Acid–Base Transport and Control in Locust Hindgut: Artefacts Caused by Short-Circuit Current

Two of the most intensively studied epithelial transport systems in insects are lepidopteran midgut (reviewed by Harvey, 1982; Harvey and Zerahn, 1972) and locust hindgut (reviewed by Phillips et al. 1986). In both cases, progress has resulted from the ability to study these epithelia in vitro under short-circuit conditions. We wish to report that the standard short-circuit current methods used in most previous studies of these insect epithelia, namely the methods of Harvey et al. (1967), Zerahn (1970), Wood (1972; Wood and Moreton, 1978; modified by Williams et al. 1978, and Hanrahan et al. 1984) and Mandel et al. (1980), are not suitable for studies of acid-base transport, which is a major activity of all these membranes.

We were led to reconsider our short-circuit methods because of paradoxical results on acid-base transfer across locust hindgut under open-circuit, as compared to short-circuit, conditions. Locust ileum and rectum both actively secrete H⁺ (J_H) into the lumen against large pH (2 units) and electrical differences under open-circuit conditions in vitro and also in situ (Thomson et al. 1988a,b; Irvine et al. 1988). This occurs against electrochemical potential differences of up to 100 mV at the apical membrane. In support of these observations, we report in this paper that, when large electrochemical differences exist, addition of cyclic AMP to the haemocoel substantially reduces rectal J_H but luminal alkalization is never observed. In contrast, Irvine et al. (1988) reported that the addition of cyclic AMP caused a dramatic shift from acid secretion (J_H of 0.4μequivcm⁻²h^_1) to alkaline secretion at high rates (J_OH of 8 _μequivcm⁻²h⁻¹) across locust ileum under shortcircuit conditions. In preliminary studies, Hanrahan (1982) and Hanrahan and Phillips (1982) observed a similar but smaller alkalization of the rectal lumen (J_OH=39 % of Cl⁻ influx) under the same stimulated short-circuit current conditions.

The question therefore arises as to whether short-circuiting modifies cell behaviour in locust hindgut (e.g. by unmasking an apical C1⁻/OH⁻ exchanger as suggested by Irvine et al. 1988), or whether luminal alkalization is a methodological artefact. We have found the latter to be true: when either Ag or Ag-AgCl current-passing electrodes are placed directly in the bathing saline of Ussing chambers, without the intervention of salt bridges, significant amounts of OH⁻ are produced at the cathode (i.e. the lumen side of all these insect epithelia) at the naturally high I_SC values observed across locust hindgut. It is worth recalling that even larger short-circuit currents (I_SC) in the same direction (i.e. anions to haemocoel, or cations to lumen) occur across lepidopteran midguts, the contents of which are very alkaline in situ (Dow, 1986). Electrode configurations that lack salt bridges have been used in most studies of lepidopteran midgut (but not all; e.g. Moffett, 1980) and locust hindgut. Fortunately, the failure to use salt bridges with current-passing electrodes does not influence other transport or electrical parameters (e.g. Cl⁻-dependent I_SC) previously reported for locust hindgut. Finally we report correct values for J_H obtained with salt bridges under short-circuit conditions in locust ileum and rectum, both before and after adding cyclic AMP.

Adult female desert locusts (Schistocerca gregaria Forskål) 14–22 days beyond their final moult were used for all experiments. They were maintained at 28°C and 60 % relative humidity on a 12 h:12 h light: dark cycle, and fed daily on fresh lettuce and a dried mixture of bran, alfalfa and powdered milk.

Isolated recta or ilea were mounted as flat sheets in miniaturized versions of the Ussing-style chambers described by Williams et al. (1978) (similar to those of Wood, 1972) with 2.0 rather than 5.0ml of solution per chamber (Thomson, 1990). Saline was constantly circulated and oxygenated in both chambers by means of a gas lift pump, which maintained constant gas tension and circulation regardless of perfusion flow rates. Provision was made for the gravity-fed perfusion (4–5 ml min⁻¹) of each chamber. Typically, recta or ilea were brought to steadystate conditions (defined by stable I_sc; after approximately 2h) under bilateral perfusion, and then perfusion was stopped unilaterally (but mixing was continued) during the experimental period when acid-base transfer was measured.

Rates of luminal acidification (J_H) and contraluminal alkalization (J_OH) were determined using a pH-stat technique (PHM84 research pH meter, TTT80 titrator, ABU80 autoburette; Radiometer, Copenhagen, Denmark). J_H and J_OH were calculated as the rate of titrant addition (0.01 mol l⁻¹ NaOH and 0.01 mol l⁻¹ HNO₃, respectively) required to maintain the initial pH.

