2022 Volume 97 Issue 5 Pages 221-227
Physiological responses to environmental changes play important roles in adaptive evolution. In particular, homeostatic regulatory systems that maintain constant circulating glucose levels are crucial in animals. However, variation in circulating glucose levels and the genetic effects on phenotypic variation in natural populations remain to be clarified. Here, we investigated the hemolymph glucose levels in natural populations of Drosophila melanogaster and its sibling species, D. simulans, in Japan. We quantified hemolymph glucose concentrations in third instar larvae of 27 lines for each species, which were reared on either glucose-free or glucose-rich food. In both species, genetic variation was not a major component of phenotypic variation on either glucose-free or glucose-rich food. The hemolymph glucose concentrations were much higher in D. simulans than in D. melanogaster. Genetic variance was larger in D. simulans than in D. melanogaster. The observed differences between the two species may be associated with the much more recent colonization history of D. simulans populations in Japan and/or the tolerance to environmental stresses. Our findings suggest that natural selection acting on hemolymph glucose levels in D. melanogaster is different from that in D. simulans.
Variation in natural populations is a primary source of evolution and most evolutionarily important phenotypes are quantitative traits (Falconer and Mackay, 1996). How much phenotypic and genetic variation exists in natural populations? Revealing the amounts of variation in natural populations is fundamental to understanding adaptive evolution.
Homeostatic regulatory systems that maintain circulating glucose levels are crucial for life. For example, extraordinarily high glucose concentrations in human blood cause serious health problems such as diabetes. Regulation of glucose levels in human blood is mediated by endocrine hormones such as insulin and glucagon. It has been revealed that regulatory mechanisms of sugar homeostasis in human resemble those in Drosophila. For example, Drosophila insulin-like peptides and adipokinetic hormone correspond functionally to human insulin and glucagon, respectively (e.g., Rulifson et al., 2002; Kim and Rulifson, 2004; Lee and Park, 2004; Musselman et al., 2011; Gáliková et al., 2015; Oh et al., 2019). However, the extent of variation in hemolymph glucose levels in natural populations of Drosophila remains to be clarified.
Now consider that hemolymph glucose concentration of an ith individual, where i = 1–n, is Xi on glucose-free food (G-) and Yi on glucose-rich food (G+). Homeostatic regulation means that the hemolymph glucose level of the ith individual should be maintained to be constant regardless of the G- or G+ food. That is, in the ith individual, Xi should be equal to Yi. When the number of individuals is k in a population and the hemolymph glucose concentration is different between individuals, we would observe a positive correlation between Xi and Yi for the k individuals. On the other hand, if there is a particular hemolymph glucose concentration at which individuals can survive better than others in a population, glucose concentration in those individuals would be similar. As a result, we would not find any correlation between Xi and Yi for the k individuals.
Because Drosophila larvae cannot move far, they utilize food within a limited area (McKenzie and McKechnie, 1979). In nature, as stable supplies of carbohydrate sources are not likely, individuals capable of adapting to changing environments would be favored. Indeed, larval feeding behaviors are influenced by nutrients such as carbohydrates and proteins (Vijendravarma et al., 2012; Hentze et al., 2015; Ugrankar et al., 2018). For example, high levels of hemolymph glucose suppress feeding in D. melanogaster larvae (Ugrankar et al., 2018).
Comparisons of the amount of variation between species may also provide insight into our understanding of adaptive evolution. Drosophila simulans is one of the most closely related species to D. melanogaster. Thus, the molecular mechanisms underlying quantitative traits should be basically the same. However, some morphological traits such as thorax length, wing length and ovariole number are different between the two species (Pétavy et al., 1997; Gibert et al., 2004), while sternopleural bristle number is similar (Gibert et al., 2004). Physiological differences are also reported in, for example, alcohol tolerance (e.g., McKenzie and McKechnie, 1979; van Herewege and David, 1980), desiccation and starvation resistance (e.g., Hoffmann and Harshman, 1999), and cold resistance (e.g., McKenzie and McKechnie, 1979). It is considered that these differences are consequences of adaptation.
The main objective of this study was to estimate the hemolymph glucose levels under different sugar conditions in D. melanogaster and D. simulans. We collected flies of D. melanogaster and D. simulans from Hikone, a temperate zone in Japan. We then examined hemolymph glucose levels in third instar larvae of 27 lines for each species, which were reared on either glucose-free or glucose-rich food.
