experimental Procedures). Most of our computational simulations focused on the rho and vnd genes, because they display consistent and reproducible differences in expression (e.g., Figure 2D). In addition, in transgenic embryos the minimal rho and vnd enhancers direct lacZ reporter gene expression profiles that accurately recapitulate the expression profiles of the endogenous genes (see Figure S3). Position effects were minimized by the use of gypsy insulator DNAs within the P element transformation vector (see Supplemental Experimental Procedures). The only significant difference between the endogenous and transgenic patterns is higher baseline expression of the lacZ reporter gene in the mesoderm, as predicted by the quantitative modeling (Figures 2D–2F, Figure S3; see below). Intronic probes were also used to assess the rho and vnd expression profiles (Figures S3A and S3B) because there is the possibility that mRNAs can diffuse across

p=

endogenous expression with exonic and intronic probes (Figure 2D and Figure S3C). In every case, the vnd pattern is narrower than rho and shifted ventrally by one nucleus, where there is weak and transient expression of the Sna repressor (Figures 2C and 2D and Figure S3C). In contrast, rho is excluded from this nucleus, but the staining pattern extends at least two nuclei beyond the dorsal limit of the vnd pattern. Quantitative Site-Occupancy Models Can Explain Distinct vNE Patterns Site-occupancy models were used to predict the vNE expression profiles at the midpoint of nuclear cleavage cycle 14. Particular efforts focused on the fractional occupancy of the Dl, Twi, and Sna binding sites contained within the DTS elements of the rho and vnd enhancers. Given the relative distributions of the Dl, Twi, and Sna concentration gradients along the DV axis of the precellular embryo, a DTS element containing one Dl site with a binding constant KDl and one Twist/Snail site with binding constants KTwi and KSna has the following probability p of achieving a successful transcriptional state:

CDl 2 Twi KDl ½Dl
KTwi ½Twi
1 + KDl ½Dl
+ KTwi ½Twi
+ KSna ½Sna
+ CDl 2 Twi KDl ½Dl
KTwi ½Twi
+ KDl ½Dl
KSna ½Sna


the syncytial embryo. Although not quantitative, intronic probes detect nascent intranuclear transcripts corresponding to sites of de novo transcription. The reproducible differences in the rho and vnd patterns are detected for reporter gene expression with probes against the vNE-lacZ fusion genes, as well as for

(1)

C refers to cooperative interactions between transcription factors (see Supplemental equations S8–S10 for a detailed description). Previous studies have shown that Dl can cooperatively interact with Daughterless (Da), a ubiquitous bHLH protein that forms heterodimers with Twi [28, 29].

Current Biology 1360

Figure 2. Fitting Expression Data with Quantitative Models (A and B) Visualization of (A) Dorsal, Twist, and Snail nuclear protein gradients and (B) mRNA distribution profiles of the indicated target genes. (C and D) Quantitative data for these gradients after background subtraction and smoothing (see Experimental Procedures). (E and F) Fractional site-occupancy models were individually fitted to rho and vnd expression data and are able to closely approximate vNE expression patterns; the best-fitting curves are shown. (G) Superposition of the best-fitting models for rho and vnd. The degree of spatial separation between the two patterns was analyzed with Pearson correlations; our results indicate that the data-to-model agreements significantly exceed the similarity between the rho and vnd quantitative data as well as between the models.

In vitro DNA binding assays suggest that the Dl, Twi, and Sna proteins bind to average recognition sequences at protein concentrations in the nanomolar range (Kd w 1029 M) ([18, 29–31] and data not shown). Computer simulations reflecting these approximate binding affinities were applied to the occupancy of individual sites in the vNE enhancers. These simulations also employed the experimentally determined distribution profiles of the Dl, Twi, and Sna proteins (Figure 2A). The predicted rho and vnd transcription profiles match the measured staining patterns only when Dl-Twi, Twi-Twi, and SnaSna cooperative DNA binding interactions are incorporated into the quantitative models (Figures 2E and 2F and Figure S6, equation S10).

