Catalytic function of Drosophila melanogaster glutathione S-transferase DmGSTS1-1 (GST-2) in conjugation of lipid peroxidation end products


P. Zimniak, VA John McClellan Medical Center, Medical Research (151/LR), 4300 West 7th Street, Little Rock, AR 72205, USA. Fax: + 1 501 257 4822, Tel.: + 1 501 257 4843, E-mail:


Drosophila melanogaster glutathione S-transferase DmGSTS1-1 (earlier designated as GST-2) is related to sigma class GSTs and was previously described as an indirect flight muscle-associated protein with no known catalytic properties. We now report that DmGSTS1-1 isolated from Drosophila or expressed in Escherichia coli is essentially inactive toward the commonly used synthetic substrate 1-chloro-2,4-dinitrobenzene (CDNB), but has relatively high glutathione-conjugating activity for 4-hydroxynonenal (4-HNE), an electrophilic aldehyde derived from lipid peroxidation. 4-HNE is thought to have signaling functions and, at higher concentrations, has been shown to be cytotoxic and involved in the etiology of various degenerative diseases. Drosophila strains carrying P-element insertions in the GstS1 gene have a reduced capacity for glutathione conjugation of 4-HNE. In flies with both, one, or none of the GstS1 alleles disrupted by P-element insertion, there is a linear correlation between DmGSTS1-1 protein content and 4-HNE-conjugating activity. This correlation indicates that in adult Drosophila 70 ± 6% of the capacity to conjugate 4-HNE is attributable to DmGSTS1-1. The high abundance of DmGSTS1-1 (approximately 2% of the soluble protein in adult flies) and its previously reported localization in tissues that are either highly aerobic (indirect flight muscle) or especially sensitive to oxidative damage (neuronal tissue) suggest that the enzyme may have a protective role against deleterious effects of oxidative stress. Such function in insects would be analogous to that carried out in mammals by specialized alpha class glutathione S-transferases (e.g. GSTA4-4). The independent emergence of 4-HNE-conjugating activity in more than one branch of the glutathione S-transferase superfamily suggests that 4-HNE catabolism may be essential for aerobic life.




glutathione S-transferase






Berkeley Drosophila Genome Project.

Glutathione S-transferases (GSTs) are ubiquitous and versatile proteins capable of a range of enzymatic and nonenzymatic functions [1]. Work on insect GSTs was initially motivated by the possible involvement of these enzymes in insecticide resistance (reviewed in [2]). Early studies revealed the presence of a group of insect GSTs originally named GST-1. According to a recent nomenclature proposal [3] to which the present paper adheres, insect GST-1 enzymes are designated as the delta class of GSTs. In a series of elegant studies [4–7], Tu and colleagues characterized the delta-class enzymes of Drosophila melanogaster and demonstrated that some members of this heterogeneous group of eight proteins have high glutathione conjugating activity for CDNB and also exhibit glutathione peroxidase activity with cumene hydroperoxide. Two additional members of the delta cluster have been recently identified [3]. Efforts to elucidate the function of all Drosophila genes through systematic P-element mutagenesis [8,9] and, more recently, the sequencing of the Drosophila genome [10] led to the discovery of additional GSTs, including a microsomal-like GST [8] and a GST that may belong to the ancestral-like class theta [11]. Preceding these systematic efforts was the serendipitous discovery of a GST that differed significantly from the delta class (then termed GST-1), and which was therefore named GST-2 [12]. This enzyme, which is represented in the Drosophila genome by a single locus, is a member of the sigma class, and its recommended name is DmGSTS1-1 [3]. As a GST, this Drosophila enzyme is unusual for several reasons: (a) it has very little activity for typical GST substrates, including CDNB (this communication); (b) it carries a N-terminal extension with a repeated hydrophobic/acidic motif [12]; (c) the majority of it is attached, probably through the N-terminal extension, to the thin filament of the indirect flight muscle [13]; (d) during development and in adults, it is found in the central nervous system, and (e) it is also localized to larval imaginal disks (J. A. Coronella & B. J. Cochrane, unpublished results). Hence, DmGSTS1-1 is primarily found to tissues normally not associated with detoxification. Because of the predominant association with the indirect flight muscle, it was proposed that the role of DmGSTS1-1 may be structural rather than catalytic, and that the protein could be part of a stretch sensor that enables the indirect flight muscle to undergo a series of rapid, self-sustaining contractions [13]. This hypothesis is consistent with known noncatalytic functions of GSTs and GST-like proteins, e.g. the binding of nonpolar ligands [1] or, in particular relevance to sigma-class GSTs, as lens crystallin in cephalopods [14]. However, in the present communication, we report that DmGSTS1-1 has significant activity in the GSH conjugation of the lipid peroxidation product 4-HNE. Indeed, the enzyme plays a central role in the 4-HNE metabolism of Drosophila as it accounts for more than two-thirds of the insect’s capacity to conjugate 4-HNE. This indicates that DmGSTS1-1 may have alternative and/or additional functions in detoxification, protection against oxidative injury, and perhaps in signaling processes.

