Concealed transepithelial potentials and current rectification in tsetse fly malpighian tubules

The tsetse fly Glossina morsitans is unusual among insects, even among blood-sucking insects, in that fluid secretion by its Malpighian tubules is driven almost entirely by the active transport of sodium ions. Cation and water movements in tsetse tubules were investigated by Gee (1976a,b), using the classical preparation of isolated tubules secreting under oil, stimulated either by extracts of thoracic ganglia or by cyclic AMP.

The present paper describes the first application of electrophysiological techniques to tsetse fly Malpighian tubules. Some preliminary data have already been reported in a review of ion transport mechanisms in insect tubules (Nicolson, 1993). Using the microperfusion technique developed for mammalian tubules, we have measured transepithelial potentials and resistances and determined current–voltage relationships. The study focuses on the tubule response to cyclic AMP and on the effect of varying chloride concentrations in the bathing fluid. An important new finding is the lack of a measurable transepithelial potential in chloride Ringer, even during cyclic-AMP-stimulated secretion. In addition, curvilinear current–voltage relationships are described in Malpighian tubules for the first time.

Tsetse flies were supplied as pupae by the Tsetse Research Laboratory, Bristol, from a colony of Glossina morsitans morsitans Westw., which originated in Zimbabwe. After emergence, flies were maintained at 10 °C and were not fed. Male and female flies between 1 and 4 days old were used in the experiments. No differences were found between the tubules of the two sexes.

The Ringer’s solutions contained either chloride or sulphate as the predominant anion. The Cl⁻ Ringer was based on the recipe of Gee (1976a) and had the following composition (mmol l⁻¹): NaCl, 120; KCl, 10; MgCl₂, 2; CaCl₂, 2; NaH₂PO₄, 1.5; malic acid, 3; citric acid, 2; alanine, 5; proline, 5; and glucose, 20. NaOH was used to adjust the pH to 7.0, and the osmolality was 285 mosmol kg⁻¹.

The SO₄²⁻ Ringer was prepared by replacing the NaCl with Na₂SO₄ (60 mmol l⁻¹) and the KCl with K₂SO₄ (5 mmol l⁻¹), and increasing the glucose concentration to 100 mmol l⁻¹. The Cl⁻ concentration of this SO₄²⁻ Ringer was thus 8 mmol l⁻¹. After adjustment of the pH to 7.0, the measured osmolality was also 285 mosmol kg⁻¹.

The specific resistances of these solutions, as measured by a laboratory conductivity meter, were: Cl⁻ Ringer 73.4 0 cm, SO₄²⁻ Ringer 88.8 0 cm.

Dibutyryl cyclic AMP (sodium salt, Sigma) was used at a concentration of 10⁻⁴ mmol l⁻¹.

Flies were cold-anaesthetized before dissection in Cl⁻ Ringer. The four Malpighian tubules were set up as in vitro preparations according to the method of Ramsay (1954). Sometimes it was possible to remove whole tubules undamaged, but usually the isolated preparations consisted of varying lengths of tubule and both open ends were secured in the liquid paraffin. Unlike Gee (1976a), we did not use ligatures. Secretion rates are expressed as nl min⁻¹ cm⁻¹ of tubule (whole tubules are approximately 2 cm long). Secretion rates are known not to vary along the length of the tubule (Gee, 1974).

In five cyclic-AMP-stimulated tubules, the apparent transtubular potential (V_t) was measured by inserting a Ag/AgCl electrode into the secreted droplet, and a Ag/AgCl microelectrode, filled with 1 mol l⁻¹ KCl, into the larger bathing drop; in each tubule the potential was measured again after changing the bathing drop from Cl⁻ to SO₄²⁻ Ringer.

The potential across the basolateral membrane (V_b) of tubules bathed in Cl⁻ Ringer was determined by insertion of intracellular microelectrodes before and after addition of cyclic AMP to the bathing solution.

Isolated tubule segments, 0.4–1 mm in length, were perfused in vitro as described by Burg et al. (1966). The height of the reservoir of perfusion fluid was held constant at 12 cm above the bath; reducing this height to zero had no effect on the measured voltages. All experiments were carried out at room temperature (22 °C).

