• We are sorry, but NCBI web applications do not support your browser and may not function properly. More information
Logo of pnasPNASInfo for AuthorsSubscriptionsAboutThis Article
Proc Natl Acad Sci U S A. Oct 31, 2006; 103(44): 16352–16357.
Published online Oct 25, 2006. doi:  10.1073/pnas.0607448103
PMCID: PMC1637586
Evolution

Genome scan for cis-regulatory DNA motifs associated with social behavior in honey bees

Abstract

Honey bees (Apis mellifera) undergo an age-related, socially regulated transition from working in the hive to foraging, which is associated with changes in the expression of thousands of genes in the brain. To begin to study the cis-regulatory code underlying this massive social regulation of gene expression, we used the newly sequenced honey bee genome to scan the promoter regions of eight sets of behaviorally related genes differentially expressed in the brain in the context of division of labor among worker bees, for 41 cis-regulatory motifs previously characterized in Drosophila melanogaster. Binding sites for the transcription factors Hairy, GAGA, Adf1, Cf1, Snail, and Dri, known to function in nervous system development, olfactory learning, or hormone binding in Drosophila, were significantly associated with one or more gene sets. The presence of some binding sites also predicted expression patterns for as many as 71% of the genes in some gene sets. These results suggest that there is a robust relationship between cis and social regulation of brain gene expression, especially considering that we studied <15% of all known transcription factors. These results also suggest that transcriptional networks involved in the regulation of development in Drosophila are used to regulate behavioral development in adult honey bees. However, differences in gene regulation between these two processes are suggested by the finding that the promoter regions for the behaviorally related bee genes differed in both motif occurrence and G/C content relative to their Drosophila orthologs.

Keywords: Apis mellifera, gene regulation, microarray, position weight matrix, transcription factors

Behavioral development in honey bees (Apis mellifera) gives rise to an intricate division of labor in the honey bee colony (1). Adult worker honey bees perform a series of tasks in the hive, such as brood care (“nursing”), when they are young and then shift to foraging for nectar and pollen outside the hive when they are 2–3 weeks of age for the remainder of their 5- to 7-week life. The transition from hive work to foraging involves changes in endocrine activity, brain chemistry, and brain structure. This transition is socially regulated and responsive to the needs of the colony, which are communicated via pheromones and other means. These social factors act directly or indirectly on physiological mechanisms that influence behavioral maturation, including mechanisms involving juvenile hormone (JH), the foraging and malvolio genes, and other genes still to be identified.

The transition from hive work to foraging in honey bee colonies also involves changes in the expression of thousands of genes in the brain (2, 3). This finding suggests that honey bees will be useful in elucidating the mechanisms by which social factors regulate gene expression in the brain, one frontier in the study of genes and social behavior.

To begin to study the cis-regulatory code underlying this massive social regulation of brain gene expression, we used the newly sequenced honey bee genome (4) to scan for cis-regulatory motifs in the promoter regions of genes previously shown to be differentially expressed in the brain in the context of socially regulated division of labor (2, 3). We searched for 41 motifs that were previously well characterized in Drosophila melanogaster, primarily for embryonic development, which represents <15% of all known transcription factors (www.godatabase.org). Starting with these experimentally validated motifs allowed us to employ solely bioinformatics methods for this study; identification of motifs de novo would require additional molecular validation. With this approach, we also were able to begin to explore the hypothesis that transcriptional networks involved in the regulation of embryonic development in Drosophila are also used to regulate adult behavioral development in honey bees. The premise of this hypothesis is that many transcription factors show high levels of functional conservation across broad evolutionary distances (5).

Microarray analysis has revealed that nurse and forager honey bees show differences in brain mRNA abundance for about one-third of the genes analyzed, and subsequent experiments indicated that many of these genes are socially regulated and predictive of behavioral status (2). Whitfield et al. (3) generated additional sets of genes implicated in social regulation by performing microarray analyses to identify genes regulated in the brain by physiological, ontogenetic, and genetic factors that are also known to influence the age at onset of foraging. We used these sets of “behaviorally related” genes in our study.

