Transepithelial transport of fluorescent p-glycoprotein and MRP2 substrates by insect Malpighian tubules: confocal microscopic analysis of secreted fluid droplets

P-glycoprotein (P-gp) and multidrug resistance-associated proteins (MRPs)are integral membrane proteins that act as ATP-dependent efflux pumps in tissues such as the gut, kidney, liver and blood–brain barrier. They have been implicated in resistance of many organisms to a vast and chemically diverse range of toxic molecules. P-gps are the products of multidrug resistance (MDR) genes and they transport neutral or cationic amphiphilic compounds. MRPs transport neutral, cationic or anionic compounds and are also important in the transport of anionic conjugates of amphiphilic compounds with glutathione, glucuronide or sulphate(Bard, 2000).

P-glycoproteins in insects merit study because such transporters may contribute to insecticide resistance. Organochlorine and organophosphorous pesticides such as chlorpyrifos bind to P-gp, and exposure to such compounds increases MDR gene expression(Bain and LeBlanc, 1996; Lanning et al., 1996). Exposure to the P-gp inhibitor verapamil increases the toxicity of ivermectin in chironomids (Podsiadlowski et al.,1998) and the toxicity of three insecticide classes (cypermethrin,ivermectin and endosulphan) in mosquitoes(Buss et al., 2002).

A few studies have examined the possible role of P-gps and MRPs in the Malpighian tubules (MTs) of insects. Although there are, to date, no functional studies of P-gps in the MTs of the fruit fly Drosophila melanogaster, three MDR genes have been identified(Wu et al., 1991; Gerrard et al., 1993), and deduced amino acid sequences are ∼40% identical to those of mammalian homologs. P-glycoprotein-like transporters in the MTs of the tobacco hornworm, Manduca sexta, transport P-gp substrates such as nicotine and vinblastine into the lumen, and transport is reduced by the P-gp inhibitor verapamil (Gaertner et al.,1998). Immunofluorescence studies have also indicated the presence of P-gp-like proteins in the MTs (Murray et al., 1994). However, although fluorescent P-gp substrates such as daunorubicin (daunomycin) and rhodamine 123 are taken up into the tubule cells of Manduca, these fluorochromes do not appear in the lumen of isolated tubules examined using confocal laser scanning microscopy (CLSM; Gaertner and Morris, 1999). These authors suggested that reflection, refraction or absorption of exciting or emitted wavelengths by structures such as uric acid crystals in the lumen may have prevented detection of the fluorescent compounds in the lumen.

MRPs have been identified in the MTs of the cockroach Periplaneta americana and the cricket Acheta domesticus (Karnaky et al., 2000, 2001, 2003). MTs of both species transport the fluorescent MRP2 substrate Texas Red (sulphorhodamine 101)through the cells and into the lumen. Confocal microscopic studies of isolated MTs indicate that the concentration of the fluorochrome in the lumen is dependent upon metabolism and is inhibited by the non-fluorescent MRP2 substrate chlorodinitrobenzene. An ortholog of human MRP genes has been identified in Drosophila, although transport of MRP substrates by the MTs has not been tested. The Drosophila MRP (dMRP)gene contains 19 exons that are alternatively spliced, resulting in multiple isoforms (Tarnay et al.,2004).

The present study describes secretion of fluorescent P-gp and MRP substrates by the MTs of two insect species. Isolated tubules were set up in Ramsay secretion assays and nanolitre samples of secreted fluid were collected in optically flat glass capillaries. The concentration of fluorescent P-gp and MRP substrates was determined using CLSM. The analysis of secreted droplets circumvents many of the problems associated with analysis of isolated tubules,in which the presence of opaque concretions in the cells and/or lumen interferes with CLSM. In addition, measurements of fluid secretion rate during flux measurements permit non-specific toxicity of inhibitors of P-gps and MRPs, or of high concentrations of the fluorescent compounds themselves, to be distinguished from specific inhibition of multidrug-resistant transporters. We report here the first measurements of the transepithelial flux for P-gp and MRP2 substrates across isolated MTs. Flux is calculated as the product of fluid secretion rate, determined in the Ramsay assay, and the concentration of the compound of interest in the secreted fluid. We have determined kinetic parameters (J_max and K_t) for transport of P-gp and MRP2 substrates. We also show that the method is suitable for fluorescent substrates of other transporters, such as those involved in transepithelial secretion of organic anions such as fluorescein and organic cations such as quinacrine.

Drosophila melanogaster (Meigen) were raised on standard yeast medium using protocols described by Ashburner(1989). Malpighian tubules were dissected under saline from adult females aged 3–14 days, as described by Dow et al. (1994). Experiments were performed at room temperature (17–23°C). The control saline contained (in mmol l^–1); NaCl (117.5), KCl(20), CaCl₂ (2), MgCl₂ (8.5), NaHCO₃ (10.2),NaH₂PO₄ (4.3), Hepes (8.6), l-glutamine (10)and glucose (20). Saline was titrated to pH 7 with NaOH. A sodium-free saline contained N-methyl d-glucamine chloride (117.5), KCl (5),KHCO₃ (10.2), KH₂PO₄ (4.3) and the same concentrations of CaCl₂, MgCl₂, Hepes, l-glutamine and glucose as control saline. Na-free saline was titrated to pH 7 with HCl.

