Active Transport by the Cecropia Midgut: IV. Specificity of the Transport Mechanism for Potassium

In the preceding paper of this series evidence was presented that the sole ionic requirement for the maintenance of the midgut potential is the presence of potassium in the solution bathing the blood-side of the isolated organ (Harvey, Haskell & Nedergaard, 1968). The conclusion was reached that the potential arises directly as a consequence of the active transport of potassium from blood-side to lumen. This conclusion depends on the demonstration that potassium is the only ion actively transported under conditions such that the mid-gut potential can be demonstrated. Harvey & Nedergaard (1964) had already shown that at least 83 % of the current produced by the midgut when the potential is short-circuited is carried by potassium. Using an improved method of short-circuiting, including high potassium concentrations in the bathing fluids, Harvey, Haskell & Zerahn (1967) found that all of the current is carried by potassium. In the present paper the specificity of the midgut for potassium is examined in further detail. It is shown that rubidium, which so closely resembles potassium chemically, can substitute fully for potassium but that other ions such as sodium or lithium and even the chemically similar ammonium ion are unable to sustain a current. Moreover, the magnitude of the current depends on the potassium concentration in the blood-side bathing solution.

Mature fifth-instar larvae of Hyalophora cecropia(Linnaeus) were reared in the vicinity of Amherst, Massachusetts, during the spring and summer of 1963 and 1964. During March, April and May the larvae were fed with leaves harvested from branches of willow (Salix babylonica) forced in the greenhouse. As weather conditions improved the larvae were moved outside to netted trees of this same species. Larvae which hatched after June were reared outdoors on wild black cherry trees (Prunus serótina). The fifth-instar larvae used in the experiments reported here weighed from 6 to 18 g. with an average weight of 13 g.

Prior to dissection each larva was weighed, isolated in a small cardboard box, and immobilized by chilling at 5° C. for 0·5−1 hr. It was then placed on a cooled paraffin block and covered with a sturdy wire screen bent to allow a 1 in. to $112$ in. space between the screen and the paraffin. A cube of frozen carbon dioxide (approximately 3 in. × 2 in. × 1 in.) was placed on the screen and the assembly was covered loosely with paper tissue. In this way the larva was simultaneously chilled and narcotized by the cold carbon dioxide flowing down through the wire mesh. Ordinarily the larva was narcotized for from 5 to 10 min. or until it has regurgitated a small amount of yellow fluid and become very flaccid.

The larva was removed from the carbon dioxide but kept on the chilled paraffin block during dissection. The portion of the insect from one segment anterior to the prolegs to between the second and third prolegs was removed with scissors, and the abdominal tergites were cut longitudinally along the dorsolateral surface to expose the midgut. In order to standardize the preparation as much as possible only the middle part of the midgut was used in experiments reported in this paper. Although there is not much difference in the fine structure of the midgut cells from the anterior to the posterior part (Anderson & Harvey, 1966) there is a greater difference between the pH of midgut contents and the blood in the hind part of the midgut than in the front part. Likewise the potential difference across the isolated organ is largest in the hind part of the midgut. However, potassium is actively transported along the entire length of the organ. The peritrophic membrane containing all of the gut contents was removed by gently pulling. Small lengths of nylon twine were looped around the free ends of the gut. These ends were slipped over the ends of the glass tubing which formed the limits of the gap of the inner chamber (Fig. 1) and the twine was tightened. The chamber was lifted slightly and the ventral attachments of the midgut to the integument were severed. With practice, the dissection and cannulation could be accomplished in three minutes.

The chamber, blown from Pyrex glass, consists of an inner and outer section. Because there is no connexion between the solutions in these sections except through the tissue of the midgut mounted in the inner section, it is possible to perfuse the two sides of the midgut independently. The gap in the inner part of the chamber, where the gut tissue is placed, is 1 cm long and 0·6 cm wide, which gives an area of about 1·9 cm². To establish the same fluid level in each chamber, and thereby avoid an hydrostatic pressure gradient across the gut, the fluid volume of the inner chamber was 10−12 ml. and that of the outer chamber 60−65 ml. The two solutions were stirred and aerated by bubbles of air (A and A’ in Fig. 1) which had been filtered through a Koby filter and saturated with water vapour. The temperature was kept constant during the experiments by inserting the chamber into a beaker of water whose temperature was under thermostatic control.