Transepithelial potential (V_t), short-circuit current (I_sc) and calculated resistance (R_t) were determined as described by Hanrahan et al. (1984). Briefly, V_t was measured between 3 mol l⁻¹ KC1 agar bridges located in the two chambers close to the epithelium. Short-circuit current was applied with a dual-channel automatic voltage clamp which allowed for compensation of saline resistance (see Hanrahan et al. 1984, for circuit description) through flat-sheet Ag or Ag-AgCl electrodes (0.29 cm² area) at the ends of each chamber. In other experiments these electrodes were replaced by salt bridges leading to separate containers for AgCl electrodes. These three arrangements are referred to as the Ag, Ag-AgCl and salt-bridge electrode configurations, respectively. V_t and I_sc were recorded on dual-channel strip-chart recorders (1242: Soltec, Sun Valley, CA).

The composition of the experimental saline, which lacked phosphate and bicarbonate, was otherwise based on that of locust haemolymph as described by Chamberlin and Phillips (1982) and Hanrahan et al. (1984). Salines were buffered with 2 mmol l⁻¹ Mops (pK_a=7.20 at 20°C) and vigorously aerated with 100% O₂ for at least 2h prior to use. This protocol maintained CO₂/HCO₃⁻ at the trace levels necessary for precise estimation of J_H and J_OH (Thomson, 1990). The pH electrodes were calibrated with Radiometer precision buffer solutions (pH±0.005) and the saline was manually titrated to pH7.00 prior to experiments at 23±1°C. All values reported are means±standard errors. Statistical significance was determined using paired or non-paired r-tests. Differences were considered statistically significant if P<0.05.

We previously reported (Thomson et al. 1988a,b)that, under open-circuit conditions in Ussing chambers, unstimulated recta maintain a steady J_H of 1.54μequiv cm⁻²h⁻¹ for at least 8h and that this secretion is completely and quickly abolished by 1 mmol l⁻¹ KCN. This in vitro rate compares favourably with in situ estimates of J_H. Short-circuiting (Ag electrode configuration) reduced J_H significantly to 0.93μequivcm⁻²h⁻¹. This was somewhat puzzling since the treatment should have very slightly enhanced J_H, given that the small opposing V_t of 6–8mV was abolished. In the present study, when we used the salt-bridge configuration to pass current, we observed a J_H under short-circuit of 1.51±0.10μequivcm⁻²h⁻¹(N=6), which is not significantly different from opencircuit results, suggesting an experimental artefact in earlier short-circuit experiments.

We then investigated preliminary reports by Hanrahan (1982) that cyclic AMP initiated alkalization on the lumen side of short-circuited recta. When Hanrahan’s (1982) conditions were duplicated exactly (apart from chamber size) high rates of luminal alkalization were observed after adding 1 mmol l⁻¹ cyclic AMP to the haemocoel side (Fig. 1). This J_OH was approximately 80 % of the simultaneous I_sc and rates of alkalization were virtually identical to the net Cl ⁻ fluxes (J_C1) reported by Hanrahan (1982) under similar conditions. (We reconfirmed that stimulated I_sc=J_CI); data not shown.) Moreover, when luminal Cl⁻ was removed (gluconate substitution) bilaterally from the bath, Δ I_SC, Δ I_C1 and J_OH were all abolished (Fig. 1). Other perturbations that inhibited stimulated J_CI [e.g. bilateral 1 mmol l⁻¹ KCN, or 1 mmol l⁻¹ DPC (N-phenylanthranilic acid) on the haemocoel side] also inhibited J_OH (data not shown).

Superficially these data appeared to provide very convincing evidence for a Cl⁻-dependent alkalization mechanism (e.g. C1⁻ /OH⁻ or C1⁻/HCO₃⁻ exchange), as suggested earlier for locust ileum, where similar results were observed (Irvine et al. 1988). However, this interpretation (i.e. an apical neutral anion exchanger) implies that active CP transport cannot account for the observed stimulated I_sc, contrary to considerable earlier evidence (reviewed by Phillips et al. 1986).

Upon closer inspection this interpretation proved to be false. If base equivalents were being transported transepithelially, the increase in luminal pH should be accompanied by a concomitant decrease in haemocoel pH (i.e. under steady-state conditions, the change in H⁺/OH⁻ activity on both sides of the epithelium should be of similar magnitude but opposite direction). When the above experiments were repeated, significant changes in rates of contraluminal alkalization were observed, but they were several orders of magnitude less than the rate of acidification observed in the lumen. There was never acid movement to the haemocoel side under any conditions (Table 1). This suggested that hydroxyl equivalents might have been added to the lumen from an external source rather than from the epithelium itself. To test this, Ussing chambers were set up as above without recta and 80 μA of current (a typical current under cyclic-AMP-stimulated conditions) was passed across the current-sending electrodes. As predicted, the luminal bathing saline alkalized at the same rate whether a tissue was present or not (Fig. 2B, Ag electrode configuration, filled circles). Therefore, the luminal alkalization observed by Hanrahan (1982) for the rectum and by Irvine et al. (1988) for the ileum under short-circuit conditions was clearly an experimental artefact.