We have previously described in detail (Isaacson et al. 1989) our methods of measurement of both spontaneous and current-induced transtubular potentials in perfused Malpighian tubules. In brief, the bath was held at ground potential, while V_o and V_l (the potentials at the proximal and distal ends of the tubule) were detected by saturated KCl calomel electrodes connected to high input impedance unity-gain voltage followers. The voltage follower at the proximal end incorporated a bridge circuit, thus permitting measurement of V_o during passage of current. The spontaneous transtubular potential (V_t) was equated with V_o at passage of zero current. Cable analysis or determination of the current–voltage relationship was effected by injection of current pulses of 600 ms duration and varying amplitude into the perfusion pipette (see below); the resulting changes in V_o and V_l were determined and the cable parameters, or current–voltage relationship (I/V plot), was calculated by a microcomputer. Cable analysis was initiated by measurement of the increments in V_o and V_l (to 0.5 mV maximal resolution) induced by passage of a single 200 nA hyperpolarising current pulse. These measurements yielded values for the transepithelial resistance (R_trans) and virtual short-circuit current (SCCv). The short-circuit current is ‘virtual’ as it was calculated as V_t/R_trans (that is, it was not measured at V_t=0 mV), and, owing to the difficulty of visualising the tubule lumen, it was not corrected for surface area.

Current–voltage relationships were determined by measurement of the changes induced in V_o on passage of a series of current pulses, incrementing in 20 nA steps from -200 nA to 200 nA; consecutive pulses were spaced 1 s apart. The resultant changes in V_o, from -50 mV to 200 mV (to a maximal resolution of 1 mV), were used to generate the I/V plot.

The following conventions and nomenclature are based on those of Helman and Fisher (1977; see Discussion). Current is plotted on the x-axis, voltage on the y-axis (Fig. 1A). The slope resistance of a linear I/V plot is referred to as R2. To facilitate description of curvilinear I/V plots, the slope resistance at depolarised values (the linear lower part of the curve, at negative currents and negative or low positive voltages) is similarly referred to as R2; that at hyperpolarised values (the linear upper part of the curve, at positive currents and larger positive voltages), as R1. Thus, where R1>R2, the curve is concave upwards; if R1<R2, the curve is concave downwards. The intersection of the regression lines describing R1 and R2 is the ‘inflection point’, the voltage and current coordinates of which are denoted E1 and I_sh respectively (Fig. 1A). The shunt resistance (R_shunt) of the equivalent electrical circuit is calculated as E1/I_sh. E1, I_sh and R_shunt cannot, of course, be determined in linear I/V plots, where R1 is necessarily equal to R2.

The simple equivalent electrical circuit (Fig. 1B) referred to in this study consists of only a Thevenin voltage source (E_Na) in series with a Thevenin resistance (R_series), both in parallel with a shunt resistance (R_shunt). E_Na is the electromotive force of the active Na⁺ transport mechanism. R_series, the resistance encountered by ions moving through the active transport pathway, is a function of the direction of current flow (from the lumen to the bath or vice versa); in the equivalent electrical circuit, the diodes select the appropriate value of R_series for a given direction of current flow. R_shunt is the resistance encountered by ions moving passively through paracellular and possibly transcellular ‘leak’ pathways.

In this circuit, V_t=E_NaXR_shunt/(R_shunt+R_series); thus, assuming some non-zero value for E_Na,V_t can approach zero only when R_shunt is some small fraction of R_series. The direction of current flow through the active transport path is reversed once the voltage induced by current injection opposes and exceeds E_Na.

In each experiment, the tubule was both bathed and perfused by either Cl⁻ or SO₄²⁻ Ringer. V_t was recorded continuously on a chart recorder, whereas cable parameters and I/V plots were obtained at 4–6 min intervals throughout the experiment. Once ‘equilibration’ had been attained, the tubule was exposed to cyclic AMP. In six instances, the bathing fluid was changed from SO₄²⁻ to Cl⁻ Ringer, or vice versa, the composition of the perfusate remaining unchanged.

Addition or removal of cyclic AMP, or changes in the Cl⁻ concentration of the bathing fluid, were effected by threefold washout of the bath (2 ml) with the new solution. As electrical transients occasionally accompanied such fluid changes, a few minutes were allowed to elapse before continuing with current injection.