Experimental procedures

Drosophila strains

The following strains, containing the P{w+mC} element were obtained from the Bloomington Indiana Stock Center: P874 (or y1w67c23; P{w+mC}l(2)k09303k09303/CyO) and P803 (or y1w67c23; P{w+mC}l(2)k08805k08805/CyO). Each strain is heterozygous for a mutant GstS1 allele; the mutant chromosomes are termed 874 or 803 for simplicity and both contain the P{w+mC} element inserted at the same position in the GstS1 gene (see Fig. 1). To obtain flies carrying both of these mutant GstS1 alleles, strains P874 and P803 were crossed and noncurly winged male flies (missing the CyO balancer chromosome) were selected for enzyme analysis. The GS2160 line carrying the P{w+wC} element [15] and the appropriate control stock (y w or Df(1)yw) were obtained from T. Aigaki (Tokyo Metropolitan University, Japan). GS2160 is homozygous viable but in our hands male-sterile at 25 °C. The exact sites of insertion of the P-elements were obtained from published reports [15] or by consultation of the Berkeley Drosophila Genome Project (BDGP, The y w1118 stock was kindly provided by J. Tower (University of Southern California, Pasadena, CA, USA); the y1w67c23 stock by J.-A. Lepesant (Institut J. Monod, CNRS, Paris, France); and the y1w67c23; CyO/Sp stock by Y. Hiromi (Princeton University, Princeton, NJ, USA). The wild-type Canton S strain is maintained in our laboratory. All strains were maintained on standard cornmeal/yeast/sugar medium at 25 °C; heterozygous GstS1 mutants were raised at 20–22 °C as indicated.

Figure 1.

Positions of selected P-elements disrupting the GstS1 gene. The TATA box is indicated by a frame, the transcribed sequence is shown in upper-case letters, the intron is underlined, and the translated sequence is marked by bold double-underlined type. GS2160, k08805 and k09303 correspond to the sites of insertion of a P-element in the strains GS2160, P803 and P874, respectively. The genomic sequence is from cosmids AC020007 and AC007520 (

Purification of DmGSTS1-1 from Drosophila

Adult flies of mixed sexes were used as the starting material. Strains y w1118 and y1w67c23 yielded identical purification results. Flies were collected and stored at −20 °C until required. The flies (10–50 mL packed volume) were homogenized in an equal volume of 20 mm potassium phosphate, pH 7.0, 1.4 mm 2-mercaptoethanol, 2 mm EDTA using a Teflon-glass homogenizer, and centrifuged at 12 000 g for 30 min. The supernatant was used for the purification of a total GST fraction by glutathione affinity chromatography [16] on a 1-mL GSTrap column (Amersham-Pharmacia, Uppsala, Sweden). The resulting GST pool was dialyzed against 10 mm Tris/HCl, pH 7.5, and adjusted to pH 6.5 with Bistris/HCl buffer (final concentration: 20 mm) immediately before loading onto a 0.7 × 4 cm Macro-Prep DEAE Support column (Bio-Rad, Hercules, CA, USA) pre-equilibrated with 10 mm Bistris/HCl, pH 6.5. After washing with 10 column volumes of the above buffer, the column was eluted with a step gradient of NaCl. The salt concentrations and volumes used are shown in the Results section. Fractions were assayed for protein (Bradford reagent, Bio-Rad) and for enzymatic activity with CDNB [17]. Even though the activity of DmGSTS1-1 for CDNB is low (see Results), it is sufficient for peak localization, and obviates the need for large amounts of 4-HNE for the screening of multiple fractions. Peaks containing active fractions were pooled, dialyzed, and used for further characterization.

Bacterial expression and purification of DmGSTS1-1

GstS1 cDNA was isolated and subcloned between the NdeI and SalI sites of the bacterial expression vector pET-21c(+) (Novagen, Madison, WI, USA; J. A. Coronella & B. J. Cochrane, unpublished results). In addition to a plasmid encoding the full-length protein, a second vector was prepared that lacked sequences for the N-terminal 44 amino acids, and in which the 45th amino acid was converted to an initiator methionine. In both plasmids, a His-tag was fused to the C-terminus of the protein (J. A. Coronella & B. J. Cochrane, unpublished results). For the purposes of the present work, the plasmids encoding the full-length DmGSTS1-1 and the N-terminally truncated protein (denoted DmGSTS1-1[tr]) were modified as follows. In both vectors, stop codons were introduced in positions corresponding to the termination of the native protein, thus eliminating the C-terminal His-tag. In the available clone of the full-length protein, the third codon was GGT (encoding Gly), while GAT (encoding Asp) is present both in the published cDNA [12] and in the genomic sequences of GstS1 (cosmids AC020007 and AC007520, Therefore, the GGT codon was converted to GAT by site-directed mutagenesis. It should be also noted that the region located between nucleotides 180 and 250 (relative to the initiator ATG) in the originally published cDNA sequence ([12] GenBank accession no. M95198) differs by six gaps and two mismatches from the available EST and genomic GstS1 sequences (e.g. nucleotides 91628–91704 in AC007520) ([10] and the BDGP, The differences result in the replacement of amino acids PSPCATCSDGNQEYEDVAHPRRV (61–83) of the original sequence [12] by AEPLRYLFAYGNQEYEDVRVTRDEW (61–85), which is longer by two residues, extending the total length of the protein to 249 amino acids. The cDNA clone used by us for expression is identical in this region to the genomic and EST consensus sequence. Thus, the sequence of the bacterially expressed full-length protein corresponds exactly to that of the native DmGSTS1-1, and DmGSTS1-1[tr] differs from the naturally occurring enzyme only by the lack of the N-terminal hydrophobic extension. Escherichia coliBL21(DE3)pLysS was transformed with the two expression vectors. For expression, 500 mL of Luria–Bertani medium containing 200 µg carbenicillin and 34 µg chloramphenicol per mL were inoculated with 5 mL of an overnight culture of the transformed bacteria, and cultured at 30 °C until D600 reached 0.6. The culture was then induced with 1 mm isopropyl thio-β-d-galactoside. After 6 h at 30 °C, the bacterial cells were harvested by centrifugation, and the resulting pellet was frozen at − 20 °C. For purification, the frozen cell pellet was sonicated (tip sonicator, four 20-s bursts on ice) in 50 mL of 20 mm potassium phosphate, pH 7.0, 1.4 mm 2-mercaptoethanol, 2 mm EDTA, and was centrifuged for 12 000 g for 30 min. The supernatant was saved, and the pellet was re-extracted by sonication with another 50 mL of the above buffer. The combined supernatants were loaded onto a 1 × 6 cm column of glutathione agarose (Sigma, cat. no. G-4510) for GST purification [16]. The yields from 500 mL bacterial culture were typically 4 mg for DmGSTS1-1 and 20 mg for DmGSTS1-1[tr].

