ABSTRACT
The perfused cockroach midgut ventriculus in vitro maintains a p.d. of approximately 12 mV. (lumen negative to haemolymph) in chloride Ringer. The p.d. comprises chiefly a lumen-side Na diffusion potential and a haemolymph-side K diffusion potential.
The size of the p.d. is greatly enhanced in sulphate Ringer, but reduced by various metabolic inhibitors, e.g. N2 and 2,4-dinitrophenol. Ouabain inhibits the p.d. from the haemolymph side but is ineffective on the lumen side.
The ventriculus preparation also exhibits a net Na efflux of approximately 0 · 3 µ-equiv./hr. Ouabain inhibits Na efflux by almost 55 % and K influx by less than 40 %.
It is suggested that a linked Na-K pump is indirectly responsible for the p.d. and directly involved in the Na transport into the haemolymph.
INTRODUCTION
In the only detailed, quantitative investigation of electrolyte movement in an insect midgut, Harvey and his associates (Harvey & Nedergaard, 1964; Harvey, Haskell & Nedergaard, 1968) have discovered an electrogenic potassium pump in the larva of Hyalophora cecropia. This insect is phytophagous and has a very specialized high K-low Na ratio in its haemolymph, unlike the vast majority of insects including the cockroach. Thus it may be that this particular pump is an adaptation to the high K content of the diet, rather than a feature characteristic of insect midguts in general.
Treherne (1957) showed that in the cockroach the midgut caeca were the major sites of glucose uptake, and that absorption occurred down a concentration gradient created by the rapid removal of water from the midgut lumen. In subsequent studies it was found that the changes in ionic concentration which resulted from ingestion of isosmotic solutions of sucrose, NaCl or KC1 were not entirely accounted for by the observed changes in volume of the midgut (O’Riordan, 1968). It appeared that the midgut wall in vivo was permeable to Na, K and Cl ions, and to water, and that the regulation of cations in the midgut fluid was energy-dependent. It therefore seemed possible that the water movement which Treherne (1957) observed could have resulted secondarily and passively from a net haemolymph-directed solute flux, perhaps brought about by ion pump(s) situated in the water-permeable midgut wall.
Recent experiments by Sauer, Schlenz-True & Mills (1969) using an in vitro cockroach midgut ‘sac’ indicate that rapid solute absorption takes place in the absence of significant net water movement. A DNP-sensitive mechanism appears to control water passage across the midgut wall, and by means of this the danger of haemolymph and tissue desiccation is considerably reduced. These authors identify the midgut as a most important osmoregulatory organ playing a vital homeostatic role in the intact animal.
Ionic movements across the cockroach midgut have now been studied in more detail by employing a perfused isolated preparation of the ventriculus. Radioactive flux measurements, and the electrical potentials recorded under various conditions, are described in this paper.
MATERIALS AND METHODS
Electrical potentials
Adult male cockroaches reared on a mixed diet and water were taken directly from culture for experimental use. After decapitation the entire midgut was dissected out from the dorsal side, the caeca were cut off, and the detached ventriculus was transferred to Ringer solution where the peritrophic membrane was gently removed from the lumen. The perfusion apparatus (Fig. 1) consisted of a wax block 4 x 4 x 1·5; cm. This had a central chamber of 1· 5 ml. capacity through which Ringer solution, constantly aerated and stirred, flowed at 12 ml./hr. After thorough rinsing the isolated ventriculus was mounted in its natural curved position across this large chamber by tying both ends with chiffon thread onto small (approximately 500 μ internal diameter) microelectrode glass nozzles. Ringer solution containing 0·01 mole/l. of the dye amaranth was next briefly and gently flushed through the inside of the ventriculus to test for leaks, and then both sides of the preparation were rinsed with normal Ringer. Following this a motor-driven hypodermic syringe was connected to the lumen-side inlet and Ringer was moved through the isolated midgut at a rate of 5 μl./min. into a small chamber. Two Ringer-3 % agar bridges made contact with the two bathing Ringer solutions as shown; they were connected through two KCl-calomel half-cells, and the potential difference (p.d.) across the ventriculus wall was measured on a Philips GM 6020 millivoltmeter having a 100 MΩ input impedance. Ringer solution levels in the small (lumen-side, R1,) and large (haemolymph-side, Rh) chambers were kept constant, and evaporation was reduced by a glass cover. After a minimum equilibration period of 30 min. at room temperature (23°C) in normal Ringer (ÆC1), an experimental Ringer was substituted onone or both sides of the gut, and the new p.d. was recorded for some minutes ; then RCl was restored and measurements continued.
