19 Gene Duplication and Subsequent Differentiation of Esterases in Cactophilic Drosophila Species

Phytophagous insects are excellent model systems to study the genetic and ecological bases of adaptation and population differentiation because the host plant constitutes an immediate environmental factor that can affect the early stages of the life cycle (Matzkin, 2005; Matzkin et al., 2006). New host plant exploitation can result in genetic and biochemical adjustments to the new resource and to chemically distinct niches, which can include potentially toxic compounds, new mating environments, parasitoids, bacteria and fungi (Kircher, 1982; Fogleman & Abril, 1990; Via, 1990; Fogleman & Danielson, 2001). These adjustments are the result of a number of physiological changes, including those related to biochemical systems associated with adaptation to the new environment. The species of the Drosophila repleta group occupy different habitats, but their common feature is that they are phytophagous; that is, they lay eggs in rotting cacti cladodes. The developing larvae feed on the yeast that are part of the rotting process (Starmer & Gilbert, 1982; Pereira et al., 1983; Starmer et al., 1986), according to the cactus-Drosophila-yeast system; therefore, they are considered specialists. However, adults are generalists because they visit other food sources in their environment (Morais et al., 1994). This ecological specificity of cactophilic Drosophila directly influences species distribution, as they are always associated with the host cactus distribution (Tidon-Sklorz & Sene, 1995; Manfrin & Sene, 2006; Mateus and Sene, 2007). Drosophila has been used as a research model for more than a century, and the first report of gene duplication was described by Bridges for the Bar locus in D. melanogaster over 70 years ago (Bridges, 1936). Since that time, mainly after the advent of biochemical and molecular biology techniques, several other examples of duplicated genes have been presented, and pathways of evolution by gene duplication have been proposed (for example, Stephens, 1951; Nei, 1969). These pathways were thoroughly discussed in 1970 in Ohno’s book “Evolution by gene duplication” (Ohno, 1970). Subsequently, several other works have reviewed the mechanisms and roles of gene duplication in the evolutionary process (A. Wagner, 2002; Kondrashov et al., 2002; and Zhang, 2003). Currently, the genomes of twelve Drosophila species have been completely sequenced (Tweedie et al., 2009), but many aspects of the functional divergence of the products of a

stocks were multifemales, except for the NA-FS stock, which was isofemale. The laboratory stocks were obtained from Prof. Dr. Hermione E. M. C. Bicudo (Department of Biology, IBILCE/UNESP, São José do Rio Preto, Brazil), who brought them to Brazil from stocks of the Genetics Foundation (University of Texas, Austin, TX, US). The two line stocks were prepared from laboratory stocks through endogamic crosses. All laboratory and line stocks were maintained as mass cultures at a constant temperature of 20ºC ±1ºC in culture vials with standard banana medium. The origin of each stock is listed in Table 1.

Obtaining late third instar larvae and adult flies for electrophoresis
Late third instar larvae and adult flies were collected directly from the vials and immediately frozen at -20º C for further electrophoretic analyses. The larvae in that phase show yellowish spiraculum and maximum EST-4 activity. Female adult flies were collected at 5-10 days old and were used in electrophoresis for comparative analysis.

Obtaining hybrids
Mass crossings in both directions were performed in population boxes (16 cm 3 ), using 200 couples, between NA-FS and MU-S, NA-FS and MO, NA-FS and AR and MU-S and MO. After setting up a cross, the courtship behavior was observed for 10 minutes, as described by Markow (1981). The culture media were placed in Petri plates at the bottom of the boxes and were changed every three days. After every plate change, the plates were inspected to detect eggs. The plates were maintained at a constant temperature of 20ºC ±1ºC until late third instar hybrid larvae were observed. These larvae were obtained directly from the plates and frozen at -20ºC for further electrophoretic analyses.

