2016 年 91 巻 2 号 p. 127-137
Molecular mechanisms underlying standing genetic variation of an ecologically relevant trait such as pigmentation trait variation in a model insect, Drosophila melanogaster, are relevant to our understanding of different kinds of intergenomic interactions. In this study, we focused on the association between body pigmentation and stress resistance, and on genotype-by-environment interaction, both of which are likely to contribute to the persistence of phenotypic variation in a natural population. First, we detected a significant association between pigmentation traits in females and starvation resistance (darker strains were weaker) and a weak association between pigmentation and chill coma recovery time (darker strains showed shorter recovery time) among 20 inbred strains from the Drosophila melanogaster Genetic Reference Panel (DGRP), which originated from a natural population in North America. These associations revealed a complex relationship between body pigmentation and physiological traits that may give rise to balanced selective forces acting on the traits under fluctuating environmental conditions. Second, using four of the DGRP strains, a substantial degree of genotype (strain) × environment (rearing temperature) interaction was detected among expression levels of the genes encoding effector enzymes in the melanin biosynthesis pathway. These interactions can potentially reduce the efficiency of purifying selection on the pigmentation traits over a wide range of temperature conditions. Finally, we discuss possible mechanisms that contribute to the maintenance of the standing pigmentation variation in this species.
Phenotypic variation arises as a realization of various types of genomic variation. How intraspecific phenotypic variation is maintained is a fundamental question in understanding different kinds of intergenomic interactions. Among many different phenotypic traits, diversity in body color provides a clearly visible example of an ecologically relevant trait. Specifically, pigmentation traits in a model insect, Drosophila melanogaster, exhibit a substantial degree of intraspecific variation (David et al., 1985; Munjal et al., 1997; Pool and Aquadro, 2007; Parkash et al., 2008a, b, 2009, 2010; Rebeiz et al., 2009; Takahashi and Takano-Shimizu, 2011; Telonis-Scott et al., 2011; Bastide et al., 2013, 2014; Dembeck et al., 2015). Repeatedly observed clines along latitude (David et al., 1985; Munjal et al., 1997; Parkash et al., 2008a; Telonis-Scott et al., 2011; Bastide et al., 2014) and altitude (Pool and Aquadro, 2007; Parkash et al., 2009, 2010; Bastide et al., 2014) indicate that these traits are involved in adaptation to clinal environmental factors. Furthermore, even within local populations, a considerable degree of variation exists in populations from many regions of the world (Pool and Aquadro, 2007; Parkash et al., 2008b; Rebeiz et al., 2009; Takahashi and Takano-Shimizu, 2011; Bastide et al., 2013, 2014; Dembeck et al., 2015; Miyagi et al., 2015), which suggests that these traits are controlled by some balanced forces.
One of the factors underlying the maintenance of pigmentation polymorphisms is natural selection on physiological traits associated with pigmentation. Indeed, there are studies suggesting an association between pigmentation and UV resistance (Bastide et al., 2014) or desiccation tolerance (Parkash et al., 2008a, b, 2009; Ramniwas et al., 2013) within this species. Also, the “thermal budget (melanism) hypothesis” suggests a relationship between body color and thermoregulation (reviewed in Clusella Trullas et al., 2007). Although Drosophila may be too small to effectively control body temperature by its color (Willmer and Unwin, 1981), recent studies suggest that temperature is a significant predictor of body pigmentation differences (Rajpurohit and Nedved, 2013; Bastide et al., 2014). Since these environmental factors fluctuate seasonally and annually, the phenotype with the highest fitness may vary to a large extent. Therefore, an association between pigmentation and physiology would contribute to the maintenance of variation in the extent of dark pigmentation.
