Sulphonylurea sensitivity and enriched expression implicate inward rectifier K⁺ channels in Drosophila melanogaster renal function

Fluid-secreting epithelia must achieve high fluxes of cations, anions,water and other solutes. A striking exemplar of such transport is offered by the Malpighian (renal) tubules of insects, which, when maximally stimulated,are capable of transporting their own cell volumes of fluid every 10 s - rates unmatched elsewhere in biology (Dow and Davies, 2003; Maddrell,1991). In insects, an apical plasma membrane V-ATPase energises a proton/alkali-metal cation exchanger, thought to prefer potassium over sodium(Wieczorek et al., 1999). Potassium enters the cell from the basolateral surface through a combination of a ouabain-sensitive Na⁺/K⁺-ATPase(Torrie et al., 2004), a Na⁺/K⁺/2Cl^- cotransporter(Ianowski and O'Donnell, 2004)and barium-sensitive potassium channels(Beyenbach and Masia, 2002; Masia et al., 2000; Weltens et al., 1992); their relative importance may vary between insects. Present estimates suggest that,in Drosophila, each contributes approximately equally to the K⁺ flux (Torrie et al.,2004). At a transepithelial level, therefore, potassium appears to be actively transported. The route for the passive anion shunt is controversial; but in Drosophila, it is regulated by stellate cells(Denholm et al., 2003; O'Donnell et al., 1996; Radford et al., 2002); there is evidence that these cells also contain aquaporins(Dow et al., 1995; Kaufmann et al., 2004), so providing a low-resistance shunt pathway for both anions and water.

Drosophila melanogaster, despite its small size, offers unique advantages for the study of epithelial physiology, and accordingly, the molecular basis of D. melanogaster renal function is now understood in considerable detail (Dow and Davies, 2001, 2003). Among the tools available to the Drosophila community is a comprehensive Affymetrix microarray, which allows gene expression to be studied quantitatively over space and time. A recent such study, comparing adult Drosophilatubule with whole flies, identified genes that are either abundantly expressed or highly enriched in tubule (Wang et al.,2004). Thus, this study both identifies candidate genes for known transport and signalling functions and suggests new, previously unsuspected functions. Among the significant findings of this study was the observation that, of the very large K⁺ channel gene family of Drosophila, all three of the inward rectifier Drosophilagenes (ir, irk2 and irk3) were both highly abundant and strongly enriched (to much greater extents than any other K⁺channels), clearly implicating inward rectifier potassium channels (IRKs) in epithelial function.

Mammalian IRKs form both homo- and hetero-tetramers and are sensitive to barium and to the antidiabetic sulphonylureas such as glibenclamide(Bryan and Aguilar-Bryan,1997). Possible orthologues expressed in mammalian kidney include Kir6.x and Kir1.x. These channels are often associated with ABC transporters,such as cystic fibrosis transmembrane conductance regulator (CFTR) and sulphonylurea receptor (SUR), and complexes with these receptors render the channels more sensitive to glibenclamide(Ruknudin et al., 1998).

The highly upregulated expression pattern of the IRKs prompted our investigation into the effect of sulphonylureas on the Malpighian tubules. Not only do sulphonylureas inhibit fluid secretion, but glibenclamide blocks organic dye excretion by the tubule, and glibenclamide itself is both excreted and modified by the tubule.

Oregon R flies were kept on standard medium in tubes at 25°C, 12 h:12 h photoperiod and 55% relative humidity. For dissection, flies were anaesthetized by chilling on ice and decapitated before dissecting out tubules in Schneider's medium (Invitrogen, Paisley, Scotland, UK). All chemicals and drugs were obtained from Sigma, unless otherwise stated.

Fluid secretion assays were performed as described(Dow et al., 1994), taking readings every 10 min. Glibenclamide was added after the tubule basal secretion rate had been measured for 30 min. The secretion rate continued to be recorded for a further hour. Inhibition was calculated as the difference between the basal rate (measured over the first 30 min) and the secretion rate at 50 min. To study the effects of glibenclamide on neuropeptide-stimulated secretion, glibenclamide was added at 30 min, and the diuretic peptides Drosophila leucokinin (dLK; Radford et al., 2002; Terhzaz et al., 1999) and Drosophila corticotropin releasing factor-like peptide (dCRF, DH; Cabrero et al., 2002) were added at 60 min, then the secretion rate recorded for a further hour. In this way, the impact of sulphonylureas could be assessed both on resting and maximally stimulated fluid secretion.