We divided eight gene sets (2, 3), drawn from a total of 3,129 genes (estimated to represent ≈25% of the genes in the bee genome), into those sets that were up-regulated or down-regulated in the brain in response to a specific condition (Table 1). Each gene set contained 100–871 genes. Three additional pairs of gene sets were derived from these gene sets (see Methods) to capture specific combinations of conditions. We identified and quantified patterns of occurrence of the 41 cis-regulatory motifs, listed in Table 5, which is published as supporting information on the PNAS web site, in the promoter regions of these gene sets.

Table 1.
Gene sets associated with honey bee socially regulated division of labor, derived from microarray experiments on brain gene expression

We scanned for these motifs by modeling the binding specificity of each transcription factor by using a position-specific weight matrix (PWM). We used the computer algorithm Stubb (6) to score a promoter (2,000-bp region upstream of the translation start site for each gene) for matches to PWMs. The Stubb algorithm was previously found to accurately predict cis-regulatory modules involved in the segmentation pathway in Drosophila (7).

The general scheme for this study is illustrated in Fig. 1. We performed “enrichment analysis” by using hypergeometric tests to analyze each gene set for statistical enrichment for each of the 41 motifs, compared with the entire complement of 3,129 genes. We also performed “correlation analysis,” analyzing each pair of up- and down-regulated gene sets for correlation between motif counts and gene expression. Because all of the gene sets are related to socially regulated behavioral maturation, in either a correlative or causal sense, an association between a motif and a gene set discovered with either enrichment or correlation analysis would implicate the corresponding transcription factor as being active in a regulatory pathway in the brain that is related to social behavior. In addition, because the PWMs for the motifs are derived from studies of Drosophila (with ≈300 Myr divergence time to honey bee; see ref. 8), an association between a motif and a gene set would also provide evidence for considerable evolutionary conservation of that motif.

Fig. 1.
Flow chart for the discovery of motif–gene set associations using four statistical tests. We identified and quantified patterns of occurrence of transcription factor binding motifs in the promoter regions of selected sets of genes (DATA1) related ...

Results and Discussion

Behaviorally Related Genes in Honey Bees Have High G/C Content Promoters.

Our initial analyses revealed a surprisingly large number of motif–gene set associations. We detected 41 associations, compared with an expected 1 under the null hypothesis (see Table 6, which is published as supporting information on the PNAS web site). Most of these associations involved G/C-rich motifs (see Table 7, which is published as supporting information on the PNAS web site), which led us to explore whether the promoters for these gene sets were, in general, high in G/C content.

Three gene sets were indeed significantly enriched for G/C-rich promoters (Table 2, hypergeometric test). In addition, G/C content correlated significantly with brain expression levels in four pairs of gene sets (Table 2, correlation coefficient). By using a supervised learning method (2-fold cross-validation with a simple threshold-based classifier), promoter G/C content correctly classified a significant fraction of genes in some gene sets as being either up- or down-regulated. The strongest result was obtained for manganese-responsive genes; classification accuracy was 74.6% (P = 1E-38, binomial test; see Table 8, which is published as supporting information on the PNAS web site)

Table 2.
Some gene sets are enriched for high G/C content (fraction of sequence length that is G or C) in their promoters

We tested whether the G/C enrichment of the above gene sets also occurred in the Drosophila genome and thus perhaps reflects some regulatory phenomenon common to both species. Hypergeometric tests showed no association between Drosophila orthologs of the bee gene sets and high G/C promoter content (see Table 9, which is published as supporting information on the PNAS web site). This difference between honey bee and Drosophila promoter regions was further substantiated with Gene Ontology (GO) analysis (www.godatabase.org). The bee genes with the highest G/C content promoters are significantly overrepresented in the categories of transcriptional regulation and ectoderm, midgut, heart, and nervous system development (based on experimental evidence for their Drosophila orthologs; see Table 10, which is published as supporting information on the PNAS web site). However, Drosophila genes with the highest G/C content promoters were not overrepresented in any GO category (see Table 11, which is published as supporting information on the PNAS web site).

The reasons for these bee/fly differences are not known but might involve enhanced transcription factor binding to known G/C-rich motifs in bees, the presence of additional as yet unidentified G/C-rich motifs in bees, species differences in methylation, or species differences in gene expression. Genes with brain-specific patterns of expression are known to have high G/C content promoters in humans (9).