Another factor contributing to the maintenance of pigmentation polymorphisms is the plasticity of the trait. Pigmentation intensity shows a sensitive response to changes in rearing temperature: the lower the temperature, the darker the body pigmentation (David et al., 1985, 1990; Das et al., 1994; Gibert et al., 1996, 2000, 2009; Munjal et al., 1997; Gibert et al., 2007). Gibert et al. (2007) have proposed that the phenotypic plasticity of the extent of pigmentation on the posterior abdominal segments A7 under different rearing temperatures is a side effect of larger-scale changes in a chromatin regulatory network involving the homeotic gene Abdominal-B, inferring that the temperature response is mostly passive. Therefore, selection under fluctuating temperature conditions may not be capable of efficiently reducing the genetic variation underlying pigmentation traits. Also, there are reports on significant genotype-by-environment interaction between populations (Gibert et al., 1996, 2000, 2009; Munjal et al., 1997) or even among strains within populations regarding body pigmentation and rearing temperature (David et al., 1990; Das et al., 1994). This type of interaction can further complicate the target genotype of selection, and thus may play a role in maintaining variation.
The major gene causing the intraspecific variation in overall intensity of body pigmentation is known to be ebony (Pool and Aquadro, 2007; Takahashi et al., 2007; Rebeiz et al., 2009; Takahashi and Takano-Shimizu, 2011; Telonis-Scott et al., 2011; Miyagi et al., 2015). Variations in the extent of the dark stripe on the abdominal segments are associated with cis-regulatory SNPs of the bric-à-brac and tan genes (Kopp et al., 2003; True et al., 2005; Bastide et al., 2013; Dembeck et al., 2015). Therefore, it is likely that changes in pigmentation traits due to different rearing temperatures are reflected in differential expression of ebony, tan, or some of the other effector genes in the melanin biosynthesis pathway (Wright, 1987; Wittkopp et al., 2003). Detecting genotype-by-environment interactions at the gene expression level would strengthen the argument that phenotypic plasticity can prevent the fixation of a particular genotype underlying pigmentation traits.
In this study, we first analyze the association between body pigmentation and several stress resistance traits by comparing body pigmentation measurements (Miyagi et al., 2015) and stress resistance scores (Ayroles et al., 2009; Mackay et al., 2012) using strains sampled from the DGRP. Next, we quantify changes in the expression levels of genes encoding enzymes in the melanin biosynthesis pathway under different rearing temperatures and assess the degrees of genotype-by-environment interaction in these genes. Finally, we discuss the possible contributions of different factors, including natural selection on the associated physiological traits and phenotypic plasticity in response to different rearing temperatures, to the maintenance of genetic variation underlying this trait.
Four inbred D. melanogaster strains from the Drosophila melanogaster Genetic Reference Panel (DGRP; Ayroles et al., 2009; Mackay et al., 2012; Huang et al., 2014), RAL-324, RAL-358, RAL-799 and RAL-852, were used for the temperature treatment assay. These flies were kept at 25 ℃ with a 12-h light-dark cycle on standard corn-meal fly medium until the experiment. Standard corn-meal fly medium was prepared by adding the following ingredients per liter of tap water: 90 g corn-meal, 40 g dry yeast, 100 g glucose, 8 g agar, 3 ml propionic acid, 10 ml butyl 4-hydroxybenzoate. After the food mixture was sufficiently heated and stirred, ~15 ml was dispensed into a glass vial (diameter 30 mm, height 103 mm), which was then plugged by a cotton ball wrapped in a piece of cotton cloth. Data on the following DGRP strains (Ayroles et al., 2009; Mackay et al., 2012; Miyagi et al., 2015) were used for the association study: RAL-208, RAL-303, RAL-324, RAL-335, RAL-357, RAL-358, RAL-360, RAL-365, RAL-380, RAL-399, RAL-517, RAL-555, RAL-705, RAL-707, RAL-732, RAL-774, RAL-786, RAL-799, RAL-820 and RAL-852.
