Substantial phenotypic variation among lines was observed for DT (Figure 1), wherein the most commons effect was an increase of DT relative to the control. This is not unexpected as the most likely effect of mutations on a fitness-related trait is reasonable to be deleterious, although, we found a significant positive correlation (r: 0.3, p < 3 × 10⁵) between DT and viability [see Additional file 1]. This pattern revealed that the result observed of the heterochornic effect of the P[GT1] insertion lines can not be explain as a consequence of unfit flies. The most extreme phenotypes were exhibited by lines BG01543 and BG01412. The former developed 60 and 50 hours (males and females, respectively) faster and the latter 119 and 146 hours (males and females, respectively) slower than the control. We observed a significant line effect indicating mutational variance among single P-element insertion lines (Table 1). Indeed, the line effect accounts for 84% of total variance, five times higher than the error. In the ANOVA, differences between sexes were also significant, with males developing faster than females. In addition, the contribution of the line-by-sex interaction to the total phenotypic variance was not significant.

Dunnett analyses revealed that 107 out of 179 lines (60%) differed significantly from the control. Mean values of all significant P[GT1] insertion lines tested at 25°C are shown in Additional file 2. In seventy-four lines (40%) the insertion of the P-element caused an increase in DT, while in the rest of the significant lines DT was shortened as compared to the control (20%) (Figure 1). Those lines that exhibited significant DT differences relative to the control (heterochronic mutants) were considered as lines bearing an insertion in a candidate DT gene. Among fast developing lines, BG01543 and BG01902 stand out, since DT was reduced in each by more than 2 days. The P-element in line BG01902 disrupts mastermind (mam), a gene encoding a glutamine-rich nuclear protein 26 that is a positive transcriptional regulator of the Notch signaling pathway 27282930. Notch is a central element in the cell signaling mechanism that controls a broad spectrum of cell fate choices during development across metazoans 313233. Interestingly, in humans, abnormalities in Notch signaling have been linked to a number of diseases, such as T cell acute lymphoblastic leukemia 34 and aortic valve disease 35. In line BG01543, the P-element insertion occurs in Merlin (Mer), a well-known negative regulator required for cell proliferation in developing imaginal discs of Drosophila 25. It is a component of the Hippo signaling pathway, which is essential for the regulation of organ size during development 36. Mer might control tissue growth by regulating endocytosis of membrane receptors in imaginal epithelia 37. Interestingly, Mer is the homolog of the human tumor suppressor gene Neurofibromatosis Type 2 (NT2) involved in the de-regulation of cell proliferation in tumors of central nervous system pathologies 2538.

Another gene involved in a cell proliferation pathway identified in our screen is forkhead box, sub-group O (foxo). The insertion line BG01573 bearing a P-element in foxo increased DT by 24 hours in both sexes. foxo is a key regulatory component of the insulin-signaling pathway, which in Drosophila regulates the control of growth size of cells, organs, and the entire body in response to nutrient availability 39. Puig et al. 40 established that foxo activates transcription in downstream as well as upstream targets of this signaling cascade by a transcriptional feedback mechanism that regulates cell growth and proliferation. It has been suggested that the insulin-signaling pathway regulates cell proliferation in imaginal discs, though the duration of the proliferation phases are controlled by Juvenile Hormone (JH) and ecdysteroids, that are themselves unaffected by the insulin-signaling pathway 41. Premature ecdysone release leads to the rapid development of small adults, while the delay in ecdysone release extends developmental time and produces larger adults 42. Recent studies have shown that the activity of the insulin-signaling pathway in the prothoracic gland modulates ecdysone release and influence both the length and the rate of larval growth 434445. It has been suggested that the neuropeptide Amnesiac (Amn) participates in regulating ecdysone synthesis in the prothoracic gland of Drosophila 43. Interestingly, the P-element insertion mutant at Amn (line BG02286) also extended DT by more than two days in both sexes [see Additional file 2]. However, ecdysone can directly inhibit insulin and growth in the fat body and other peripheral tissues, an event that triggers metamorphosis. Furthermore, the ecdysone suppression of growth rate is lost in foxo mutants 45, indicating complex cross-talk between the ecdysone and the insulin signaling pathway during Drosophila development.