We collected flies of D. melanogaster and D. simulans from natural populations in Hikone, located in the northern temperate zone of Japan. In this study we used 27 lines for both D. melanogaster and D. simulans. We quantified hemolymph glucose concentrations of third instar larvae, which were reared on either glucose-free or glucose-rich food. Hereafter, we call glucose-free and glucose-rich food G- and G+, respectively.
Hemolymph glucose levels were similar in larvae reared on both foods in D. melanogaster, but not in D. simulans (Fig. 1). On the G- food, hemolymph glucose concentration in D. melanogaster ranged from 11.3 mg/dl to 74.0 mg/dl. On G+, it ranged from 15.4 mg/dl to 64.2 mg/dl. In this species, the difference of glucose levels was not significant between G- and G+. On the other hand, the glucose levels were significantly higher on G- than on G+ in D. simulans (Fig. 1, P < 0.05). On G- hemolymph glucose concentration in D. simulans ranged from 68.8 mg/dl to 286.1 mg/dl, while on G+ it ranged from 37.1 mg/dl to 379.7 mg/dl. In comparisons between the species, the glucose levels were significantly higher in D. simulans than in D. melanogaster on both G- and G+ (Fig. 1, P < 0.001).
Distribution of hemolymph glucose concentrations. The number of lines is 27 in both species. G- and G+ indicate glucose-free and glucose-rich foods, respectively. A horizontal bar and a cross in a box show median and average values, respectively. ns: not significant, *P < 0.05, ***P < 0.001 after Bonferroni correction.
To compare the degree of phenotypic variation, the coefficient of variation (CV) was estimated. In terms of CV, phenotypic variation in D. melanogaster on G- (0.429) was similar to that on G+ (0.407). In contrast, in D. simulans, phenotypic variation on G+ (0.616) was ~1.6 times larger than that on G- (0.382).
Response to dietary glucoseAs described above (see Introduction), when each Drosophila line has a particular glucose concentration, Xi on G- and Yi on G+, and in addition, Xi = Yi, glucose concentrations in the 27 lines should show positive correlations between G- and G+. In this study, no significant correlations were found in either D. melanogaster (r = 0.118) or D. simulans (r = 0.288) (Fig. 2). In D. melanogaster, values of glucose concentration appeared to gather around ~30 mg/dl. Seventeen of the 27 lines (~63%) were within the range of 20–40 mg/dl. There was a single line of D. simulans with a strikingly high concentration on G+ (~380 mg/dl) (Fig. 2B). Excluding this line, significant positive correlation was found (r = 0.51, P < 0.01), indicating that D. simulans lines with low glucose levels on G- have low glucose levels on G+. In 21 of the 27 lines (~78%) of D. simulans, glucose concentrations on G- were higher than those on G+.
Scatter plots of hemolymph glucose concentration on G- and G+ foods. X- and Y-axes represent the glucose concentrations on G- and G+ foods, respectively. The diagonal line indicates that the concentration on G- equals that on G+. The correlation coefficient (r) was 0.118 in D. melanogaster (A) and 0.288 in D. simulans (B).
To find the extent to which the concentration changes, the ratio of glucose concentration on G- to that on G+ (G-/G+) was calculated for each line. Figure 3 shows the distribution of the G-/G+ ratios in D. melanogaster and in D. simulans. The average ratio was ~1.3 and ~1.7, respectively, in D. melanogaster and D. simulans. The ratio was significantly higher in D. simulans than in D. melanogaster (P < 0.05).
Ratio of glucose concentration on G- to that on G+. A horizontal bar and a cross in a box show the median and average values, respectively. The number of lines is 27 in both species. *P < 0.05.