Temperature treatmentFlies were allowed to lay eggs on the standard corn-meal fly medium in a glass vial at 25 ℃ for 1−2 days. After adult flies were removed, the vial was kept at 25 ℃ for 7 days. On day 7, the vials with white pupae were moved to either 20 ℃, 25 ℃ or 29 ℃. Photographs were taken between 7 and 28 days later.
Measuring pigmentation phenotypesPhoto images were captured of three adult females from each strain and temperature treatment. To measure pigmentation intensity, flies were placed in 10% glycerol in ethanol for 30 min at room temperature and gently rocked on a shaker. The samples were then moved to 10% glycerol in PBS and incubated for 15 min with gentle rocking. After removal of legs and wings, the samples were transferred to 10% glycerol in PBS and dorsal images were captured using a digital camera (DP73, Olympus) connected to a stereoscopic microscope (SZX16, Olympus). All images were captured under constant light condition with the reference grayscale (Brightness = 128; ColorChecker, X-rite). White balance was corrected using the white scale (Brightness = 255; ColorChecker, X-rite) with CellSens standard 1.6 software (Olympus). Images were captured using the RGB mode. Pigmentation intensity at the flanking area of the thoracic trident (Fig. 1F) was measured using ImageJ software (Schneider et al., 2012). The mode % grayscale brightness value from each area (0–255) was corrected using the reference grayscale as:
(A)–(D) Dorsal images of the thoracic and abdominal segments of adult females reared at 20 ℃, 25 ℃ or 29 ℃ after 7 days at 25 ℃. Individuals from RAL-799 (A), RAL-852 (B), RAL-358 (C) and RAL-324 (D) are shown. (E) Mean % brightness scores of the thoracic segment (n = 3) in each of the four strains reared at 20 ℃, 25 ℃ or 29 ℃ after 7 days at 25 ℃. One-way ANOVA results for each of the four strains are indicated by asterisks inside the graph. Two-way ANOVA results for testing the genotype (strain) × environment (temperature) interaction are indicated above the graph. ** and *** indicate P < 0.01 and P < 0.001, respectively. (F) The manually selected area in which % brightness was scored is indicated by a black rectangle.
For the % dark-pigmented area measurement, flies were placed in 10% glycerol in ethanol and dissected. The abdominal cuticle was flattened after cutting the entire ventral midline and dorsal midline of the A1−5 segments, leaving the dorsal side of the A6−7 segments intact, and was mounted on a slide. Images were captured as described above. The dark-pigmented area of A6 and A7 (Fig. 2A) in the captured image was extracted by setting the threshold to grayscale brightness value 1−100 (Fig. 2B) using ImageJ software. The borders of the left or right side of A6 and A7 were manually marked and the % area of the dark-pigmented stripe was quantified.
(A) Posterior abdominal segments A6 and A7 are those encompassed by the black box. (B) An example of a modified image of the abdominal segments mounted on a slide. The image was modified by ImageJ software to mark the dark stripe areas of the segments in black. Left and right lateral halves of abdominal segments A6 and A7, respectively, are bordered with red lines. (C) Photo images of slide-mounted abdominal segments from RAL-799, RAL-852, RAL-358 and RAL-324 reared at 20 ℃, 25 ℃ or 29 ℃ after 7 days at 25 ℃. A6 and A7 are encompassed by black boxes. (D) and (E) Mean % dark-pigmented area of A6 (D) and A7 (E) (n = 3) in each of the four strains reared at 20 ℃, 25 ℃ or 29 ℃ after 7 days at 25 ℃. One-way ANOVA results for each of the four strains are indicated by asterisks inside the graphs. Two-way ANOVA results for testing the genotype (strain) × environment (temperature) interaction are indicated above the graphs. *, ** and *** indicate P < 0.05, P < 0.01 and P < 0.001, respectively.