qPCR was performed as described previously(McGettigan et al., 2005). Briefly, mRNA was prepared from 7-day-old Oregon R flies or their tubules using Qiagen RNAeasy column, following the manufacturer's methods (Qiagen,Crawley, West Sussex, UK). Reverse transcription was carried out using Superscript II (Invitrogen) using oligo-dT primer. For each sample, 500 ng of cDNA was added to 25 μl of SYBR Green reaction mix (Finnzymes, GRI,Braintree, Essex) and appropriate primers(Table 1). An Opticon 2 thermocycler (MJ Research, now Bio-Rad, Hemel Hempstead, Herts, UK) was set as follows; the recommended 15 min HotStart Taq activation time, then 40 cycles of (denaturing at 94°C for 30 s, annealing for 30 s at primer-dependent temperature, 30 s of extension at 72°C, and 10 s absorption reading at 80°C), followed by a 5 min final extension at 72°C and a melting curve from 70°C to 90°C. The ribosomal rp49 gene was used as a standard in all experiments. Data were then expressed as fold difference of tubule cDNA compared with whole fly cDNA (± s.e.m.).

Tubules were dissected, fixed and subjected to in situhybridization according to published protocols(Siviter et al., 2000) and the Berkeley Drosophila Genome Project (BDGP) 96-well in situprotocol(http://www.fruitfly.org/about/methods/RNAinsitu.html). In situ probes were directed towards the 3′ UTR of each gene to minimise cross-hybridisation between related genes. PCR products derived from the 3′ UTR of each gene were cloned into pBluescript or pCRII vectors(Invitrogen), and DIG-labelled RNA probes were generated by in vitrotranscription. Adult tissues comprising gut, testes, ovaries and Malpighian tubules were dissected in Schneider's medium (Invitrogen) and placed into wells of a Millipore 96-well plate (MAGVN22 or MAGVS22) with 100 μl Schneider's medium. Schneider's medium was removed using a vacuum pump, and postfix solution [10 mmol l^-1 potassium phosphate buffer (pH 7.0)containing 140 mmol l^-1 NaCl, 0.1% Tween 20 and 5% v/v formaldehyde] was added for 20 min, followed by three washes with PBT [10 mmol l^-1 potassium phosphate buffer (pH 7.0) containing 140 mmol l^-1 NaCl and 0.1% v/v Tween 20]. The tissues were incubated with proteinase K in PBT (4 μg ml^-1) for 3 min at room temperature. The reaction was stopped with two washes of PBT containing 2 mg ml^-1 glycine. The samples were washed twice with PBT before incubating with postfix for a further 20 min at room temperature. The tissues were washed with five changes of PBT, followed by one wash with 50%hybridisation buffer [5× SSC containing 50% v/v formamide, 10 mmol l^-1 kPO₄, 140 mmol l^-1 NaCl, 1 mg ml^-1 glycogen, 0.2 mg ml^-1 sheared salmon sperm DNA and 0.1% v/v Tween 20 (pH 7.0)] plus 50% PBT. The samples were washed once with hybridisation buffer, prior to a 1 h preincubation with hybridization buffer at 55°C, and subsequently incubated for 43 h at 55°C with 100 μl of hybridisation buffer containing 10-500 ng of either the sense or antisense riboprobe, taking care to seal the wells with Parafilm™ to prevent evaporation. Following hybridisation, the samples were washed four times with hybridization buffer at 55°C, followed by a final wash overnight with hybridization buffer at 55°C. Samples were washed once with 50% v/v hybridisation buffer and 50% v/v PBT, followed by four washes with PBT and then incubated overnight at room temperature with 100 μl of pre-absorbed alkaline phosphatase-conjugated anti-digoxigenin Fab fragment (Roche Molecular Biochemicals, Lewes, East Sussex, UK) diluted 1:2000 with PBT. The unbound antibody was removed with extensive washing in PBT (at least 10 times for∼5-10 min). The samples were incubated with DIG detection buffer [100 mmol l^-1 Tris-HCl (pH 9.5), 100 mmol l^-1 NaCl, 50 mmol l^-1 MgCl₂] for 5 min then repeated again. The colour reaction was initiated by the addition of DIG detection buffer +5-bromo-4-chloro-3-indolyl phosphate (BCIP) and nitroblue tetrazolium (NBT)and left for 10 min-2 h at room temperature. Development was stopped with extensive washing with PBT containing 50 mmol l^-1 EDTA, and the tissues were removed from the wells, mounted on slides with 70% glycerol and viewed on an Axioscope microscope equipped with an Axiocam imaging system(Zeiss, Welwyn Garden City, UK).