Results are presented as means ± S.D. Paired and independent sample t-tests were used to assess differences between means. Where the variances differed markedly (as in comparison of the means obtained in the Cl⁻ or SO₄²⁻ bathing solutions), the Mann–Whitney U-test was employed. Differences mentioned in the text were significant at the P<0.05 level. The slope resistances of linear regions of the I/V plots were calculated by regression analysis.

Fluid secretion by unperfused tubules bathed in Cl⁻ Ringer was less than 0.1 nl min⁻¹ cm⁻¹. If tubules were bathed in SO₄²⁻ Ringer, no secretion was detectable. Within seconds of exposure to cyclic AMP, fluid secretion in Cl⁻ Ringer increased dramatically (Fig. 2), levelling off at 25 nl min⁻¹ cm⁻¹. Reducing the Cl⁻ concentration from 138 to 8 mmol l⁻¹ (SO₄²⁻ Ringer), without removing cyclic AMP, reduced the rate of fluid secretion by 60 % to about 10 nl min⁻¹ cm⁻¹ (Fig. 2).

In five cyclic-AMP-stimulated tubules, V_t in Cl⁻ Ringer was -1.0±3.4 mV. On replacing the bathing Cl⁻ with SO₄²⁻ Ringer, V_t rose promptly and significantly to 10.8±5.1 mV.

In seven tubules bathed in Cl⁻ Ringer, exposure to cyclic AMP caused V_b to fall from -38.7±10.9 mV to -15.3±8.6 mV (P<0.01) (i.e. V_b became less negative by 23.4±9.9 mV).

In those tubules bathed and perfused in Cl⁻ Ringer, V_t and SCCv were at or close to zero (Table 1). After addition of cyclic AMP, V_t sometimes rose promptly to 3–5 mV, only to fall towards 0 mV within a few minutes. Similar minor excursions in SCCv accompanied these fluctuations in V_t. R_trans always fell.

Tubules bathed and perfused in SO₄²⁻ Ringer displayed initial values of V_t ranging from -1.7 mV to 16 mV, and SCCv was close to zero. After administration of cyclic AMP (Fig. 3), both V_t and SCCv increased promptly to attain peak values 3–30 min later (13±9 min) and remained relatively constant thereafter. R_trans fell progressively, but never to the levels seen in the tubules bathed in Cl⁻ Ringer, either before or after exposure to cyclic AMP. This electrical response is consistent with the secretory response (above), revealing that the transport mechanism is inactive or barely operative in SO₄²⁻ Ringer prior to the addition of cyclic AMP to the bath.

In Cl⁻ Ringer, some 5–10 min after exposure to cyclic AMP, the number of tubules displaying curvilinear I/V plots increased from five to seven (N=9); an example is shown in Fig. 4. In some instances, R1 (N=6) and R1/R2 (N=5) fell, while E1 increased (N=4), but overall these changes fell short of attaining statistical significance (Table 2). R_shunt remained unchanged.

In SO₄²⁻ Ringer, the number of tubules displaying curvilinear I/V plots increased from two to seven (N=10) some 5–20 min after exposure to cyclic AMP; an example is shown in Fig. 5. Both R1 and R2 fell markedly, but never to the levels measured in Cl⁻ Ringer. R1/R2 increased in three, fell in four, and remained unchanged in three tubules. While it was possible to estimate initial values of E1 and R_shunt in only two tubules, these values were not significantly different to those in the seven tubules in which they could be measured after the addition of cyclic AMP.

In short, tubules in Cl⁻ Ringer displayed values of R1, R2, E1 and R_shunt lower than those in SO₄²⁻ Ringer; there was no difference in the mean value of R1/R2 (Table 2). In both solutions, exposure to cyclic AMP was followed by an increased frequency of appearance of curvilinearity, and in SO₄²⁻ Ringer, by falls in both R1 and R2.