Three to five virgin females from each strain and treatment were collected within 30 min of eclosion. Collected flies were placed in 50% RNAlater (Ambion) in PBS with 0.1% Tween20 (MP Biomedicals) and their heads were removed. The residual body samples were transferred to RNAlater and kept at 4 ℃ until use. Total RNA was extracted using a TRIzol Plus RNA Purification Kit (Life Technologies) and quantified using a Nanodrop 2000c spectrophotometer (Thermo Fisher Scientific). First-strand cDNA was reverse-transcribed from 1 μg of total RNA using a PrimeScript RT Reagent Kit with gDNA Eraser (TaKaRa). Three biological replicate samples were obtained for each strain and treatment.
Real-time qRT-PCR was performed in a 25-μl reaction volume with SYBR Premix Ex Taq II (Tli RNaseH Plus, TaKaRa) on a Thermal Cycler Dice TP800 (TaKaRa). Primers used for the reactions were ebony: 5′−CTTAGTGTGAAACGGCCACAG−3′ and 5′−GCAGCGAACCCATCTTGAA−3′; tan: 5′−GTTGAGGGGCTTCGATAAGA−3′ and 5′−GTCCTCCGGAAAGATCCTG−3′; black: 5′−GGGTGATGGCATCTTCTGTC−3′ and 5′−GCGTCCTCCGAGGTGAAG−3′; yellow: 5′−TTTGGCAATCGGTTATTCGT−3′ and 5′−GTCAAACTGCGGTCCATGT−3′; pale: 5′−CAGCAGCCCAAAAGAACCGC−3′ and 5′−AACGGGCATCATCCACCAGG−3′; RP49: 5′−TCGGATCGATATGCTAA-GCTG−3′ and 5′−TCGATCCGTAACCGATGTTG−3′; and Act57B: 5′−CGTGTCATCCTTGGTTCGAGA−3′ and 5′−ACCGCGAGCGATTAACAAGTG−3′. RP49 and Act57B were used for internal controls, and standard curves drawn from a 5-fold dilution series of cDNA from RAL-852 were used to correct for PCR efficiencies. Two qRT-PCR reactions were conducted for each cDNA.
Associations between variable pigmentation trait values and stress resistance phenotypes in adult females were assessed using 20 strains sampled from the DGRP. Table 1 shows the correlation coefficients between pigmentation traits quantified in Miyagi et al. (2015) and two stress-related phenotypes, starvation resistance and chill coma recovery time, from Ayroles et al. (2009) and Mackay et al. (2012). The results indicated a negative correlation between the % dark-pigmented area of the A4 abdominal segment and starvation resistance, which indicated that darker individuals are less resistant to starvation. Percent darkness of the body may also have a marginally significant association with starvation resistance, although the relationship is weak in our data. Chill coma recovery time may also be weakly associated with the % dark-pigmented area of A4. This is the time (in minutes) to recover at room temperature from an immobilized condition at 0 ℃ for 3 h, and serves as an index of cold resistance (Ayroles et al., 2009; Mackay et al., 2012). The original data used for the analyses in Table 1 are summarized in Supplementary Table S1.
| Tested traits | Pearson’s r | Spearman’s r | ||
|---|---|---|---|---|
| Starvation resistance in females (sex-average) a | ||||
| % darkness in females b | –0.460* | (–0.458*) | –0.468* | (–0.474*) |
| % dark-pigmented area of A4 in females c | –0.566** | (–0.546*) | –0.692** | (–0.669**) |
| % dark-pigmented area of A6 in females d | –0.198 | (–0.209) | –0.164 | (–0.102) |
| Chill coma recovery time in females (sex-average) a | ||||
| % darkness in females b | –0.172 | (–0.075) | –0.278 | (–0.239) |
| % dark-pigmented area of A4 in females c | –0.451* | (–0.492*) | –0.441 | (–0.568*) |
| % dark-pigmented area of A6 in females d | –0.164 | (–0.041) | –0.054 | (0.002) |
*: P < 0.05; **: P < 0.01 (not corrected for multiple tests).