Because there may be discrepancies between the in vitro DNA binding constants and actual in vivo affinities, modeling solutions were obtained for a spectrum of affinity values [log(Ka) = 8.5–9.75, see Table S2]. The results consistently identify cooperative DNA binding interactions as a crucial factor for the observed vNE expression patterns [log(C) > 0, see Table S2]. A key underlying assumption is that the Dl, Twi, and Sna proteins bind to the rho and vnd enhancers with similar affinities, regardless of the absolute binding constants. Indeed, the overall quality of the Dl, Twi, and Sna binding sites is very similar in the vnd and rho enhancers (see Table S1, Figures S7A–S7C) [7]. A number of computer

Quantitative Modeling of Enhancer Response 1361

Figure 3. Parameters Responsible for Differences in rho and vnd Expression Stepwise adjustment of parameters obtained for rho and vnd demonstrates the predicted effects of structural changes on expression patterns. The analysis shows rho to vnd transitions for the five best parameter combinations (solutions, see Table S2). Solutions for rho are shown in blue (A), and solutions for vnd are in red (D). Reduction of Dorsal-Twist cooperativity is responsible for the shift of the dorsal expression border (blue to light blue in [A]). Reduction of Twist-Twist cooperativity changes the shape of the dorsal border and shifts it ventrally (blue to green in [B]); an increase in the number of modules is responsible for an overall ‘‘boost’’ in expression and an increase in mesodermal expression (green to orange in [C]); Snail-Snail cooperativity predominantly affects the level of mesodermal expression (orange to red in [D]).

simulations were done to investigate the outcome of altering the binding affinities. Even over a large range of affinity values [7 < log(Ka) < 11], the computer

simulations cannot account for the measured differences in the rho and vnd expression patterns (Figure S5A, compare with Figure S5B). The same is

Current Biology 1362

Figure 4. Conserved Structural Differences between Drosophilid vnd and rho Enhancers Shown are conserved structural differences between the rho (A) and vnd (B) DTS modules in seven drosophilids. The rho enhancer contains a single tightly spaced DTS element with an additional E box nearby, presumably fostering high Dl-Twi and Twi-Twi cooperative interactions. The vnd enhancer contains two more loosely spaced DTS elements. Binding motifs are indicated by color (Dl, red; Sna, blue; optimal Twi, dark green; and potential Twi E-box, light green).

true when different numbers of binding sites (or DTS modules) are used in the simulations (Figures S5C and S5D). The vnd enhancer has two DTS elements rather than one, and contains an overall larger number of Dl, Twi, and Sna sites than the rho enhancer (Table S1, Figure S7). Nonetheless, computer simulations indicate that these differences alone are not sufficient to account for the distinctive rho and vnd expression profiles (Figures S5C and S5D). Accurate predictions (Figure 2G) are obtained only when cooperativity is invoked. Cooperativity is not only required to recapitulate the known vnd and rho expression patterns, but it is also needed to explain why the two patterns are slightly different from one another (Figure 3). The computer simulations that best fit the measured expression profiles of vnd and rho depend on the following parameters (summarized in Figure 3, see parameter values in Table S2). First, the rho models require 5–10fold-higher Dl-Twi cooperativity than vnd, as well as higher Twi-Twi cooperativity. Second, the vnd models require more DTS modules than rho. Third, the vnd models require higher Sna-Sna cooperativity. Enhanced Dl-Twi cooperativity is the most important determinant for the shift in the dorsal border of the rho expression pattern, as compared with the vnd pattern (Figure 3A). Twi-Twi cooperativity largely influences the shape of the dorsal border (Figure 3B). Computer simulations suggest that the ventral shift in the vnd expression pattern, as compared with the rho pattern, is

best explained by an increase in the number of DTS elements (Figure 3C). However, this increase should produce a general boost in the ‘‘baseline’’ expression of vnd in the ventral mesoderm. In principle, Sna-Sna cooperativity would diminish such expression (Figure 3D). In the preceding discussion, we have emphasized classical protein-protein cooperative DNA binding interactions. However, it is conceivable that the cooperativity could be indirect. Perhaps transcriptional synergy depends on the recruitment of coactivator complexes such as the mediator complex. Certain arrangements of binding sites, such as linked Dl and Twi sites, might lead to more efficient recruitment of coactivator complexes than other arrangements. Distinct Dl, Twi, and Sna Linkages in the rho and vnd Enhancers The different cooperativity estimates provided by the computer simulations might be explained by the detailed arrangement of binding sites in the rho and vnd enhancers (see Figure 4, Table S1, and Figures S2 and S7). Transcription factor cooperativity, which is crucial for embryonic development [32], has been shown to depend on spacing of binding sites, where closely linked and helically phased sites can provide higher cooperative interactions (e.g., [33, 34]). Orthologous enhancer sequences were identified in seven divergent drosophilids. The conservation of Dl, Twi, and Sna binding sites was examined in all 14 putative enhancers, as