Enzyme activity determination

GST activities were measured spectrophotometrically in a microtiter plate reader (SpectraMax Plus, Molecular Devices, Sunnyvale, CA, USA) or in 1-mL quartz cuvettes in a Shimadzu UV-2410PC spectrophotometer. Conjugation of CDNB was measured at 25 °C according to [18]. Enzyme activity with 4-HNE was calculated from the rate of consumption of this substrate (measured by decrease of absorbance at 224 nm; this absorbance is due to the α,β-unsaturated carbonyl structure in 4-HNE) at 30 °C [19]. For the determination of kinetic constants of 4-HNE conjugation, the GST activities of the enzymes were measured by varying the concentrations of 4-HNE at fixed concentrations of GSH, and the results were analyzed by least-squares nonlinear fitting of a hyperbola. Specific activities of glutathione conjugation of 3,4-dichloronitrobenzene, 4-nitrobenzyl chloride, 1,2-epoxy-3-(p-nitrophenoxy)propane, 4-nitropyridine-N-oxide, trans-4-phenyl-3-buten-2-one, and ethacrynic acid were determined essentially according to [18], and that of trans-non-2-enal according to [20]. Phospholipid hydroperoxide substrates were prepared, and glutathione peroxidase activity toward these substrates and towards cumene hydroperoxide was measured according to [21].

Gel electrophoresis and Western blotting analysis

Laemmli SDS/PAGE was carried out in 15% separating gels. The gels were either stained with Coomassie blue or electroblotted to nitrocellulose membranes and probed with antibodies. Polyclonal antibodies (raised against bacterially expressed DmGSTS1-1 with glycine in amino-acid position 3, and with a C-terminal His tag; J. A. Coronella & B. J. Cochrane, unpublished results) were used for detection of DmGSTS1-1. A peroxidase-coupled secondary antibody and the ECL Plus kit (Amersham-Pharmacia) with fluorescence detection were used for visualization and quantitation of bands on a Storm 860 imager (Molecular Dynamics, Sunnyvale, CA). For the determination of DmGSTS1-1 content in fly lysates, aliquots of the lysates and known amounts of bacterially expressed DmGSTS1-1 were loaded on the same gel. The calibration curve constructed by quantitation of bands resulting from the standards was used to determine the unknowns (whose loading amount was adjusted to be within the range of the calibration curve). The following antibodies were employed for detection of phosphorylated proteins: anti-phosphotyrosine Ig PY20 (Transduction Laboratories, Lexington, KY, USA), anti-phosphoserine Ig Z-PS1, and anti-phosphothreonine Ig Z-PT1 (Zymed, San Francisco, CA, USA). A polyclonal antibody against protein-4-HNE adducts was a kind gift of L. Szweda (Case Western Reserve University, Cleveland, OH, USA). An alkaline phosphatase-coupled secondary antibody with chromogenic development was used for PY20. For the other antibodies, peroxidase-coupled secondary antibodies were used in conjunction with the ECL Plus kit as described above.

Statistical methods

Comparisons of enzyme activity and DmGSTS1-1 content of three groups of flies (carrying two, one, and no intact GstS1 alleles) were performed by anova followed by pairwise t-tests with the Bonferroni correction for multiple comparisons. The standard deviation of the ratio of two means (with their associated variances) was calculated according to [22].


4-HNE-conjugating activity in Drosophila

A total homogenate of adult D. melanogaster had the ability to conjugate 4-HNE with glutathione. The specific activity for this reaction in the crude homogenate was 128 ± 17 nmol·mg−1·min−1 for the Canton S wild-type strain and 234 ± 49 nmol·mg−1·min−1 in strain y1w67c23; CyO/Sp which was subsequently used as the appropriate control in further analysis. These values are comparable to that obtained in a parallel measurement for rat liver (192 ± 40 nmol·mg−1·min−1).

4-HNE-conjugating activity in Drosophila with a disrupted GstS1 gene

A correlation of P-element disruption of the GstS1 gene with changes in 4-HNE conjugating activity in total fly lysates provided a quantitative estimate of the contribution of DmGSTS1-1 to the overall capacity for 4-HNE metabolism in Drosophila. Three mutants, all with insertions in the 5′ untranslated region of the gene (Fig. 1) were tested. Two of these P-element insertions (strains P803 and P874) are lethal when homozygous. Although the lethality is probably not due to the disruption of the GstS1 gene but to the presence of additional P-elements elsewhere on the chromosome (, it precludes an analysis of flies homozygous for the disruption. However, crosses between these strains produced flies which have both GstS1 alleles disrupted and were viable, although they grew very poorly at 25 °C, providing few if any noncurly flies at this temperature. Homozygous mutant flies (874/803 and 803/874) were therefore generated by crosses in both directions (male A × female B and female A × male B) at 20–22 °C. The GST activities of the homozygous mutant flies were compared to those of heterozygous flies (874/CyO, CyO/874, 874/Sp, Sp/874, 803/CyO, CyO/803, 803/Sp, Sp/803) generated by crosses to the stock, y1w67c23 ; CyO/Sp as the P-element insertions were recovered in a y1 w67c23 background and the chromosome 2 lethalities are maintained over the CyO balancer. This stock may carry a different GstS1 allele on the Sp chromosome. For a more traditional genetic analysis, there are currently no deficiency stocks uncovering the chromosomal location of GstS1 (53F3/4), or a small region around 53F of chromosome 2, available at any of the Drosophila stock centers. Heterozygotes carrying either the CyO or Sp chromosome were compared.

