Introduction

Since the findings of the unexpectedly high amount of protein polymorphism (allozymes) in various organisms in nature (eg Harris, 1966; Hubby and Lewontin, 1966; Lewontin and Hubby, 1966), the mechanisms explaining its maintenance have been extensively discussed. At the protein level, most observations appear to be consistent with the neutral hypothesis. As it has become easy to determine nucleotide sequences, the main interest has shifted to understanding maintenance mechanisms of DNA variation rather than those of protein polymorphism. As a result, the adaptive significance of allozyme polymorphism remains unclear.

One approach to evaluate adaptive significance of allozyme polymorphism is to understand the molecular basis of functional differences between the allelic classes. In this context, the most advanced study is Fast and Slow allozyme polymorphism of alcohol dehydrogenase (Adh) gene in Drosophila melanogaster. Association studies and experimental approaches in vivo clearly showed differences in catalytic efficiency as well as in noncoding regions between the two allelic classes (eg Aquadro et al, 1986; Laurie and Stam, 1988, 1994; Choudhary and Laurie, 1991; Laurie et al, 1991; Berry and Kreitman, 1993; Stam and Laurie, 1996).

In contrast to extensive progress of the studies on allozyme polymorphism from a single locus such as the Adh gene in Drosophila, examinations of the adaptive significance of isozymes, which are proteins produced by different loci, have been lacking. Since the level of activity and protein content of an individual (strain) is the sum over all the isozymes, most experimental approaches that have been used in single-gene analysis cannot be applied directly to the products of duplicated genes. Instead, we have made use of P-element-mediated transformation to enable us to dissect effects of individual isozymes produced by such genes.

The starch degradation of Drosophila is one of the most extensively studied processes in the evolutionary research. The α-amylase (AMY, EC 3.2.1.1) breaks down starch into glucose and maltose. D. melanogaster has two amylase (Amy) gene copies, composed of the Amy-proximal and Amy-distal genes (Gemmill et al, 1985; Levy et al, 1985). The AMY activity is repressed by its products, glucose and maltose (Doane, 1969; Hickey and Benkel, 1982; Inomata et al, 1995a) and induced by the substrate, starch (Inomata et al, 1995a). Variation in AMY activity and food-response ability (inducibility) has been found (Matsuo and Yamazaki, 1984, Yamazaki and Matsuo, 1984, Inomata et al, 1995a). Moreover, food-response ability positively correlates with fitness components (Matsuo and Yamazaki, 1984; Yamazaki and Matsuo, 1984), indicating that regulatory factors controlling Amy gene expression are important for adaptation.

In natural populations of D. melanogaster, there are six major and many minor AMY isozymes (Kikkawa, 1964; Doane, 1969; Dainou et al, 1987; Inomata et al, 1995b). There are considerable differences in the level of AMY activity among strains with different isozymes (Kikkawa, 1964; Doane, 1969). Hoorn and Scharloo (1978) showed differences in the Km (enzyme–substrate affinity, the Michaelis constant) and Vmax (maximal activity) between AMY1 and AMY4,6 strains, and suggested some of the activity differences were due to the catalytic efficiency of the enzymes. On the other hand, Hickey (1981) showed a difference in the amount of amylase protein between high- and low-activity AMY1 variants. Yamate and Yamazaki (1999) found a positive correlation between specific activity and the amount of mRNA among strains with different AMY isozymes. On the basis of these observations, the variation of specific activity between AMY isozymes is likely to be largely explained by the differences in the amount of enzymes. The previous studies did not control the genetic backgrounds of the strains with different AMY isozymes and did not examine each isozyme and gene copy separately. Therefore, it is possible that they could not properly evaluate effects of individual isozymes. Moreover, molecular dissection of the effects of activity variation between the AMY isozymes has not been done yet.