Boldface: P < 0.01 after Bonferroni correction for multiple (six) tests.
The association between starvation resistance and pigmentation has been observed among populations in India (Parkash and Munjal, 2000), where many traits are reported to correlate with latitude (reviewed in Rajpurohit and Nedved, 2013). Body darkness and latitude show a positive correlation, whereas starvation resistance and latitude show a negative correlation. Therefore, these two traits may be independently associated with latitude in the Indian subcontinent. However, the association found in this study within a population from North America, in which there is no evidence for global population structure (Mackay et al., 2012), suggests that these phenotypes share common genetic factors. Recently, it has been shown that cuticular pigmentation is linked to the nutrient-sensing pathway, and that nutrient deprivation during the larval stage results in decreased pigmentation in adults (Zitserman et al., 2012; Shakhmantsir et al., 2014). Pigmentation trait variation may reflect genetic variation underlying some metabolic properties that are associated with starvation resistance in adults. Although the mechanism is not yet clear, an indication of a weak association between pigmentation and two stress resistance traits (Table 1) suggests that body color is indeed genetically associated with many different physiological traits. If this is the case, an inefficient stabilizing selection acting on those pigmentation-associated traits may contribute to the maintenance of genetic variation underlying body pigmentation.
Despite the signature of a possible association between A4 pigmentation and stress resistance, the % dark-pigmented area of A6 did not show any sign of association with either starvation resistance or chill coma recovery time (Table 1). The color patterns in posterior abdominal segments A5−7 are sexually dimorphic: males have a totally dark cuticle, whereas dark pigments on females are restricted to posterior stripes on these segments. The stripe pattern in A5−7 is known to be mainly under the control of bric-à-brac, Abdominal-B and doublesex (Kopp et al., 2000; Williams et al., 2008), and that in non-sexually dimorphic A1−4 is most likely to be controlled by other factors such as optomotor-blind, wingless and decapentaplegic (reviewed in Wittkopp et al., 2003). Therefore, the differential association of A4 and A6 with the stress resistance traits may be due to the effects of different regulating factors, as also pointed out by Gibert et al. (2009).
Furthermore, it has been reported that rearing temperature affects the size of the dark-pigmented area in abdominal segments, most pronouncedly in the posterior A6 and A7 segments (Gibert et al., 1998, 2000; Gibert et al., 2007). One reason for the susceptibility of most posterior segments to temperature changes may be that pigmentation changes in these segments are less prone to fitness reduction through association with vital physiological traits. In this case, purifying selection on the pigmentation status of A6 and A7 may be weak, and its genotype-by-environment interaction may not contribute substantially to the existing pigmentation variation. However, because of the unique regulatory systems in these sexually dimorphic segments, they may be associated with physiological properties that are different from that in sexually monomorphic segments. The two traits investigated in Table 1 are not sex-specific and since males have completely dark pigmentation in these segments, it is possible that the extent of dark pigmentation in these segments in females is associated with a female-specific trait. A broader investigation on various other physiological traits is awaited, but if this is the case, genotype-by-environment interaction still has a consequence on variation maintenance through natural selection on the associated trait.
Temperature effect on pigmentationSome natural populations of D. melanogaster harbor genetic variation for differential phenotypic plasticity, i.e., genotype-by-environment interaction, in regard to body pigmentation and rearing temperature (David et al., 1990; Das et al., 1994). To investigate how much genotype-by-environment interaction for pigmentation traits is present in the DGRP, four sampled strains were subjected to three different temperature treatments: 1) Reared at 25 ℃ and moved to 20 ℃ on day 7; 2) Reared at 25 ℃ for the whole period; 3) Reared at 25 ℃ and moved to 29 ℃ on day 7. Day 7 corresponds to the onset of pupation, after which temperature plays a critical role in adult body pigmentation intensity (Chakir et al., 2002). Adult female flies were sampled on the 7th to 28th day after starting the temperature treatment and pigmentation traits were scored according to the procedure in MATERIALS AND METHODS.