Behaviorally Related Genes in Honey Bees Are Enriched for cis-Regulatory Motifs of Hairy, Cf1, Adf1, Dri, and Snail.

To identify motif–gene set associations on the basis of the specific identity of the motifs, rather than the G/C content of the motifs and promoters, we repeated the analyses while controlling for the effects of G/C content. First, Stubb was run with a local, rather than global, background model based on each gene's promoter. This conservative strategy likely resulted in a failure to detect some true motif–gene set associations (G/C-rich motifs in G/C-rich promoters will have suppressed Stubb scores), but it enhances the reliability of our results. Second, the enrichment analysis was done with the additional requirement that the gene set be more enriched for the motif than for G/C content. Third, the correlation analysis used partial correlation coefficients to factor out the correlation between G/C content and expression (10).

Adf1, Hairy, Snail, Dri, and Cf1 were found to be associated with several different gene sets (Table 3; and Tables 12 and 13, which are published as supporting information on the PNAS web site). Each of these motif–gene set associations was the most statistically significant (P < 0.001) among all motifs for that set. Three of these associations (involving Adf1, CF1, and Hairy) were significant by both enrichment and correlation analyses (Table 3).

Table 3.
Identification of motif–gene set associations by enrichment or correlation analysis

As an additional negative control apart from that provided inherently by our statistical framework, we performed the same enrichment analysis on a randomized promoter sequence obtained by permuting the bases of each of the 3,129 promoters. Hypergeometric tests revealed no statistically significant associations between these five (or any other) motifs and gene sets (data not shown).

For these five motifs, the genome has more promoters with extremely high Stubb scores than expected by chance (in a randomized genome) (see Fig. 2, which is published as supporting information on the PNAS web site). These statistical results suggest that the motifs have some functional relevance in the honey bee genome. Additional evidence is provided by GO analysis. The 200 genes with the strongest association with Adf1 and Dri (in terms of Stubb scores) are significantly associated with the biological process “nervous system development,” whereas the genes for Hairy are significantly associated with “sensory organ development” (see Table 14, which is published as supporting information on the PNAS web site), a process that is JH-dependent (11, 12).

The known functions of the transcription factors associated with these five motifs are consistent with the possibility that these motifs are involved in the regulation of neural and behavioral plasticity in the honey bee. Adf1, Hairy (13, 14), and Snail (15) proteins have roles in nervous system development in Drosophila, and Adf1 is also implicated in olfactory learning (16) and synapse formation in adults (17). Cf1 (also called “Ultraspiracle” or “USP”) binds specifically to JH in vitro (18), is rapidly up-regulated by JH in the adult honey bee (19), and is expressed throughout the adult bee brain, especially in the mushroom bodies, a region of multimodal sensory integration and memory (20). JH blood levels increase during honey bee behavioral development; levels are lower in nurse bees than in foragers, and methoprene (JH analog) treatment causes precocious foraging (1). Thus, JH is likely to be a crucial mechanism by which social cues interact with the genome (3).

We were able to find orthologs in the bee genome (http://racerx00.tamu.edu/bee_resources.html) for four of five corresponding Drosophila transcription factors (all but Adf1). We also found a high degree of sequence identity for the DNA binding domains of these four orthologs, suggesting that they bind to the same motifs as in Drosophila (see Fig. 3, which is published as supporting information on the PNAS web site). For example, the basic N-box-specific region of the Hairy protein and the P box of the Cf1 protein are completely conserved across insects and humans.

However, despite these similarities, the motif–gene set associations we found for honey bee were not detected in fly. Hypergeometric tests on orthologs of the bee gene sets revealed no strongly significant associations in Drosophila; for the six significant associations in Table 12, P values in Drosophila were 0.33, 0.54, 0.69, 0.91, 0.02, and 0.54, respectively. These results suggest that regulatory processes related to the observed motif–gene set associations might be associated with some specific aspects of honey bee social behavior that differs from Drosophila behavior. Consistent with this finding is the observation that although ≈76% of the genes annotated in the honey bee genome have detectable orthologs in D. melanogaster (4), orthologs were present for a significantly lower fraction of up-regulated genes in our gene sets (69%, P < 0.006; and 77%, P = 0.67, for down-regulated genes). Some of the gene sets studied here may encode a slightly higher proportion of “novel” or rapidly evolving proteins than does the whole genome.