To clarify the causes of the AMY activity differences among fly strains with different AMY isozymes, we examined the effects of differences in the nucleotide sequence between two alleles, Amy1-distal and Amy6-distal, of one of the duplicated Amy genes. We show that nucleotide polymorphism in both cis-regulatory and coding regions affects AMY activity. We also assessed the effect of nucleotide polymorphism in cis-regulatory region on two fitness components, developmental time and productivity.

Materials and methods

Chimeric Amy genes

Chimeric Amy genes were constructed based on Amy1 and Amy6 alleles, which originated from the Amy-distal gene of the KO140 and 1420#1 strains of D. melanogaster, respectively (see Inomata et al, 1995b). The chimeric Amy gene consisted of three regions: 616 bp of 5′-flanking region (cis-regulatory region), 1482 bp of coding region and 260 bp of 3′-flanking region. Its structure is shown in Figure 1a. There were 16 nucleotide differences in the 5′-flanking region between the Amy1 and Amy6 alleles. In all, 21 nucleotide differences, resulting in five amino-acid changes, occurred in the coding region. First, each gene region was amplified by the PCR method using the primers and Amy1 or Amy6 gene as a template DNA. The PCR primers used were as follows; Notd−600, 5′-CGGGCGGCCGCTGTAGCGTGAGATTCCTAA-3′ and Sped+, 5′-GTCACTAGTGTGCCTCGCCCT-3′ for amplification of the 5′-flanking region, Sped−, 5′-CCGACTAGTGACTTGGCCAGAAACATG-3′ and Sma+1500, 5′-TCCCCCGGGACAACTGGGGCAAAGACC-3′ for amplification of the coding region, and Sma+1485, 5′-TCCCCCGGGTTACAACTTGGCGTTGAC-3′ and Eco+1730, 5′-CCGGAATTCATCACGCTGAATACCGAG-3′ for amplification of the 3′-flanking region. Each primer contains a restriction enzyme site that enables us to make any combinations of Amy1- or Amy6-originated regions in the 5′-flanking, coding and 3′-flanking regions. NotI, SpeI, SmaI and EcoRI restriction sites were introduced at nucleotide positions −616, +16, +1482 and +1742, respectively (the numbering scheme starts at +1 at the adenine of the first codon (Met)). Note that introduction of the SpeI site results in one amino-acid replacement (Ile to Leu) at the seventh amino-acid position. It is considered that this replacement has no effect on AMY activity, since it is located in the region coding for signal peptide (Boer and Hickey, 1986). The PCR fragments were subcloned into pGEM T-easy cloning vector (Promega), and then sequenced to check for PCR errors. Each region was combined with restriction enzyme sites, and three chimeric Amy genes, Amyc111, Amyc161 and Amyfc661, were constructed (Figure 1a). The Amyc111 gene contains cis-regulatory, coding and 3′-flanking regions originated from the Amy1 gene. Similarly, the Amyc161 gene contains coding region originated from the Amy6 gene and the other regions from the Amy1 gene. The Amyfc661 gene contains 3′-flanking region originated from the Amy1 gene and the other regions from the Amy6 gene.

Figure 1
figure 1

(a) Structure of the chimeric Amy genes. An adenine of the first codon (Met) was numbered as +1. Numbers, −616, +16, +1482 and +1742, indicate nucleotide position at which NotI, SpeI, SmaI and EcoRI restriction sites were introduced, respectively. Top: Horizontal lines indicate 5′- and 3′-flanking regions and black bar indicates coding region. White and black boxes represent the region originated from Amy1 and Amy6 genes, respectively. (b) Structure of two transformation constructs, pP[wfl:Amyc111, Amyfc661] and pP[wfl:Amyc111, Amyc161], used in this study. The chimeric Amy genes were tandemly inserted into the pP[wfl] vector. Gene A indicates the Amyc111 gene in both the Amyfc661 and Amyc161 strains. Gene B indicates the Amyfc661 and Amyc161 genes in the Amyfc661 and Amyc161 strains, respectively. w+ indicates mini-white gene in the vector. White and black boxes represent the region originated from Amy1 and Amy6 genes, respectively.