The potential difference across the isolated midgut was measured by means of two calomel electrodes (C and C′) which were connected to corresponding bathing solutions by bridges (E and E’), one on either side of the midgut tissue. Each bridge consisted of a 10 cm. length of glass tubing (O.D. 1·5 mm.) inserted into a section of no. 160 Intramedic polyethylene tubing and filled with a 3 % agar solution of the bathing fluid being used. The bridges were fixed so that the glass tips were directly opposite each other on either side of the midgut tissue at the midpoint in the gap of the inner chamber. The potential was read by a PHM4 Radiometer potentiometer (mV in Fig. 1).

The short-circuiting technique of Ussing & Zerahn (1951) was adapted to the midgut system. Two silver-silver chloride electrodes (S and S’ in Fig. 1) were placed in the bathing solutions one on each side of the tissue. Unfortunately the conductivity of the standard bathing solution (SI) is low, because of the low concentration of ions, the osmolarity being made equal to that of larval haemolymph by using sucrose.

As a consequence of this and of the geometry of the apparatus the current density is greater at one end of the midgut than at the other. However, if the glass bridges of the potential-measuring electrodes are placed carefully in the middle and fixed there it is reasonable to assume that something close to the actual short-circuit current is measured (see also Harvey, Haskell & Zerahn, 1967).

The osmolarity of blood of mature fifth-instar larvae was measured by comparing its freezing-point depression with that of standard solutions (Gross, 1954). For this purpose haemolymph was drawn from prolegs of chilled, narcotized larvae into glass capillaries which were then sealed with petroleum ‘jelly’ and quickly frozen in a mixture of alcohol and solid carbon dioxide. The thawing time of the capillaries filled with whole blood was compared with that of capillaries filled with solutions of known osmolarity. The average osmolarity of haemolymph samples taken from sixteen fifth-instar Cecropia larvae was 259 ± 4·3 (standard error) m-osmoles/l. Although the osmolar activity of cell-free blood was not determined, preliminary experiments revealed that a bathing solution with a concentration at least 260 m-osmoles/l. was adequate to preserve high potentials over long periods of time. All of the solutions were brought to this osmolarity by adding sucrose.

The amounts of potassium (32 mM/l.), calcium (5 mM/l.) and magnesium (5 mM/l.) in the standard solution were chosen from an analysis of the concentrations of these ions in haemolymph, midgut tissue, and midgut contents performed by Quatrale (1966), and from preliminary experiments designed to investigate the stability and longevity of the potential in solutions with varying concentrations of these ions. Calcium was included because of the well-known effects of the omission of this ion on membrane structure and permeability (e.g. Nakas, Higashino & Loewenstein, 1966). Because the midgut sometimes tended to contact rather strongly in solutions containing calcium but not magnesium, because magnesium is present in rather high concentrations in the midgut tissue (14 mM/l.) and in the blood (35 mM/l.), and because its concentration in the tissue changes markedly during the larval-prepupal-transformations (Quatrale, 1966), it seemed prudent to include magnesium in the bathing solution. Sodium was omitted from the standard solution because there is very little sodium in either haemolymph, midgut tissue, or midgut contents, and because its presence or absence had no effect on the maintenance of the potential.

All of the cations were introduced into the solutions as chloride salts except that the solutions were buffered with 2 mM/l. bicarbonate at pH 8·0, which is intermediate between the value for haemolymph (6 ·3) and midgut contents (9·4) (W. R. Harvey & R. Hamelin, unpublished results). The composition of the standard solution (SI) is given in Table 1. Fisher certified reagents and glass-distilled water were used for all solutions.

In double-label experiments the beta emissions of ⁴²K were counted with an aluminium shield to exclude most of the weaker ⁸⁶Rb emissions. After ten half lives of ⁴²K (half life 12·5 hr.) had elapsed the samples were counted again with the shield in place so that rubidium counts could be subtracted from the potassium counts. At this time there was still approximately 85 % of the ⁸⁶Rb (half life 18·6 days) and it was then counted without the shield.

Concentrations of potassium, sodium, lithium and rubidium were measured by atomic absorption spectrophotometry.