Several gene sets (e.g., those related to “genetic variation”) were not found to be associated with particular cis-regulatory motifs. Possible reasons for this finding include associations with motifs other than those studied here or extensive regulation at levels other than transcription.

Patterns of Occurrence of Hairy, Snail, and GAGA Binding Factors Classify Expression Patterns of Honey Bee Behaviorally Related Genes.

Our results provide evidence for a connection between cis and social regulation of brain gene expression. If this connection is strong, then the patterns of motif occurrence we detected should have sufficient discriminative power to classify the expression pattern of genes in our gene sets as up- or down-regulated. We tested this hypothesis by looking for cases in which motif-based classification accuracy is statistically significant and G/C content-based classification accuracy is less so.

We first tested each motif separately, using a threshold-based classifier, and found four such cases (Table 4; and Tables 15 and 16, which are published as supporting information on the PNAS web site). These cases involved three different motifs: Hairy and Snail (again) and GAGA. The presence or absence of the GAGA motif alone correctly classified the expression (up- or down-regulated) of 71% of 100 of the most differentially expressed genes in the Hive bee-to-forager transition↑ vs. Hive bee-to-forager transition↓ set, compared with 50% expected on the basis of chance alone (P = 1.6E-5, binomial test). Similar results were obtained for Hairy and the response of genes to methoprene (JH analog), and for Snail and the response of genes to cGMP (related to the foraging gene; see ref. 1).

Table 4.
Classification of brain gene expression (up- or down-regulated) on the basis of the pattern of (single) motif occurrences

Using all 41 motifs in our compendium and a support vector machine (SVM) classifier (http://svmlight.joachims.org), we found one more case: 2-fold cross-validation analysis correctly classified the expression pattern of 71% of the genes up-regulated in foragers and by methoprene compared with genes up-regulated in nurses and down-regulated by methoprene (n = 78 genes, 39 of each; P = 0.00018, binomial test). G/C content alone had insignificant predictive power (P > 0.1, binomial test). These findings are remarkable, considering that our motif compendium comprises <15% of the estimated transcription factors in the Drosophila (www.godatabase.org) or honey bee genomes.

cis-Regulatory Motifs Show Combinatorial Regulation of Honey Bee Behaviorally Related Genes.

The transcriptional pathways involved in embryonic development are well known for their combined action of multiple transcription factors on common gene targets (21). The combinatorial nature of transcriptional pathways is a highly conserved feature of cis-regulation, so we expected to be able to detect this in the context of social regulation. Pairs of cis-regulatory binding sites do indeed co-occur in the promoter regions of genes in our gene sets.

Focusing on the Forager↑ and Nurse↑ gene sets, we counted the common gene targets of each pair of transcription factors and found this number to be significantly higher than by chance for 78 of the 820 pairs (P < 0.001, Q = 0.01, hypergeometric test; see ref. 22), even after controlling for the effect of local background composition. In comparison, a negative control experiment involving randomly permuted versions of the promoters produced only five pairs of transcription factors. After generating a pairwise interaction graph from these data and performing further analysis, we found a cohesive set of 7 transcription factors (Adf1, Hairy, Cf1 and GAGA again, plus AbdB, Zeste, and Eve) of the 41 studied, with almost every pair having significant patterns of interaction (see Fig. 4, which is published as supporting information on the PNAS web site).

Conclusions

These results demonstrate a robust association for social behavior, brain gene expression, and distributions of transcription factor binding sites throughout the honey bee genome and are consistent with results from studies of social behavior in rodents, which have revealed important roles for cis regulation for single genes (23, 24). Our finding of motifs by using PWMs primarily derived from studies of Drosophila developments demonstrates that transcriptional networks involved in the regulation of embryonic development are reused by nature for adult behavioral functions. We also detected differences in motif–gene set associations and promoter G/C content between honey bee and Drosophila that might reflect unique aspects of gene regulation associated with social regulation. Social behavior is a highly derived trait, and we predict that the evolution of transcriptional combinatorial codes for socially regulated gene expression in the brain will involve both conserved and novel motifs that await discovery.

Methods

Scoring Genes for Motif Occurrence.