Esterase detection
Esterase detection was performed using 10% polyacrylamide gel electrophoresis (PAGE), adapted by Ceron (1988) from Davis (1964) and Laemmli (1970). All samples were prepared in 25 µL of 0.1 M Tris-HCl (pH 8.8) buffer containing 10% glycerol, where 10 µL was used in the gels. After electrophoresis, all gels were soaked in 0.1 M phosphate buffer (pH 6.2) for 1 hour at 25ºC. After this period, the gels were stained in solution containing 100 mL of phosphate buffer, 10 ml of n-propanol and 120 mg of Fast Blue RR Salt, where 40 mg of αnaphthyl acetate and 30 mg of β-naphthyl acetate, previously dissolved in 2 ml of acetone, were added. After approximately 1 hour, the staining reactions were stopped in a solution of acetic acid:ethanol:water (1:2.5:6.5 by v:v:v). Because the esterases hydrolyze substrates differently, the bands in the gel stain differently: black when they hydrolyze α-naphthyl acetate, red when they hydrolyze β-naphthyl acetate and magenta when they hydrolyze both αand β-naphthyl acetates. Polyacrylamide gels were air dried at room temperature using gelatin and cellophane wound slab gels in an embroidering hoop (Ceron et al., 1992).

Determination of isoelectric point (I.P.)
The I.P. was determined in 10% PAGE containing 5% ampholyte solution (Sigma). The first ampholyte formed a pH gradient between 3.0 and 10.0 after 1 hour of constant 100 V prefocusing. This experiment was performed to verify the best gradient to determine the I.P. values of all enzymes in all species. After this verification, another ampholyte was used that formed a pH gradient between 6.0 and 8.0 after 1 hour of constant 100 V pre-focusing. In both cases, ampholyte solutions were added before gel polymerization. Samples of the six Drosophila species and of the I.P. marker (hemoglobin) were prepared in a 10% glycerol in water solution. Esterase isoelectric focusing was performed under constant 100 V conditions in the power supply for 3 hours. After focusing, the gels were soaked in buffer for 1 hour, followed by esterase staining for the same period, as described in section 2.5. Following esterase identification, the gels were stained for total protein with Coomassie Blue G250 overnight. The staining reaction was stopped, and the gels were dried as described in section 2.5. The I.P. was estimated by comparing the positions of EST-4 and EST-5 with the position of human hemoglobin (I.P. = 7.1) after focusing.

Molecular weight (MW) estimation
The MW estimation was performed using the method adapted by Mateus et al. (2009) from Hedrick and Smith (1968). The following standard MW proteins were used: myoglobin (17.8 kD), soybean trypsin inhibitor (24 kD), carbonic anhydrase (29 kD), ovalbumin (45 kD), human serum albumin (66 kD) and phosphorylase-b (97.4 kD). All graphics were constructed using Microcal Origin software, version 3.5 (Scientific and Technical Graphics in Windows -copyright 1991. Figure 1 shows the esterase patterns of larvae and adults (females) from six Drosophila species. For all species, EST-4 always migrated slower than EST-5. The D. navojoa stock was the only one that had more than one band for EST-4. EST-4 was more strongly stained than EST-5 in D. mulleri, D. aldrichi and D. wheeleri (Figure 2A and B). The opposite was observed for D. arizonae, with EST-5 more strongly stained than EST-4 ( Figure 2A). Differences in the staining intensity among EST-4 bands were also observed, with the D. mojavensis cluster species (D. mojavensis, D. arizonae and D. navojoa) showing fainter bands than the species of the D. mulleri cluster (D. mulleri, D. aldrichi and D. wheeleri).  Despite the observation of homozygotes for EST-4 in five out of six species analyzed, the quaternary structure for this enzyme as a dimer could be deduced from the presence of a heterodimer between EST-4 and EST-5 in D. mojavensis and D. arizonae (Figure 2A). This dimeric structure for EST-4 and EST-5 was confirmed by hybrid analyses. In Drosophila navojoa, in addition to the presence of the heterodimer, three phenotypes were observed in gels for EST-4 and EST-5 ( Figure 2B): homozygous for a slower band (EST-4 S and EST-5 S , respectively); homozygous for a faster band (EST-4 F and EST-5 F , respectively); and heterozygous, with a three-band pattern. These patterns reinforce the quaternary structure of these enzymes for this species. The same results were observed for EST-5 of D. mulleri (data not shown).  Table 2. Esterase activity patterns of EST-4 and EST-5 for the six Drosophila species analyzed in the presence of different inhibitors. PMSF = phenylmethylsulfonyl fluoride; Eserine = eserine sulfate; CuSO 4 = copper sulfate; IAC = iodoacetamide; E-64 = trans-epoxysuccinyl-Lleucyl-amido(4-guanidino) butane; pCMB = p-chloromercuribenzoate; HgCl 2 = mercuric chloride. ++ activity inhibited; ⊗ activity not affected.