Even though the active transport of potassium is completely independent of sodium in the bathing solutions (Harvey & Nedergaard, 1964) the midgut might be able to transport sodium when it is present in a high concentration. To test this possibility 32 mM/l. sodium chloride was added to the standard solution and a corresponding amount of sucrose was left out to keep the osmolarity constant (Table 1). To test the effect of increasing the ionic concentration 32 mM/l. of choline chloride was added to the bathing solutions. In other systems such as the frog skin choline is an indifferent ion except in high concentrations. The short-circuit current was reversibly decreased about 25 % with added choline chloride (Fig. 2). The effect might be attributed to the increased choline, chloride, or ionic concentration (and thereby on K+ activity) or the decreased sucrose concentration. For present purposes it is sufficient to note that the replacement of the choline chloride by either sodium or lithium chloride had no further effect on the short-circuit current (Fig. 2) so that we can conclude that neither sodium nor lithium affects the current.

Following the sodium and lithium experiments samples were taken from the lumen-side and analysed by atomic absorption spectrophotometry. Quantitatively the analyses were not very accurate because the error is very high when the background concentration of ion is much larger than the expected increase in concentration. However, there was invariably a net flux of potassium toward the lumen in each of seven experiments, whereas there was no net flux of either sodium or lithium (Table 2). Therefore sodium and lithium are not transported actively across the midgut tissue even when they are present in the same concentration as potassium.

When all the potassium was replaced by ammonium, the short-circuit current dropped almost to zero and the recovery when the potassium was restored was rather limited. When half (16 mM/l.) of the potassium was replaced by ammonium the current dropped significantly (Fig. 3). Ammonium does not seem to substitute for potassium, the current observed in experiments with 16 mM/l. ammonium and 16 mM/l. potassium probably being carried by just the potassium.

The midgut potential drops to the same extent when the concentration of rubidium on the blood side is changed from 32 mM/l. to 2 mM/l. as when the concentration of potassium is similarly changed (compare Harvey & Nedergaard (1964) and Harvey, Haskell and Nedergaard (1968)) nor is there any effect on the potential when all the potassium in the bathing solutions is replaced by rubidium. Similarly, replacement of potassium by rubidium had no effect on the short-circuit current (Fig. 4). With 16 mM/l. potassium and 16 mM/l. rubidium in the bathing solutions the short circuitcurrent was carried to an equal extent by rubidium and potassium. The ratio of net flux of potassium to rubidium in samples taken from the lumen and analysed by absorption photometry was 1·4, which is not significantly different from the ratio of the concentration of potassium to rubidium, which was 1·1 (Table 3).

The dependence of the short-circuit current on the potassium concentration in the bathing solutions was investigated. When lowered potassium concentrations were desired sodium was substituted for potassium and when high concentrations were desired part of the sucrose was substituted by potassium so that the osmolarity remained constant (Table 1). However, in the high-potassium solutions the total ionic concentration was increased above its value in the standard solution. As described above, sodium itself has no influence on the short-circuit current in a concentration as high as 32 mM/l. and therefore can be used as an inert ion.

When the lumen-side concentration was held constant at 32 mM/l. potassium and the blood-side concentration was varied from 2 to 100 mM/l. potassium the current varied with the potassium concentration. Because the current normally falls with time it is not possible to obtain the same amount of current for the same potassium concentrations when the latter is increased again. This is illustrated in Fig. 5, where the two sets of current values are plotted against the blood-side potassium concentration. However, when the percentage of the short-circuit current generated with 32 mM/l. potassium on the blood side is plotted against the blood-side potassium concentration (Fig. 6) the dependence of current on potassium concentration is seen to be the same whether the potassium level is being dropped or raised. The results of twelve similar experiments are summarized in Table 4 and plotted as Fig. 7. It is seen that the short-circuit current increases with increasing potassium concentration on the blood-side up to at least 64 mM/l. potassium. Figure 7 also shows that the short-circuit current is not affected by the potassium concentration on the lumen-side within the concentration range studied. The blood-side potassium concentration was 32 mM/l. during the potassium concentration changes on the lumen-side.