Dorsal images of the flies captured under constant light condition showed that the temperature response of body pigmentation varied among strains (Fig. 1, A–D, Supplementary Table S2). Quantification of the % brightness of the flanking area of the thoracic trident (Fig. 1F) in individuals subjected to different temperature treatment showed a significant genotype (strain) × environment (temperature) effect (Fig. 1E, two-way ANOVA, P < 10–4). Among the four strains, only RAL-358 showed a significant temperature effect on the brightness value, which was in the direction to become lighter in response to increased rearing temperature (Fig. 1, C and E, one-way ANOVA, P < 0.01 after Bonferroni correction). Some other studies showed a more consistent temperature effect on the darkness of the thoracic segment across wider temperature ranges (David et al., 1985; Gibert et al., 2000, 2009). The thoracic color response to the temperature range tested in this study may be variable compared to the abdominal color response in Fig. 2.
Percent dark-pigmented areas of the A6 and A7 segments (Fig. 2A) were quantified after dissecting the abdomen and flattening the dorsal cuticle (Fig. 2B, details in MATERIALS AND METHODS). The measurements of A6 and A7 abdominal segments are listed in Supplementary Table S3. All four strains showed a significant temperature effect on the % dark-pigmented area of A6 (Fig. 2D, one-way ANOVA results after Bonferroni correction) and A7 (Fig. 2E, one-way ANOVA results after Bonferroni correction). A significant genotype (strain) × environment (temperature) effect was detected in both A6 (Fig. 2D, two-way ANOVA, P < 10–4) and A7 (Fig. 2E, two-way ANOVA, P < 10–4), indicating that the response to temperature varied among strains (Fig. 2C). Most notably, the % dark-pigmented areas of A6 and A7 in RAL-852 showed a weak response to temperature changes, in contrast to those of the other three strains.
Our results showed considerable degrees of genotype-by-environment interaction in the body pigmentation and rearing temperature comparisons (% brightness of the thoracic trident (Fig. 1E), and % pigmented area of A6 and A7 (Fig. 2, D and E, respectively)) among strains in the DGRP. The genetic bases for this genotype-by-environment interaction are not fully understood. However, Gibert et al. (2007) have suggested that expression of ebony and pale is likely to be affected by temperature. Thus, we next quantified the expression level of selected genes by real-time qRT-PCR.
Temperature effect on genes encoding enzymes in the melanin biosynthesis pathwayCuticle pigmentation takes place during the early adult stage immediately after eclosion, when the melanin biosynthesis pathway is activated. Transcript abundance of five genes encoding enzymes in this pathway (Fig. 3F) was quantified from headless body samples of females within 30 min after eclosion. Standardized expression levels of these genes quantified by qRT-PCR are shown in Supplementary Table S4. The abundance of each transcript showed a different response to temperature changes, except that of black (Fig. 3, A–E). The expected direction of change in expression levels can be inferred from the simplified pathway model shown in Fig. 3F. An increase in the expression levels of ebony and black should make the body color lighter by acting in the direction to increase yellowish pigment, whereas a reduction in the expression levels of tan and yellow is expected to make the body color darker by acting in the direction to increase darker pigments. Since body color becomes lighter in response to increased rearing temperature, changes in the expression levels of tan in RAL-799 and RAL-852 (Fig. 3B), and of ebony in RAL-358 (Fig. 3A), are in the expected direction. Expression level changes in black and yellow genes (Fig. 3, C and D, respectively) are in the opposite direction.