Full size image

Plasmid constructs

For P-element-mediated germline transformation of D. melanogaster, chimeric Amy genes were introduced into the pP[wFl] vector (Siegal and Hartl, 1996). It is well known that expression among transformants of the same construct differs according to its genomic position, the so-called position effect (Spradling and Rubin, 1983; Laurie-Ahlberg and Stam, 1987; Brennan and Dickinson, 1988; Wilson et al, 1990). However, we overcame the problem by introducing two chimeric Amy genes into the pP[wFl] vector (see Figure 1b) to simultaneously insert the two constructs into a particular genomic position. In this study, the Amyc111 gene was used as a genomic-position control gene, and the Amyc161 or Amyfc661 genes were used as test genes. Two transformation constructs, pP[wfl:Amyc111, Amyfc661] and pP[wfl:Amyc111, Amyc161], were generated as shown in Figure 1b.

Drosophila transformation and genetic manipulations

P-element-mediated transformation of D. melanogaster was carried out using the standard method (Rubin and Spradling, 1982; Spradling and Rubin, 1982) with helper plasmid pπwc (Nitasaka, personal communication) and an isogenic host strain w1118, which was full-sib mated for 14 generations. Hereafter, we denote this host strain as w-sib14. Three independent transformants with P[wfl:Amyc111, Amyc161] insertion and two independent transformants with P[wfl:Amyc111, Amyfc661] insertion were obtained. Since the Amy genes are located on the second chromosome and the host strain has AMY1,3 isozymes, the second chromosome of the transformants was finally replaced by the Amynull chromosome. The transformants containing the insertion on the third chromosome were used in this study. To increase the number of independent transformants containing insertion on the third chromosome, the insertion on X and second chromosome were mobilized to new location through genetic cross using the w; CyO/Sp; TM3, Ubx/Sb P[ry+ Δ2–3](99B) strain as source of transposase (P[ry+ Δ2–3](99B), Robertson et al, 1988). Following transformation and mobilization, all strains were crossed to the w-sib14; c Amynull; TM3, Sb, Ser/Dr strain to exchange genetic background. Finally, we obtained transformants containing insertion on the third chromosome with the isogenic genetic background (w-sib14 for the X and third chromosomes, and c Amynull for the second chromosome). This included eight and nine independent lines of the Amyc111,fc661 and Amyc111,c161 strains, respectively. Hereafter, we call the former and latter the Amyfc661 and Amyc161 strains, respectively.

Food media

Two different food media, glucose and starch, were used in this study. Glucose medium contains glucose as a carbohydrate source, and starch medium contains starch as a carbohydrate source. The components of food media were 5% ebios (killed yeast) (w/v), 0.6% agar (w/v), 0.4% propionic acid (v/v) and 10% specific carbohydrate source, glucose or starch. Cornmeal medium was used to maintain stocks. Its components were 7% cornmeal (w/v), 10% glucose (w/v), 5% ebios (w/v), 0.6% agar (w/v) and 0.4% propionic acid (w/v).

Sample collection for AMY activity assay

Five male and five female adult flies were transferred to a new vial containing each test food medium, and allowed to lay eggs at 25°C. After 3 days, adults were collected from the vials, and then frozen at −80°C. After three additional days, three third-instar larvae, chosen at random, were collected from the vials without distinguishing sexes, washed with distilled water, and then kept at −80°C. Three replicates of larvae and adults were prepared per line for each test food medium.

