Insecticide resistance is a major problem for both medicine and agriculture and is frequently associated with overexpression of metabolic enzymes that catalyze the breakdown of pesticides, leading to broad-spectrum resistance. However, the insect tissues within which these metabolic enzymes normally reside remain unclear. Microarray analysis of nine adult tissues from Drosophila melanogaster reveals that cytochrome P-450s and glutathione-S-transferases show highly tissue-specific expression patterns; most were confined to one or more epithelial tissues, and half showed dominant expression in a single tissue. The particular detoxifying enzymes encountered by a xenobiotic thus depend critically on the route of administration. In particular, known insecticide metabolism genes are highly enriched in insect Malpighian (renal) tubules, implicating them in xenobiotic metabolism. The tubules thus display, with the fat body, roles analogous to the vertebrate liver and immune system, as well as its acknowledged renal function. To illustrate this, when levels of a single gene, Cyp6g1, were manipulated in just the Malpighian tubules of adult Drosophila, the survival of the whole insect after 1,1,1-trichloro-2,2-bis(4-chlorophenyl)ethane (DDT) challenge was altered, whereas corresponding manipulations in the nervous system or the fat body were without effect. This shows that, although detoxification enzymes are widely distributed, baseline protection against DDT resides primarily in the insect excretory system, corresponding to less than 0.1% of the mass of the organism.

  • transcriptomics
  • insecticide resistance
  • transgenics

insects are major vectors of transmissible disease and pests of major crops. Effective insect control is thus vital, and insecticide resistance is a continuing problem, both in developed and developing worlds. Insecticide resistance across many species has been attributed to upregulation of enzymes associated with xenobiotic detoxification and metabolism (12), whereas the unusual sensitivity of the honey bee to insecticides may reflect underrepresentation of these genes in its genome (8). In Drosophila melanogaster several different cytochrome P-450s have been implicated in conferring resistance to 1,1,1-trichloro-2,2-bis(4-chlorophenyl)ethane (DDT) and a range of more recent insecticides such as the neonicotinoids: for example, CYP6G1, CYP6A2, CYP12D1, and CYP12A4 (1, 10). In the case of the P-450 Cyp6g1, the insertion of an Accord transposable element into the 5′ end of the gene has led to its elevated expression (16). DDT resistance in mosquitoes has also been associated with upregulation of glutathione-S-transferases (12), such as GSTE2 (18). These broad-specificity mechanisms produce cross-resistance to other novel classes of insecticide, such as pyrethroids (12) and the recently introduced neonicotinoids (10). However, despite the extensive literature on the number and types of P-450s and GSTs expressed in resistant strains, there is little information on the tissues in which xenobiotic metabolism is effected.

Given the large number of P-450 and GST encoding genes in the Drosophila genome and the large number of studies implicating different genes in resistance in different Drosophila strains, there is a lively debate as to the relative importance of single vs. multiple genes in insecticide resistance (5, 1012, 16, 17). Despite the observation that single metabolic genes are upregulated in resistant strains recently isolated from the field, it is a matter of biochemical fact that several of the candidate enzymes are capable of detoxifying xenobiotics, and so it is possible that different genes play key roles in different circumstances. For example, different tissues might provide the first line of defense against ingested or topical insecticides, and if detoxifying enzymes are tissue-specific in their distribution, then apparently conflicting results might be obtained. It is thus critical to obtain a clear view of gene expression in multiple tissues.

Although the biochemistry of insecticide resistance is well understood, the physiology is much less so. Is insecticide metabolism a “housekeeping” role found in all cells, or are specific tissues (like the nervous system) responsible for local or global defense against xenobiotics? Such data could affect strategies for efficacious and environmentally sparing insecticide use. Previously, attention has focused on the midgut, fat body, and Malpighian tubules for metabolism of both insecticides and plant secondary metabolites. For example, phenobarbital administration induced Cyp6a2 overexpression in Drosophila midgut, the pericuticular fat bodies, and the Malpighian (renal) tubules (4), whereas the detoxification of furanocoumarin by Cyp6b1 and Cyp6b3 depended on midgut and fat body (22). These data are consistent with a classical view that the fat body performs as an insect “liver.” Reanalysis of a detailed microarray study of gene expression in adult Drosophila Malpighian (renal) tubule (26) revealed substantial upregulation of several members of the cytochrome P-450 and glutathione-S-transferase families (data not shown). In particular, the known insecticide resistance genes Cyp6g1 and Cyp6a2 were found to be upregulated 9.4-fold and 8-fold, respectively, in tubule compared with whole fly (26). These data clearly suggest that on an organismal scale, xenobiotic metabolism may be primarily a renal function, and so a more detailed investigation was undertaken.