(A)–(E) Standardized relative expression levels of ebony (A), tan (B), black (C), yellow (D) and pale (E) quantified by real-time qPCR from headless body samples of females within 30 min of eclosion. Mean expression levels obtained from RAL-799, RAL-852, RAL-358 and RAL-324 reared at 20 ℃, 25 ℃ or 29 ℃ after 7 days at 25 ℃ are shown. One-way ANOVA results for each of the four strains are indicated by asterisks inside the graphs. Two-way ANOVA results for testing genotype (strain) × environment (temperature) interaction are indicated above the graphs. *, ** and *** indicate P < 0.05, P < 0.01 and P < 0.001, respectively; NS, not significant. (F) Schematic representation of the simplified metabolic pathway for melanin biosynthesis. Genes analyzed in this study are shown in blue. Yellow arrows show reactions enhancing the lighter cuticle color and black arrows show reactions producing darker cuticle color. Gray arrows show reactions synthesizing precursors of both light- and dark-colored pigments.
Substantial changes in the darkness of the whole body as well as the % brightness of the flanking area of the thoracic trident were detected in RAL-358 (Fig. 1, C and F). Notably, this strain showed a slight but significant increase in the ebony gene expression level at higher temperatures (Fig. 3A). ebony is known to be the major locus controlling variation in whole-body pigmentation intensity. However, the modest increase in ebony expression does not seem sufficient to explain the degree of pigmentation change in response to temperature changes in RAL-358. Despite a wide variation in expression level among strains under all temperature conditions (Fig. 3A), this gene may not be playing a major role in controlling body color changes in response to different temperature conditions in some genotypes.
A significant genotype (strain) × environment (temperature) effect was detected in tan and yellow genes after Bonferroni corrections for multiple (five) tests, which indicated that the response to temperature changes varied among strains in these genes. tan is associated with the extent of dark stripes in A5−A7 segments (Kopp et al., 2003; True et al., 2005; Bastide et al., 2013; Dembeck et al., 2015). Thus, the reduced expression of tan at high temperature, at which stripe areas are reduced, is consistent with the expectation. However, RAL-852, which showed a small response in abdominal pigmentation to temperature changes, showed a significantly reduced expression level of tan under high-temperature conditions (Fig. 3B), which is unexpected. A possible explanation is that the changes in tan expression levels are overridden by high expression levels of ebony at all temperatures (Fig. 3A).
Moreover, curiously, this strain showed no significant changes due to temperature in the yellow gene expression level (Fig. 3D). The yellow gene is expressed maximally on the second and third day of the pupa stage (Gelbart and Emmert, 2013). Transcript distribution is thereafter confined to the posterior region of each abdominal segment, where a dark stripe is going to form, and eventually the secreted Yellow protein enters the cuticle by the time pigmentation reactions actually take place (Wittkopp et al., 2002); therefore, this gene is also known to play an important role in abdominal stripe formation. It is thus possible that in RAL-799, RAL-358 and RAL-324, the confinement process is somehow delayed relative to other metamorphosis processes, and the subsequent protein secretion step is still insufficient at eclosion under high-temperature conditions. This might explain why there is higher yellow expression at higher temperatures in these strains. yellow gene expression in RAL-852 might be robust and not show this delayed process under high-temperature conditions, thus accounting for the weak response in the extent of abdominal pigmentation to temperature changes.
The temperature was switched from 25 ℃ to three different temperatures on day 7 (the 7th day after the eggs were laid). This is the time when pupation initiates, and the period between late pupa and early adult is known to be the critical period for determining the cuticle darkness of the adult body (Chakir et al., 2002). This temperature switch may have altered the expression level as well as the timing of the expression peaks for these genes. A precise assessment of the periodic expression profile changes by a broader stage sampling of these genes would clarify this issue.