In six additional experiments, the bath fluid was changed from Cl⁻ to SO₄²⁻ Ringer or vice versa, the perfusate remaining unchanged. The changes described below were not dependent on the composition of the perfusate, so that the effects of possible diffusion potentials between the perfusate and the bath fluid can be discounted. In all but one of these experiments, cyclic AMP was present throughout. The changes detected on addition of cyclic AMP to the bath, on replacing Cl⁻ with SO₄²⁻ Ringer or vice versa (in the continued presence of cyclic AMP) and on removal of the cyclic AMP, are exemplified in the following. Fig. 6 depicts the sequence of changes observed in a tubule bathed initially in Cl⁻ Ringer. Addition of cyclic AMP led to a fall in R1; replacing the Cl⁻ with SO₄²⁻ Ringer increased R1, R1/R2 and V_t; finally, removal of the cyclic AMP caused R1 and R2 to increase sharply, with little change in R1/R2, while V_t fell sharply. Fig. 7 depicts the changes detected when the bathing SO₄²⁻ Ringer was replaced with Cl⁻ Ringer, in a cyclic-AMP-stimulated tubule: R2 remained unchanged, but R1 fell sharply, with the appearance of a clear inflection point.

The secretory response of tsetse fly tubules to cyclic AMP is reduced by 60 % when Cl⁻ Ringer is replaced by SO₄²⁻ Ringer (Cl⁻ 8 mmol l⁻¹) and is absent in Cl⁻-free Ringer (L. C. Isaacson and S. W. Nicolson, unpublished data). Fluid secretion is also Cl⁻-dependent in Malpighian tubules of other species, but the sensitivity varies. For example, Calliphora erythrocephala species tubules do not secrete at all until the Cl⁻ concentration is 40 mmol l⁻¹ or higher (Berridge, 1969), whereas Rhodnius prolixus tubules secrete at maximum rates when the Cl⁻ concentration is 30 mmol l⁻¹ (Maddrell, 1969).

In both non-perfused and perfused tubules bathed in Cl⁻ Ringer, V_t (and in perfused tubules, SCCv) was at or close to zero, and, apart from occasional small rapid transients, remained so even after exposure of the tubules to cyclic AMP. (As SCCv is derived from V_t, see Materials and methods, the absence of a measurable SCCv is meaningless when V_t is at or close to zero.) Yet cyclic AMP elicited a brisk secretory response in tubules bathed in Cl⁻ Ringer. The apparent lack of change in V_t in response to cyclic AMP, in the face of stimulated secretion, has not previously been reported in insect Malpighian tubules.

The absence of response in V_t (and SCCv) in cyclic-AMP-stimulated tubules bathed in Cl⁻ Ringer was in sharp contrast to that observed in the tubules bathed in SO₄²⁻ Ringer (Table 1). Similar increments in V_t on reduction of the Cl⁻ concentration of the bathing fluid have been reported by others. Exposure to Cl⁻-free Ringer causes a dramatic rise in V_t in cyclic-AMP-stimulated tubules of Rhodnius (O’Donnell and Maddrell, 1984), and in Locusta migratoria tubules, especially in the presence of stimulants (Fogg et al. 1989). The absence of a measurable response in V_t in stimulated tubules bathed in Cl⁻ Ringer, in contrast to those bathed in SO₄²⁻ Ringer, was all the more striking in that the accompanying fall in V_b (of some 23 mV) would, in itself, be expected to contribute to a positive V_t. This discrepancy suggests that a Cl⁻ shunt was short-circuiting the transtubular potential. The values of R_trans found here, both before, and particularly after, exposure to cyclic AMP, are consistent with the suggestion of a shunt path of unusually low resistance. R_trans in tsetse fly tubules is very low compared with values measured in Onymacris plana, Aedes aegypti and Formica polyctena (7, 11 and 23 k0 cm respectively; Isaacson et al. 1989; Pannabecker et al. 1992; Leyssens et al. 1992). Similarly the prompt increases in V_t and SCCv seen on changing the bath fluid from Cl⁻ to SO₄²⁻ Ringer were accompanied by sharp rises in R_trans and R2 (Figs 3, 6, 7).

We have, however, been unable to find any morphological evidence of such a shunt. Preliminary electron microscopical examination of both control and cyclic-AMP-stimulated tubules showed that adjacent cells were tightly apposed to each other, with barely discernible intercellular spaces. We could not detect a second cell type, as has been suggested as a possible Cl⁻ shunt path in Aedes (Pannabecker et al. 1993).

Cyclic AMP has been shown to increase the Na⁺ permeability of the basolateral membrane in mosquito Malpighian tubules (Sawyer and Beyenbach, 1985). A similar effect may have contributed to the fall in R_trans observed here, in both Cl⁻ and SO₄²⁻ Ringer.