Wild-type and mutant Drosophila stocks were kept on standard Drosophila diet at 22°C and 55% relative humidity on a 12:12 h photoperiod. Stocks used were as follows: Canton S, wild-type controls; 5′CS-Cyp6g1-GFP, flies carrying the control region of Cyp6g1, corresponding to bases −1 to −1200 from the transcriptional start site, cloned from Canton S flies and placed upstream of an enhanced green fluorescent protein (GFP) open-reading frame (ORF); c42, a GAL4 driver specific to the Malpighian tubule principal cells (3, 23, 24); elav-GAL4, a GAL4 driver specific to the nervous system; c564, a GAL4 driver specific to the fat body (kind gift of Dr. Shoichiro Kurata, Sendai, Japan); UAS-Cyp6g1-RNAi(I), a transgenic line carrying a hairpin dsRNA designed against the Cyp6g1 cDNA (from NIG-Fly, Shizuoka, Japan); UAS-Cyp6g1-3A and UAS-Cyp6g1-8A, two transgenic lines carrying independent insertions of the Cyp6g1 ORF inserted downstream of the UAS promoter. The strategy of designing RNAi hairpins against nonconserved sequence in 3′-UTRs appears to protect against off-target effects in Drosophila (25). Additionally, none of the genes with greatest sequence similarity to Cyp6g1 (6g2, 6t1, 6t3, 6u1, 6v1, or 6w1) are abundantly expressed in tubule.

Fluorescence microscopy.

Flies were anesthetized briefly on ice, and dissected in sterile Schneider's medium. Tissues were mounted on a microscope slide under a #0 coverslip in a drop of Schneider's solution and viewed immediately on a Zeiss Axioplan microscope. Color images were recorded with a Zeiss camera and Axiophot software.

Quantitative PCR validation of modulated gene expression.

Tubules were dissected from 1-wk-old adult progeny of GAL4/UAS crosses as described above, together with matched parental controls, and pooled in groups of 20 in TRIzol reagent on ice. mRNA was extracted, and cDNA was reverse transcribed and quantified according to standard procedures. For quantitative PCR (qPCR), samples were split and amplified with Cyp6g1 primers, using Drosophila rp49 as a reference gene, as described previously (21, 26). Ct values were converted to nanograms using a standard curve derived from parallel standards on the same plates.

DDT toxicology.

DDT knockdown was assessed by standard methods. Briefly, 1-wk-old transgenic or control flies were taken from diet tubes and placed in glass vials in which the indicated quantities of DDT (Sigma) had been introduced in acetone vehicle, after allowing the acetone to fully evaporate. Knockdown efficiency was measured after 24 h. Mortality was calculated from experimental replicates, and probit analysis performed with GraphPad Prism software.


The FlyAtlas data set (http://flyatlas.org) provides the most comprehensive database of Drosophila gene expression to date, as it is based on the Affymetrix Dros2 GeneChip, with 18,770 probe sets covering the estimated 13,500 genes (6). The atlas is based on four biological replicates of pooled samples of each of the following: brain, head, crop, midgut, Malpighian tubules, hindgut, testis, ovary, and male accessory glands of 7-day old adult D. melanogaster (Oregon R strain), together with data for larval Malpighian tubules and larval fat body. Although no database can be comprehensive, FlyAtlas provides the best available quantitative expression data for a range of insect neural, epithelial, and other tissues. The database was searched on the terms “Cyp|P450”, and “glutathione transferase|glutathione-s-transferase|GST”, and the data was transcribed to tables. Where there was detectable expression in at least some tissues, the tissue with highest expression was marked in bold (see Tables 1 and 2), to aid study.

View this table:
Table 1.

The tubule is a major organ for detoxification and metabolism

View this table:
Table 2.

Glutathione transferase family


Expression profiles of xenobiotic metabolism gene families.