Isoelectric point (I.P.) determination
The I.P. determination was performed in two phases. In the first phase, we verified that the best range for I.P. determination was 6.0 to 8.0. In the second phase, an ampholyte solution was used for this pH range. Table 3 shows that all esterases presented I.P. between 6.0 and 7.0. As expected, the I.P. values for EST-5 in both larvae and adults of the same species were equal, ranging from 6.47 (D. navojoa) to 6.64 (D. aldrichi). EST-4 showed a wider range of I.P. variation than EST-5, with D. mulleri and D. navojoa showing the highest and lowest I.P. values (6.88 and 6.37, respectively). EST-5 (larvae and adult) 6.53 EST-5 (larvae and adult) 6.47 Table 3. Isoelectric points for EST-4 and EST-5 of larvae and adults of the six analyzed Drosophila species, obtained through the comparison of esterase band mobility in gels with an I.P. marker (hemoglobin; I.P. = 7.1) in a pH range between 6.0 and 8.0.

MW determination
To determine the MW of both enzymes in all six Drosophila species analyzed, the technique described by Mateus et al. (2009) was applied using 6% to 12% PAGE and the same MW markers. The results presented there are part of this study. Therefore, in this study, we present the results that were not shown in Mateus et al. (2009) Figure 3. The plots for both esterases were parallel in all species, indicating that these enzymes have different charges and/or tridimensional structures but very similar molecular weights. From these graphs, the slope was obtained for each enzyme in each species. These values were used to estimate the MW in each case, using the equation Y = A + BX, where A is the intercept of the Y-axis (2.18766), and B is the slope (0.09452). The slopes and molecular weights are presented in Table 4.
The slope values for both enzymes in all species were similar. EST-5 had more variation, ranging from -10.05407 ± 0.29546 for D. navojoa to -11.03429 ± 0.30178 for D. mulleri. EST-4 was less variable, ranging from -10.08361 ± 0.33581 for D. wheeleri to -10.52607 ± 0.44878 for D. mulleri. The MWs, estimated from these slope values (     Table 3). The same was not observed for EST-5, as this enzyme has nearly the same value for both of these species.

Crosses between D. navojoa and D. mojavensis
Asymmetric isolation was also observed in the cross between D. navojoa and D. mojavensis.
No offspring were obtained in the direction of D. mojavensis females and D. navojoa males, despite the fact that courtship between couples and eggs on the plate were observed. The cross between D. navojoa females and D. mojavensis males was very fertile. The hybrid larvae from this cross were analyzed in 10% PAGE and showed the same threeband patterns observed for D. mulleri and D. mojavensis ( Figure 5). For EST-4, the slower band corresponded to the enzyme encoded by D. mojavensis, the faster band corresponded to the enzyme encoded by D. navojoa, and the intermediate band corresponded to a hybrid enzyme. The opposite was observed for EST-5: the slower band was from D. navojoa, the faster band was from D. mojavensis, and an intermediate band was a hybrid enzyme. Again, these results confirm the quaternary structure of both enzymes of these species. An interesting observation was the absence of EST-5 expression in two samples (samples 12 and 13; Figure 5).