The CDNB activities of the heterozygous mutant flies averaged 93 ± 18% of control (data not shown), indicating that disruption of the GstS1 gene does not significantly affect the overall GST activity for CDNB. It can be thus concluded that DmGSTS1-1 has minimal conjugating activity for CDNB, in agreement with our direct measurements on the purified enzyme (see below).

In contrast to CDNB, the 4-HNE-conjugating activity differed among flies carrying two intact GstS1 alleles, one intact and one disrupted allele, and two disrupted alleles. On average, the activities in homogenates from these genetically different groups of flies were 0.204 ± 0.009, 0.131 ± 0.011, and 0.091 ± 0.006 µmol·min−1·mg−1, respectively. These values are significantly different from each other (P < 0.0001). The positive correlation between the number of intact GstS1 alleles (‘gene dosage’) and 4-HNE-conjugating activity indicates that DmGSTS1-1 contributes significantly to the metabolic capacity for 4-HNE in Drosophila.

Correlation of 4-HNE-conjugating activity with DmGSTS1-1 content in Drosophila

To characterize quantitatively the correlation between DmGSTS1-1 and 4-HNE-conjugating activity, we determined the amount of DmGSTS1-1 protein in fly extracts by Western blotting (see Experimental procedures). Lysates of wild-type Drosophila carrying two intact GstS1 alleles contained 18.5 ± 1.7 µg DmGSTS1-1 per mg total protein. Therefore, DmGSTS1-1 accounts for 1.9 ± 0.2% of the total protein mass in adult wild-type Drosophila. Flies with one and two disrupted GstS1 alleles contained 8.9 ± 2.2 and 5.4 ± 0.9 µg DmGSTS1-1 per mg total protein, respectively. Each of the three values is significantly different from the remaining two (P < 0.02). The results show that flies with one disrupted GstS1 allele contained 48 ± 13% of the DmGSTS1-1 protein present in wild-type flies, consistent with the expected 50%. It is noteworthy that flies with P-element insertions in both GstS1 alleles still contained DmGSTS1-1 protein (29 ± 6% of wild-type, vs. the expected 0%). This indicates that the P-element insertions used in the present study do not completely prevent the expression of the GstS1 gene.

There was a good linear correlation (correlation coefficient R = 0.887 ± 0.018) between the DmGSTS1-1 protein content and the level of 4-HNE-conjugating activity in all Drosophila lines and crosses used in the present work (Fig. 2). The slope of the regression line, which corresponds to the specific activity of DmGSTS1-1, was 7.3 ± 1.1 µmol·mg−1·min−1, in good agreement with the value measured directly on the purified enzyme (see below). The intercept of the regression line was 0.062 ± 0.011 µmol·min−1·(mg total protein)−1, or 30 ± 6% of the activity found in wild-type flies. The intercept represents 4-HNE-conjugating activity that remains when the amount of DmGSTS1-1 is extrapolated to zero, and indicates that 70 ± 6% of the total capacity for 4-HNE conjugation in normal adult Drosophila is attributable to DmGSTS1-1, while the residual 30 ± 6% may be due to other GSTs.

Figure 2.

Correlation between 4-HNE-conjugating activity (ordinate) and amount of DmGSTS1-1 protein (abscissa) in various strains and crosses of Drosophila. Flies were homogenized, and extracts were used for activity determination and for quantitative Western blotting as detailed in the Experimental procedures. Closed symbols: strains carrying two intact/wildtype GstS1 alleles (y1w67c23; +/Sp and y w). Half-filled symbols: strains carrying one intact and one P-element-disrupted GstS1 allele (874/Sp;803/Sp;803/CyO;Sp/803;CyO/803;Sp/874;874/CyO;CyO/874). Open symbols: strains carrying P-element insertions in both GstS1 alleles (874/803;GS2160;803/874). In each of the three groups described above, the strains are listed in increasing order of DmGSTS1-1 content. Linear regression was carried out by the unweighted least squares method. The correlation coefficient was 0.887 ± 0.018.

Purification of DmGSTS1-1 from Drosophila

Genetic analysis revealed that loss of DmGSTS1-1 expression correlated with a decrease in 4-HNE conjugation (Fig. 2), and thus provided indirect evidence that DmGSTS1-1 is responsible for the latter enzymatic activity. To confirm this directly, we purified DmGSTS1-1 from adult Drosophila by a combination of affinity and ion-exchange chromatography (Table 1), and characterized it biochemically. As DmGSTS1-1 has been reported to be tightly associated with the indirect flight muscle of Drosophila[13], in initial experiments we verified that the protein could be extracted into buffer during homogenization of flies. All of the extracted DmGSTS1-1 was retained on the GSH affinity column, as no enzyme was detected by immunoblotting in the flow-through (Fig. 3C).

Table 1.  Purification of DmGSTS1-1 from Drosophila. DmGSTS1-1 protein was isolated from adult Drosophila (50 mL packed volume of flies) as described in Experimental procedures. DmGSTS1-1 is found in peaks b1 and b2 (see text for more details).



Protein conc.

Total protein
Specific activity
for 4-HNE
Total activity
for 4-HNE

  1. a  Soluble Drosophila protein (12 000 g supernatant, see Experimental procedures). b Values shown for peak b represent the arithmetic sum of experimentally determined values for peaks b1 and b2 (shown in parentheses in the table). DmGSTS1-1 elutes from the DEAE column as two peaks (see Fig. 3A and text).

Homogenate a1006.76700.18 ± 0.04121.0 ± 26.8100
GSH column250.8822.04.1 ± 0.490.0 ± 9.574
DEAE column
 peak a14.80.314.69.4 ± 0.642.6 ± 2.635
 peak b b29.3 14.0 69.5 ± 0.657
 (peak b1)(15.3)(0.50)(7.6)(4.9 ± 0.1)(37.4 ± 1.1)(30)
 (peak b2)(14.0)(0.46)(6.4)(5.0 ± 0.1)(32.1 ± 0.8)(27)
 peak c10.50.212.2000
Figure 3.