Several years ago Helman and co-workers demonstrated nonlinear I/V relationships (‘rectification’) in a variety of Na⁺-transporting epithelia – frog skin, toad urinary bladder and colon, snake distal tubules and rabbit collecting ducts (Helman et al. 1975; Helman and O’Neil, 1977; Helman and Fisher, 1977; Macchia and Helman, 1979; Koeppen et al. 1980). In all of these studies, the I/V plots showed two straight lines of different slope, sometimes meeting at a clear inflection point, or alternatively joined by a short curved line. In the latter case, the inflection point was taken as the intersection of the extrapolated linear regions of the plot. Helman and co-workers provided convincing evidence, confirming an earlier suggestion by Civan (1970), that E1, the voltage coordinate of this inflection point, was identical to the Thevenin electromotive force of the active Na⁺ transport mechanism (E_Na). As no current can flow through the active transport pathway in the presence of an external voltage equal and opposite to E_Na, the current coordinate of the inflection point, I_sh, had to be that current which passed only through the shunt pathways. The total resistance of all possible shunt paths (R_shunt) could therefore be calculated as E1/I_sh. It should be stressed that this electrical analysis provides no information about thermodynamic or metabolic events within the tissue.

As mentioned earlier, the Malpighian tubule of the tsetse fly Glossina morsitans is unusual in that fluid secretion is driven almost entirely by the active transport of sodium ions (Gee, 1976a). It was therefore of interest to ascertain whether this Na⁺-secreting epithelium also possessed a non-linear I/V relationship, from which its Thevenin E_Na and R_shunt might be derived.

Assuming R_shunt to be a simple ohmic resistance, non-linear I/V behaviour within R_series would be more easily seen in the I/V plot, for greater ratios of R_shunt to R_series (because more current would now pass through the latter). This was just the pattern of events found here. Thus, in unstimulated tubules, non-linear I/V plots were seen more frequently in tubules bathed in Cl⁻ than in SO₄²⁻ Ringer, and this frequency increased as R_series fell after exposure to cyclic AMP, in both Cl⁻ and SO₄²⁻ Ringer (Tables 1, 2). (A fall in R_trans with no change in R_shunt implies a fall in R_series, because R_trans is equal to R_series and R_shunt in parallel.)

E1, as found in the tubules bathed in Cl⁻ Ringer, was considerably less than that found in tubules bathed in SO₄²⁻ Ringer (Table 2). In Cl⁻ Ringer, as seen above, V_t was effectively short-circuited; presumably E1 was similarly attenuated. Consequently, the value of E1 (approximately 50–60 mV) as found in tubules bathed in SO₄²⁻ Ringer, both in the absence or the presence of cyclic AMP, is presumably a closer approximation to the value of E_Na in the tsetse fly Malpighian tubule. This is about half the value of E_Na found in different tissues in vertebrates in many of the studies cited above.

However, we have yet to confirm that the relationship between E1 and E_Na found in epithelia of other species pertains also in the insect Malpighian tubule. In principle, an electromotive force is least attenuated if measured under equilibrium conditions; that is, in the present context, with zero net Na⁺ transport, as would occur in SO₄²⁻ Ringer that is Cl⁻-free. The figure given above for the E_Na of the tsetse fly Malpighian tubule is therefore at best only a first approximation.

The basis of the curvilinear I/V plots is not immediately obvious. The phenomenon presumably reflects a mix of anionic and cationic Goldman rectification in two or more channel types in the basolateral and apical membranes, in parallel with a linear I/V response through the shunt resistance. R1/R2 was sometimes greater and sometimes less than unity, in both Cl⁻ and SO₄²⁻ Ringer. However, the data suggest some correlation between a fall in R1, R1/R2 and Cl⁻ permeability; thus, in Cl⁻ Ringer, addition of cyclic AMP was usually followed by a fall in R1/R2 (Fig. 6), whereas these variables were unchanged or changed inconsistently in SO₄²⁻ Ringer. Similarly, in those instances in which SO₄²⁻ Ringer in the bath was replaced by Cl⁻ Ringer (in the presence of cyclic AMP), R1 and R1/R2 fell (Fig. 7). On replacing the SO₄²⁻ Ringer, these values returned towards their original levels (Fig. 6). Thus, unlike mammalian and amphibian epithelia in which no explanation has been found for changes in R1/R2 in the I/V plot, such changes in the tsetse fly Malpighian tubule appear to be a function of the Cl⁻ concentration of the bathing fluid, and presumably reflect changes in Cl⁻ conductance.