Crickets (Teleogryllus commodus Walker) were obtained from Biosuppliers (Auckland, New Zealand) and were maintained on ground Purina rat chow and water ad libitum, supplemented with lettuce leaves. Tubules were dissected from last instar larvae and adult males under saline containing(in mmol l^–1) NaCl (100), KCl (8.6), CaCl₂ (1.5),MgCl₂ (8.5), NaHCO₃ (4), NaH₂PO₄(4.0), Hepes (25), l-glutamate (10), sucrose (56) and glucose (24). Saline was titrated to pH 7 with NaOH.

Isolated tubules of each species were transferred to 20 μl droplets of saline under paraffin oil for Ramsay assays. One tubule of each pair of isolated Drosophila tubules was pulled out of the bathing droplet and wrapped around a steel minuten pin (0.15 mm diameter) stuck into the Sylgard™ bottom of the assay dish. The lower tubule and ureter were positioned outside of the bathing saline, so that the composition of the secreted fluid was determined by transport activity of the main segment only. The lower tubule was readily identified by the absence of stellate cells. Secreted droplets formed at the end of the ureter and were collected after 45–60 min with a fine glass probe. Droplet diameters (d) were measured using an ocular micrometer, and droplet volume (nl) was calculated asπ d³/6. Secretion rate (nl min^–1) was calculated by dividing the droplet volume by the time (min) over which the droplet formed.

The blind end of each cricket tubule was placed in the saline droplet and the open end, formed where the tubule was broken from its junction with the ampulla, was wrapped around a steel pin. The tubule was then squeezed between forceps at a point halfway between the pin and the droplet, thereby rupturing the tubule wall and permitting the escape of secreted fluid.

For analysis by CLSM, secreted droplets were collected in optically flat hollow rectangle glass capillaries (VitroTubes; VitroCom, Mountain Lakes, NJ,USA). Most of the experiments described below used borosilicate capillaries with a path length of 50 μm, a wall thickness of 50 μm and a width of 500 μm. Capillaries were supplied in 50 mm lengths and were scored with the edge of a carborundum stone and broken into lengths of 8–10 mm. Some experiments used capillaries with a path length of 20 μm, wall thickness of 20 μm and width of 200 μm. Capillaries were held with forceps and inserted through the paraffin oil of the Ramsay assay dish and into a droplet of secreted fluid or calibration solution. The aqueous sample was taken up by capillarity and was thus enclosed by glass on four sides and by paraffin oil on two sides. Luminal concretions were sometimes expelled into the droplets secreted during the Ramsay assay. The concretions settled to the bottom of the droplet and were not collected when the fluid was taken up into the hollow rectangle glass capillary.

For Drosophila, transepithelial dye flux (fmol min^–1 tubule^–1) was calculated by multiplying fluid secretion rate (nl min^–1tubule^–1) by the dye concentration (μmol l^–1) measured by CLSM. Cricket tubules varied in length by>25% within an individual and in different insects. Moreover, vigorous contractions of the muscles in each cricket tubule prevented precise measurement of tubule length, so we could not normalize fluid secretion rates per unit length. We therefore report only secreted fluid dye concentrations rather than dye flux for individual cricket tubules. However, approximate values for transepithelial dye flux across cricket tubules could be estimated from secreted fluid dye concentration and a mean secretion rate based on measurements of >50 tubules from several insects.

A chamber made from a 6-well tissue culture dish was used for CLSM of MTs or optically flat glass capillaries. A 15 mm hole was drilled in the centre of each well and a 22 mm square glass cover slip was secured across the opening with melted dental wax. The cover slips were pre-coated with 100 μl droplets of 125 μg ml^–1 poly-l-lysine (70 kDa)to facilitate adherence of isolated MTs. Tubules were transferred into 2 ml saline in the bottom of each well and Texas Red was added from stock solutions to achieve concentrations of 0.5 μmol l^–1 to 50 μmol l^–1. Glass capillaries containing secreted fluid or calibration solutions were placed in empty wells on uncoated glass cover slips.

Confocal fluorescent images were obtained with a Zeiss LSM 510 confocal microscope (Carl Zeiss, Oberkochen, Germany). The software used to obtain and analyse the images was Zeiss LSM5 (Carl Zeiss). A 40× water-immersion objective was used for isolated tubules, and a 20× objective was used for samples collected in hollow rectangle glass capillaries. The system consisted of an inverted microscope, a mixed argon/krypton-ion laser with 458,477, 488 and 514 nm lines and helium–neon lasers with 543 and 633 nm lines. For measurement of rhodamine 123, the 514 nm laser line, a 458/514 nm dichroic filter and a 530–600 nm band pass emission filter were employed. For measurement of daunorubicin, the 543 nm laser line, a 488/543 nm dichroic filter and a 560 nm long-pass emission filter were employed. For measurement of fluorescein, the 488 nm laser line, a 488 nm dichroic filter and a 505–550 nm band pass emission filter were employed. For measurement of quinacrine, the 458 nm laser line, a 488 nm dichroic filter and a 505 nm long-pass emission filter were employed. The Texas Red(sulphorhodamine 101) absorbance spectrum extends between 500 nm and 650 nm,with a primary peak at ∼600 nm and a secondary peak at ∼550 nm. Previous studies of Texas Red transport using CLSM have used either the 529 nm Ar ion laser line (Miller et al.,1998) or the 568 nm krypton ion laser line(Miller et al., 2001). These lines were not available on our system, and we used the 543 nm He–Ne laser line with a 488/543 nm dichroic filter and 560 nm long-pass emission filter. Because of the high concentrations of Texas Red in secreted fluid droplets (4–400 μmol l^–1; see Results), the 514 nm Ar/Kr laser line with a 458/514 nm dichroic filter could also be used.