The combined effect of these reaction sequences is the replacement of bath Cl⁻ by an equal number of OH⁻, thus explaining the similarity between rates of Cl⁻ flux, I_sc and alkalization. The alkalization appeared to be a cellular event modulated by cyclic AMP simply because cyclic AMP stimulated active electrogenic Cl⁻ transport, which in turn was accompanied by a concomitant increase in the amount of current passed between the Ag electrodes (and hence the amount of OH⁻ formed luminally).

This is not a new or novel observation. The problems associated with passing current across bare silver electrodes have been known to electrophysiologists for decades. Under most circumstances, problems with unwanted products of electrolysis can be eliminated by coating the silver electrode with AgCl (either electrolytically or by dipping it in molten AgCl). If Ag-AgCl electrodes are used, the reaction at the anode is the same as above, but Cl⁻ is electrolytically released at the cathode from the AgCl and Ag is deposited on the electrode. Theoretically, with Ag-AgCl electrodes, the quantity of Cl⁻ removed from the bath at the anode will be exactly matched by the quantity of Cl⁻ (rather than OH⁻) released at the cathode. In practice, however, we found this not to be the case.

The problem with Ag-AgCl electrodes is that the formation of unwanted electrolytic by-products (e.g. OH⁻ at the cathode) is directly related to the quality and quantity of the AgCl coating on the electrode. Intuitively this is obvious, but empirically the quality of the coating is generally very difficult to judge, or to control completely, and it changes with time during usage. We found that protection from electrolytic by-products, as judged by OH⁻ production, varied significantly amongst Ag-AgCl electrodes that appeared visually very similar and which should have been very well coated with AgCl (data not shown). Moreover, the rate of pH change varied significantly (and often dramatically) with the intensity and length of time for which the current was passed (Fig. 2B; Ag-AgCl electrodes, open circles). Although freshly prepared Ag-AgCl electrodes are adequate for long periods at low current levels (10 μA; similar to unstimulated, steady-state levels for locust hindgut; Fig. 2A), the degree of protection from OH⁻ production is very limited and of very short duration at the high currents (80μA, or 400μAcm⁻²) typical of stimulated locust hindgut. The I_sc is even greater for unstimulated lepidopteran midgut. Moreover, short-term control experiments without epithelia present in the Ussing chambers may not reveal the problem of OH⁻ production, which only appeared after 0.5 h. Clearly Ag-AgCl electrodes by themselves are not suitable for passing large amounts of current in weakly buffered bathing solutions where exogenously induced changes in bath pH cannot be tolerated. Adequate protection can only be obtained by agar bridges (Fig. 2B; triangles). Unfortunately, salt bridges have not been used in most previous studies of insect epithelia.

We found that electrode by-products do not influence the other transport parameters (i.e. other than J_H/J_OH) that we reported previously for locust rectum and ileum. For example, when we repeated Hanrahan’s (1982) and Hanrahan and Phillips’ (1984) experiments on short-circuited locust rectum using agar bridges, the measured V_t, I_sc (electrogenic Cl⁻ transport) and R_t were completely unaffected by the change in electrode configuration, both before and after stimulation with cyclic AMP (Table 2). This was expected, because Hanrahan used Ag electrodes only after he had observed no difference from preliminary experiments using Ag-AgCl electrodes (J. W. Hanrahan, unpublished data). Ag-AgCl electrodes were used previously in our laboratory (e.g. Williams et al. 1978). Moreover, with the standard well-buffered saline used in previous experiments on locust hindgut (5 % CO₂,10 mmol l⁻¹ HCO₃⁻) we found that saline pH changes by 0.1 unit, at most, after several hours at high current densities, when Ag electrodes are used to pass current (Fig. 3). Cl⁻ transport, for example, is not affected by changes in saline pH between 6.0 and 8.0 (Hanrahan and Phillips, 1982). In summary, we conclude that previous results for insect hindgut obtained using Ag or Ag-AgCl current-passing electrodes are valid unless they involved determinations of acid-base transfer, or unless mechanisms very sensitive to slight changes in luminal pH were studied.