Tubules were dissected and mounted on poly-lysine-coated slides in Schneider's medium. Glibenclamide labelled with Texas Red (Molecular Probes,now Invitrogen) was added to a final concentration of 0.1 mmol l^-1. A time series was generated under fluorescence with a Zeiss 510 Meta Confocal system. All images were taken at the same gain and exposure and processed using Axiovision 3.0.6 software.

Glibenclamide, at 10^-6-10^-3 mol l^-1, was labelled with 0.05 μCi ¹²⁵I (Amersham Pharmacia, Amersham, UK,specific activity, 2000 Ci mmol l^-1) and added to tubule secretion assays, performed as described above. Secreted fluid was collected for 1 h,the volume measured and counted in Optiflow SAFE scintillant (Fisher Scientific UK, Loughborough, Leicestershire, UK). Transport ratios were calculated as the ratio of specific activities of secreted drop: reservoir bubble (values >1 thus imply concentration of the label by the tubule).

Samples of 1 μl of both authentic [¹²⁵I]glibenclamide (0.05μCi) and secreted fluid, obtained by the same method as the glibenclamide transport assay, were dried onto a 10×20 cm Polygram Sil G/UV254 plate(Machery-Nagel GmbH, Düren, Germany) and run out using an eluent of 85%v/v ethanol, 15% v/v PBS (pH 7) and 0.1% w/v SDS. Plates were visualized with a Fuji PhosphorImager.

Where errors are shown, these represent the standard error of the mean(s.e.m.). Where appropriate, the significance of differences was tested with Student's t-test (two-tailed), taking a critical level of P<0.05.

Although microarray data provide a valuable first-pass survey of gene expression levels, it is prudent to validate array `hits' independently before engaging in detailed study. The mean ratio results from qPCR analysis(Table 2) showed that all three IRK-encoding genes were enriched in tubule, compared with whole fly, cDNA,confirming previous Affymetrix microarray results(Wang et al., 2004). While the actual values produced from the qPCR and Affymetrix MAS5 analyses differ slightly, the relative levels between the genes are maintained; this represents further validation of the microarray dataset beyond the 12 genes originally subjected to qPCR (Wang et al.,2004).

One of these genes, irk2, was known to have two splice forms(Fig. 1), although the Affymetrix probes (against the 3′ end of the mRNA) would not have distinguished them. qPCR measurement allowed the identification of irk2-RA as the major transcript (Table 1). These results also suggested the presence of an additional splice form,verified by cloning and sequencing the PCR product. This new splice form,which will be denoted RC, has an extra 21 bases prior to the translated region of the RA splice form (Fig. 1). It appears to be a low-abundance RNA product in the tubules, comprising only 6.2±0.6% of the expression of Irk2-RA. Irk2-RC did not reliably produce a qPCR product from the whole-fly cDNA because of its very low copy number. No other D. melanogaster IRK-encoding genes are known to have alternative splice forms.

In situ hybridisation confirmed that that all three IRKs were expressed abundantly in the tubule (Figs 2, 3, 4). The tubule is divided into six regions, with two predominant cell types(Sözen et al., 1997). For the most abundant and enriched gene, irk3(Table 2), there was widespread expression in all regions except the initial and transitional segments of the anterior tubules; interestingly, the equivalent regions of the posterior tubules stained throughout (Fig. 4). Additionally, within the main (secretory) segment of the tubule, only the principal cells and not the stellate cells expressed irk3. This is consistent with the principal cell being the major route for active potassium flux through the tubule: V-ATPases are also highly enriched in this cell type (Sözen et al., 1997), as in another dipteran, Aedes aegypti(Weng et al., 2003). Irk3 was also observed in developing oocytes (Fig. 4F).

The other two genes gave fainter signals, but both were still clearly expressed in tubule principal cells. However, ir was most abundantly expressed in the ureter (Fig. 2), whereas irk2 was most abundant in spermatheca(Fig. 3A). It is thus clear that, although all genes contribute to main segment function, ir and irk3 are also found in lower tubule, a region associated with fluid reabsorption (O'Donnell and Maddrell,1995). IRKs may thus play a role both in secretion and reabsorption in distinct spatial regions of the tubule. As the genes are extensively co-expressed in the tubule, and as IRKs can form hetero-tetramers,there is a possibility that the properties of the channels in vivomay not resemble those measured by heterologous expression in vitro(Doring et al., 2002).