Tubules in the chamber were viewed under reduced, transmitted light illumination, and a field containing 20–30 cells was selected. The focus was adjusted so that the optical slice was through the centre of the lumen. Then, in confocal fluorescence mode, two successive 8-s scans of the field were collected. Laser power was adjusted to minimize photobleaching, as determined from a decline in fluorescence intensity of the second image relative to the first. The confocal image (512×512 pixels) was viewed on a high-resolution monitor, and fluorescence intensity was determined as the mean value for five regions of interest of 200–500 pixelsselected in the cellular and luminal compartments and adjacent bathing saline. Detector gain and offset were adjusted so that fluorescence intensity was ∼2000 in the lumen on a scale of 0–4095 (12 bits) and the pinhole was set to ∼75μm, corresponding to an optical slice thickness of ∼10 μm. Laser power was reduced to minimize photobleaching, and detector gain was adjusted so that when the average pixel intensity in the tubule lumen was ∼2000, MT autofluorescence was undetectable.

Data are expressed as means ± s.e.m., unless otherwise indicated. Student's t-test was used to evaluate statistical significance, and means were considered significantly different when P<0.05. Kinetic parameters describing maximal dye flux (J_max), maximal secreted fluid dye concentration ([Dye]_sf,max), and the bathing saline dye concentrations corresponding to half maximal flux or secreted fluid dye concentration (K_t) were calculated from concentration–response curves fitted by non-linear regression to the Michaelis–Menten equation (SigmaPlot 2000, SPSS Inc., Chicago, IL,USA).

Representative confocal microscopic images of cricket and Drosophila MTs bathed in saline containing Texas Red are shown in Fig. 1. Approximately 5–10% of the tubules dissected from crickets were similar to those shown in Fig. 1A. The fluorescence intensity (FI) profile showed that concentrations of Texas Red in the cells and lumen were elevated above the bath concentration(Fig. 1B). Both cellular and luminal compartments were clearly visible in Fig. 1A, and FI could be measured readily in selected regions of interest. In most tubules, however,either the lumen (Fig. 1C) or both the cells and the lumen (Fig. 1D) were not clearly visible because of the presence of numerous opaque concretions. The presence of concretions and extensive apical microvilli also complicated the selection of suitable regions of interest in the lumen. As seen in Fig. 1C,D, FI at the periphery of the lumen was often distinctly greater (i.e. brighter) than that in more central regions. Images of Drosophila MTs isolated from either adults(Fig. 1E) or larvae(Fig. 1F) were also generally unsuitable for measurements of FI because of the presence of cellular and/or luminal concretions.

For cricket tubules similar to those shown in Fig. 1A, FI was measured in the bath, cells and lumen. Plots of FI ratio indicated that Texas Red was approximately 40-fold more concentrated in the lumen relative to the bath when the concentration of Texas Red in the bath was 0.5 or 5 μmol l^–1 (Fig. 2). The concentration of Texas Red in the cells was approximately 15–20-fold more concentrated than the bath over the same range of bathing saline Texas Red concentrations. Both lumen/bath and cell/bath FI ratios declined to ∼4 at bathing saline concentrations of 50μmol l^–1, which is consistent with the presence of saturable mechanisms of Texas Red transport in the MTs.

The widespread occurrence of cellular and luminal concretions in the MTs of crickets, and variations in FI between central and peripheral regions of the lumen, complicated the use of isolated tubules for quantitative analysis of Texas Red transport. Analysis of isolated tubules of crickets and Drosophila in the presence of the P-gp substrates rhodamine 123 and daunorubicin showed no clear evidence of dye accumulation in the lumen,although the dyes were taken up into the cells (data not shown).

Because of concerns that the concretions in the cells or lumen of the tubules of Drosophila and many other species (e.g. Wessing and Zierold, 1999)would confound both quantitative and qualitative analysis of dye accumulation in the tubule lumen, we developed a method for measuring the concentration of Texas Red, daunorubicin and other fluorochromes in the fluid secreted by MTs set up in Ramsay assays. Secreted fluid droplets were collected in optically flat glass capillaries, as described in the methods. Fig. 3A shows a sample of fluid containing Texas Red. The plane of focus was adjusted so that the optical slice comprised approximately the middle 50% of the light path along the Z-axis. For capillaries with a path length of 50 μm, a slice thickness of 28 μm was used. For capillaries with a 20 μm path length,pinhole diameter was reduced for a slice thickness of 10 μm. The sensitivity of FI measurement was maximized by using thick optical slices for CLSM analysis of secreted fluid samples collected in hollow rectangle capillaries. The choice of slice thickness represented a compromise between two factors. Reducing the thickness of optical slices by decreasing pinhole diameter facilitated positioning of the optical slice at the Z-axis midpoint of the light path through the glass capillary, but FI also declined linearly as the optical slice thickness was reduced.

Two methods were used to position the optical slice at the Z-axis midpoint of the fluid in the capillary. In the first method, a series of X–Y images was collected at intervals along the X-axis using the Z-stack feature of the LSM 510 software. The first and last slices of the stack were selected by manually focusing just outside the upper and lower surfaces of the capillary. A pinhole corresponding to a slice thickness of 28 μm was selected, and a Z-stack of 10 images at 5 μm intervals was then collected. A region of interest (ROI)corresponding to 50–100% of the fluorescent region within the image was then selected, and the FI of each slice of the Z-stack was determined. Fluorescence intensity was maximal for 2–3 slices near the middle of the stack, and the maximal values were used to calculate the fluorochrome concentration from the corresponding calibration curve (described below). The first method typically required 60–100 s to scan a complete stack.