The Stubb algorithm (6) computes a statistical (log-likelihood ratio) score for the clustering of binding sites of one or more transcription factors in a sequence, accounting for varying numbers and strengths of the sites. The software was run with a window size of 200, shift of 100, and a zeroth-order background Markov model. The training sequence for the Markov model was either the entire complement of 2-kbp promoters (“global background”) or the sequence of the promoter being analyzed (“local background”). (The local background of 2 kbp provides sufficient statistics to train a zeroth-order Markov model of four states.) The best-scoring 200-bp window in a gene's promoter was used to score that gene. We masked short tandem repeats (putative micro- and minisatellites) in the sequence to prevent the Stubb algorithm from confounding these repeats with weak binding sites. Tandem repeats were masked using the program Tandem Repeat Finder (25), with recommended parameters 2 7 7 80 10 25 500. The heat shock element was represented by a set of eight PWMs from TRANSFAC (www.gene-regulation.com/pub/databases.html#transfac), corresponding to trimers of the AGAAN site in all possible orientations (26). Other PWM motifs, corresponding to binding sites in fruitfly, were obtained from TRANSFAC, JASPAR (27), and Schroeder et al. (28), and redundant matrices were removed manually to obtain a set of 41 motifs for 37 distinct transcription factors. Unless specified otherwise, a gene was considered as containing binding sites for a particular transcription factor if it was among the top 200 genes ranked by Stubb score for the corresponding motif; these genes are referred to as “targets” of a motif. Thus, motif targets are decided based on clustering of strong and weak motif occurrences in their promoters. The G/C content of a promoter was measured as the fraction of the promoter that is Gs or Cs.

Gene Sets.

Eight sets of genes established as being significantly up-regulated or down-regulated in the brain in response to specific conditions (behavioral development, genetic variation, or neuroactive treatment) were taken from refs. 2 and 3. Each set was partitioned into up- and down-regulated subsets, as listed in Table 1. Treatments with cGMP, methoprene (JH analog), or manganese are known to induce precocious foraging in the honey bee. Hence, we intersected the Forager↑ set with each of the cGMP↑, Methoprene↑, and Manganese↑ sets, and likewise the Nurse↑ set with each of cGMP↓, Methoprene↓, and Manganese↓ sets to obtain six additional sets (set sizes as given in Table 17, which is published as supporting information on the PNAS web site), for a total of 22 sets. Enrichment analysis was done for each of these 22 sets separately.

Enrichment Analyses.

Hypergeometric test.

The set of genes targeted by a motif was intersected with a given gene set, and the P value of the intersection size was calculated with the hypergeometric test: H(N, g, m, i), where N is the total number of genes (3,129), g is the size of the given gene set, m is the number of genes targeted by motif (200), and i is the intersection size. For motif enrichment vs. G/C enrichment, let iG/C be the size of the intersection of the gene set with the top m G + C-rich genes (promoters). We heuristically computed the P value of the hypergeometric distribution H(N, iG/CN/m, m, i) and required this P value to be ≤0.05 as an additional criterion for reporting the motif as significantly associated with the gene set. Note that the original hypergeometric test H(N, g, m, i) compares the fraction i/m (the specificity of predicting gene set membership based on motif score) with the fraction g/N, which is the baseline specificity of a random predictor. In the second statistical test, we change the baseline performance to be that achieved by a G/C content-based predictor, iG/C/m, and compare the fraction i/m with this fraction, using H(N, iG/C N/m, m, i).

Partial correlation analysis.

The sets of up- and down-regulated genes for a particular condition (e.g., methoprene response) were combined into one set, and for each gene we noted the expression value (E), the Stubb score (M) for a motif, and the G + C percentage (GC) of its promoter. A partial correlation analysis was performed to assess the correlation between variables E and M, while factoring out the effect of variable GC, and a P value was computed. The partial Pearson's correlation between two variables X and Y, after controlling for a third variable Z, was computed from the pairwise Pearson's correlation coefficients by the formula

equation image

The significance of a partial correlation rXY,Z with n data points was assessed with a two-tailed t test on t = rXY,Z ·(n3)(1rXY,Z)2, with n − 3 degrees of freedom.

Q Value.

To account for multiple-hypothesis testing, Q values (22) (estimated false discovery rates) were calculated as Q = (estimated no. of false positives)/(no. of called positives at a given P value) = (P × n)/i, where P is the P value, N is the total number of tests, and i is the sorted rank of the P value.