Measurement of AMY activity

In this study, AMY activity was represented by the intensity of AMY isozyme bands separated on acrylamide gels. Three larvae and three adult flies were sampled without distinguishing the sexes. They were homogenized with 100 μl of a buffer (pH 8.9) (0.1 M Tris-borate, 5 mM MgCl2 and 10% sucrose (w/v)) by sonication. The homogenates were centrifuged at 10 000 rpm for 5 min and the supernatants were used for assay. To equalize samples applied to the gels, the protein content of each sample was quantified using the commercially available kit, BCA protein assay reagent (Pierce). The procedure essentially followed the manufacturer's protocol. In total, 1 ml of the BCA working reagent made from reagent A (sodium carbonate, sodium bicarbonate, BCA detection reagent and sodium tartrate in 0.1 N sodium hydroxide) and reagent B (4% copper sulfate solution) in the ratio of 50:1 was added to 25 μl distilled water and 25 μl of the supernatant. After incubation for 30 min at 37°C, the absorbance was measured at 562 nm using spectrophotometer.

The samples with equal protein content (0.52 mg) were applied to the polyacrylamide gels (5% acrylamide (w/v), 0.2% bis-acrylamide (w/v), 20 mM CaCl2 and 0.1 M Tris-borate) in a 0.1 M Tris-borate (pH 8.9) buffer. After running for 3 h at 4°C under 300 V constant condition, the gel was kept in starch solution (2% soluble starch (w/v), 0.1 M Tris-HCl (pH 7.4) and 20 mM CaCl2) for 30 min at 37°C. The gels were briefly washed with water, and incubated in solution without starch (0.1 M Tris-HCl (pH 7.4) and 20 mM CaCl2) for 30 min at 37°C. Subsequently, they were briefly washed with water, and stained in I2–KI solution. Gel images were captured into computer by a scanner, and then the band intensity of AMY1 and AMY6 isozymes on a gel was measured, as an index of AMY isozyme activity, using NIH image software (http://rsb.info.nih.gov/nih-image/).

Measurement of the fitness components

Five virgin females and five males from the cornmeal food vials were crossed on starch food. Three virgin females and three males from the cornmeal food were crossed on glucose food. When the parental flies died, they were replaced by flies of the same condition. After 3 days, the parents were discarded. The developmental time index was obtained by calculating the average number of days of eclosion of F1 progeny since the time of crosses. Productivity was calculated as the ratio of the number of progeny to parent input flies. Counting for developmental time and productivity was carried out from the 9th day to the 17th day for both food media.

Data analysis

Statistical analysis was carried out using the StatView software version 4.5. Three replicates of AMY activity and fitness components for each line were used in analyses.

The following two comparisons were performed by two-way analysis of variance (ANOVA). First, to evaluate effect of polymorphism in both 5′-flanking (cis-regulatory) and coding regions, activity of the AMY6 isozyme produced by the chimeric Amyfc661 gene was compared with that of AMY1 isozyme produced by the chimeric Amyc111 gene in the eight Amyfc661 strains. Second, to evaluate effect of polymorphism in coding region, activity of the AMY6 isozyme produced by the chimeric Amyc161 gene was compared with that of AMY1 isozyme produced by the chimeric Amyc111 gene in the nine Amyc161 strains. In these comparisons, the position effect of transgenes is canceled out or minimized, since the two genes under comparison have the same position in the genome. The ANOVA model was as follows, Yijk=u+Ai+Lj+(A*L)ij+eijk, where Y is AMY activity, u was the overall mean, Ai is the ith Amy allele effect (either Amy1 or Amy6), Lj is the jth independent transformant line effect in the strain (j=1–8 or 1–9), (A*L)ij is their interaction effect and eijk is the error term. All effects were considered random except for the allele effect. The sum of the AMY1 and AMY6 activity was regarded as total AMY activity. The difference in the total AMY activity between the Amyfc661 and Amyc161 strains should reflect effect of polymorphism in the 5′-flanking (cis-regulatory) region between the chimeric Amyfc661 and Amyc161 genes. The total AMY activity was analyzed by the nested ANOVA. The model of the nested ANOVA was as follows, Yijk=u+Si+Bij+eijk, where Y is total AMY activity, u is the overall mean, Si is the ith transformant strain effect (either Amyfc661 or Amyc161 strain), Bij was the random contribution for the jth independent transformant line (j=1–8 or 1–9) within the ith transformant strain and eijk is the error term. The strain effect was considered as the fixed effect. As in the case of the total AMY activity, the difference in fitness between the Amyfc661 and Amyc161 strains should reflect effect of polymorphism in the 5′-flanking (cis-regulatory) region between the chimeric Amyfc661 and Amyc161 genes. Fitness components data were analyzed by the nested ANOVA. The model of the nested ANOVA was as follows, Yijk=u+Si+Bij+eijk, where Y is fitness component, u is the overall mean, Si is the ith transformant strain effect (either Amyfc661 or Amyc161 strain), Bij is the random contribution for the jth independent transformant line (j=1–8 or 1–9) within the ith transformant strain and eijk is the error term. The strain effect was considered to be a fixed effect.