Crosses between D. navojoa and D. arizonae
This cross was very fertile in both directions. However, larvae from the cross in the direction of D. navojoa females and D. arizonae males had very slow development and took much longer to achieve the late third instar stage; they also had a high mortality rate. The larvae analyzed in 10% PAGE from both cross directions presented the same three-band patterns for EST-5. For EST-4, as in both species of the cross, the enzymes had almost the same migration speed under these electrophoretic conditions. One thicker band was observed in the hybrid larvae, which must be the agglomeration of the three bands expected for this enzyme ( Figure 6).  enzymes in these species (Figure 7). Again, the EST-4 hybrid band migrated closer to the slower band from D. mulleri, which could be a consequence of the different I.P. values of these enzymes. The same was not observed for EST-5, as they show similar I.P. values for both species.

Discussion
Isozymes are very important in insects and have been used to understand biological problems in several fields of research, including population genetics and systematics, tissue organization, development, metamorphosis, gene regulation and protein synthesis and gene duplication (R. P. Wagner & Selander, 1974). The set of proteins known as esterases constitute one of the most heavily studied groups of isozymes. In the Drosophila mulleri complex, which is the subject of this study, esterases have been extensively studied in several species, including D. serido (Lapenta et al., 1995(Lapenta et al., , 1998, D. buzzatii (East, 1982;Barker, 1994;Lapenta et al., 1995Lapenta et al., , 1998, D. mojavensis (Zouros et al., 1982;Zouros & Van Delden, 1982;Pen et al., 1984Pen et al., , 1986aPen et al., , 1986bMateus et al., 2009), D. arizonae (Zouros et al., 1982;Ceron, 1988;Mateus et al., 2009), D. aldrichi (F. M. Johnson et al., 1968;Kambysellis et al., 1968) and D. mulleri (F. M. Johnson et al., 1968;Kambysellis et al., 1968;Richardson & Smouse, 1976;Ceron, 1988). Zouros et al. (1982) detected two esterases with different patterns of temporal and tissuespecific expression in Drosophila mojavensis and D. arizonae (formerly D. arizonensis). They detected a specific -esterase of the late third instar phase of larval development and in the carcass, named EST-4, in contrast to another -esterase, named EST-5, which is expressed during all developmental phases and is found predominantly in hemolymph and the fat body. They proposed that the most likely hypothesis is that both enzymes are products of a gene duplication that occurred prior to the speciation of the D. repleta group, and their patterns of tissue-specific and temporal expression diverged more recently. This hypothesis was suggested because the enzymes show interlocus heterodimers, different patterns of expression (Zouros et al., 1982) and 82% identity in their N-terminal amino acid sequences (Pen et al., 1986a;Pen et al., 1990). More recently, Robin et al. (2009) proposed that, in D. mojavensis, these enzymes are most likely encoded by two genes, Est-2a (EST-5) and Est-2c (EST-4), which are products of one gene duplication out of a total of eleven duplications that explain the evolution of the catalytic -esterase cluster in the Drosophila genus (five in the Sophophora and 6 in the Drosophila subgenus).
Our results reinforce the hypothesis proposed by Zouros et al. (1982), extending the knowledge about these enzymes as products of a gene duplication to other D. mulleri complex species. All six analyzed species show distinct temporal expression patterns for  showing activity only at the end of the third instar larval stage (Figure 1). The inhibition experiments (Table 2) showed that EST-4 has the same pattern for all six species: it is inhibited by PMSF and not affected by malathion. The opposite was observed for EST-5 in all six species: it was inhibited by malathion and not affected by PMSF. The other inhibitors tested (eserine sulfate, copper sulfate, iodoacetamide and E-64) had no effect on the activity of either enzyme. Moreover, the presence of homozygotes and heterozygotes for EST-5 independent of the EST-4 genotype in D. navojoa ( Figure 2) and D. mulleri (data not shown) support the idea of an independent origin of these enzymes from two distinct loci. Despite these differences, these enzymes display similar features, such as structural similarities (Pen et al., 1986a;Pen et al. 1990) that allow the formation of dimers in D. mojavensis, D. arizonae and D. navojoa (Figure 2). The gene duplication process is considered one of the most important mechanisms of the generation of new genes and functions during the evolutionary process. Jeffreys & Harris (1982) suggested gene duplication mechanisms that could happen to genes during evolution. Among the mechanisms presented, the most likely mechanism that could have generated the EST-4 and EST-5 