Purification of DmGSTS1-1 from Drosophila. The purification (from 10 mL packed volume of flies) was carried out as described in Experimental procedures. (A) The elution profile of the DEAE ion-exchange column. Closed symbols, protein concentration; open symbols, CDNB-conjugating activity; broken line, NaCl concentration used for elution. Peak a contains mostly delta-class GSTs, peaks b1 and b2 contain DmGSTS1-1 (see text for further discussion), and peak c contains an unidentified protein without GST activity for the substrates tested. (B) Coomassie blue-stained SDS/PAGE gel of the GST pool obtained by GSH affinity chromatography (lane ‘total’), and fractions recovered from the DEAE column labeled as in panel A (lanes ‘a’, ‘b1’, ‘b2’, and ‘c’). (C) Western blot probed with anti-(DmGSTS1-1) Ig. Lane ‘total’, the GST pool obtained by GSH affinity chromatography; lane ‘FT’, flow-through from the DEAE column after loading the above GST pool; lanes ‘a’, ‘b1’, and ‘b2’, fractions from the DEAE column labeled as shown in (A).

As expected [23], the total GST pool purified by GSH affinity chromatography was heterogeneous (Fig. 3B, lane ‘total’), and could be further separated by ion-exchange chromatography. Because of the very acidic isoelectric point of DmGSTS1-1 predicted from its sequence (calculated pI = 4.4), we used an anion-exchange column, and chose for the separation a pH that was low but still compatible with stability of enzyme activity. Under these conditions, a large peak of CDNB-conjugating activity emerged at the void volume of the column (Fig. 3A, peak a). This fraction consisted of a mixture of GSTs (Fig. 3B, lane a), probably mostly of the delta class [5,6], but contained no DmGSTS1-1 (Fig. 3C, lane a). DmGSTS1-1 was eluted with increasing NaCl concentration as two separate peaks (Fig. 3A, peaks b1 and b2). Both peaks were recognized by the anti-(DmGSTS1-1) Ig (Fig. 3C, lanes b1 and b2), establishing their identity. Both peaks migrated irregularly on SDS/PAGE at an apparent molecular mass of 34 kDa, a position that is however, characteristic for the 27.6-kDa DmGSTS1-1 protein [12,13,23]. Finally, both peaks had identical catalytic properties for CDNB (Fig. 3A) and 4-HNE (Table 1). The split of the DmGSTS1-1 protein into two fractions separable according to their charge was observed previously [13]. This phenomenon is probably not due to phosphorylation, as neither of the two peaks was recognized by antibodies against phosphoserine, phosphothreonine, and phosphotyrosine (results not shown). In independent preparations from two different strains of Drosophila (y w1118 and y1w67c23), the two DmGSTS1-1 peaks contained approximately equal amounts of protein. While this could indicate the presence of two different GstS1 alleles, such explanation is unlikely because both strains of Drosophila are inbred and highly homozygous. Further work will be required to explain this observation.

Elution of the ion-exchange column with 0.25 m NaCl yielded a protein with an apparent molecular mass of 100 kDa (Fig. 3A, peak c, and Fig. 3B, lane c). The protein, which was present in the total GST fraction obtained by GSH affinity chromatography (Fig. 3B, lane ‘total’) and thus presumably binds GSH, had no GST activity with either CDNB or 4-HNE, and was not recognized by anti-(DmGSTS1-1) Ig (data not shown). This protein was not investigated further.

In a preparation starting with a 50-mL packed volume of flies, the amount of DmGSTS1-1 protein recovered in peaks b1 + b2 (Fig. 3A) was 14 mg, while 4.6 mg of protein were recovered in peak a (Table 1). Thus, DmGSTS1-1 accounts for approximately 75% of the total mass of those Drosophila GSTs that bind to GSH agarose.

It should be noted that not only the DmGSTS1-1 fractions, but also the heterogeneous GST peak a (Fig. 3A) had 4-HNE-conjugating activity (Table 1). While the specific activity in peak a was in fact higher than in peaks b1 and b2, the total activity in the latter fractions (which contain DmGSTS1-1) accounted for approximately 60% of the overall 4-HNE-conjugating activity recovered from the DEAE column (Table 1). This agrees well with our previously discussed conclusion (based on disruption of GstS1 alleles) that DmGSTS1-1 accounts for 70 ± 6% of the 4-HNE-conjugating activity in Drosophila homogenates.

Expression of DmGSTS1-1 in bacteria and purification of the protein

To rule out the possibility that the enzymatic activity associated with DmGSTS1-1 isolated from Drosophila may not be an intrinsic property of the protein but is due to a putative minor impurity, we expressed the enzyme in E. coli. In addition to the above experimental objective, we wanted to establish whether the N-terminal extension of the protein which is essential for its anchoring to the indirect flight muscle [13] has an effect on enzyme activity. We thus expressed the full-length protein (identical in its sequence to the native DmGSTS1-1) as well as its truncated version lacking the 44-amino-acid N-terminal extension and hence corresponding to a ‘core’ GST. Both proteins were purified by GSH affinity chromatography. The full-length bacterially expressed DmGSTS1-1 migrated on SDS/PAGE at a position corresponding to 34 kDa (Fig. 4), which is considerably higher than the calculated molecular mass of 27.6 kDa but indistinguishable from the migration of the native protein isolated from Drosophila. The truncated protein migrated at an apparent molecular mass of 25 kDa (Fig. 4), close to the value of 23.6 kDa calculated from its sequence. Thus, the irregular behavior on SDS/PAGE appears to be largely due to the N-terminal extension which is rich in hydrophobic and in negatively charged amino acids, and is probably not the result of post-translational modifications as it equally affects DmGSTS1-1 synthesized in eukaryotic and in prokaryotic cells.

Figure 4.