We have yet to attempt to localise the anatomical site of origin (apical or basolateral membrane) of the non-linearity in the I/V plot, by repeating these studies with an intracellular microelectrode in situ; however, it presumably originates in the apical membrane, as in frog skin (Helman and Fisher, 1977).

The phenomena detected here are comparable to those previously described in vertebrate epithelia. For example, antidiuretic hormone, acting via increased intracellular production of cyclic AMP, increases Na⁺ transport through frog skin and toad bladder by reducing R_series, and is without effect on E_Na (Yonath and Civan, 1971). In addition, both R1 and R2 fall; the decrease of R1 in Cl⁻ Ringer frequently exceeds that in R2 (that is, R1/R2 falls), whereas E1 remains constant (Macchia and Helman, 1979). These effects are also observed in our study, following exposure of the tubules to cyclic AMP. This similarity is remarkable, given that the mechanism of Na⁺ transport in the tsetse tubule is very different to that in vertebrate epithelia.

As argued above, the absence of response in V_t (and SCCv) in stimulated tubules bathed in Cl⁻ Ringer is consistent with the presence of a Cl⁻ shunt that short-circuits the transtubular potential. In terms of the equivalent electrical circuit, however, R_shunt could not have been the site of such a short circuit. Although R_shunt was many times lower in Cl⁻ than in SO₄²⁻ Ringer, it was far too high, in comparison to R_series, to account for a zero V_t. (As R_trans is simply R_shunt and R_series in parallel, R_series is readily calculated from the data in Tables 1 and 2. Thus, in Cl⁻ Ringer, R_series was 3.8 kΩ cm before, and 1.7 kΩ cm after, exposure to cyclic AMP. Both these values of R_series are small fractions of the concomitant values of R_shunt.) However, a shunt within the active transport path, and across E_Na, could also account for a zero V_t. If E_Na is largely or solely located at the apical membrane, apical Cl⁻ channels could conceivably fill this role. Thus, as argued above, the falls in R1 and R1/R2 in stimulated tubules bathed in Cl⁻ Ringer are suggestive of a cyclic-AMP-enhanced Cl⁻ permeability in the apical membrane, and Cl⁻ channels have been found on the apical membrane in Aedes (Wright and Beyenbach, 1987). Measurement of the fractional resistance of the apical membrane, before and after exposure to cyclic AMP, would assist in establishing this possibility. However, we have yet to measure these variables in the tsetse fly Malpighian tubule.

We have been unable to find studies of I/V relationships in insect epithelia, other than that of Moffet (1980). This study of K⁺ secretion in the midgut of larval tobacco hornworm (Manduca sexta) showed a non-linear I/V relationship in this tissue. In an earlier study on the Malpighian tubules of Onymacris plana (Isaacson et al. 1989), a preliminary survey of I/V relationships revealed no evidence of rectification. We have not examined Malpighian tubules of other species. However, the findings presented here suggest that measurement of I/V relationships may be of value in the elucidation of transport mechanisms in insect Malpighian tubules.

In this study, V_t was determined using both the Ramsay and the tubular perfusion techniques, primarily to confirm that some perfusion-induced artefact was not responsible for the absence of an appreciable V_t in Cl⁻ Ringer in the latter. With SO₄²⁻ Ringer and cyclic AMP in the bathing fluid, the transtubular potentials detected using the Ramsay technique were significantly lower than those found in perfused tubules (P<0.005). This was not surprising since measurement of the transtubular potential by the Ramsay technique is prone to error (Isaacson and Nicolson, 1989; Aneshansley et al. 1989). The striking discrepancy between the transtubular potentials in Cl⁻ and SO₄²⁻ Ringer, as found here, further emphasises that conclusions based upon the absolute values of these potentials may be grossly misleading, unless interpreted in the context of an equivalent electrical circuit.

Tsetse flies were supplied by the Tsetse Research Laboratory at the University of Bristol; we are grateful for Dr Peter Langley’s help. Assistance with electron microscopy was given by the Electron Microscope Unit, University of Cape Town. This research was supported by the Medical Research Council, the Foundation for Research Development and the University of Cape Town.