Recently, a comprehensive microarray-based atlas of adult gene expression in multiple Drosophila tissues has become available (http://flyatlas.org) (6), and so it is possible to establish whether detoxification and metabolism genes are indeed widely expressed “housekeeping” genes or have more specific patterns of expression. The results (Tables 1 and 2), provide a uniquely comprehensive overview of the P-450 and GST gene families (those mainly implicated in xenobiotic metabolism), and, while confirming previous reports of expression in midgut and Malpighian tubules, provide detailed evidence that the majority of the genes in each family are expressed in defined subsets of tissues. Remarkably, none of the P-450 family, and only a few of the GST family (for example gstd1) show ubiquitous, “housekeeping”-like expression patterns; a very few show widespread expression except for a particular tissue (for example, Cyp6d5, which is very abundant in all tissues except ovary); the large majority are confined to epithelial tissues (midgut, hindgut, Malpighian tubules); and about half are extremely specific, showing overwhelming expression in a single tissue. For example, Cyp6g2 is unique to brain and head, whereas Cyp6d4 is predominantly expressed in midgut, Cyp6a18 in tubules, and Cyp6a19 in ovary.

Unusual P-450s.

The P-450 gene family has been described as having a relatively large number of pseudogenes, given that such genes are relatively rare in the genome as a whole (13). However, as the Drosophila genome sequence has been successively refined, many gene calls have been revised, leading to several pseudogenes being reannotated as expressed genes. Intriguingly, one of the P-450 pseudogenes (Cyp6a16psi) shows strong tissue-specific expression in tubule (Table 1), although the other pseudogenes are transcriptionally almost silent. Although still annotated in Flybase as a pseudogene with at most a 392-amino-acid ORF (with a truncated match to the pfam00067 p450 conserved domain), Swiss-Prot accession Q9VMN8 reports a 496-amino-acid ORF for cyp6a16 that includes a full-length match to the pfam p450 domain. It seems likely that Cyp6a16 is a genuinely transcribed gene; its highly specific expression pattern may have militated against its earlier identification in expressed sequence tag (EST) projects. By contrast, no expression was detected for the pseudogenes Cyp6t2psi or Cyp6a15psi. The position with respect to Cyp9f3psi remains unclear; it is clearly expressed, both from our data and from Berkeley Drosophila Genome Project (BDGP, http://www.fruitfly.org/) embryonic in situ data, but it lacks the COOH-terminal heme domain and abuts so closely between its neighbors that there is little space in which to discover such a domain. It is also extremely close (<10 kb) to Cyp9f2, to which it is almost identical over shared regions. On balance, it does seem to be a transcribed pseudogene based on a recent local duplication.

Another gene, Cyp4g1, which is the best represented P-450 in the Drosophila EST database (http://drnelson.utmem.edu/droso.special.P450s.html), has a whole-fly signal that massively exceeds any individual tissue studied, implying intense expression in some tissue yet to be identified (Table 1).

Tissue specificity.

Overall, the picture thus emerges that, as well as evolving new genes to deal with specific threats (for example, the expansion of the CYP6A family in Lepidoptera exposed to plant secondary metabolites; Ref. 9), these metabolic enzymes perform highly tissue-specific tasks in Drosophila. This has important implications for our understanding of xenobiotic handling, because the first tissue to encounter a xenobiotic will vary with the route of administration: ingested toxins, like secondary plant metabolites or systemic insecticides, will move from the impermeable foregut to the highly permeable midgut, whereas topical xenobiotics will meet the cuticle, then, on entering the hemolymph, the fat body, and tubules. There could also be a life-stage effect, because the fat body and tubules both ramify throughout the hemocoel of larval Drosophila, whereas fat body is largely confined to the head in adults, leaving the tubules with a more dominant presence.