loci is the same mechanism that might have generated the globin family, that is, in tandem gene duplication by pairing errors during meiosis that cause unequal crossing-over because of the presence of short repeat sequences located in the 3' and 5' ends of the unduplicated ancestral gene. The EST-5 gene in D. pseudoobscura is a good example of gene duplication with later divergence (Brady & Richmond, 1992). The EST-5 enzyme is encoded by the Est-5B gene, which is expressed during the life cycles of all insects and is linked to two other genes, Est-5A and Est-5C, on the X chromosome (Brady et al., 1990). In D. melanogaster, the homologous gene is Est-6, which codes for the enzyme EST-6 during the insect's life cycle and has only one grouped gene, Est-P (Collet et al., 1990). Both Est-5A of D. pseudoobscura and Est-P of D. melanogaster are expressed only at the third instar larval stage, producing only one transcript. On the other hand, Est-5C of D. pseudoobscura is not expressed in any developmental phase (Brady et al., 1990). According to Brady & Richmond (1992), who compared the DNA sequences of coding and flanking regions of all three D. pseudoobscura and two D. melanogaster genes, only two genes, which are already products of a gene duplication, were present before these two species diverged. These two ancestral genes were probably Est-5A and Est-5B in the first species and Est-6 and Est-P in the second species. A second duplication occurred later in D. pseudoobscura, giving rise to the Est-5C gene. However, the findings of Robin et al. (2009) contrast with the evolutionary model proposed by Brady & Richmond (1992); in their analyses, the Est-5A/Est-5B duplication (which they call Est6/7) occurred after the melanogaster/obscura group divergence, whereas Brady & Richmond (1992) place this duplication prior to the divergence.
In our case, Zouros et al. (1982) proposed that the genes coding for EST-4 and EST-5 (Est-2c and Est-2a, respectively, according to Robin et al., 2009) were also products of a duplication event prior to the divergence of the species that belong to the D. repleta group and that the EST-4 gene was later inactivated in some species of this group, including D. tira, D. hydei and D. eohydei. Moreover, the lower activity of EST-4 in D. mulleri, D. aldrichi, D. repleta and D. peninsularis could indicate this EST-4 inactivation process. However, our results showed that D. mulleri, D. aldrichi and D. wheeleri had high EST-4 activities compared to the other species (Figure 2A). This difference in the level of activity of EST-4 for the same species in these studies could be the result of differences in the origins of the lines used in each study. Therefore, the populations of D. mulleri and D. aldrichi that were analyzed by Zouros et al. (1982) could have a certain degree of EST-4 inactivation that was not observed in the present study.
The enzymes analyzed in this study had biochemical differences compared to other esterases of other Drosophila species. For example, the I.P. values for EST-4 and EST-5 for the six Drosophila species analyzed were between 6.0 and 7.0 (Table 3). These values were different from those of D. melanogaster obtained by Healy et al. (1991), as only 2 out of 15 esterases had I.P. values close to the values obtained in this study (between 6.0 and 7.0). All others showed values below 6.0, with the majority of values between 4.0 and 5.0. Regarding the MWs of these enzymes, our results are in agreement with previous studies that estimated this parameter. EST-4 had MW values between 83 and 89 kD, and EST-5 had MW values between 83 and 94 kD (Table 4), which are very close to the MWs obtained by Pen et al. (1984), which were between 85 and 95 kD for a variant of the EST-4 (with altered specificity to -naphthyl acetate) using gel filtration chromatography. Pen et al. (1984) also used denaturing gel electrophoresis (SDS-PAGE) and obtained the MWs of the subunits of EST-4 as 62-64 kD. In another study, Pen et al. (1986a) determined the MWs for the subunits of EST-5 as 64-66 kD. Regardless of the method used and the different results obtained (for the entire protein or for subunits), EST-4 had a smaller MW than EST-5, as observed in this study.