Phenotypic variation (total variance) was divided into variance components by ANOVA. Table 1 shows the results of the two-way ANOVA. The line effect was significant in both D. melanogaster (P < 0.05) and D. simulans (P < 0.001), indicating that there was genetic variation in hemolymph glucose levels. A significant interaction effect was also found in both species. On the other hand, the food effect was significant in D. simulans (P < 0.01), but not in D. melanogaster.
| Source | df | Mean square | F |
|---|---|---|---|
| D. melanogaster | |||
| Food | 1 | 0.01771 | 0.13 |
| Line | 26 | 0.21064 | 2.89*** |
| Interaction | 26 | 0.13729 | 1.88** |
| Error | 356 | 0.07290 | |
| D. simulans | |||
| Food | 1 | 2.25592 | 11.32** |
| Line | 26 | 0.33723 | 4.44*** |
| Interaction | 26 | 0.19920 | 2.62*** |
| Error | 256 | 0.07594 |
** P < 0.01, *** P < 0.001.
Based on the results of one-way ANOVA (Table 2), broad-sense heritability (H2) was estimated for each food. In D. melanogaster, H2 values on G- and G+ were 0.113 and 0.226, respectively, suggesting that genetic variation is not a major component of phenotypic variation on either G- or G+. In D. melanogaster, genetic variation (σ2L) was not significantly different between G- and G+. On the other hand, in D. simulans, H2 values on G- and G+ were 0.249 and 0.366, respectively, suggesting that ~25% and ~37% of phenotypic variation on G- and G+, respectively, can be attributed to genetic variation. The values of σ2L were not significantly different between foods. In comparisons of the two species, σ2L was larger in D. simulans than in D. melanogaster; on G+, σ2L was significantly larger even after Bonferroni correction (P < 0.05).
| Source | df | Mean square | F |
|---|---|---|---|
| D. melanogaster G- | |||
| Line | 26 | 0.18774 | 1.97* |
| Error | 176 | 0.09544 | |
| D. melanogaster G+ | |||
| Line | 26 | 0.16039 | 3.15** |
| Error | 180 | 0.05085 | |
| D. simulans G- | |||
| Line | 26 | 0.23956 | 2.87*** |
| Error | 131 | 0.08350 | |
| D. simulans G+ | |||
| Line | 26 | 0.28226 | 4.15*** |
| Error | 125 | 0.06802 |
* P < 0.05, ** P < 0.01, *** P < 0.001.
Proper homeostatic regulation should keep the circulating glucose levels of an individual constant under any dietary sugar environment. If our assumptions are valid, we should observe a positive correlation in hemolymph glucose concentrations on glucose-free food (G-) and on glucose-rich food (G+). However, a significant correlation between them was not found in the D. melanogaster population from Hikone. In addition, glucose concentrations in 17 of 27 fly lines (~63%) were within a comparatively narrow range of 20–40 mg/dl. These results imply that there is an optimal concentration in the D. melanogaster population from Hikone, Japan. If so, the glucose concentration may be maintained by stabilizing selection in D. melanogaster. It is considered that stabilizing selection acts on most quantitative traits, and a number of empirical studies have reported evidence of stabilizing selection (e.g., Karn and Penrose, 1951; García-Dorado and González, 1996). It is well known that stabilizing selection does not directly act on the number of sternopeurals, which are bristles on the strernoepisternum of the adult thorax, in D. melanogaster; rather, the optimal number of bristles is the consequence of natural selection acting on larval viability (apparent stabilizing selection: Kearsey and Barnes, 1970). Recently, Matsushita and Nishimura (2020) reported that proper circulating glucose levels are essential for developmental homeostasis, suggesting that glucose levels associate with development. Therefore, like the sternopleural number in Drosophila, the glucose level may not be a direct target of selection. To find out whether stabilizing selection actually directly or indirectly acts on hemolymph glucose levels, it is necessary to examine fitness components such as viability and life-span.