Assessment of relative expression in fat body and tubules in adults must be made indirectly, because the data set does not include a specific fat body sample (its relative inaccessibility in adults made it too difficult to dissect to the quality required for FlyAtlas). However, the head sample can be taken as an indicator of fat body activity, as this is where fat body is most abundant in the adult fly. [To confirm that the head sample acted as a reasonable surrogate for adult fat body, we surveyed known immunity genes which are considered to be markers of fat body (14, 20) and found that they showed very high levels of expression in the head sample: drosomycin-5 has a signal of 4464, and turandot A, a signal of 3847; both signals in head are at least 10 times higher than for any other tissue.] However, FlyAtlas contains data for larval fat body and larval tubule, and so permits a general comparison with other tissues, although across different life stages. It is thus significant that each of the loci implicated in xenobiotic resistance (Cyp6g1, Cyp6a2, Cyp6a8, Cyp6a9, Cyp4e2, Cyp12a4) all show maximum signals in the adult tubule. Consistent with these results, flies transgenic for a Cyp6g1 promoter>GFP reporter construct showed highest levels of expression in tubule (Fig. 1). In addition, CG16936, the Drosophila gene most similar (BLASTP e value 1e-58) to Aedes GSTE2 (another insecticide-resistance locus, Ref. 11) is also most abundant in tubule. Several lines of evidence thus suggest that the tubule may thus be the dominant tissue for xenobiotic metabolism in the adult. This has implications for whole-organism microarray studies of insecticide resistance, as such studies are biased to follow concerted changes in expression across multiple tissues and so are predicated on detecting global changes in expression of widely expressed genes (6).

Fig. 1.

Mapping expression of Cyp6g1 expression within adult Drosophila. The control region of Cyp6g1, corresponding to bases −1 to −1200 from the transcriptional start site, was cloned from Canton S flies and placed upstream of an enhanced green fluorescent protein (GFP) open-reading frame, and transformed into Drosophila via the germ line. Expression in adult Canton S flies was mapped by fluorescence microscopy of acutely dissected live tissues. A: punctate expression in cuticle. B: salivary glands. C: proventriculus and part of midgut. D: midgut. E: intense staining in Malpighian tubule main segment. F: no expression in tubule initial segment. G: expression in lower tubule and ureter. H: rectum. I: fat body.

A major role for tubule?

Given that several genes and multiple tissues might impinge on insecticide detoxification, how can the importance of the tubule be tested experimentally? We decided to adopt a systems approach in Drosophila, as it is possible to modulate expression levels of particular genes in a cell-specific fashion (2). Thus, if Cyp6g1 expression in the Malpighian tubule is limiting for insecticide susceptibility to xenobiotics, then small perturbations in levels of Cyp6g1 in just the tubule should impact detectably on survival of the whole organism. Such experiments are possible in Drosophila by using the GAL4/UAS binary expression system (2).

Several transgenic Drosophila UAS-RNAi lines were generated, and further lines obtained from the NIG-Fly project (http://shigen.lab.nig.ac.jp/fly/nigfly/). They were crossed to the c42 GAL4 driver line that drives expression specifically in tubule principal cells (3) and screened by qPCR for effective suppression of Cyp6g1 expression. One such line (UAS-Cyp6g1-RNAi-I) proved highly effective; when driven in Malpighian tubules with c42, a principal-cell-specific GAL4 driver line (3, 23, 24), Cyp6g1 mRNA expression (measured by qPCR) was 0.64% of the level in parental UAS-Cyp6g1-RNAi-I flies. Such knockdown flies are significantly (P < 0.0001) more susceptible to DDT intoxication, showing clearly that reduction of Cyp6g1 expression in just the principal cell of just the Malpighian tubule significantly reduces insecticide resistance of the whole organism (Fig. 2A).

Fig. 2.

Cell-specific manipulation of Cyp6g1 expression in intact flies. A: knockdown of Cyp6g1 in just principal cells of Malpighian tubules confers increased insecticide susceptibility to the whole animal. Flies were exposed to 1,1,1-trichloro-2,2-bis(4-chlorophenyl)ethane (DDT), and was mortality calculated over 24 h. Principal cell-specific knockdowns [c42/UAS-Cyp6g1-RNAi-(I)] were compared with the parental UAS-Cyp6g1-RNAi(I) line. The slopes are not significantly different, but the intercepts differ significantly (P = 0.0001). B: overexpression of Cyp6g1 in just principal cells of Malpighian tubules confers increased insecticide resistance to the whole animal. Flies were exposed to DDT, and mortality was calculated over 24 h. Principal cell overexpressors (c42/UAS-Cyp6g1) were compared with the parental UAS-Cyp6g1-3A line. The slopes are not significantly different, but the intercepts differ significantly (P = 0.0032). C: as B, except that an independent UAS line, 8A, was used. The slopes are not significantly different, but the intercepts differ significantly (P = 0.0001).