Quantitative Modeling of Enhancer Response 1363

Figure 5. Mutations in the vnd Enhancer Cause Predicted Changes in Gene Expression (A–F) Expression patterns (A–C) and quantitative profiles (D) driven by wild-type and mutant vnd enhancers indicated in (F). Binding sites are color coded, light green boxes indicate lower-affinity Twist sites, and green arrows indicate the orientation of optimal Twist sites in DTS elements. (E) shows quantitative vnd models based on Table S2, top row; the vnd models adequately predict changes of expression upon lowering the affinity for Twist. Expression (D) and model data (E) were normalized to 1 to allow comparison. Without normalization, the models predict lower expression levels for the mutants as compared to wild-type expression, which agrees with enzymatic staining for expression in embryos (A–C) for equivalent times (not shown).

were the distances separating neighboring sites. Formal alignments are shown in supplementary Table S1. The Dl and Twi sites are separated by 3–7 bp in the seven rho enhancers, but vary in distance between 10 and 13 bp and more than 20 bp for each of the two DTS modules in the orthologous vnd enhancers (Figure 4, Figures S2B and S7G, and Table S1). Closer linkage might foster stronger cooperative Dl-Twi proteinprotein interactions in the rho enhancer. The stronger Twi-Twi cooperativity predicted for the rho enhancer correlates with the conserved linkage of a second Twi site located just 4 bp 50 of the DTS Twist site (Figure 4 and Figure S2B). In contrast, there is weak and variable linkage (20 bp or more) between a suboptimal Twist site (CAGATG) and a DTS Twist site in the vnd enhancer among divergent drosophilids (Figures 1D and 4 and Table S1). The orthologous vnd enhancers contain two DTS modules, whereas the rho enhancers contain just one (see Figures S2 and S6 and Table S1). Computer simulations predict that this results in a shift in the ventral limit of the vnd pattern as compared with rho. The presence of two DTS modules is also predicted to augment ectopic expression in the ventral mesoderm. However, Sna-Sna cooperative DNA binding interactions could diminish this expression [log(C) 0.7 versus 0.2, see Table S2]. Stronger Sna-Sna cooperativity for vnd versus rho

might be explained by the distinct arrangements of Sna sites in the two enhancers. Many of the sites in the vnd enhancer exhibit 11 bp spacing (Figures S7J and S7K), and helical phasing has been shown to foster cooperative binding [e.g., 22]. In contrast, the rho enhancers display apparently nonperiodic arrangements of Sna sites (Figures S7H and S7I). The precision of the expression patterns produced by the rho and vnd enhancers suggest that there might be a strict ‘‘code’’ relating the primary DNA sequence to predicted patterns of gene expression. However, different strategies may be used to produce similar vNE expression patterns. For example, the brinker (brk) gene encodes a transcriptional repressor that restricts dpp expression to the dorsal ectoderm. brk is initially expressed within the vNE, but the pattern expands throughout the neurogenic ectoderm during development. The brk enhancer does not contain compact DTS modules like those seen in the rho and vnd enhancers. Thus, Dl-Twi cooperativity does not appear to be a likely basis for the expression pattern. Perhaps an unknown, ubiquitous activator fosters cooperative occupancy of Dl, Twi, or other critical binding sites in the brk enhancer. Experimental Tests of Computational Predictions A critical test of the fractional-occupancy model is to predict gene-expression changes caused by selective