To facilitate more rapid analysis of 25–50 capillaries typically used in each experiment, a second method was developed. The microscope was initially focused approximately in the middle of the fluid within the capillary. An X–Z scan was then done, based on 20 slices at 4 μm or 5 μm intervals. The images were thus collected from some distance below, through and above the column of fluid in the capillary. Using the LSM software, the plane of focus was then adjusted to the Z-axis midpoint of the column of fluid by moving the horizontal line corresponding to the plane of focus until the band of fluorescence in the X–Z image was divided into two mirror images. Fig. 3B shows the images of X–Z scans made at the apparent midline of the image,defined as a Z-position of 0 μm. Fig. 3B also shows the X–Z scans made when the plane of focus was adjusted 10μm above or below the midline. A series of X–Yimages were then collected over a range of Z-positions, and a plot of FI as a function of Z-position is shown in Fig. 3C. The plot shows that FI in the X–Y images varied by less than 5%when the plane of focus was set 5 μm above or below the apparent midline. With practice, the plane of focus could be adjusted to within 1–2 μm of the position corresponding to the maximum FI. Fluorescence intensity declined at distances greater than 10 μm from the Z-axis middle of the column of fluid within the capillary(Fig. 3C), consistent with inclusion of part of the upper or lower glass wall of the capillary in the optical section.

After the optical slice was adjusted to the Z-axis midpoint of the fluid column within the glass capillary, a time series of two X–Y images was collected. Using the 20×objective, the samples typically filled 25–75% of the X–Y image, corresponding to ∼60 000–200 000 pixels. The mean FI for the sample was subsequently calculated using the ROI feature of the LSM software. Laser strength was adjusted to avoid any photobleaching, evident as a decline in mean FI of the second image relative to the first.

Concentrations of dyes in secreted fluid samples collected in hollow rectangle glass capillaries were calculated from a calibration curve constructed for each experiment. The curve was based on FImeasurements of samples of known dye concentrations that bracketed the range of interest. For most experiments, four calibration droplets whose concentrations extended over an 8-fold range were used. Detector gain and offset were adjusted so that the FI (0–4095, 12 bit) in the highest concentration was ∼3900. For concentration–response curves,several sets of calibration samples were required to bracket the wide range of dye concentrations in the secreted droplets. Fig. 3D shows a representative calibration curve for Texas Red. Each curve relating FI to dye concentration was fitted by regression analysis. Although the relationship of FI to dye intensity was approximately linear(r²>0.8), fits with r² values near unity (r²>0.999) were obtained for 2nd or 3rd order polynomials. The non-linearity was attributable to self-quenching of the dye at higher concentrations. Preliminary experiments revealed that fluid secretion rates and secreted fluid dye concentrations could vary as much as twofold on different days due to variations in room temperature(17–23°C). Concentration–response curves in the Results below were therefore done in the same assay dish with equal numbers of tubules at each concentration on each day to avoid any bias as a result of differences in temperature.

Concentration–response curves for secreted fluid Texas Red concentration as a function of bathing saline Texas Red concentration for cricket tubules are shown in Fig. 4. The inset in Fig. 4 shows the ratio of Texas Red concentration in the secreted fluid to that in the bathing medium (S/M) as a function of bathing saline Texas Red concentration. The value of S/M for bath Texas Red concentrations of 0.1–1 μmol l^–1 was in the range of 30–40,similar to the value of ∼40 for the lumen/bath FI ratio in isolated MTs (Fig. 2). Secreted fluid Texas Red concentration declined at a bath concentration of 50 μmol l^–1, and experiments with Drosophila tubules described below suggest that high concentrations of the dye are toxic. The data point at 50 μmol l^–1 was therefore excluded from the curve-fitting procedure used to estimate the kinetic parameters. The length of cricket tubules varies within and between animals, so fluid secretion rates were not calculated in most experiments. However, an approximate flux can be calculated from an estimate of fluid secretion rates of 53 tubules from three insects (0.40±0.04 nl min^–1). Using the latter value,a maximum flux of 163 fmol min^–1 tubule^–1 is estimated at the maximum secreted fluid concentration of 405 μmol l^–1.

Concentration–response curves for Texas Red flux across Drosophila MTs and corresponding S/M ratios are shown as functions of bathing saline Texas Red concentration in Fig. 5. Both fluid secretion rate and secreted fluid dye concentration declined in saline containing 50μmol l^–1 Texas Red relative to the values at lower concentrations, so the data point at 50 μmol l^–1 Texas Red was excluded from the curve-fitting analysis used to estimate the kinetic parameters J_max and K_t. The S/M ratios for Drosophila tubules were somewhat lower than those observed for crickets, in the range of 20–30 for bathing saline Texas Red concentrations between 0.1 μmol l^–1 and 5 μmol l^–1.

S/M ratios and concentration–response curves for secreted fluid rhodamine 123 concentration in fluid secreted by isolated cricket tubules are shown in Fig. 6. Cricket tubules stopped secreting fluid after 15–30 min in saline containing 10μmol l^–1 or 50 μmol l^–1 rhodamine 123. For Drosophila tubules, it was not feasible to calculate J_max and K_t for rhodamine 123 flux because fluid secretion rates declined for all bathing saline concentrations of rhodamine 123 above 0.1 μmol l^–1. Fluid secretion by Drosophila tubules was completely inhibited by 10 μmol l^–1 rhodamine 123 and was reduced to 0.09±0.02 nl min^–1 (N=10) in 0.5 μmol l^–1rhodamine 123, approximately 20% of the control value. The S/M ratios for Drosophila tubules in 0.1 and 0.5 μmol l^–1rhodamine 123 were 35.3±3.6 (N=36) and 11.9±2.7(N=10), respectively.