As the electrochemical gradient at the apical membrane does not favour net K⁺ secretion (Beyenbach et al.,2000), it is also likely that IRKs will be located on the basolateral membrane. Could they then constitute a major route for K⁺ entry?

The cardinal inhibitors of IRKs are the antidiabetic sulphonylureas such as glibenclamide (Inagaki et al.,1996). Consistent with a major role for IRKs in tubule, fluid secretion was found to be inhibited by sulphonylureas. Glibenclamide inhibited fluid secretion with an IC₅₀ of 0.78±0.03 mmol l^-1 (Fig. 5). Maximal inhibition was observed 20-40 min after the drug was added, depending on concentration. The IC₅₀ value is comparable with those found in mammalian Kir1.1 channels in the absence of SUR2B(Konstas et al., 2002) and Tenebrio molitor tubules (Wiehart et al., 2003).

To characterize the sulphonylurea sensitivity of the tubule further,tolbutamide, minoxidil and diazoxide were also tested and shown to inhibit secretion (Fig. 6). Approximate IC₅₀ values for these drugs were estimated as 5 mmol l^-1, 0.01 mmol l^-1 and 0.1 mmol l^-1,respectively. Tolbutamide has a lower binding efficiency to SURs than glibenclamide (Dorschner et al.,1999) and is often shown to require approximately 10 times the concentration of glibenclamide in order to produce the same effects. Although Minoxidil is a potassium channel activator, it has also been shown to decrease urine produce in humans (Wang,2003), as was found here. It is thus clear that tubule secretion is inhibited by a range of sulphonylureas, with very similar pharmacology to humans. On balance, however, the relatively high IC₅₀s suggest that either the tubule IRKs are of the lower affinity class that do not need to associate closely with an SUR or that there are additional targets for sulphonylureas in tubule.

Diflubenzuron is a competitor for glibenclamide binding to SUR in Drosophila (Abo-Elghar et al.,2004). In secretion assays where diflubenzuron was added, followed 30 minlater by glibenclamide, no change in secretion rate was observed(Fig. 7). Inhibition by glibenclamide was unaffected, suggesting that glibenclamide is not binding to SUR. Microarray results (Wang et al.,2004) showed that SUR was not abundant in the Malpighian tubule;however, it was envisaged that there might be a functional homologue. The results from the diflubenzuron and IC₅₀ experiments suggest that the IRK channels in the tubule do not form complexes similar to the Kir6.x/SUR complexes in mammals, although they do not exclude the possibility that glibenclamide is acting on a target distinct from IRK, such as CFTR. However,no Drosophila orthologues for CFTR were identified in a survey of Drosophila ABC transporters (Dean et al., 2001).

Amaranth, an innocuous red food dye, was observed to be concentrated in secreted fluid (Maddrell and Gardiner,1975) and is widely used to help view the secreted bubbles in secretion assays (Dow et al.,1994); however, the mechanism of active amaranth secretion is unknown. During the secretion assay, amaranth transport was observed to be inhibited by glibenclamide. Amaranth transport was abolished after 35 min when the concentration of glibenclamide was greater than 0.1 mmol l^-1,and some inhibition was observed above 0.01 mmol l^-1(Fig. 8). Inhibition of amaranth transport thus occurs at glibenclamide concentrations tenfold lower than the IC₅₀ for glibenclamide action on fluid secretion, and so shows that the two effects are independent actions of glibenclamide. Glibenclamide was further found to block Orange R and Phenol Red transport(data not shown). The inhibition of amaranth transport suggested that glibenclamide may be preferentially transported via the same mechanism as amaranth.

Further secretion assays demonstrated that Phenol Red and amaranth did not impact on the inhibition of fluid secretion by glibenclamide. In secretion assays where no dye was added to the reservoir bubble, the reduction of secretion caused by glibenclamide was not significantly different(Fig. 9).