Conversely, does localized overexpression directly confer increased survival? Two Cyp6g1 overexpression UAS lines, Cyp6g1-3A and Cyp6g1-8A, were also crossed with the c42 driver, and overexpression of Cyp6g1 mRNA validated by qPCR. Both induced overexpression compared with the UAS parental line: UAS-3A by 4.0×, and for UAS-8A by 3.5×. As predicted, overexpression of either UAS-Cyp6g1 construct in just tubule principal cells confers increased survival upon insecticide challenge (Fig. 2, B and C). These results are exciting and clearly show that the resistance of a whole, live insect to insecticide is related to the expression level of Cyp6g1 in just the tubule principal cells (less than 0.1% of the insect body mass).

If the tubule were limiting in insecticide metabolism, then selective knockdown of Cyp6g1 in other tissues should not show a pronounced effect. Accordingly, we drove the Cyp6g1-RNAi construct in two plausible tissues, the fat body and brain, and assessed mortality after DDT exposure. Remarkably, although each tissue is massive by comparison with the tubule, neither was limiting for survival on exposure to DDT (Fig. 3). So although Cyp6g1 is expressed in several tissues, the fact that manipulation of expression levels in just the tubule modulates the survival of the insect implies that the tubule is the limiting tissue in control of mortality to DDT and thus, by inference, other insecticides. Classically, this tissue has been seen as critical for osmoregulation, analogous to the vertebrate kidney. However, the enrichment of the tubule transcriptome for broad spectrum organic solute transporters and for metabolic enzymes (26) suggests that it also plays roles associated with those of vertebrate liver. Additionally, the tubule has been shown to play a robust role in immune response (15, 21). Insect organ systems may thus perform roles analogous to several distinct vertebrate organs. As the tubules generate a primary urine at an exceptional rate (19), they are ideally placed to sense and respond to a range of external insults. Our results thus emphasize the importance of considering specific tissues when interpreting the whole organism's response to xenobiotics.

Fig. 3.

Selective knockdown of Cyp6g1 in brain and fat body does not increase sensitivity to DDT. The Cyp6g1-RNAi(I) line was driven in either central nervous system (A) or fat body (B), using the GAL4 driver lines Elav-GAL4 or c564, respectively, and mortality was assessed by comparison with parental lines. In A, the driven RNAi regression differs significantly from the GAL4 parental in intercept and from the UAS parental in slope; however, it actually survives slightly better than either. In B, there is no significant difference in slope or intercept of the regressions.

We propose that some of the conflicts in the literature may turn on the differing conditions applied. For example, fat body is far more extensive in larval than adult Drosophila, and so the genes that are particularly abundant in fat body may play a more significant role in larval toxicity studies. We have direct experimental evidence to support this model; the key insecticide resistance gene, Cyp6g1, shows a signal of 5542 in adult tubule, but only 88 in larval tubule; by contrast, the signal in larval fat body is 4886 (Table 1). Major Cyp6g1 activity thus appears to reside in adult tubule, but larval fat body. For ingested insecticides, the midgut is likely to be the first line of defense; however, the tubules are more likely to handle topically applied agents that appear in the hemocoel. In addition, our experiments were performed on susceptible (Canton S), naive insects, and so measure the initial response to first presentation of insecticide, rather than monitor the development of long-term resistance. Nonetheless, our results have bearing on xenobiotic handling in resistant flies, because global (whole organism) Cyp6g1 levels are known to be upregulated in DDT-resistant flies (10). Furthermore, in resistant flies the transposon insertion drives overexpression in proportion to the basal expression levels in different tissues (7); in other words, the already high tubule levels of Cyp6g1 in susceptible flies are higher still in resistant strains, and so the tubule is likely to play a role in resistance mechanisms as well as detoxification. Ultimately, it would be interesting to manipulate Cyp6g1 transgenically in resistant strains, to test this idea.

Now that a major tissue responsible for baseline insect defense has been identified, understanding of tubule function may thus help both in modeling the fate of xenobiotics such as insecticides and in designing new strategies that target tubule function to enhance susceptibility to known insecticides, so controlling or blocking insecticide detoxification.


We are most grateful to Dr. Shireen Davies for critical reading of the manuscript.


  • Address for reprint requests and other correspondence: J. A. T. Dow, Division of Molecular Genetics, Univ. of Glasgow, Glasgow G11 6NU, UK (e-mail: j.a.t.dow{at}bio.gla.ac.uk).

    Article published online before print. See web site for date of publication (http://physiolgenomics.physiology.org).


View Abstract