Our survey revealed extreme heterochronic mutants, particularly in the direction of increased DT [see Additional file 2]. For instance, the P-element insertion in the gene Karl (BG01010) extended DT by more than 100 hours in both sexes. The molecular function inferred from protein sequence suggests that Karl is a retinol-binding protein with no clear association to a biological process. In line BG00372 we registered a delay of 93 hours and 108 hours for males and females, respectively. In this case the P-element has inserted 1498 bp downstream of the 3' end of the CG1678 gene. The molecular function and the biological processes in which this gene is involved are unknown. However, there is evidence that CG1678 interacts with other genes such as And, ewg, CG1472, rl, Bsg25D, CG11275 and msb1l 46. Line BG01412 exhibited the most extreme phenotype in our screen, the P-element insertion was associated with an extended developmental time of 119 and 146 hours in males and females, respectively. Unfortunately, information about the nucleotide sequences flanking the P-element insertion site is not available.

Candidate DT genes are involved in a wide range of biological processes according to their gene-ontology (GO) terms (Table 2). Interestingly, DT genes not only are involved in processes associated to organismal development but also to biosynthetic and cellular metabolic processes. We did not find significant differences between groups of candidate genes accelerating development vs those that extended development in the distribution among GO terms, suggesting that similar ontogenetic pathways may be responsible for both types of heterochronic phenotypes.

Growth and development of ectotherms are determined in part by their thermal environment 2223. In particular, temperature during ontogeny exerts a strong influence shaping the evolution of larval traits 244748. In fact, phenotypic responses result of adaptation to different thermal environments and/or may be an unavoidable consequence of the effect of temperature on the organism's physiology during development 49. In this context, the pattern of phenotypic effects of P-element insertion lines reared at different developmental temperatures would provide new insights in the study of phenotypic evolution of larval traits. The ANOVA showed that differences among lines and between thermal treatments were significant (Table 3). More importantly, our screen revealed that the line-by-sex and the line-by-temperature interactions were also highly significant, indicating that the behavior of each line depended on the temperature at which it was reared and the sex. However, there was a large difference in the magnitude of these genotype-by-environment interactions. The former accounts for only 1% of the total phenotypic variance (a percentage similar to that obtained in the 25°C assay) while the line-by-temperature interaction explained 52% of the variation. Moreover, note that in the general assay, the percentage of total phenotypic variance explained by differences among lines was 84%, whereas in our assays of phenotypic plasticity this percentage dropped to 30%. It may be argued that part of the effect was obscured by the high value of the line-by-temperature interaction term. This observation opens an excellent opportunity for studying the genetic basis of phenotypic plasticity of developmental time. Thus, we decided to analyze whether the significant line-by-temperature interaction can be explained by changes in magnitude of among-line variance across thermal treatments or changes in the rank order among lines, i.e. a cross-temperature genetic correlation lower than unity. Our results showed that about half of the interaction variance can be explained by a greater among-line variance observed at 17°C than 25°C and the other half by temperature-specific effects on DT of the lines (Figure 2).

Thirty out of 42 lines tested at 17°C and 25°C showed significant differences relative to the control at one or both temperatures (Table 4). In addition, DT measured at different temperatures were positive correlated for the set of 30 heterochronic lines in both sexes (males: r: 0.61, p < 0.05; females: r: 0.55, p < 0.05; Figure 3). These results imply that most of the mutations that affected DT at both temperatures did so in the same phenotypic direction (either increasing or decreasing DT). Forty seven percent of the lines showed this pattern (bottom left and top right quadrants in Figure 3). Nevertheless, we also identified a set of lines in which the effect of the mutation on DT was temperature-dependent, suggesting that the mutated loci may be possible candidates for temperature plasticity genes. Indeed, forty percent of the lines affected DT in only one of the temperatures. We observed three lines showing a heterochronic phenotype only at 25°C. The line bearing an insertion in CG14478 is an example of this pattern. In contrast, nine lines showed a heterochronic phenotype only at 17°C. For example, the insertion in Imp exhibited significant differences at 17°C in both sexes, but not at 25°C. Surprisingly, four lines (13%) affected DT in opposite phenotypic directions in the two rearing temperatures. On the one hand, line BG02159, bearing a mutation in CG32666, decreased and increased DT at 17°C and 25°C, respectively. On the other hand, mutant lines for genes Nmdar, sugarless and CG11226 exhibited the opposite pattern (Figure 4).