We considered two aspects related to the universally observed body color variation in this species: 1) association with stress resistance traits, and 2) genotype-by-environment interaction. We found that in the DGRP, there were indications of weak associations between body pigmentation and at least two stress resistance traits, starvation resistance and chill coma recovery time (Table 1). The molecular mechanisms underlying these associations are not clear; however, it was interesting to find that even in our limited comparisons, the correlation between extent of body pigmentation and resistance was in the opposite direction in the two traits tested. Darker strains showed weaker starvation resistance, but stronger cold resistance. Because the correlation is in the opposite direction, natural selection on these traits is likely to fluctuate depending on the conditions and may act in a balanced manner. Our results, together with the universally observed polymorphisms in pigmentation traits (David et al., 1985; Munjal et al., 1997; Pool and Aquadro, 2007; Parkash et al., 2008a, b, 2009, 2010; Rebeiz et al., 2009; Takahashi and Takano-Shimizu, 2011; Telonis-Scott et al., 2011; Bastide et al., 2013, 2014; Dembeck et al., 2015), support this view.
There are many other traits or factors that potentially associate with body pigmentation. Bastide et al. (2014) have analyzed different factors associated with pigmentation in African populations and found that UV radiation as well as temperature, humidity, soil conditions and some more factors are significant. Testing other physiological traits, together with investigating the molecular mechanisms underlying the associations by genetic and functional assays, should lead to a comprehensive understanding of the pigmentation-associated traits.
Among the four strains from the DGRP that were subjected to temperature treatments, only one, RAL-358, showed a clear increase in thoracic darkness at higher temperatures (Fig. 1). A more consistent response appeared in the extent of dark-pigmented stripes in the abdominal segments A6 and A7, which showed reduced values in all four strains at higher temperatures with a relatively minor response in RAL-852 (Fig. 2). The associated changes in expression levels of genes encoding effector enzymes in the melanin biosynthesis pathway were not straightforward. An increase in the expression level of ebony at high temperatures in RAL-358 but not in other strains (Fig. 3A) was consistent with the pattern of thoracic darkness (Fig. 1); however, the degree of expression changes did not seem to explain the extent of changes in pigmentation. A role for ebony in phenotypic plasticity is suggested from the data, but it may be only a limited one. The pattern of changes among the four strains in A6/A7 pigmentation following temperature change was most consistent with the changes in yellow gene expression, although in the opposite direction to that predicted from the metabolic pathway. Since Yellow is a secretory protein and its encoding gene has a unique developmental profile of changes in transcript locality, transcript abundance at one time point may not be a good predictor of the effective enzyme abundance within the time window when pigmentation actually takes place. Periodical measurements of transcript abundance may clarify whether or not the peak expression level of this gene has shifted.
Although the changes in effector gene expression that are responsible for the pigmentation changes due to different rearing temperatures were not fully identified, it was clear that all five genes except black showed a significant genotype-by-environment interaction effect. This means that purifying selection under one temperature condition will not effectively reduce phenotypic variation under other conditions. Therefore, our data at least indicate that genetic variation in expression level regulation of these pigmentation effector genes (due to cis- or trans-regulatory effects, or both) is potentially maintained through genotype-by-environment interaction assuming that the trait is under selection. Extending the analyses to trans-regulatory factors such as Abdominal-B, bric-à-brac and optomotor-blind, focusing on limited body positions may provide a clearer view of the molecular mechanism for the observed interaction.
Pigmentation trait variation in D. melanogaster is a long-standing variation observed in many regions of the world (reviewed in Takahashi, 2013). Maintenance of the genetic variation underlying this phenotype is likely to involve many complex, intermingled factors such as association with multiple physiological and behavioral traits as well as substantial genotype-by-environment interaction. How these ecologically relevant phenotypic variations persist in a population and how they diverge between species are old but challenging questions to be pursued. Undertaking more detailed investigation of the molecular mechanisms and genetics underlying pigmentation in Drosophila should certainly shed light on these fundamental questions.
We thank the Bloomington Stock Center for fly stocks and Koichiro Tamura for valuable discussions. This work was supported by a Grant-in-Aid for Scientific Research on Innovative Areas from the Ministry of Education, Culture, Sports, Science and Technology of Japan.