Results

Effect of polymorphism in both the 5′-flanking and coding regions on the AMY activity

To assess the effect of polymorphism on the AMY activity in both 5′-flanking and coding regions, we compared AMY1 activity with AMY6 activity in the eight Amyfc661 strains on the glucose and starch media at the larval and adult stages (Figure 2). Compared with the AMY1 activity, the AMY6 activity was slightly lower at the larval stage, but the difference was not significant on both media (F1,32=1.844, P=0.1840 and F1,32=0.284, P=0.5980 on the glucose and the starch media, respectively, see Figure 2a). At the adult stage, on the other hand, the AMY6 activity was significantly higher than the AMY1 activity on both media (F1,32=23.655, P<0.0001 and F1,32=26.505, P<0.0001 on the glucose and the starch media, respectively, see Figure 2b). Other effects were not significant at 5% level in either media at both stages. These results indicate that the effect of polymorphism on AMY activity in both the 5′-flanking and coding regions is different between the two stages.

Figure 2
figure 2

Effect of polymorphism on AMY activity of the 5′-flanking and coding regions in larvae (a) and adults (b) on glucose and starch media. White and black bars indicate average AMY1 and AMY6 activity, respectively, with standard error in the Amyfc661 strain. *** Indicates P<0.001. AMY activity was represented by the band intensity of the AMY isozyme separated on acrylamide gels.

Full size image

Effect of polymorphism in the coding region on the AMY activity

To evaluate the effect of polymorphism in the coding region on the AMY activity, we compared the AMY6 activity with the AMY1 activity in the nine Amyc161 strains on the glucose and starch media at the larval and adult stages (Figure 3). In both media, the AMY6 activity was significantly lower than the AMY1 activity at the larval stage (F1,35=77.684, P<0.0001 and F1,36=54.461, P<0.0001 on the glucose and the starch media, respectively, see Figure 3a), while the difference was not significant at the adult stage on both media (F1,36=0.209, P=0.6503 and F1,36=0.075, P=0.7851 on the glucose and the starch media, respectively, see Figure 3b). In larvae, the line effect and interaction effect between line and Amy allele were also significant in both media (line effect: F8,35=17.484, P<0.0001 and F8,36=7.912, P<0.0001 on the glucose and the starch media, respectively, and interaction effect: F8,35=5.455, P=0.0002 and F8,36=3.446, P=0.0048 on the glucose and the starch media, respectively). In adults, the line effect was significant in glucose medium (F8,36=2.231, P=0.0479). Other effects were not significant at the 5% level. These results indicate that the effect of the 21 nucleotide differences in the coding region is different between the two stages, even though the same isozymes are encoded.

Figure 3
figure 3

Effect of polymorphism on AMY activity of the coding region in larvae (a) and adults (b) on glucose and starch media. White and black bars indicate average AMY1 and AMY6 activity, respectively, with standard error in the Amyc161 strain. *** Indicates P<0.001. AMY activity was represented by the band intensity of the AMY isozyme separated on acrylamide gels.