Expression of recombinant DmGSTS1-1 in E. coli. Full-length DmGSTS1-1 (lanes labeled ‘full’) and truncated DmGSTS1-1 lacking the N-terminal hydrophobic/acidic extension (lanes labeled ‘tr’) were expressed and purified by GSH affinity chromatography. (A) Coomassie blue-stained SDS/PAGE gel. (B) Western blot using anti-(DmGSTS1-1) Ig.

Glutathione transferase activity of DmGSTS1-1

The customary monitoring of GST purification by assaying fractions for CDNB-conjugating activity indicated that, in contrast to the delta class GSTs (Fig. 3A, peak a), DmGSTS1-1 has very little activity for CDNB (Fig. 3A, peaks b1 and b2, open symbols). This observation was confirmed by direct characterization of the purified enzyme. DmGSTS1-1 had a specific activity of CDNB conjugation of less than 0.5 µmol·min−1·mg−1(Table 2), i.e. one to two orders of magnitude less than most GSTs [1]. This marginal level of activity was insufficient for a reliable determination of kinetic parameters. In contrast, the enzyme had moderately high activity for the lipid peroxidation product 4-HNE (see Table 2 for kinetic constants). There were no significant differences in the kinetic parameters of DmGSTS1-1 isolated from Drosophila or expressed in E. coli (Table 2), confirming that the activity is attributable to DmGSTS1-1 and not to a copurifying contaminant. Furthermore, the kinetic constants for 4-HNE were similar for the full-length and the truncated forms of DmGSTS1-1 (Table 2), indicating that the N-terminal hydrophobic/acidic extension that anchors the protein to the flight muscle has only a slight effect on its catalytic properties.

Table 2.  Glutathione transferase activity of DmGSTS1-1 for 4-HNE and CDNB. DmGSTS1-1 protein was isolated from adult Drosophila or expressed in E. coli. The recombinant enzyme was either full-length or was truncated to remove the hydrophobic N-terminal stretch of 44 amino acids (denoted DmGSTS1-1[tr]).


Kinetic parameter

isolated from flies
expressed in
E. coli
expressed in
E. coli
  1. a  Calculated using molecular masses of 55 224 Da and 47 146 Da for the full-length and truncated forms of dimeric DmGSTS1-1, respectively. b  Specific activity was measured at 1 m m CDNB. The activity with this substrate was too low for reliable determination of kinetic parameters. c With 4-HNE as the second substrate.

4-HNE V max (µmol·mg−1·min−1)7.8 ± 0.38.4 ± 0.48.4 ± 0.6
k cat (s−1)a7.2 ± 0.37.7 ± 0.46.6 ± 0.5
K mm)146 ± 12123 ± 12178 ± 23
k cat/Km (s−1·m−1)4.9 × 1046.3 × 1043.7 × 104
CDNBSpec. act. (µmol·mg−1·min−1) b0.450.390.40
GSH K m (mm)c1.1 ± 0.10.6 ± 0.21.7 ± 0.3

The virtual lack of activity for CDNB in DmGSTS1-1 is unusual for this class of enzymes. For example, the prototypical squid Sigma-class GST has the strikingly high activity with CDNB of almost 1000 µmol·min−1·mg−1[24], compared with approx. 0.4 µmol·min−1·mg−1 for DmGSTS1-1 (Table 2). The vertebrate hematopoietic-type prostaglandin D synthases which belong to the sigma class of GSTs have lower but still significant activities with CDNB, especially the avian enzyme [25,26]. To determine whether DmGSTS1-1 is specifically deficient for the conjugation of CDNB or whether it is narrowly specialized for conjugating 4-HNE, we examined the ability of the enzyme to accept representative GST substrates. As shown in Table 3, DmGSTS1-1 had generally low activity (in comparison with typical mammalian GSTs [18]) for aryl and arylalkyl halides, an epoxide, a nitroaromatic compound, and two bulky α,β-unsaturated carbonyl compounds. In contrast, the enzyme had significant activity for trans-nonenal. Thus, DmGSTS1-1 appears to be well-adapted for conjugation of α,β-unsaturated carbonyl compounds with an elongated structure lacking aromatic rings. It could be hypothesized that the electrophile binding site of DmGSTS1-1 is narrow and thus excludes bulky compounds.

Table 3.  Specific activity of DmGSTS1-1 for glutathione conjugation of representative substrates. DmGSTS1-1 expressed in E. coli either as the full-length protein (DmGSTS1-1) or lacking the hydrophobic N-terminal stretch of 44 amino acids (denoted DmGSTS1-1[tr]) was assayed as described in Experimental procedures. ND, not detectable.
 Specific activity (µmol·mg−1·min−1)
3,4-Dichloronitrobenzene0.040 ± 0.0050.031 ± 0.004
4-Nitrobenzyl chloride0.25 ± 0.020.28 ± 0.02
1,2-Epoxy-3-(p-nitrophenoxy)propane0.70 ± 0.020.65 ± 0.02
trans-4-Phenyl-3-buten-2-one0.013 ± 0.0010.015 ± 0.001
Ethacrynic acid0.58 ± 0.020.53 ± 0.01
trans-Non-2-enal1.68 ± 0.092.08 ± 0.06

Formation of 4-HNE-protein adducts in Drosophila

4-HNE was originally identified in mammalian tissues [27]. While a selective modification of mitochondrial adenine nucleotide translocase by 4-HNE has been described for the house fly [28], there are no reports in the literature on 4-HNE in Drosophila. Therefore, we probed a blot of total Drosophila proteins with an antibody specific for 4-HNE adducts of histidine, lysine, and cysteine side chains in proteins [29]. As shown in Fig. 5, multiple proteins were recognized by the antibody, indicating that 4-HNE is indeed formed in Drosophila.

Figure 5.