By contrast, fluid secretion rates for tubules of both species were maintained in the presence of daunorubicin at concentrations as high as 50μmol l^–1 for Drosophila and 100 μmol l^–1 for crickets. S/M ratios and concentration–response curves for secreted fluid daunorubicin concentration in fluid secreted by cricket tubules as a function of bathing saline daunorubicin concentration are shown in Fig. 7. S/M ratios and concentration–response curves for daunorubicin flux across Drosophila tubules are shown in Fig. 8.

Fluorescein is a substrate of organic anion transporters in tubules of Drosophila and other species(Neufeld et al., 2005; Bresler et al., 1990). We wished to compare transport of smaller, more hydrophilic organic anions such as fluorescein with the transport of larger, amphiphilic MRP2 substrates such as Texas Red. S/M ratios and concentration–response curves for fluorescein flux are shown in Fig. 9. It is worth noting that J_max for fluorescein transport is ∼2.5-fold higher than that for Texas Red, but the value of K_t is ∼4.5-fold higher than that of Texas Red. As a consequence, the transport efficiency(J_max/K_t) for Texas Red is 1.8-fold higher than that for fluorescein.

Transport of Texas Red was reduced by the MRP2-specific inhibitor MK-571(van Aubel et al., 1998) in tubules of both Drosophila (Fig. 10A) and crickets. High concentrations of MK-571 (≥10 μmol l^–1) inhibited fluid secretion by Drosophilatubules. There was no effect of 2 μmol l^–1 MK-571 on fluid secretion rates of 0.67±0.12 nl min^–1 (N=7)and 0.60±0.05 nl min^–1 (N=9) in control and experimental groups, respectively. Texas Red flux for tubules bathed in 0.1μmol l^–1 Texas Red was reduced 37% by 2 μmol l^–1 MK-571 (Fig. 10A). For cricket tubules, fluid secretion rates of control(0.48±0.07 nl min^–1; N=8) and experimental(0.63±0.10 nl min^–1; N=7) groups were unaffected by 10 μmol l^–1 MK-571. However, secreted fluid Texas Red concentration of cricket tubules bathed in 0.5 μmol l^–1 Texas Red was reduced by 68%, from 18±2.1 μmol l^–1 in control saline to 5.7±1.7 μmol l^–1 in the presence of 10 μmol l^–1MK-571.

Transport of Texas Red was also inhibited by probenecid(Fig. 10B,C). The concentration of 114.7±5.4 μmol l^–1 (N=10)Texas Red in fluid secreted by Drosophila tubules bathed in saline containing 5 μmol l^–1 Texas Red was reduced by 63%, to 42.0±7.2 μmol l^–1, in the presence of 1000 μmol l^–1 probenecid (Fig. 10B). However, fluid secretion rate was also reduced significantly, from 0.39±0.06 nl min^–1 in the controls to 0.26±0.05 nl min^–1 in the presence of probenecid. The 78% reduction in Texas Red flux was thus partly due to the effects of probenecid on fluid secretion rate, which suggested non-specific effects of the drug at this concentration. At a lower concentration of 200 μmol l^–1 probenecid, the flux of 0.5 μmol l^–1Texas Red was reduced by 43% (Fig. 10C) and there was no significant effect on fluid secretion rate.

Although probenecid is a known inhibitor of both MRP2 and also Na⁺-dependent organic anion transport systems(Horikawa et al., 2002),transport of Texas Red was not reduced in Na⁺-free saline. In salines containing 5 μmol l^–1 Texas Red, the concentration of the dye in the secreted fluid was 92±16 μmol l^–1(N=10) for tubules in control saline and 105±5 μmol l^–1 for tubules in Na⁺-free saline. Texas Red flux was 62.9±14.5 fmol min^–1 tubule^–1 in control saline and 52.3±5.0 fmol min^–1tubule^–1 in Na⁺-free saline. The differences are not significant (P>0.05).

We examined transport of quinacrine by Drosophila tubules because it is both a P-gp modulator and it is also a substrate or inhibitor of organic cation transporters such as rOCT2 (Sweet and Pritchard, 1999; Dohgu et al., 2004). Kinetic analysis showed that the J_max for transport was 27.7 fmol min^–1tubule^–1 and the associated K_t was 21.4μmol l^–1 (Fig. 11).

Although quinacrine may itself be transported by the organic cation transport system or by P-gps in other cells, neither the P-gp inhibitor verapamil nor the OCT substrate tetraethylammonium (TEA) reduced the flux of quinacrine by Drosophila tubules. Transepithelial quinacrine flux across tubules bathed in saline containing 5 μmol l^–1quinacrine and 500 μmol l^–1 verapamil (5.99±0.72 fmol min^–1 tubule^–1; N=10) did not differ from those of the corresponding controls in 5 μmol l^–1 quinacrine alone (6.31±0.77 fmol min^–1 tubule^–1; N=9). Similarly,transepithelial quinacrine flux across tubules bathed in saline containing 5μmol l^–1 quinacrine and 500 μmol l^–1TEA (4.28±0.78 fmol min^–1 tubule^–1; N=10) did not differ from those of the corresponding controls(3.72±0.66 fmol min^–1 tubule^–1; N=10). There was no effect of verapamil or TEA on fluid secretion rate or the concentration of quinacrine in the secreted fluid (data not shown).