The genetic architecture of a trait determines the variational properties of the phenotype, that is, its evolutionary potential. Our study revealed 116 heterochronic independent mutants involved in the expression of DT (107 lines showed a mutant phenotype at 25°C, while 9 lines were heterochronic only at 17°C). In this sense, we were capable to identify components of different kind of pathways involved in DT expression. On the one hand, some components correspond to molecular mechanisms directly involved in timekeeping processes, such as, the ecdysone and the insulin-signaling pathway 415051. It has been established that environmental inputs during ontogeny influences multiple steps of the these kind of heterochronic gene pathways, such as, the nutrition status and the light: dark regimes. A paradigmatic example of this kind of environmental input is the circadian rhythm period mutants that alter DT in Drosophila 52. On the other hand, we also observed heterochronic phenotypes that seem to be a by-product of the disruption of a pathway that is not explicitly involved in the control of temporal process but plays a major role in organ growth This is the case of Merlin, a negative regulator of the Hippo signaling pathway, required for cell proliferation in developing imaginal discs. Interestingly, Merlin is a component of a tumour-suppressor network associated with human tumorigenesis 36. All in all, our results support the hypothesis raised by Moss 53 that timing of development is control directly and indirectly by different ontogenetic pathways.

Since the same co-isogenic lines were also screened for abdominal and sternopleural bristle number, starvation resistance and olfactory behavior 175455 we also examined whether DT candidate genes have pleiotropic effects on adult fitness-related traits. Several DT heterochronic lines (genes) affect starvation resistance (4 genes), bristle number (8 genes) and olfactory behavior (13 genes) [see Additional file 3]. Once again, Merlin seems to plays a role not only in egg-to-adult DT but also in starvation resistance and adult olfactory behavior. In BG01543 (Mer mutant), the P[GT1] insertion in exon 1 caused a reduction in the RNA expression levels in embryos, third instar larvae and adults but not in the pupal stage compared to the control 17 indicating that anomalies caused by the disruption of Merlin on DT took place before metamorphosis. Since insects do not grow as adults, their final size is a product of growth rate and length at larval stages 56. Surprisingly, the mutated Merlin line did not differ significantly from the control in several morphological traits such as face width, head width, thorax length and wing size suggesting a decoupled mechanism of timing control and size development (Carreira, Mensch & Fanara, unpublished results). Regarding the pleiotropic effect of Mer on starvation resistance and adult olfactory behavior, it is not easy to discern whether aberrant phenotypes are the consequence of early developmental events or are directly related to the functional disruption of the gene product in the adult. As another example, the insertion line for tramtrack (ttk), a transcription repressor gene involved in cell-fate determination 57, increased DT by two days in both sexes and also affected starvation resistance and olfactory behavior. These interesting examples constitute clear evidence of an intricate network of pleiotropic effects of key developmental genes throughout the life-cycle of a holometabolous organism. Such pervasive pleiotropy of key genes that control development, affecting early and late fitness-related traits, show an integrated picture of the evolution of life-history traits at the molecular level.