The interspecies crosses performed in this study had results that were in accordance with the known phylogenetic relationships among the species analyzed. This information is based on the morphological work of Throckmorton (1982) and Vilela (1983), the cytological work of Wasserman (1982Wasserman ( , 1992 for reviews), several allozyme studies (Zouros, 1973;Richardson et al., 1975;Richardson and Smouse, 1976;Richardson et al., 1977;Heed et al., 1990), molecular studies (Sullivan et al., 1990;Russo et al., 1995;Spicer, 1995Spicer, , 1996 and an analysis using multiple sources of characters (Durando et al., 2000). The crosses between D. mulleri and D. mojavensis showed the same results of those of Patterson & Crow (1940) and Bicudo (1982), with offspring obtained only in the direction of D. mulleri females and D. mojavensis males. For D. navojoa crossed with D. mojavensis, an F1 was produced only in the direction of D. navojoa females and D. mojavensis males. In this case, Ruiz et al. (1990) observed descendants in both directions but a very low percentage of offspring, depending on the geographic lineage used, in the direction in which we detected isolation. In crosses between D. navojoa and D. arizonae, both directions were fertile, which was also found by Ruiz et al. (1990). Finally, in crosses between D. navojoa and D. mulleri, we detected descendants in both directions, in contrast to the results of Bicudo (1982), who found fertility only in the direction of D. mulleri females and D. navojoa males.
In all of these crosses, the phenotypic observations of the esterase patterns from late third instar hybrid larvae produced three bands for both 5,6 and 7), except for larvae from the cross between D. navojoa and D. arizonae, which produced a thicker band because the parental bands have almost the same migration pattern under the electrophoretic conditions used in this study. These results indicate that in all six Drosophila species, EST-4 and EST-5 have dimeric quaternary structures. Another important observation from some of these crosses was the presence of hybrid larvae with no EST-5 activity (D. navojoa x D. mojavensis - Figure 5; D. navojoa x D. arizonae -data not shown).
These results indicate that some hybrid larvae had problems with the regulation of Est-2a gene expression, which most likely codes for the EST-5 enzyme, without affecting the expression of its homologous gene, Est-2c, which most likely codes for the EST-4 enzyme (Robin et al., 2009). These results reinforce the idea that these two loci are independent. The possible role of EST-4 in these Drosophila species remains an open question. According to Holmes & Masters (1967, as cited in Oakeshott et al., 1993, esterases can be classified into four types through their specific inhibition patterns. Carboxylesterases are esterases that are inhibited only by organophosphates, such as paraoxon, fenitrooxon and DFP (diisopropylfluorophosphate). Cholinesterases are inhibited by organophosphates and carbamates, such as eserine sulfate. Arylesterases are inhibited only by sulfhydrylic agents, such as p-chloromercuribenzoate (pCMB). Acetylesterases are not inhibited by any of these agents. Inhibition of EST-5 only by malathion, an organophosphate, suggests that this enzyme belongs to the class of carboxylesterases. Inhibition of EST-4 by PMSF and the absence of inhibition in the presence of all other inhibitors tested suggest that this enzyme probably belongs to the class of acetylesterases. According to Augustinsson (1968), esterases are closely related to the class of serineproteases, forming a multigenic family of serine-hydrolases. The main features that support this hypothesis are the three consensus amino acid residues that are present in the active site of esterases and serine-proteases, including an invariant serine, enzymatic inactivation by DFP, which binds irreversibly to the serine r e s i d u e o f b o t h e n z y m e s , i n h i b i t i o n b y organophosphates and carbamates and the superposition of substrate preference (Augusteyn et al., 1969;Krisch, 1971;Dayhoff et al., 1972;Heymann, 1980;Previero et al., 1983; as cited in Myers et al., 1988). However, Myers et al. (1988) showed that some esterases cannot be included in this multigenic family because they do not have the same amino acid residues in the charge exchange system of the enzyme active site. The absence of EST-4 and Est-5 inhibition by copper sulfate and iodoacetamide, combined with data for E-64, which is a diagnostic inhibitor of cysteine-proteases, indicate that neither enzyme has an essential cysteine residue in its active site. On the other hand, the inhibition of EST-4 by PMSF, which is a diagnostic compound for serine-proteases and other enzymes with a serine residue in the active site, and of EST-5 only by malathion indicated that both enzymes have an important serine residue in the active site, suggesting that they belong to the class of serine-hydrolases. As these enzymes display high esterase activity, we can postulate that they are serine-esterases (Holmes & Masters, 1967). The multigenic family of serine-esterases includes several enzymes wit h a w i d e r a n g e o f functions, including cholinesterases, lipases, lysophospholipases, cholesterol-esterases, non-specific carboxylesterases and juvenile hormone esterases (Ryger et al., 1989;Doctor et al., 1990;Shimada et al., 1990; as cited in Myers et al., 1993). Therefore, EST-4 and EST-5 probably belong to this multigenic family, with EST-4 as an acetylesterase (E.C. 3.1.1.6) and EST-5 as a non-specific carboxylesterase (E.C. 3.1.1.1).