In contrast to D. melanogaster, in D. simulans a significant positive correlation was found between the hemolymph glucose concentrations on G+ and G- when a single line with extremely high concentration on G+ was excluded, but the glucose level in each fly line on G- was not equal to that on G+ (Fig. 2B and Fig. 3). Regulation of the hemolymph glucose levels in Drosophila is mediated by complex neuronal and endocrine systems. For example, Drosophila insulin-like peptides promote a decrease of hemolymph glucose concentration, while adipokinetic hormone increases hemolymph glucose concentration (e.g., Rulifson et al., 2002; Kim and Rulifson, 2004; Lee and Park, 2004; Musselman et al., 2011; Gáliková et al., 2015; Oh et al., 2019). Glucose homeostasis is maintained by such antagonistic regulation. If antagonistic regulation does not work well because of functional deficiency in the system, it is possible that hemolymph glucose concentrations on G- become higher than on G+, or vice versa. The difference in hemolymph glucose concentration between the foods in D. simulans may be caused by loss of function in the antagonistic regulatory system. In Drosophila, nutrients such as polysaccharides, proteins and lipids are digested by many enzymes in the midgut. Digested molecules are absorbed though the midgut and transferred into the hemolymph. When sugars (monosaccharides and disaccharides) in the diet are poor, proteins and lipids are utilized as an energy source. In a sugar-rich environment, expression of the genes that are necessary to obtain sugars is repressed. This is known as glucose repression. A well-known example is α-amylase genes in D. melanogaster (e.g., Hickey and Benkel, 1982; Yamazaki and Matsuo, 1984). Many Drosophila species show suppression of amylase gene expression, but the degree of glucose repression, which is represented by the ratio G-/G+, varies between species, and in most species is > 1 (Inomata et al., 1995). Suppression of the expression of these genes is considered to be adaptive because it prevents excessive energy consumption when the environment is sugar-rich. Yamazaki and Matsuo (1984) found a positive correlation between fitness and glucose repression. They concluded that genetic variation in the degree of glucose repression is an important source of adaptive evolution. The ratio of hemolymph glucose concentration on G- to that on G+ tends to be > 1 (Fig. 3). If the extent of glucose repression is associated with the ratio in hemolymph glucose levels, it is plausible that D. simulans is in the course of adaptation to the environment at Hikone, while D. melanogaster has already adapted to the environment.
What is the cause of the difference in hemolymph glucose levels between D. melanogaster and D. simulans populations at Hikone in Japan? One of the possible explanations is associated with the founder effect. Drosophila simulans invaded the Japanese main islands in the 1970s and rapidly spread across Japan (Watanabe and Kawanishi, 1976, 1978). Actually, among the populations in the main islands, genetic differentiation in D. simulans is lower than that in D. melanogaster (Watada et al., 1986b). Population bottlenecks can increase additive genetic variance (e.g., Carson, 1990; van Heerwaarden et al., 2008; Andersson et al., 2010). Thus, the larger σ2L in hemolymph glucose concentration of D. simulans is likely to be explained by the much more recent colonization history of Japanese D. simulans populations. After the bottleneck event, the number of alleles at quantitative loci associated with the regulation of glucose homeostasis may have been reduced in Japanese D. simulans populations. As a result, the antagonistic regulation might not have worked because of a deficit of some function(s) in the system. As for the much more recent invasion of D. simulans into Japan, Watada et al. (1986a) pointed out the possibility of less divergence in quantitative morphological traits between Japanese D. simulans populations. They investigated eight quantitative morphological traits in both sexes in D. melanogaster populations and sympatric populations of D. simulans in Japan. In D. simulans, only three morphological traits were different between the populations, while in D. melanogaster most traits were differentiated between the populations. They suggested that the very recent colonization by D. simulans in Japan is likely to result in different patterns of geographical divergence.
The other possible cause is differences in tolerance to environmental stresses. For example, Parkash et al. (2013) reported that desiccation tolerance was positively correlated with carbohydrate energy budget (carbohydrate stores) in D. melanogaster and D. simulans inhabiting Fagu in the western Himalayas. However, the carbohydrate stores were different between the two species. At present, we do not know whether hemolymph glucose concentrations are associated with carbohydrate stores.
In conclusion, we found the following: (1) hemolymph glucose concentrations were much higher in D. simulans than in D. melanogaster; (2) the concentrations were similar between glucose-free (G-) and glucose-rich (G+) foods in D. melanogaster, while those of D. simulans were higher on G- than on G+; (3) in D. simulans, a positive correlation was found in the glucose concentrations between the G- and G+ foods, but not in D. melanogaster; (4) there was no difference in genetic variation (σ2L) of the hemolymph glucose concentrations between the foods in either D. melanogaster or D. simulans, but σ2L in D. simulans was larger than that in D. melanogaster; and (5) genetic variation in terms of H2 was not a major component of the phenotypic variation in either species. Taken together, our present results suggest that stabilizing selection acts on hemolymph glucose levels in D. melanogaster, but not in D. simulans. In addition, the observed differences in the hemolymph glucose levels between the two species may have resulted from recent colonization history in D. simulans, and/or different tolerance to environmental stresses between the two species.