Current Biology 1364

point mutations in individual binding sites. The vnd enhancer was analyzed because it is inherently more robust than the rho or vn enhancers; vnd expression is not predicted to depend on Dl-Twi cooperative binding to the same extent as rho and vn. Instead, vnd contains more Dl and Twi sites with less stringent linkage. As a result, mutations in binding sites should lead to relatively subtle disruptions in vnd expression as compared with vn or rho. Simply rotating the Twi/Sna site in the vn DTS caused a catastrophic reduction in activity (see Figures 1A–1C), presumably as a result of strong Dl-Twi cooperativity. Computer simulations predict that nucleotide substitutions causing a reduction or loss of the Twi binding sites in the vnd DTS elements should cause a progressive narrowing in the expression driven by the modified enhancers (Figure 5E). This was tested by creating point mutations in the Twi sites (Figure 5F). In one set of mutations, the core CA residues in the optimal Twi binding sites (caCAtgt) were converted into AT residues, thereby reducing the quality of the sites. Fractional-occupancy simulations predict a w30% narrowing in the vnd expression pattern (Figure 5E). Normally, the lateral stripes of vnd expression encompass an average of six cells (Figure 5A). Lowering the Twi binding affinities is predicted to reduce these stripes to an average of four cells (Figure 5E). This is exactly what is observed when the mutagenized vnd enhancer is attached to a lacZ reporter gene and expressed in transgenic embryos (Figures 5B and 5D; compare with Figure 5A). Additionally, the Twi binding sites in the DTS elements were completely eliminated by base substitutions. Activation of the mutagenized enhancer now relies on ‘‘cryptic’’ weak Twi sites, including a CAGATG motif located between the two DTS elements. Computer simulations predict a severe narrowing in the expression pattern produced by the mutagenized enhancer, just two cells in width rather than six (Figure 5E). Again, this is exactly what is observed when the mutagenized enhancer is expressed in transgenic embryos (Figures 5C and 5D; compare with Figure 5A). There are several potential rate-limiting steps governing de novo transcription in vivo, including enhancer occupancy, enhancer-promoter communication, and chromatin modifications (see also Supplemental Data) [35–38]. The concordance of the computer simulations and in vivo expression patterns suggests that enhancer occupancy is the key rate-limiting step, at least during the initial phases of cell-fate specification in the early Drosophila embryo. The primary DNA sequences of vNE enhancers control lateral stripes of gene expression with high precision—the exact orientation, arrangement, and linkage of binding sites control nuances of the final expression patterns. It is conceivable that other transcriptional control mechanisms, particularly chromatin remodeling and histone modification, become rate limiting at later stages of development when transient regulators such as Dl are no longer present.

Experimental Procedures Motifs and Sequence Analysis Full annotations of binding motifs and Dorsal target enhancer sequences are available online (http://www.dvex.org). See also the Supplemental Data for more information.

Expression Data Embryo fixation, in situ hybridizations, and antibody stainings were performed as previously described (see Supplemental Data). Transcription factor gradients were detected with preabsorbed antibodies. RNA in situ hybridization was done with DIG- and DNPlabeled probes and with primary and secondary antibodies (see Supplemental Data). RNA transcripts were always detected in combination with rho mRNA and Snail protein as DV positional markers. Confocal stacks of 12- to 17-cell-wide mid-trunk sections of properly oriented early to mid-nuclear cleavage cycle 14 embryos were acquired on a Leica LS confocal microscope with fixed settings (see Supplemental Data for details); gain was adjusted to remain well below signal saturation. Sum projections of confocal stacks were assembled and protein/RNA concentration along the DV axis was measured with the ImageJ software (see Supplemental Data for details). Datasets were averaged across embryos and normalized as described in the Supplemental Data. Distribution data and sample pictures are available (http://www.dvex.org). Mutagenesis The 497 bp vn [16, 17] and the 743 bp vnd enhancers [11] were mutated via site-directed mutagenesis with the primers below; results were consistent across multiple lines (see Supplemental Data for details). vn(flip): 50 -GGACAGGTAACGGGCgACATGTgTGGCCGGAAATT CCCCG-30 vnd(CA/AT)1: 50 -GAAACCCCAATCGGGAATGACAatTGTACG ACAGACATGGGACTC-30 vnd(CA/AT)2: 50 -GCAAGCAAACTTGCGCCAACAatTGGCACG ACCTGTTTCGACC-30 vnd(DE-box)1: 50 -GAAACCCCAATCGGGAATGgatccaaTACGAC AGACATGGGACTCAG-30 vnd(DE-box)2: 50 -GCAAGCAAACTTGCGCCAtggaaaaGCACGA CCTGTTTCGACCCG-30