Orthology of Genes.

Predicted honey bee genes were assigned to orthology groups with D. melanogaster genes on the basis of reciprocal best BLASTX match.

GO Analysis.

GO analysis was done using the program GeneMerge (29) that computes the significance (E value) of the enrichment of a particular set of genes for a GO term. We used the biological processes ontology (D_melanogaster.BP) and an E value threshold (computed using Bonferroni correction) of 0.1.

Threshold-Based Classification.

Up- and down-regulated genes for a particular condition were labeled as positive and negative, respectively, and for any given motif, and a score threshold was sought that maximized the correctly classified genes (motif score above threshold predicts positive and below threshold predicts negative). This technique is known as the TNoM score (30) and has been used in feature selection tasks related to cancer tissue classification. Two-fold cross-validation was done for each pair of gene sets and each motif, and the fraction of correctly classified genes on the test set was used to evaluate the classification accuracy of the motif. In 2-fold cross-validation, both the positively and negatively labeled genes are partitioned into two equal parts: the training set and the test set. Parameters of the classifier are trained on the training set, and the results are evaluated on the test set. The roles of the training and test sets are then reversed, and the overall results are the average of the two experiments.

SVM Classification.

The SVM classifier used was SVMlight (http://svmlight.joachims.org), and 2-fold cross-validation was done. The fraction of correctly classified genes in the test set gives the accuracy of the classifier.

Data Sets Used in Classification.

For both classification exercises (threshold-based and SVM), the data sets were modified from the original, as follows. Each pair of up- and down-regulated gene sets (eight pairs listed in Table 1, as well as the three additional pairs derived from these, as described above) was taken, and the larger of the two was shrunk to match the size of the smaller set, retaining the most up- or down-regulated genes. Thus, every classification test was done on positive and negative sets of the same size. The test was also repeated with each of these gene sets further shrunk to half of its original size, retaining the most up- or down-regulated genes (“top 50%”). Threshold-based classification was also done on the original gene sets (before the above modifications), and results are shown in Fig. 5, which is published as supporting information on the PNAS web site.

Pairwise Associations Between Motifs.

Association among motifs was measured as follows. For every pair of motifs, the intersection set X of their target gene sets was assessed for significance, using the hypergeometric test. To factor out the effect of G/C content, (i) the Stubb score of each promoter sequence was computed by randomly shuffling the bases in the sequence, (ii) the intersection of the two motifs' target gene sets according to these newly computed randomized Stubb scores was obtained, and (iii) this intersection set was subtracted from the original intersection set X. Hence, the hypergeometric test on the shrunk intersection is made stronger by removing certain common gene targets that might be due to similar G/C content.

Supplementary Material

Supporting Information:

Acknowledgments

We thank P. Kheradpour for early assistance; R. Velarde for the USP alignment; and Y. Ben-Shahar, J. H. Hunt, S. Zhang, members of the Robinson laboratory, and G. D. Stormo for comments that improved the manuscript. This work was supported by the University of Illinois Sociogenomic Initiative (G.E.R.) and by a National Science Foundation Frontiers in Integrative Biological Research grant.

Abbreviations

GO
Gene Ontology
JH
juvenile hormone
PWM
position-specific weight matrix
SVM
support vector machine.

Footnotes

The authors declare no conflict of interest.