Gene Duplication and Subsequent Differentiation of Esterases in Cactophilic Drosophila Species 367
To establish the possible role of EST-4, the following information must be considered. Healy et al. (1991) observed that all D. melanogaster acetylesterases are inhibited by OTFP (3octylthio-1,1,1-trifluoropropan-2-one), which is a powerful inhibitor of juvenile hormone esterase activity in Lepidoptera, suggesting that all acetylesterases from this species have similar properties as juvenile hormone esterase. Moreover, East (1982) proposed that esterase-J from D. buzzatii, which is supposedly the enzyme from this species that corresponds to EST-4 in this study, is a juvenile hormone esterase, acting together with EST-1 in the larval phase to control the levels of this hormone. In the adult phase, only EST-1 would be responsible for this control. On the other hand, EST-2 could be the enzyme responsible for digestive and detoxification processes and ester absorption in adults. EST-4 has a very tissue-specific and temporal pattern of expression, which indicates that there is a specific regulatory system that controls its expression at a specific tissue (carcass) and period of time (at the end of the larval phase, when all of the processes for pupation have been initiated). Therefore, as an acetylesterase with a very specific temporal expression pattern, EST-4 could be involved in these transformation processes, acting as an auxiliary enzyme for the degradation of juvenile hormone esterase. The degradation of this hormone in this phase allows the liberation of prothoracicotropic hormone by the brain, which stimulates ecdysone production by the prothoracic gland, initiating metamorphosis (Coundron et al., 1981). However, analyzing the EST-4 inhibition data alone could lead to the hypothesis that this enzyme could be a serine-protease that also has esterase activity and is involved in a proteolytic activity during the larva-pupae conversion process; it is likely to be involved in this process. Regarding EST-5, considering the fact that it is expressed during the entire life cycle of the insect and is found mainly in the hemolymph and fat body, it is a non-specific carboxylesterase that is probably involved in digestive processes.

Conclusions
This study contributes to a better understanding of the differentiation of two enzymes that are products of a gene duplication in six cactophilic Drosophila species. We present additional evidence to support the gene duplication event that gave rise to the genes responsible for the EST-4 and EST-5 enzymes, which are the main -esterases found in several species of the D. mulleri complex of the D. repleta group. We also contribute to the elucidation of the possible physiological roles of these esterases in this group. Further steps in this investigation will be to determine specific biochemical parameters of both enzymes after purification. We are also interested in identifying the changes that occur in the regulatory system of gene expression that lead to differentiation in the patterns of tissuespecific and temporal expression of these enzymes; that is, understanding what triggers EST-4 expression only in the late third instar larvae and at the larval carcass. We are also interested in determining the intra-and/or extracellular processes in which these enzymes are involved and their interacting molecules. Thus, we will be able to complement this initial step with an increased understanding of the differentiation of these two genes that result from a gene duplication event.

Acknowledgments
We would like to thank CNPq for funding Rogério P. Mateus (Master's degree fellowship). We would also like to thank CAPES, FINEP and FUNDUNESP for supporting this work,