Supplemental Data Supplemental Data include Supplemental Experimental Procedures, three tables, and details regarding the modeling and are available with this article online at: http://www.current-biology.com/cgi/ content/full/16/13/1358/DC1/. Acknowledgments We thank Matt Ronshaugen for advice on confocal microscopy, as well as John Reinitz, Johannes Ja¨ger, Karen Zinzen, and Fred Biemar for critical reading of the manuscript and helpful discussions. This study was supported by a grant from the National Institutes of Health (GM46638) and the Center for Integrative Genomics at UC Berkeley. Received: January 6, 2006 Revised: May 13, 2006 Accepted: May 15, 2006 Published online: June 1, 2006 References 1. Crauk, O., and Dostatni, N. (2005). Bicoid determines sharp and precise target gene expression in the Drosophila embryo. Curr. Biol. 15, 1888–1898. 2. Houchmandzadeh, B., Wieschaus, E., and Leibler, S. (2005). Precise domain specification in the developing Drosophila embryo. Phys. Rev. E Stat. Nonlin. Soft Matter Phys. 72, 061920. 3. Jaeger, J., Blagov, M., Kosman, D., Kozlov, K.N., Manu, Myasnikova, E., Surkova, S., Vanario-Alonso, C.E., Samsonova, M., Sharp, D.H., and Reinitz, J. (2004). Dynamical analysis of regulatory interactions in the gap gene system of Drosophila melanogaster. Genetics 167, 1721–1737. 4. Jaeger, J., Surkova, S., Blagov, M., Janssens, H., Kosman, D., Kozlov, K.N., Manu, Myasnikova, E., Vanario-Alonso, C.E., Samsonova, M., et al. (2004). Dynamic control of positional information in the early Drosophila embryo. Nature 430, 368–371.

Quantitative Modeling of Enhancer Response 1365

5. Laubenbacher, R., and Stigler, B. (2004). A computational algebra approach to the reverse engineering of gene regulatory networks. J. Theor. Biol. 229, 523–537. 6. Levine, M., and Davidson, E.H. (2005). Gene regulatory networks for development. Proc. Natl. Acad. Sci. USA 102, 4936–4942. 7. Papatsenko, D., and Levine, M. (2005). Quantitative analysis of binding motifs mediating diverse spatial readouts of the Dorsal gradient in the Drosophila embryo. Proc. Natl. Acad. Sci. USA 102, 4966–4971. Published online March 28, 2005. 10.1073/ pnas.0409414102. 8. Schroeder, M.D., Pearce, M., Fak, J., Fan, H., Unnerstall, U., Emberly, E., Rajewsky, N., Siggia, E.D., and Gaul, U. (2004). Transcriptional control in the segmentation gene network of Drosophila. PLoS Biol. 2, e271 10.1371/journal.pbio.0020271. 9. Scott, M.P., and Carroll, S.B. (1987). The segmentation and homeotic gene network in early Drosophila development. Cell 51, 689–698. 10. Stathopoulos, A., and Levine, M. (2005). Genomic regulatory networks and animal development. Dev. Cell 9, 449–462. 11. Stathopoulos, A., Van Drenth, M., Erives, A., Markstein, M., and Levine, M. (2002). Whole-genome analysis of dorsal-ventral patterning in the Drosophila embryo. Cell 111, 687–701. 12. Ackers, G.K., Johnson, A.D., and Shea, M.A. (1982). Quantitative model for gene regulation by lambda phage repressor. Proc. Natl. Acad. Sci. USA 79, 1129–1133. 13. Laiken, S.L., Gross, C.A., and Von Hippel, P.H. (1972). Equilibrium and kinetic studies of Escherichia coli lac repressorinducer interactions. J. Mol. Biol. 66, 143–155. 14. Oppenheim, A.B., Kobiler, O., Stavans, J., Court, D.L., and Adhya, S. (2005). Switches in bacteriophage lambda development. Annu. Rev. Genet. 39, 409–429. 15. Ptashne, M. (2005). Regulation of transcription: From lambda to eukaryotes. Trends Biochem. Sci. 30, 275–279. 16. Erives, A., and Levine, M. (2004). Coordinate enhancers share common organizational features in the Drosophila genome. Proc. Natl. Acad. Sci. USA 101, 3851–3856. Published online March 16, 2004. 10.1073/pnas.0400611101. 17. Markstein, M., Zinzen, R., Markstein, P., Yee, K.P., Erives, A., Stathopoulos, A., and Levine, M. (2004). A regulatory code for neurogenic gene expression in the Drosophila embryo. Development 131, 2387–2394. 18. Ip, Y.T., Park, R.E., Kosman, D., Bier, E., and Levine, M. (1992). The dorsal gradient morphogen regulates stripes of rhomboid expression in the presumptive neuroectoderm of the Drosophila embryo. Genes Dev. 6, 1728–1739. 19. Kulkarni, M.M., and Arnosti, D.N. (2005). cis-regulatory logic of short-range transcriptional repression in Drosophila melanogaster. Mol. Cell. Biol. 25, 3411–3420. 20. Arnosti, D.N., Barolo, S., Levine, M., and Small, S. (1996). The eve stripe 2 enhancer employs multiple modes of transcriptional synergy. Development 122, 205–214. 21. Cave, J.W., Loh, F., Surpris, J.W., Xia, L., and Caudy, M.A. (2005). A DNA transcription code for cell-specific gene activation by notch signaling. Curr. Biol. 15, 94–104. 22. Kim, T.K., and Maniatis, T. (1997). The mechanism of transcriptional synergy of an in vitro assembled interferon-beta enhanceosome. Mol. Cell 1, 119–129. 23. Merika, M., and Thanos, D. (2001). Enhanceosomes. Curr. Opin. Genet. Dev. 11, 205–208. 24. Senger, K., Armstrong, G.W., Rowell, W.J., Kwan, J.M., Markstein, M., and Levine, M. (2004). Immunity regulatory DNAs share common organizational features in Drosophila. Mol. Cell 13, 19–32. 25. Kosman, D., Mizutani, C.M., Lemons, D., Cox, W.G., McGinnis, W., and Bier, E. (2004). Multiplex detection of RNA expression in Drosophila embryos. Science 305, 846. 26. Myasnikova, E., Samsonova, A., Kozlov, K., Samsonova, M., and Reinitz, J. (2001). Registration of the expression patterns of Drosophila segmentation genes by two independent methods. Bioinformatics 17, 3–12. 27. Myasnikova, E., Samsonova, M., Kosman, D., and Reinitz, J. (2005). Removal of background signal from in situ data on the expression of segmentation genes in Drosophila. Dev. Genes