Finally, we have re-investigated acid-base transfer across locust rectum and ileum under short-circuit conditions using the salt-bridge configuration at a bilateral pH of 7.00. Under CO₂/HCO₃⁻-free conditions, the acidification of the rectal lumen equals the alkalization of the haemocoel side under both open-and short-circuit conditions and these rates are not significantly changed by shortcircuiting (Table 3). Moreover, under both open-and short-circuit conditions, cyclic AMP caused a significant reduction in J_H of 66 % and 42 %, respectively (Table 4). This suggests that rectal J_H is probably under hormonal control in situ. In contrast, 5 mmol l⁻¹ cyclic AMP did not change locust ileal J_H (Table 5), which is similar in rate to unstimulated rectal J_H. N. Audsley (personal communication) in our laboratory has found that extracts of both corpora cardiaca and ventral abdominal ganglia 4–7 completely inhibit ileal J_H. This implies that ileal acid secretion is also under hormonal control, but that a second-messenger system other than cyclic AMP is probably involved. The nature of the putative control system in situ remains to be investigated.

Both lepidopteran midgut (Dow, 1984) and locust hindgut (Thomson et al. 1988a) maintain large pH differences by transport of acid-base equivalents. Short-circuited preparations provide a means of quantitatively measuring and characterizing active transport. It is perhaps fortunate that these acid-base transfer processes have been relatively little studied by this method, because we have shown that the procedure commonly used in the past of omitting salt bridges for current-passing electrodes leads to serious errors. Based on our recent experience, Chamberlin (1990) used agar bridges during measurements of alkaline secretion by Manduca sexta midgut under short-circuited conditions (see also Dow and O’Donnell, 1990). However, there is no reason to believe that the production of reaction products at current-passing electrodes in the absence of salt bridges necessarily affects previous estimates of other transport processes in these tissues, at least when well-buffered solutions were used so that saline pH did not change significantly.

When salt bridges are used, estimates of acid secretion by locust ileum or rectum under short-circuit conditions are similar to those in the open-circuit state and in situ (Thomson, 1990). This is accompanied by equal movement of base equivalents to the haemocoel side. Contrary to earlier reports (Irvine et al. 1988), alkalization of the lumen is never observed, even after Cl⁻ transport is stimulated 10-fold to about 10μequivcm⁻²h⁻¹ by addition of cyclic AMP. Cyclic AMP addition merely reduces rectal J_H-The cyclic-AMP-induced change in J_H is an order of magnitude lower than the concomitant increase in J_C1, indicating that H⁺ and Cl⁻ fluxes are not coupled. In support of this view, complete bilateral Cl⁻ replacement does not affect rectal J_H (Thomson, 1990; Phillips et al. 1986). Moreover, stimulated I_sc is independent of external pH over a wide range from pH 6.0 to pH 8.0 (Hanrahan and Phillips, 1982). Hanrahan and Phillips (1984) excluded Cl⁻ transport by exchange for HCO₃⁻ at the apical border of locust rectum. Our observation that, in the absence of CO₂/HCO₃⁻ in the saline, stimulated I_sc is not accompanied by any movement of base equivalents into the lumen removes the previous nagging possibility that some component of Cl⁻ transport might occur by OH⁻/C1⁻ exchange in locust ileum and rectum. In summary, the results described in this paper greatly strengthen the conclusions of Hanrahan and Phillips (1984) concerning the nature of the Cl⁻ pump in locust hindgut and make it much less likely that this anion transport could be driven secondarily by primary active proton secretion and recycling (i.e. by HC1 cotransport).

Changes in luminal pH of the hindgut in situ, in response to acid injection into the haemolymph and to starvation, previously suggested that acid secretion in locust hindgut in situ is controlled so as to aid in whole-body pH regulation (Thomson et al. 1988a). Thomson et al. (1988a) showed that this control of rectal is not mediated directly through changes in haemocoel pH. More recently, Thomson (1990) has also shown that this control is not mediated directly by changes in haemolymph or [HCO₃⁻]. He has concluded that a significant fraction of the acid-base transport observed in the rectum must be under hormonal control. In support of this hypothesis, we found that a common second messenger of neuropeptide hormones, namely cyclic AMP, inhibits rectal acid secretion in vitro (Table 4). Similarly, extracts of both corpus cardiacum and ventral ganglia have been found to reduce greatly the normally basic pH and bicarbonate levels in absorbate from locust ileal sacs (Lechleitner et al. 1989), while N. Audsley (personal communication) recently found that the same locust glandular extracts inhibit acid secretion by locust ileum in Ussing chambers. These studies from our laboratory provide the first evidence that acid-base transport in locust hindgut is normally under hormonal control.

This work was supported by operating grants to JEP and postgraduate fellowships to RBT from NSERC, Canada.