Full size image

Effect of polymorphism in the 5′-flanking region on AMY activity

To evaluate the effect of polymorphism in the 5′-flanking (cis-regulatory) region on the AMY activity, we compared the total AMY activity, which is the sum of AMY1 and AMY6 activity, between the Amyfc661 and Amyc161 strains on the glucose and starch media at the larval and adult stages (Figure 4). At the larval stage, total AMY activity in the Amyfc661 strain was significantly higher than that in the Amyc161 strain on the starch medium (Fs1,15=9.590, P<0.01, see Figure 4a). At the adult stage, total AMY activity in the Amyfc661 strain was significantly higher than that in the Amyc161 strain on both the glucose and starch media (Fs1,15=82.983, P<0.001 and Fs1,15=67.863, P<0.001 on the glucose and the starch media, respectively, see Figure 4b). Other effects were not significant at the 5% level. These results indicate that 16 nucleotide differences in the cis-regulatory region have effect on the AMY activity between the two allelic classes.

Figure 4
figure 4

Effect of polymorphism on AMY activity of the 5′-flanking region in larvae (a) and adults (b) on glucose and starch media. White and black bars indicate average AMY activity in the Amyc161 and Amyfc661 strains, respectively, with standard error. ** and *** indicate P<0.01 and P<0.001, respectively. AMY activity was represented by the band intensity of the AMY isozymes separated on acrylamide gels.

Full size image

Effect of polymorphism in cis-regulatory region on fitness components

Significant effect of polymorphism in the cis-regulatory region on AMY activity was found. Therefore, if the effect is sufficiently large, it may be possible to detect the effect on fitness. To test whether the difference in cis-regulatory elements causes the difference in fitness, we measured two fitness components, developmental time and productivity, on the glucose and starch media in the Amyfc661 and Amyc161 strains. Average developmental time was 13.21 and 13.13 days on glucose media in the Amyfc661 and Amyc161 strains, respectively. On the starch media, average developmental time was 12.89 and 12.72 days in the Amyfc661 and Amyc161 strains, respectively (see Figure 5a). No significant effect on developmental time was found between the Amyfc661 and Amyc161 strains on both media at 5% level (Fs1,15=0.155 and Fs1,15=0.206 on the glucose and the starch media, respectively). Average productivity was 4.93 and 4.16 individuals on the glucose media in the Amyc161 and Amyfc661 strains, respectively. On the starch media, average productivity was 2.16 and 5.07 individuals in the Amyc161 and Amyfc661 strains, respectively (see Figure 5b). We found 2.35 times difference in productivity between the Amyfc661 and Amyc161 strains on the starch media at the face value. However, there was no significant effect on developmental time between the Amyfc661 and Amyc161 strains on both media at the 5% level (Fs1,15=0.188 and Fs1,15=0.240 on the glucose and the starch media, respectively). Significant effect was found between the lines for each strain (Fs15,34=2.2751, P<0.05 and Fs15,34=9.021, P<0.001 on the glucose and the starch media, respectively).

Figure 5
figure 5

Effect of polymorphism on two fitness components, developmental time (a) and productivity (b) on glucose and starch media. White and black bars indicate average fitness components in the Amyc161 and Amyfc661 strains, respectively, with standard error.