Formation of 4-HNE-modified proteins in Drosophila. A blot of a total homogenate of 19-day-old flies (22 µg protein, separated by SDS/PAGE on a 15% gel) was probed with antibodies against 4-HNE-protein adducts.

Glutathione peroxidase activity of DmGSTS1-1

Insects do not express a selenium-dependent glutathione peroxidase [30,31]; instead, they depend on GSTs for reduction of organic hydroperoxides. We therefore checked whether DmGSTS1-1 possesses glutathione peroxidase activity. Although the enzyme was active with both a model substrate (cumene hydroperoxide) and putative natural substrates (phospholipid hydroperoxides; see Table 4), the activity was low compared to that of mammalian alpha-class GSTs [32]. However, for cumene hydroperoxide for which comparisons can be made with literature data [6,33], the glutathione peroxidase activity of DmGSTS1-1 was similar to that of Drosophila delta class DmGSTD2-2 (originally GST D21 [6]) and DmGSTD8-8 (originally GST D27 [33]).

Table 4.  Glutathione peroxidase activity (nmol·mg−1·min−1) of DmGSTS1-1 with various substrates. DmGSTS1-1 was expressed in E. coli either as the full-length protein (DmGSTS1-1) or lacking the hydrophobic N-terminal stretch of 44 amino acids (denoted DmGSTS1-1[tr]).
EnzymeCumene hydroperoxidePhosphatidylcholine hydroperoxidePhosphatidylethanolamine hydroperoxide
DmGSTS1-1173 ± 1692 ± 584 ± 4
DmGSTS1-1[tr]156 ± 685 ± 359 ± 3


Previously, DmGSTS1-1 has been localized primarily to the indirect flight muscle of Musca domestica and D. melanogaster[13,34]. It is thought that DmGSTS1-1 is anchored to troponin-H 34 in the thin filament of the muscle via the hydrophobic/acidic N-terminal extension of the GST protein [13]. As no enzymatic activity has been previously attributed to DmGSTS1-1, it was postulated that the protein has a structural function, or perhaps is involved in the stretch activation of the indirect flight muscle [13]. However, in some insects, including Choristoneura fumiferana, Manduca sexta, and Anopheles gambiae, the orthologs of DmGSTS1-1 lack the N-terminal extension ([35] and J. A. Coronella & B. J. Cochrane, unpublished results). This indicates that anchoring of this GST to the flight muscle is not universal. Moreover, lesser but still significant amounts of DmGSTS1-1 and its orthologs in other insect species are found in nonmuscle tissues ([13,34], and J. A. Coronella & B. J. Cochrane, unpublished results). This argues for a function of the protein other than, or in addition to, a role in indirect flight muscle structure or contraction. Our present finding that DmGSTS1-1 has an enzymatic activity is consistent with this interpretation.

In the present communication we report for the first time that DmGSTS1-1 has the ability to catalyze the conjugation of GSH to 4-HNE, an electrophilic end-product of lipid peroxidation and a major diffusible mediator of oxidative damage [36,37]. In mammals, this function is largely carried out by a specialized group of GSTs belonging to the alpha class, e.g. the rodent GSTA4-4 [38–41] and the human hGSTA4-4 [20,42,43] and hGST5.8 [44]. Although the kcat for 4-HNE was approximately 20-fold lower for DmGSTS1-1 than for these mammalian alpha class GSTs [45], the Km values of the Drosophila and the mammalian enzymes were similar and were of the order of 100 µm. Because, at least in mammals, the prevailing serum and tissue concentrations of 4-HNE are submicromolar [27] and thus much lower than the Km, intracellular 4-HNE conjugation catalyzed by DmGSTS1-1 would be ≈ 20-fold slower than that catalyzed by mammalian GSTs exemplified by mGSTA4-4. The lower catalytic efficiency of DmGSTS1-1 is compensated by its greater abundance. In mammalian liver, all GSTs account for up to 5% of the soluble protein [46]. Given that in mouse liver mGSTA4-4 constitutes 2% of the total GSTs [47], the overall abundance of mGSTA4-4 is approximately 0.1% of the soluble hepatic protein. In contrast, 14 mg of DmGSTS1-1 could be purified from 670 mg crude Drosophila homogenate (Table 1), translating into an abundance of 2.1% of the total soluble protein. This estimate agrees very closely with the abundance of 1.9 ± 0.2% calculated independently from calibrated Western blots of fly homogenates (see above). The 19- to 21-fold higher abundance of DmGSTS1-1 in Drosophila compared to mGSTA4-4 in mouse liver offsets the approximately 20-fold higher catalytic efficiency of the murine enzyme. Thus, in total homogenates of Drosophila and of mouse liver, the specific activities for 4-HNE conjugation should be similar, as is indeed the case (see Results). Perhaps even more importantly, the local concentration of DmGSTS1-1 (and thus its catalytic effectiveness) may be further increased in the indirect flight muscle by anchoring the protein through its N-terminal extension to troponin-H 34 [13]. According to this interpretation, the interaction of DmGSTS1-1 with muscle fibers may not necessarily indicate a role of the protein in muscle structure or in triggering stretch-activated contraction. Rather, the anchoring would localize DmGSTS1-1 in a highly aerobic tissue subject to oxidative stress due to rapid mitochondrial respiration [48].

The postulated role of DmGSTS1-1 in counteracting the effects of oxidative stress raises the question of the mechanism of the protective effect. DmGSTS1-1 has glutathione peroxidase activity for lipid hydroperoxides (Table 4) which, although low in comparison with many mammalian GSTs, is similar to that of the most active delta class insect GSTs [6,33]. As insects do not express a separate glutathione peroxidase, the higher abundance of DmGSTS1-1 compared to delta class enzymes could indicate that DmGSTS1-1 is of greater physiological importance with regard to this activity. By analogy with mammalian enzymes [49], it is also likely that the newly discovered microsomal-like GST [8] contributes to glutathione peroxidase activity in Drosophila, although the enzyme has not yet been characterized biochemically. The finding that disruption of the microsomal-like GST reduces the life span of Drosophila[50] is consistent with a role for GST enzymes in the cellular antioxidant defense system.