References

1. Robinson GE, Grozinger CM, Whitfield CW. Nat Rev Genet. 2005;6:257–270. [PubMed]
2. Whitfield CW, Cziko AM, Robinson GE. Science. 2003;302:296–299. [PubMed]
3. Whitfield CW, Ben-Shahar Y, Brillet C, Leoncini I, Caruser D, LeConte Y, Rodriguez-Zas S, Robinson GE. Proc Natl Acad Sci USA. 2006;103:16068–16075. [PMC free article] [PubMed]
4. Honeybee Genome Sequencing Consortium. Nature. 2006;443:931–949. [PMC free article] [PubMed]
5. Carroll SB, Grenier JK, Weatherbee SD. From DNA to Diversity: Molecular Genetics and the Evolution of Animal Design. Malden, MA: Blackwell Science; 2001.
6. Sinha S, van Nimwegen E, Siggia ED. Bioinformatics. 2003;19(Suppl 1):i292–i301. [PubMed]
7. Sinha S, Schroeder M, Unnerstall U, Gaul U, Siggia ED. BMC Bioinformatics. 2004;5:E129. [PMC free article] [PubMed]
8. Grimaldi D, Engel MS. Evolution of the Insects. Cambridge, UK: Cambridge Univ Press; 2005.
9. Vinogradov AE. Nucleic Acids Res. 2003;31:5212–5220. [PMC free article] [PubMed]
10. Lee S, Kohane I, Kasif S. BMC Genomics. 2005;6:E168.
11. Orenic TV, Jr, Held LI, Paddock SW, Carroll SB. Development (Cambridge, UK) 1993;118:9–20. [PubMed]
12. Zhou X, Riddiford LM. Development (Cambridge, UK) 2002;129:2259–2269. [PubMed]
13. Heng JIT, Tang SS. Bioessays. 2003;25:709–716. [PubMed]
14. Frankfort BJ, Mardon G. Development (Cambridge, UK) 2002;129:1295–1306. [PubMed]
15. Ashraf SI, Hu X, Roote J, Ip YT. EMBO J. 1999;18:6426–6438. [PMC free article] [PubMed]
16. Roman G, Davis RL. Bioessays. 2001;23:571–581. [PubMed]
17. DeZazzo J, Sandstrom D, de Belle S, Velinzon K, Smith P, Grady L, DelVecchio M, Ramaswami M, Tully T. Neuron. 2000;27:145–158. [PubMed]
18. Jones G, Sharp PA. Proc Natl Acad Sci USA. 1997;94:13499–13503. [PMC free article] [PubMed]
19. Barchuk AR, Maleszka R, Simoes ZL. Insect Mol Biol. 2004;13:459–467. [PubMed]
20. Velarde R, Robinson GE. Fahrbach SE Insect Mol Biol. in press. [PMC free article] [PubMed]
21. Stathopoulos A, Levine M. Dev Cell. 2005;9:449–462. [PubMed]
22. Rhodes DR, Yu J, Shanker K, Deshpande N, Varambally R, Ghosh D, Barrette T, Pandey A, Chinnaiyan AM. Proc Natl Acad Sci USA. 2004;101:9309–9314. [PMC free article] [PubMed]
23. Insel TR, Young LJ. Nat Rev Neurosci. 2001;2:129–136. [PubMed]
24. Weaver IC, Szyf M, Meaney MJ. Endocr Res. 2002;28:699. [PubMed]
25. Benson G. Nucleic Acids Res. 1999;27:573–580. [PMC free article] [PubMed]
26. Fernandes M, Xiao H, Lis JT. Nucleic Acids Res. 1994;22:167–173. [PMC free article] [PubMed]
27. Sandelin A, Alkema W, Engstrom P, Wasserman WW, Lenhard B. Nucleic Acids Res. 2004;32:D91–D94. [PMC free article] [PubMed]
28. Schroeder MD, Pearce M, Fak J, Fan H, Unnerstall U, Emberly E, Rajewsky N, Siggia ED, Gaul U. PLoS Biol. 2004;2:E271. [PMC free article] [PubMed]
29. Castillo-Davis CI, Hartl DL. Bioinformatics. 2003;19:891–892. [PubMed]
30. Ben-Dor A, Bruhn L, Friedman N, Nachman I, Schummer M, Yakhini Z. J Comput Biol. 200;7:559–583. [PubMed]
31. Bloch G, Wheeler DL, Robinson GE. In: Hormones, Brain, and Behavior. Pfaff DW, Arnold AP, Etgen AM, Fahrbach SE, Rubin RT, editors. Vol 3. New York: Academic; 2002. pp. 195–236.
32. Ben-Shahar Y, Robichon A, Sokolowski MB, Robinson GE. Science. 2002;296:741–744. [PubMed]
33. Ben-Shahar Y, Dudek NL, Robinson GE. J Exp Biol. 2004;207:3281–3288. [PubMed]

Articles from Proceedings of the National Academy of Sciences of the United States of America are provided here courtesy of National Academy of Sciences
PubReader format: click here to try

Formats:

Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...

Links