Full size image

Discussion

Effect of polymorphism on AMY activity

The effects of the coding region variants on AMY activity were different between the larvae and adults; similarly, the combined effect of the 5′-flanking and coding regions differed between larvae and adults. On the other hand, the effect of polymorphism in the 5′-flanking region alone did not differ between the two stages. In larvae, it is likely that higher AMY activity caused by the 5′-flanking sequence derived from the Amy6 allele (see Figure 4a), whereas lower AMY activity was caused by the coding sequence derived from the Amy6 allele (see Figure 3a). The combined result was, therefore, no significant difference in AMY activity (see Figure 2a). On the other hand, in adults the higher AMY6 activity caused by the combination of 5′-flanking and coding sequences (see Figure 2b) is most likely to be due to the 5′-flanking sequence (see Figure 4b), because no significant effect on AMY activity was found between the coding sequences of the two alleles (see Figure 3b). If this is true, our results suggest that effect of polymorphism in the coding region is different between larvae and adults, even though the same AMY isozymes are encoded. In other words, AMY isozymes show stage-specific activity. One plausible explanation for this observation is that the post-transcriptional processing of the mature protein, such as the stability of protein and translational efficiency, may differ between larvae and adults.

In the Amyc161 strain, AMY1 activity was significantly higher than AMY6 activity in larvae in both glucose and starch media (see Figure 3a), indicating that the coding sequence of the Amy1 allele results in higher AMY activity than that of the Amy6 allele. As both the genetic background and the cis-regulatory region is the same for the chimeric genes (Amyc111 and Amyc161), the amount of mRNA would be expected to be the same. Therefore, the difference in AMY activity seems to be due to the post-transcriptional differences caused by 21 nucleotide differences in the coding region of the two allelic classes. Five of these differences are nonsynonymous and may cause the difference in catalytic efficiency. Stability of mRNA, stability of protein and translational efficiency may also be affected by synonymous differences.

The 5′-flanking sequence derived from the Amy6 allele always causes higher AMY activity than that from the Amy1 allele. This result indicates that the Amy6-derived 5′-flanking sequence produces more mRNA, because the level of AMY activity correlates with the amount of mRNA (Hickey, 1981; Yamate and Yamazaki, 1999). There were only 16 nucleotide differences in the 5′-flanking region between the two alleles. No nucleotide differences were located in the core sequence of the putative cis-regulatory elements. A single-nucleotide difference may have a major effect on the level of the Amy gene expression, or alternatively, the difference in the mRNA level may be caused by nonrandom associations between polymorphism at different sites.

Since a highly inbred laboratory strain was used as a host, and assays were performed with only two allelic classes, our results might not be easily extended to other isozymes. Similarly, they might not mirror the effects of polymorphism in variable genetic backgrounds and/or under natural conditions. However, our results clearly indicate that nucleotide polymorphism in the 5′-flanking and coding regions appear to produce different AMY activity.

Effect of polymorphism on fitness

Yamazaki and Matsuo (1984) and Matsuo and Yamazaki (1984) showed that fitness was positively correlated with the inducibility (response ability to different food media) of AMY activity but not with activity itself. Therefore, they proposed that variation of all regulatory factors including transcriptional and post-transcriptional factors are important for adaptive evolution. The adaptive significance of changes in regulatory factors has been pointed out by several authors (eg King and Wilson, 1975). Recently, Wittkopp, Haerum and Clark (2004) indicated that cis-acting regulatory changes are more important than trans-acting ones for interspecific expression differences. In this study, we measured two fitness components, developmental time and productivity, to examine the effect of cis-acting regulatory changes. No significant effect was detected. However, on the starch media, the Amyfc661 strain showed 2.35 times more productivity than the Amyc161 strain. The difference of such magnitude may be biologically significant. Our design may have had insufficient statistical power to obtain a statistically significant result because of the small number of transformants examined.

Evolutionary implications

Some models of evolution of multigene families assume that the ancestral gene has two or more distinct but pleiotropically constrained functions (Piatigorsky and Wistow, 1991; Hughes, 1994), while other models are free of such an assumption (eg Ohno, 1970). Our results suggest that the relative AMY activity encoded by certain Amy alleles differs among the developmental stages, which indicates that they may have distinct subfunctions in Drosophila. This observation supports the model proposed by Piatigorsky and Wistow (1991) and Hughes (1994) who suggested that genes with multiple subfunctions can be important in the evolution of at least some multigene families.