While the glutathione peroxidase activity of DmGSTS1-1 may be physiologically significant, the most striking property of the enzyme is its ability to conjugate 4-HNE. In mammals, 4-HNE is the product of nonenzymatic degradation of oxidized polyunsaturated fatty acids, primarily of the abundant arachidonic acid [27]. Drosophila lacks arachidonic acid, at least in some tissues [51,52]. However, 4-HNE can be formed from all ω-6 polyunsaturated fatty acids [27,53] which, in addition to arachidonic acid, includes linoleic acid which is present in large amounts in larval and adult stages of Drosophila[54–56]. Moreover, 4-hydroxyalkenals of chain lengths different from the nine carbon atoms of 4-HNE may arise from peroxidation of other polyunsaturated fatty acids [27]. While it is not fully known to what extent Drosophila is able to desaturate fatty acids [57], dietary polyunsaturated fatty acids are also incorporated into phospholipids. As lipid peroxidation and the subsequent decomposition of the peroxides are thought to be nonenzymatic processes, 4-HNE and/or similar 4-hydroxyalkenals are expected to be generated in Drosophila. We have confirmed this directly by detection of 4-HNE-protein adducts in Drosophila by immunoblotting, and others have previously demonstrated the formation of such adducts in the housefly [28]. Thus, detoxification of 4-HNE may be a major physiological role of DmGSTS1-1.

While the majority of DmGSTS1-1 is localized in the indirect flight muscle, lesser amounts of the protein are present in other tissues, including the nervous system [34]. Neurons are critically sensitive to oxidative damage, as demonstrated in Drosophila by the fact that overexpression of an antioxidant enzyme (superoxide dismutase) specifically in motor neurons caused a life span extension of the flies [58,59] similar to that effected by general overexpression of antioxidant enzymes in all tissues [60–62]. This suggests that neuronal localization of DmGSTS1-1 may contribute to the antioxidant defenses of these especially sensitive cells. This conclusion is strengthened by the recent finding that DmGSTS1-1 has a protective effect against oxidative and/or chemical stress in a Drosophila model of the neurodegenerative disorder, spinocerebreller ataxia type 1 [63].

The formation of 4-HNE appears to be an inevitable consequence of aerobic life. Recent evidence (reviewed in [64]) demonstrates that increases of 4-HNE above its normal steady-state level have been evolutionarily adopted as a signal of oxidative stress. The ability of 4-HNE to function as a signal is probably related to its ability to diffuse, a longer half-life that that of oxygen-based radicals, and its high reactivity with histidine, lysine, and cysteine side chains of proteins [65,66]. The resulting functional modifications of key cellular proteins have been well documented [64,67,68]. Particularly striking among the many physiological effects of 4-HNE is the modulation of the cell cycle [69–71], differentiation [72–74], and apoptosis [73,75,76]. It is intriguing to speculate that 4-HNE may act as a signal during embryogenesis and tissue remodeling when the above processes are prominent. Termination of the signal could occur through conjugation, and thus inactivation, of 4-HNE. High levels of GstS1 transcripts during the embryonal and pupal stages [12], and the localization of the protein in imaginal disks of Drosophila (J. A. Coronella & B. J. Cochrane, unpublished results), are consistent with this interpretation.

As already discussed, the conjugation of 4-HNE in mammals is mediated by specialized alpha-class GSTs, which have a significantly higher catalytic efficiency for this substrate than DmGSTS1-1. The emergence of alpha class GSTs is relatively recent on the evolutionary time scale; these enzymes are absent from arthropods. Even in vertebrates, classes of GSTs other than alpha have 4-HNE-conjugating activity. For example, mammalian mu [1] and avian sigma GSTs [25] have been reported to catalyze GSH conjugation of 4-HNE, often with activities approaching that of DmGSTS1-1. This suggests that a GST ancestral to the sigma class may have adapted to catalyze 4-HNE conjugation. The enzyme continues to provide the majority of this activity in extant dipteran insects. In vertebrates, specialized alpha class GSTs subsequently evolved to catalyze 4-HNE conjugation more efficiently. The descendants of the ancestral sigma class GST may provide redundant functionality in vertebrates, or their activity for 4-HNE could constitute an evolutionary relic that was retained in vertebrates because the enzymes acquired other essential functions.

Our finding that a sigma class GST accounts for the majority of 4-HNE conjugation in Drosophila indicates that conjugating activity for this compound arose, presumably independently, within several branches of the GST superfamily, including sigma, mu, and alpha. Other classes of GSTs may also possess this activity: according to our results, approximately 30% of total 4-HNE conjugation in Drosophila, an organism that does not express either alpha or mu classes of GSTs, is not due to DmGSTS1-1. Rather, the activity is associated with a GST fraction that contains the delta class enzymes (Fig. 3A, peak a). Further work will be required to identify the enzyme(s) responsible for this residual activity. The conclusion that 4-HNE-conjugating activity was acquired in the course of evolution more than once points to a fundamental biological importance of 4-HNE catabolism, whether in termination of inter- or intracellular signaling, or in a detoxification process, which is probably obligatory in all aerobic forms of life.


We thank Toshiro Aigaki for providing fly stocks and Neha Mehta for performing some of the genetic crosses and GST enzyme assays. This work was supported in part by National Institutes of Health (National Institute of Environmental Health Sciences) grant ES 07804 to P. Z., University of Arkansas for Medical Sciences Pilot Study Grant to H. B. and P. Z., and National Science Foundation grant DEB9317997 to B. J. C.


  1. *Present address: Arizona Cancer Center, University of Arizona, Tucson, AZ 85724, USA.

  2. Enzyme: glutathione S-transferase (EC