The concentration of daunorubicin in the fluid secreted by cricket tubules bathed in saline containing the dye (5 μmol l^–1) was reduced 71% by 100 μmol l^–1 verapamil(Fig. 12A) but was unaffected by 100 μmol l^–1 TEA(Fig. 12B). The concentration of rhodamine 123 in the fluid secreted by cricket tubules bathed in saline containing the dye (0.5 μmol l^–1) was reduced 60% by 100μmol l^–1 verapamil(Fig. 12C) but was unaffected by 100 μmol l^–1 TEA(Fig. 12D).

The fluxes of daunorubicin across Drosophila tubules bathed in salines containing 1 μmol l^–1 and 5 μmol l^–1 daunorubicin were reduced 67% and 64%, respectively, by the P-gp inhibitor verapamil at a concentration of 500 μmol l^–1 (Fig. 13A,B). Flux of daunorubicin across tubules bathed in saline containing 5 μmol l^–1 daunorubicin was reduced 44% by verapamil at 100 μmol l^–1(Fig. 13C) but was not significantly reduced by the drug at a concentration of 20 μmol l^–1 (Fig. 13D). For Drosophila tubules bathed in 0.5 μmol l^–1 daunorubicin, the transepithelial flux of the dye was reduced 46% by 50 μmol l^–1 quinacrine(Fig. 13E). Transepithelial fluxes of 5 μmol l^–1 daunorubicin were reduced 29% by 1000μmol l^–1 of the organic cation TEA(Fig. 13F). Transepithelial flux of rhodamine 123 was unaffected by high concentrations (1000 μmol l^–1) of verapamil (Fig. 13G) but was reduced 64% by 1000 μmol l^–1 TEA(Fig. 13H).

Our results show that fluorescent substrates of p-glycoproteins and multidrug-resistance associated protein 2 are secreted by the MTs of two insects, the fruit fly Drosophila melanogaster and the cricket Teleogrylluscommodus. The results are significant in two respects. Firstly, we provide the first measurements of transepithelial flux of P-gp and MRP substrates across insect MTs. Flux is calculated as the product of fluid secretion rate and secreted fluid dye concentration. Previous studies of P-gp-like transport by Manduca tubules used perfused rather than freely secreting tubules(Gaertner et al., 1998), and studies of MRP2 transporters in crickets and cockroaches measured luminal concentrations of the dye in isolated tubules but did not report corresponding fluid secretion rates (Karnaky et al., 2000, 2001, 2003). Secondly, the collection of secreted fluid samples in optically flat glass capillaries and the measurement of dye concentration in the samples by CLSM may be applicable to other secretory epithelia (salivary and lacrimal glands) or to sub-nanolitre samples of fluid collected by microperfusion or other means.

Previous studies by Karnaky et al.(2000, 2001, 2003) demonstrated active accumulation of the MRP2 substrates Texas Red and 5-chloromethylfluorescein(CMF) in the cells and lumen of the MTs of crickets (Acheta domesticus) and cockroach (Periplaneta americana). Texas Red transport is dependent on metabolism and is reduced by the MRP2 inhibitor chlorodinitrobenzene, which does not inhibit transport of chlorophenol red, a substrate for the classic organic anion transporter system. Texas Red transport is unaffected by a 50-fold excess of para-aminohippurate (PAH), a substrate of the organic anion transporter, by TEA, a substrate of the organic cation transporter, or by verapamil, an inhibitor of p-glycoprotein. An apical location of an MRP2-like transporter in cricket MTs has been demonstrated by immunocytochemical staining with an antibody to a sequence of rat MRP2(Karnaky et al., 2001). However, an apical location of an MRP2-like transporter cannot account for accumulation of Texas Red in the cells of cricket tubules. Cellular accumulation requires an additional mechanism for transport of the dye across the basolateral membrane. The characteristics of such basolateral transport could be assessed in cricket tubules with few or no intracellular concretions(as in Fig. 1A) so that the cytoplasm is relatively translucent and fluorescence intensity could be measured.

Our analyses of both isolated MTs and of secreted fluid samples collected in glass capillaries demonstrated that Texas Red concentrations are elevated 20–40-fold above those in the bathing medium when the latter contains Texas Red at concentrations near or below the K_t values of 5.5 μmol l^–1 for cricket tubules and 7.1 μmol l^–1 for Drosophila tubules. The typical transepithelial potential across Drosophila tubules is 30–45 mV lumen-positive (O'Donnell et al., 1996, 1998), which could only account for a 3–6-fold elevation of the concentration of the anion Texas Red by passive means. The higher S/M ratios measured in the present paper are therefore indicative of active transepithelial transport of Texas Red.

Possibly toxic effects of high concentrations of Texas Red are evident in the reduction of luminal Texas Red concentration in both species and the reduction in fluid secretion rate of Drosophila tubules when the concentration of Texas Red in the bathing medium is increased above 20 μmol l^–1. By contrast, fluid secretion rate was unaffected by fluorescein at concentrations as high as 100 μmol l^–1. In broad terms, the results of our kinetic analyses suggest that Texas Red transport mechanisms are of higher affinity but lower capacity than those responsible for elimination of organic anions such as fluorescein. The finding that high concentrations of Texas Red inhibit fluid secretion provides an important caveat for studies based only on measurement of dye concentrations in the lumen. Measurement of fluid secretion rate not only permits calculation of transepithelial dye flux but also provides an independent measurement of the non-specific toxicity of dye substrates or inhibitors (see below).