The ability of rubidium to replace potassium so completely suggested another way to confirm the previous findings that most of the current is carried by potassium. ⁴²K was used to measure the outflux and simultaneously ⁸⁶Rb was used to measure the influx (or vice versa) on a single piece of midgut tissue. In seven measurements the net potassium-rubidium flux accounted for about 87 % of the short-circuit current, a value close to the results obtained from ‘sandwich’ experiments in which a measurement of flux toward the lumen was sandwiched between measurements of flux toward the blood-side (Harvey & Nedergaard, 1964). The short-circuiting technique was the same as that used in the sandwich experiments and limited the accuracy of the overall determination so that the agreement between short-circuit current and net flux could not be expected to be better.

Potassium appears to be the only ion specifically needed for the generation of a substantial electrical current by the isolated midgut. Neither sodium, calcium, nor magnesium ions are necessary and a current is maintained in a solution containing only potassium chloride and sucrose (Harvey & Nedergaard, 1964). The chloride is not specifically required either, because a solution of potassium bicarbonate and sucrose will suffice (W. R. Harvey, S. Nedergaard & K. Zerahn, unpublished results).

It is more difficult to make a conclusive statement about what ions the midgut can transport without testing every known ion. The fact that most of the short-circuit current in the standard solution is carried by potassium virtually excludes the other ions present, i.e. magnesium, calcium, sodium, hydrogen, chloride and bicarbonate ions (Harvey & Nedergaard, 1964; Harvey, Haskell & Zerahn, 1967; Results 5). The failure to detect a net flux of either sodium or lithium under conditions such that a net potassium flux toward the lumen is invariably demonstrable (Table 2) shows that neither of these ions is transported even when they are present in high concentration in the bathing fluids. Although the ammonium ion is similar to the potassium ion the short-circuit current decreases rapidly when ammonium replaces all the potassium. Moreover, the current drops irreversibly when half of the potassium is substituted by ammonium (Fig. 3). At the pH of the bathing solutions (c. 8·0) about 95 % of the added 16 mM/l. ammonium is in the ionic form.

The only ion which can substitute for potassium is rubidium. When all the potassium in the bathing solutions is substituted by rubidium there is no detectable change in the short-circuit current (Fig. 4). Even when the midgut is maintained for hours with rubidium replacing potassium and the tissue is washed repeatedly to remove traces of potassium which may leak from the cells the current is maintained. Potassium and rubidium contribute to the net flux across the tissue and thereby to the shortcircuit current approximately in the ratio in which they are present in the bathing solutions (Table 3). Even with the technical limitations of these absorption measurements mentioned in the results the large net fluxes of potassium and rubidium in Table 3 contrast strongly with the negligible net fluxes of sodium and lithium in Table 2. This ability of rubidium to substitute for potassium is not surprising because in addition to their chemical and physical similarities rubidium is known to substitute for potassium in living cells such as red blood cells (Solomon, 1952).

The short-circuit current and therefore the active potassium transport is dependent on the potassium concentration in the blood-side bathing solution but is completely independent of the potassium concentration on the lumen-side (Fig. 7). The values in Table 4 for 100 mM/l. K are not included in Figure 7 because the lowered current may be due to a lowered sucrose concentration, rather than the elevated potassium concentration. Unpublished results indicate that it is not possible to maintain the shortcircuit current in the absence of sucrose. The details of curves such as Fig. 7 depend very much on exactly how the measurements are made and are valid only for the conditions stated. However, a similar sort of curve showing a dependency of current on blood-side potassium concentration would be expected under most conditions.

The ionic requirements for the short-circuit current of the isolated midgut can be summarized by saying that potassium must be present on the blood-side and perhaps in the midgut tissue. The only ion present under natural conditions which to date has been shown to be actively transported across the midgut is potassium. Taken together with the results reported in the preceding paper of this series we can state that only potassium is required for both the potential and the current and that under most conditions all of the current is carried by potassium ions. The Cecropia midgut epithelium, simpler in structure and more active in transporting ions than any other epithelial membrane hitherto described, is also simpler in its ionic requirements. Apparently it is a pure potassium pump.

This research was supported in part by a research grant (AI-04291) from the National Institute of Allergy and Infectious Diseases, U.S. Public Health Service, and a grant from the University of Massachusetts Research Council. We thank Dr K. Zerahn for criticizing the manuscript.