Regarding the effect of sex, males developed significantly faster than females, an observation that is quite striking since it is at odds with extensive literature showing that Drosophila females usually reach adulthood before males in the species studied so far 58. Moreover, a recent study by Paranjpe et al. 59 showed that females developed faster than males in D. melanogaster. However, some differences between our study and Paranjpe et al.'s in terms of the experimental design are worth mentioning. Paranjpe et al. found that females developed faster than males under several photoperiod regimes during ontogeny, including 12:12 L:D (the regime performed in our study). Nevertheless, we must note that different genetic backgrounds were analyzed in these two studies. Indeed, our control line, which shares the same genetic background with all mutant lines tested, did not show any sign of sexual dimorphism in DT, stressing the need to take into account the effect of the genetic background on the expression of the analyzed trait since genetic background × gene interactions are known to be quite pervasive 6061. At the same time the null input of the line-by-sex interaction to the total phenotypic variance adds a complementary aspect to the discussion of the absence of sexual dimorphism for DT. In the last few years a similar set of coisogenic insertion lines was studied for several adult traits for which the line-by-sex interaction explained significant portions of variance. In this context, our DT survey yielded the lowest value in this category, followed by sternopleural bristle number (5%), adult odor-behavior (13%), abdominal bristle number (21%) and starvation resistance (66%) 175455. Among the heterochronic mutants, nine had a male-specific effect and in seven only female DT was affected. It is important to note that none of the lines showed opposite phenotypic effects across sexes, meaning that we did not find any lines that decreased DT in one sex and increased the trait in the other. In conclusion, our results indicate that there are features of the genetic architecture of DT that are highly conserved in both sexes.

Although the multiple interacting genes affecting complex traits can readily be dissected, how much genotype-environment interactions contribute to variation in these traits remains elusive. The quantitative dissection of developmental time reported here reveals that almost all heterochronic lines tested at different developmental temperature presented an effect with quantitative variation between temperatures, a remarkable consequence of the large gene-by-environment interaction (Table 4). The most striking examples of these developmental reaction norms are the mutant lines for genes CG11226, CG32666, Nmdar1, and sugarless (Figure 4). CG32666 gene product is a receptor signaling protein with serine/threonine kinase activity, NMDAR1 is a ionotropic glutamate receptor and SUGARLESS is a protein with UDP-glucose 6-dehrygenase activity indicating that this routine cellular metabolic processes are not only involved in developmental time but also that are sensitive to environmental variation.

The part of the genetic diversity that has the potential to affect the phenotype, but that is not expressed under the current genotypic or environmental conditions, is referred to as "cryptic" or "hidden" genetic variation 62. Molecular mechanisms responsible for this particular genetic variation include epistasis and genotype-by-environment interactions. In our case, in addition to the plastic developmental reaction norms, large input to the gene-by-environment interaction refers to the bigger variance magnitude at 17°C (Figure 2). It is possible that at 25°C genetic variation was buffered in comparison to the variation expressed at 17°C, an unusual rearing temperature for these lines. Taken together, our results highlight the large potential of the Drosophila genome to changes in this relevant environmental factor, although this genetic plasticity was exhibited by these particular inbred lines, a pattern that it is not necessarily true for outbred populations. Future efforts will focus on elucidating the molecular mechanisms controlling the temperature plastic response observed. Since in nature insects are exposed to a wide range of environmental temperatures during the day and also in different seasons along the year with local-specific thermal regimes, we are also interested in studying the evolutionary forces shaping the variation in the candidate DT genes.

We scored 179 homozygous viable P[GT1] insertion lines, contructed in a co-isogenic Canton-S B background 63 as part of the Berkeley Drosophila Genome Project (see Availability and requirements section for URL) for DT.

Batches of 30–40 P-element insertion lines were assessed simultaneously. To account for environmental variation in DT between batches, 8 replicated vials of the control strain (a co-isogenic P-element insertion free line with the same genetic background) was run in parallel with each batch. For each line, 300 pairs of sexually mature flies were placed in oviposition chambers for 8 hours. Eggs were allowed to hatch and batches of 30 first-instar larvae were transferred to culture vials containing a cornmeal-agar-molasses medium (4 replicates vial per line). Emerged flies from each vial were collected every 12 hours and sorted by sex. We estimated DT as the time elapsed since the transfer of first-instar larvae to the vials until adult emergence. Each vial was kept in an incubator at 25°C ± 0.5, under a 12:12 h light:dark photoperiod and at 60–70% of humidity. Thirteen lines with low viability (those that reached less than 50% of control pre-adult survival) were excluded from our survey to avoid the confusing effect of high pre-adult mortality with DT. A subset of 42 lines selected at random was also screened at 17 ± 0.5°C.