28.

29.

30.

31.

32.

33.

34.

35.

36. 37.

38.

Evol. 215, 320–326. Published online February 12, 2005. 10. 1007/s00427-005-0472-2. Castanon, I., Von Stetina, S., Kass, J., and Baylies, M.K. (2001). Dimerization partners determine the activity of the Twist bHLH protein during Drosophila mesoderm development. Development 128, 3145–3159. Gonzalez-Crespo, S., and Levine, M. (1993). Interactions between dorsal and helix-loop-helix proteins initiate the differentiation of the embryonic mesoderm and neuroectoderm in Drosophila. Genes Dev. 7, 1703–1713. Ip, Y.T., Kraut, R., Levine, M., and Rushlow, C.A. (1991). The dorsal morphogen is a sequence-specific DNA-binding protein that interacts with a long-range repression element in Drosophila. Cell 64, 439–446. Jiang, J., and Levine, M. (1993). Binding affinities and cooperative interactions with bHLH activators delimit threshold responses to the dorsal gradient morphogen. Cell 72, 741–752. Lebrecht, D., Foehr, M., Smith, E., Lopes, F.J., Vanario-Alonso, C.E., Reinitz, J., Burz, D.S., and Hanes, S.D. (2005). Bicoid cooperative DNA binding is critical for embryonic patterning in Drosophila. Proc. Natl. Acad. Sci. USA 102, 13176–13181. Published online September 6, 2005. 10.1073/pnas.0506462102. D’Alonzo, R.C., Selvamurugan, N., Karsenty, G., and Partridge, N.C. (2002). Physical interaction of the activator protein-1 factors c-Fos and c-Jun with Cbfa1 for collagenase-3 promoter activation. J. Biol. Chem. 277, 816–822. Inoue, J., Sato, R., and Maeda, M. (1998). Multiple DNA elements for sterol regulatory element-binding protein and NF-Y are responsible for sterol-regulated transcription of the genes for human 3-hydroxy-3-methylglutaryl coenzyme A synthase and squalene synthase. J. Biochem. (Tokyo) 123, 1191–1198. Bolouri, H., and Davidson, E.H. (2003). Transcriptional regulatory cascades in development: Initial rates, not steady state, determine network kinetics. Proc. Natl. Acad. Sci. USA 100, 9371– 9376. Published online July 25, 2003. 10.1073/pnas.1533293100. Dean, A. (2005). On a chromosome far, far away: LCRs and gene expression. Trends Genet. 22, 38–45. von Dassow, G., Meir, E., Munro, E.M., and Odell, G.M. (2000). The segment polarity network is a robust developmental module. Nature 406, 188–192. Reinitz, J., How, S., and Sharp, D.H. (2003). Transcriptional control in Drosophila. Complexus 1, 54–64.