The serendipitous observation that glibenclamide blocked the transport of a wide range of organic dyes by tubule suggested a second action of sulphonylureas in tubule; an impact on organic solute transporters. We hypothesized that this might reflect competition for the same excretory transporters; tubules are known to express a huge repertoire of broad-specificity solute transporters(Wang et al., 2004). Accordingly, possible transport of glibenclamide was observed using confocal microscopy. The fluorescent glibenclamide appeared initially to adhere to the basolateral membrane, before entering the cells(Fig. 10). The fluorescent glibenclamide was observed in the cytoplasm of the principal cells; this occurs along the length of the tubule. The basolateral border between the principal cell and stellate cells also showed up clearly. There was a particularly strong signal from the initial segment of the anterior tubules, a region that does not contribute to fluid secretion(O'Donnell and Maddrell, 1995)but that actively excretes calcium (Dube et al., 2000). At the urethra, fluorescent secreted bubbles were visible.

The nature of the glibenclamide transport was investigated further using I¹²⁵-labelled glibenclamide. Measurements of the radioactivity in the secreted drops and reservoir bubbles revealed that the level of radioactivity was approximately 8-fold higher than the bathing solution in the secreted drops across all glibenclamide concentrations investigated(Fig. 11B). The rate of secretion rose with the glibenclamide concentration(Fig. 11A). The rate and ratio measurements show that glibenclamide transport certainly occurs and that the mechanism was saturated at the highest concentration we were able to employ. Thin layer chromatography (TLC) on the secreted and original radiolabelled glibenclamide showed that the secreted glibenclamide was modified(Fig. 11C). This modification precludes the conclusion that glibenclamide is actively transported, as we cannot confirm the existence of transport across an adverse electrochemical gradient. However, for practical purposes, glibenclamide is rapidly and avidly removed from the basolateral surface of the cell.

Glibenclamide, although a classical inhibitor of IRKs, has also been shown to act on a wide range of targets. For example, it blocks ABC transporters other than SUR, such as CFTR (Zhang et al., 2004). It has been argued to affect L-type calcium channels and the Na⁺/K⁺-ATPase(Lee and Lee, 2005) and to induce a chloride conductance in mitochondria that collapses the inner membrane potential (Fernandes et al.,2004). So, it is premature to ascribe the potent inhibition of tubule secretion by glibenclamide to blockade of IRKs. However, our confidence is increased by the fact that a range of sulphonylureas, and not just glibenclamide, are effective (Fig. 6). Is it possible to go further?

The effects of glibenclamide were compared with exposure of the tubule to K⁺-free saline. Drosophila tubule secretion has been reported to persist for some hours in K⁺-free saline(Linton and O'Donnell, 1999),so if the action of glibenclamide was at a site distinct from IRKs, then secretion in K⁺-free saline should still be sensitive to glibenclamide. However, in our experiments, K⁺-free saline led to a total inhibition of fluid secretion, with a time course very similar to that of glibenclamide addition, and so glibenclamide has no further effect on tubules in K⁺-free saline (Fig. 12). We thus consider that the action of glibenclamide is consistent with a run-down of cellular K⁺, similar to that produced by removing external K⁺, although, given the multiple potential targets of glibenclamide, we cannot rule out more complex possibilities.

Microarray analysis is a potent tool, but all the more so when it directs downstream functional analysis. Fortunately, the Drosophila renal tubule is an ideal tissue for such study. Microarray analysis clearly implicated IRKs in tubule function, and this paper has confirmed the abundance and enrichment of mRNAs for all three Drosophila IRKs and further localised them to the principal cell, which is known to be the site of active electrogenic cation transport. This builds on previous in situ data,which have variously described ir as widely expressed(MacLean et al., 2002),especially in embryonic salivary gland, and localises irk2 to embryonic hindgut and irk3 to embryonic Malpighian tubules(Doring et al., 2002). Our data show that, in adults, all three IRKs are present in tubule. As IRKs form both homo- and hetero-tetramers (Bryan and Aguilar-Bryan, 1997), there is thus ample scope for sophistication in tubule K⁺ channel expression. Interestingly, the three IRKs show overlapping, but distinct, patterns of expression; irk3 is expressed in main segment principal cells and lower tubule, implying a role in both fluid secretion and reabsorption.

As would be expected, fluid secretion was shown to be sensitive to a wide range of the antidiabetic sulphonylureas that are selective inhibitors of IRKs in vertebrates. The sensitivity to these drugs, however, was not sufficiently high to suggest that these IRKs are associated with Drosophila SUR or a related ABC transporter; in support of this, a known inhibitor of Drosophila SUR had no effect on fluid secretion. However, it was possible to completely inhibit fluid secretion with sulphonylureas(Fig. 6), whereas inhibitors of Na⁺/K⁺-ATPase and Na⁺/K⁺/2Cl^- co-transport achieve only partial inhibition of fluid secretion (Ianowski and O'Donnell, 2004; Torrie et al., 2004), and so, with this broad panel of selective inhibitors,we can confidently assert a major role for IRKs in tubule.