Transport of Texas Red was inhibited by the MRP2 inhibitors MK-571 and probenecid. Although probenecid inhibits both MRP2 and organic anion transporters, the latter pathway is Na⁺ dependent. Secretion of organic anions such as PAH is inhibited 80% by removal of Na⁺ from the bathing medium (Linton and O'Donnell,2000). By contrast, transport of Texas Red was unaffected by Na⁺-free conditions. This finding clearly distinguishes transport of Texas Red from that of compounds such as fluorescein and indicates that inhibition by probenecid must therefore reflect an effect upon a transporter distinct from the classic organic anion transport pathway. Reductions of fluid secretion rate by high (1000 μmol l^–1) concentrations of probenecid were associated with a 33% reduction in fluid secretion rate. The use of lower concentrations of probenecid permitted inhibition of Texas Red transport to be detected in the absence of any effect on fluid secretion rate. The relative concentrations of substrates and inhibitors that we used are similar to those of previous studies of MRP2. Transport of 0.05 μmol l^–1 estradiol glucuronide by rabbit MRP2 expressed in Sf9 cells is reduced to 33% of the control value by 5 μmol l^–1MK571 (van Aubel et al.,1998). Transport of 3 μmol l^–1 saquinavir is reduced to 17% of the control value by 75 μmol l^–1 MK571(Williams et al., 2002). Transport of 4 μmol l^–1N-ethylmaleimide glutathione by human MRP2 expressed in Spodoptera frugiperda ovarian cells is reduced 50% by 1000 μmol l^–1 probenecid(Bakos et al., 2000) and transport of 1 μmol l^–1 methotrexate by a human carcinoma cell line is reduced 50% by 500 μmol l^–1 probenecid(Hoijberg et al., 1999).

Previous studies have reported uptake of daunorubicin and rhodamine 123 into the cells but not the lumen of the MTs of the cricket Acheta domesticus (Karnaky et al.,2001) and the tobacco hornworm, Manduca sexta(Gaertner and Morris, 1999). The latter authors suggested that accumulation of the dyes in the cells is a form of xenobiotic scavenging unrelated to P-gp. Our results, based on collection of secreted fluid, clearly indicate that transepithelial secretions of both rhodamine 123 and daunorubicin are saturable forms of transport that are inhibitable by verapamil and quinacrine. It is possible that species differences underlie the differences between our results and those of Gaertner and Morris (1999) or that the toxic effects of rhodamine 123 noted in both Drosophila melanogasterand Teleogryllus commodus also apply to the tubules of Manduca sexta. Gaertner and Morris(1999) suggested several possible explanations to account for transepithelial transport of 50–5000 μmol l^–1 nicotine observed in their earlier study (Gaertner et al., 1998),whereas 5–10 μmol l^–1 daunorubicin and rhodamine 123 were not accumulated in the lumen. In particular, they note the possibility that `exciting and/or emitted light was quenched, reflected or absorbed by the tissue and opaque uric acid crystals present within the lumen.' We also noted that it was not feasible to clearly visualize dyes in the lumen of many cricket MTs (T. commodus) and all Drosophila tubules because of opaque concretions in the cells and/or lumen. A reexamination of the possible secretion of fluorescent P-gp substrates by lepidopteran tubules may be worthwhile, using the method developed here for analysis of secreted fluid samples collected in hollow rectangle glass capillaries. Tubules of the cabbage looper, Trichoplusia ni, might be suitable for such study because they secrete fluid at high rates and for prolonged periods (M. R. Rheault, J. Plaumann and M. J. O'Donnell, unpublished).

Transport of daunorubicin by tubules of both species was reduced by the P-gp inhibitor verapamil, consistent with involvement of a P-gp-like transporter. The relative concentrations of substrate and inhibitor that we have used are broadly similar to those used in previous studies. For example,transport of 1 μmol l^–1 rhodamine 123 or 3 μmol l^–1 doxorubicin is reduced ∼50% by 20 μmol l^–1 verapamil in several cell lines(van der Sandt et al., 2000). Transport of 1 μmol l^–1 daunorubicin is reduced 50% by 10.3 μmol l^–1 verapamil in lymphocytes(Green et al., 2001). Daunorubicin transport by Drosophila tubules is also inhibited by quinacrine, an organic cation that is also a P-gp modulator, and by high concentrations of TEA. A role for both P-gp and organic cation/H⁺exchange has been proposed to account for the effects of daunorubicin transport in flounder renal tubules(Miller, 1995). We suggest that in Drosophila tubules as well, daunorubicin may be transported both by P-gp-like transporters and by the organic cation transporters involved in transepithelial secretion of quaternary ammonium compounds such as TEA(Rheault and O'Donnell,2004).

Analysis of transepithelial transport of rhodamine 123 is complicated by the inhibitory effects of this P-gp substrate on fluid secretion, particularly in the case of isolated Drosophila tubules. Transport of rhodamine 123 is saturable in cricket tubules and is inhibited by verapamil. Fluid secretion by Drosophila tubules is inhibited by rhodamine 123 at concentrations above 0.1 μmol l^–1, and transport at such low concentrations may involve mechanisms distinct from saturable transport by P-gp-like transporters. Moreover, the transport of rhodamine 123 by Drosophila tubules is unaffected by very high concentrations of verapamil, again suggesting processes unrelated to P-gp. A further complication is that the principal cells of dipteran tubules retain rhodamine 123 for a short period and the stellate cells reabsorb the dye from the tubule lumen (Meulemans and De Loof,1992). The dye accumulates in the stellate cell mitochondria and eventually in intensely fluorescing vesicles, probably lysosomes. Endocytotic uptake has been ruled out. In addition, rhodamine 123 precipitates on the luminal concretions in the distal segment of the anterior tubules (Meulemeans and De Loof, 1992). Given the inhibitory effects of rhodamine 123 on fluid secretion rate and the complexities in the routes of transport, we suggest that it is an inappropriate substrate for further analysis of P-gp-like transporters in dipteran MTs.