We collected flies from three localities in Hikone, Japan in October 2020 using eight traps. Drosophila melanogaster and D. simulans were collected from the same traps. In total, we obtained 89 female (123 male) flies of D. melanogaster and 233 female (146 male) flies of D. simulans. We randomly selected female flies and established ~80 isofemale lines in each species. The flies of isofemale lines were reared at 18 ℃ on standard cornmeal food, whose composition was 5% cornmeal (w/v), 10% glucose (w/v), 5% killed yeast (w/v), 0.6% agar (w/v), 0.4% propionic acid (v/v) and 0.1% nipagin (w/v) in distilled water.
Larval hemolymph samples and measurement of glucose concentrationWe used two test foods that differed only in the presence or absence of glucose. Glucose-rich food (G+) contained 10% glucose (w/v), 5% killed yeast (w/v), 0.6% agar (w/v) and 0.4% propionic acid (v/v) in distilled water. Glucose-free food (G-) contained 5% killed yeast (w/v), 0.6% agar (w/v) and 0.4% propionic acid (v/v) in distilled water. The composition of yeast used in this study (Asahi, Japan) was as follows: 54.7% protein, 3.9% fat, 0–1.4% sugar, 28.1% dietary fiber. We randomly selected 27 lines out of ~80 isofemale lines in both D. melanogaster and D. simulans and measured hemolymph glucose concentration. Flies from each line were put on either G+ or G- and allowed to lay eggs, and larvae were reared at 25 ℃ for the same period. Thus, except for their food, larvae were reared under the same environmental conditions. At third instar, larvae were collected from a vial of each food. The larval cuticle was torn using forceps and hemolymph was extracted from 10–20 third instar larvae. This procedure was repeated several times per line for each food (independent replicates; n: number of replicates). Fifty microliters of PBS was added to 2 μl of larval hemolymph extracted from a line reared on either G+ or G-. The hemolymph in PBS was separately put into each well of the same 96-well microplate. Next, 150 μl of LabAssay Glucose (FUJIFILM Wako) was added to each sample, and incubated at room temperature for 20 min. After incubation, the absorbance at 505 nm was measured using a 2300 EnSpire multilabel plate reader (PerkinElmer, Japan) according to the manufacturer’s instruction. A standard curve was generated using the manufacturer’s glucose standards. Glucose concentration of each sample was calculated from the standard curve. The number of replicates (n) was eight for D. melanogaster and six for D. simulans for each food.
Data analysesStatistical analyses were performed using JMP Pro ver. 16.0.0 (SAS Institute). Differences in the hemolymph glucose levels between foods (G- and G+) and between species were tested using the Wilcoxon rank sum test. Statistical significance for multiple comparisons was corrected using the Bonferroni method. The amount of phenotypic variation was estimated as the coefficient of variation (CV). The difference of the degree of response (G-/G+) between species was also tested using the Wilcoxon rank sum test. Hemolymph glucose concentrations were transformed to logarithm values for normalization and analyzed using analysis of variance (ANOVA). In two-way ANOVA, phenotypic variation of hemolymph glucose concentration (total variance) was divided into variance components of food (σ2F, fixed), among-line (σ2L, random), interaction of food and line (σ2FL, random), and within-line (σ2E, random). For each food, one-way ANOVA was performed and the variance components of among-line (σ2L, random) and within-line (σ2E, random) were estimated. Broad-sense heritability (H2), which represents the proportion of genetic variation to phenotypic variation, was then estimated, where H2 = σ2L/ (σ2L + σ2E). To find whether genetic variation (σ2L) was different between foods in each species or between species in each food, we performed an F-test. Statistical significance for multiple comparisons was corrected using the Bonferroni method.
Conflicts of interest: None declared.
Authors’ contributions: N. I. conceived and designed the study. N. I., M. M. and M. N. performed laboratory work, and N. I., M. M. and M. N. conducted the analyses. M. I. collected flies and established isofemale lines. N. I. wrote the manuscript and N. I. and M. I. revised it. All authors read and approved the final manuscript.
Funding: This study was supported by Fukuoka Women’s University KENKYU-SHOREI-KOFUKIN (B) for N. I.
Data availability: We quantified absorbance, and so we have absorbance values only.
We thank Dr. H. Tachida for valuable comments on the manuscript, and Mr. N. Stott for his proofreading of the English.