The extraordinary abundance of mRNA for ir, irk2 and irk3in tubule (Wang et al., 2004; Table 2) suggests an important role for these channels. The sulphonylurea pharmacology, although less conclusive, shows a similar effect to K⁺-free saline, suggesting a possible role for IRKs in the entry step. This is consistent with many models for tubule function, in which basolateral K⁺ channels have featured(Beyenbach et al., 2000; Dow and Davies, 2003; Wiehart et al., 2003). However, important basolateral roles have also been shown both for Na⁺/K⁺/2Cl^- co-transport and the Na⁺/K⁺-ATPase(Ianowski and O'Donnell, 2004; Torrie et al., 2004), and a small (9 mV) outward electrochemical gradient for K⁺ has been shown for the basolateral membrane of resting Drosophila tubule(Ianowski and O'Donnell,2004). Can this conflicting data be reconciled? We suggest that IRKs play only a minor role in basal secretion: their rectification properties imply that they are unlikely to let much K⁺ out, down its electrochemical gradient. However, when the tubule is maximally stimulated,and the apical electrogenic V-ATPase is driving K⁺ exit across the apical membrane, we suggest that basolateral IRKs allow K⁺ entry to support the active apical membrane. IRKs could thus be seen as adaptations to facilitate the remarkably high per-cell secretion rates that seem to be characteristic of insect tubules (Dow and Davies, 2003; Maddrell,1991). Consistent with this, all the sulphonylureas showed a more marked effect on stimulated, rather than basal, fluid secretion(Fig. 6).

We also observed two novel effects of glibenclamide: (1) the tubule itself both excretes and modifies the compound and (2) glibenclamide inhibits the excretion of a range of organic dyes. Although the sulphonylureas are valuable human therapeutic agents, the molecular mechanisms of their clearance are not understood in detail; for example, it is not clear whether SUR is itself a sulphonylurea transporter. However, our results may impinge on clearance mechanisms. Although we did not rigorously show competition between these solutes, it would be a very plausible explanation, and this paper does identify a range of interesting dyes that may help in such studies.

Demonstration of glibenclamide transport further raises the possibility that the apparent IC₅₀s for sulphonylureas may be distorted by rapid removal from the basolateral membrane infoldings of the principal cell,as has been shown for ouabain (Torrie et al., 2004). This would have the effect of protecting basolateral IRKs by only exposing them to a concentration of sulphonylureas that was significantly lower than the macroscopic bath concentration, thus producing anomalously high IC₅₀s. We have not yet identified an inhibitor of glibenclamide transport, so we cannot unmask a higher affinity effect of glibenclamide, as was possible for ouabain(Torrie et al., 2004). We must thus be cautious about asserting an accurate IC₅₀ for glibenclamide; this will need to be established by heterologous expression in culture, where such confounding elements do not appear.

Comparison of the Drosophila and Anopheles genome projects reveal that IRKs are very similar between these dipteran species;indeed, there appears to have been a gene expansion in Anopheles(McCormack, 2003). Glibenclamide sensitivity (at relatively high concentrations: 5 mmol l^-1) has also been documented in Tenebrio molitor, a beetle (Wiehart et al., 2003). In Tenebrio, the effect of glibenclamide was similar to that of barium, another characteristic of IRKs. Barium sensitivity has also been widely documented in tubules of Drosophila(Ianowski and O'Donnell, 2004; Wessing et al., 1993), the locust Schistocerca gregaria(Hyde et al., 2001) and the yellow fever mosquito, Aedes aegypti(Beyenbach and Masia, 2002)Consistent with this, Drosophila IRKs reveal weak barium sensitivity when expressed heterologously in S2 cells(Doring et al., 2002). Our results, showing sensitivity of fluid secretion to a spectrum of sulphonylurea antidiabetics, and localising all three IRKs to the same cell type as the apical V-ATPase, may thus have broad scope.

This work was supported by a studentship and grants from the UK's Biotechnology and Biological Sciences Research Council (BBSRC). We are grateful to Dr C. T. Brett for the loan of thin layer chromatography equipment. We would like to thank a referee for suggesting the K⁺-free experiment.