Transport of the organic cation quinacrine is also saturable, with Drosophila tubules accumulating the compound in the lumen at concentrations as high as 17-fold above those in the bath. However, in contrast to quinacrine transport by the rat choroid plexus, transport by Drosophila tubules is unaffected by either verapamil or TEA. We had examined the effects of quinacrine on daunorubicin transport because it is a known P-gp modulator (Tiberghien and Loor,1996). Although we have not examined quinacrine transport in detail, these results suggest that there may be differences in the mechanisms of quinacrine transport in Drosophila MTs and vertebrate tissues. Moreover, it appears that the tubules possess a mechanism of organic cation transport distinct from that utilized for the P-gp substrate daunorubicin and the organic cation TEA (Rheault and O'Donnell, 2004).

Dye concentration in nanolitre samples of secreted fluid droplets collected in optically flat glass capillaries can readily be measured using CLSM. In conjunction with Ramsay assays, these measurements permit calculation of transepithelial dye flux. The Ramsay assays also provide a means for determining the effects of the dyes or transport inhibitors on processes other than dye transport. Rhodamine 123, for example, clearly inhibits fluid secretion by tubules of Drosophila and crickets. This finding raises the possibility that in instances where fluid secretion is reduced, any associated reduction in dye transport may reflect impairment of cellular metabolism or other transporters involved in cellular homeostasis. Rhodamine 123 is known as a mitochondrial dye, so its effects on fluid secretion transport may reflect impairment of cell metabolism. Analysis of dye transport solely by image analysis of cellular and luminal compartments would not provide this additional information about the effects of the dye on cell function.

We developed the use of optically flat glass capillaries for analysis of transepithelial dye transport because of concerns that opaque cellular or luminal concretions would block or interfere with laser light transmission. Although we have previously used CLSM for visualization of fluorescein transport by Drosophila tubules(Linton and O'Donnell, 2000),the presence of concretions alters the amount of laser light that reaches a given region of the tubule, thereby making quantitative analysis unfeasible. There are also concerns that the concretions move in the lumen during fluid secretion, thereby altering the amount of laser light passing through the lumen, and that cellular concretions may be transferred from cell to lumen under certain conditions (Hazelton et al.,2001). Another advantage of analysis of secreted fluid droplets by CLSM is that laser strength can be optimized to minimize photobleaching but without the additional concerns over the effects of the laser light on cellular components. Low concentrations of fluorochromes can be measured quickly and accurately because fluorescence intensity measurements are based on large numbers of pixels (typically >60 000) relative to isolated tubules and because the thickness of the optical slice can be increased. Estimates of dye concentration are less affected by variations in pH of the cytoplasmic milieu, and the calibration solutions can be adjusted to resemble closely the ionic strength and pH of the secreted fluid. Secreted fluid samples from large numbers of tubules (typically 20 per experiment) in each Ramsay assay can be analysed much more rapidly than is possible if multiple regions of interest within the lumen and bath are selected for analysis in isolated tubules examined by CLSM. It is important to point out that the methods described here for analysis of nanolitre samples of fluid containing fluorescent transporter substrates provide a faster and cheaper alternative to the use of radiolabelled substrates. Lastly, it is worth noting that, although we have analysed fluid samples collected in hollow rectangle glass capillaries by CLSM, fluorescence intensity could also be measured by epifluorescence microscopy of the capillaries. Transport of fluorescein by isolated tubules of the cricket A. domesticus has recently been studied by means of quantitative fluorescence microscopy(Neufeld et al., 2005).

The flux measurements reported here can be used to estimate the time required to clear the haemolymph of a given concentration of dye. Haemolymph volume in Drosophila is of the order of 0.1 μl in control flies but is as high as 0.32 μl in desiccation-resistant strains(Folk et al., 2001). For Texas Red, each tubule secretes ∼80 fmol min^–1 when bathed in saline containing the dye at a concentration of 20 μmol l^–1 (Fig. 5). The four tubules will thus clear the haemolymph content of 2 pmols (=0.1 μl× 20 μmol l^–1) of the dye in ∼6 min. Similar calculations indicate clearance times of ∼10 min for daunorubicin, ∼12 min for fluorescein and ∼25 min for quinacrine. These estimates suggest that the MTs may play an important role in the elimination of potentially toxic P-gp substrates and MRP2 substrates from the haemolymph.

In summary, our results show that transport of fluorescent substrates analysed by CLSM of secreted fluid samples collected in hollow rectangle glass capillaries indicates the presence of P-gp-like and MRP2-like transporters in the MTs of two insect species. Demonstration of secretion of P-gp and MRP2 substrates by Drosophila tubules is of particular interest because of previous molecular genetic evidence for the presence of MDR and MRP genes in this species (Wu et al., 1991; Tarnay et al.,2004). Our results set the stage for further analysis of the effects of treatments, such as cadmium exposure and heat shock(Tapadia and Lakhotia, 2005),known to alter MRP or MDR gene expression on transport of P-gp and MRP2 substrates by isolated MTs and the possible role of such transport in insecticide resistance. In view of recent evidence showing modulation of organic cation transport by dietary loading, peptides, amines and intracellular second messengers(Bijelic and O'Donnell, 2005; Bijelic et al., 2005), it will also be of interest to examine whether transport of P-gp and MRP2 substrates is also modulated by such treatments.