The compositions of the Ringer solutions employed are shown in Table 1. In every case experiments were repeated on at least 10 preparations, and the results illustrated are typical individual records.
Fluxes of 22Na and 42 K
Isolated ventricular preparations from the midguts of adult male cockroaches were mounted as described in the previous section. The perfusion apparatus used here for flux measurements was almost identical to that described above (Fig. 1). It differed only slightly in that it was mounted on a magnetic stirrer, had a tightly fitting glass fid and lacked inlet and outlet tubes for the haemolymph-side Ringer. The isotopes were obtained as chloride salts from A.E.R.E. at Harwell or the Radiochemical Centre at Amersham, and were used singly at specific activities of 10−25 μC./ml. of Ringer.
Each ventricular preparation was first equilibrated in normal chloride Ringer for 30 min., during which its transepithelial p.d. was continuously monitored. Preparations with unstable p.d.’s were discarded. For efflux measurements the outside of the ventriculus was covered by 1 ml. of previously aerated Ringer solution, while identical Ringer containing 42K or 22Na was perfused at 5μl./min. through the lumen. A magnetic flea mixed the Ringer bathing the outside (haemolymph-side) from which 25 μl. samples were periodically removed and replaced with isotope-free Ringer. Each sample was rinsed on to a small planchette with 200μl. of distilled water, dried down and assayed. The rate of isotope influx was measured by bathing the outside of the ventriculus in 1 ml. of isotope-labelled Ringer, and collecting all of the lumen-side Ringer as it passed from the outlet tube. Each experimental sample was counted to an accuracy of ±5% on a Labgear Dekatron counter under a G.E.C. Geiger-Müller window tube. Appropriate corrections were applied, including a decay factor for all 42K samples.
Between successive flux determinations the preparations and contaminated surfaces of the perfusion apparatus were rinsed for 30 min. with normal Ringer and then blotted dry. Five determinations of each flux were made. At no time was it found that an appreciable change in sodium or potassium concentration occurred in either bathing solution, nor was more than 1 % of the activity added to one compartment transferred to the other.
Each unidirectional flux was measured for 15 min. immediately following a 10 min. period in the presence of isotope during which fluxes became constant. Owing to the normal decline of both the p.d. and fluxes in all preparations, the measurement of one flux was normally sandwiched between two measurements of the flux in the opposite direction ; an average value was then taken from the latter. This arrangement was not possible, however, when inhibitor (ouabain) was used, since its effect was irreversible. In that situation the procedure was to take a normal flux measurement, rinse both sides of the ventriculus for 20 min. with normal Ringer, and then change to Ringer + 10−4 mole/1. ouabain on the outside of the preparation only. After this the ‘inhibited-flux’ was measured in the presence of 10−4 mole/1. inhibitor.
RESULTS
Electrical potentials
In a total of more than 100 experiments fresh, isolated midguts perfused on both sides with RC1 exhibited stable transepithelial potential differences ranging from 8−26 mV., with a mean value ± S.E. of 12 ± 1 mV. In every case the midgut lumen was negative to the haemolymph. Satisfactory preparations could be maintained for up to 9 hr., but were rarely used for more than 4 since the p.d. always very steadily declined at a rate of 0·4 − 0·5 mV./hr.
(a) Effects of various gases, pH and temperature
The substitution of oxygen for air on the haemolymph-side merely resulted in a short-lived 1 mV. rise, in p.d., i.e. no significant change. The presence of 95% nitrogen instead of air led only to a steady decline of 2 mV. in 15 min. (15 % fall from initial level); full p.d. was restored when aeration recommenced (Fig. 2). In contrast, 95% CO2 in the haemolymph-side Ringer induced an immediate, large (90%) drop in the p.d., though this too promptly rose again to its normal level as soon as air was re-applied. This response to CO2 is thought to result from a change in intracellular pH and perhaps also in membrane permeability, but acidified (pH 6) or alkaline (pH 8) Ringer on either side of the preparation only induced very small reversible falls in p.d. Chilled (10°C) Ringer inhibited the p.d., a prompt 30% fall gradually reversing as the Ringer warmed up to room temperature.
(b) Effects of chemical inhibitors
A substantial and rapid fall in p.d. was induced by 10−3 mole/1. 2,4-dinitrophenol when present in either of the bathing Ringer solutions, 10−4 mole/1. DNP also was effective, though slowly, on the haemolymph-side; but neither side responded to 10−5 mole/1. of this inhibitor (Fig. 3). Decreases in potential were partially reversible if not allowed to go to completion, i.e. zero p.d. lodoacetamide at a concentration of only 10−5 mole/1. inhibited the p.d. from the haemolymph-side, but even 10−3 mole/1. was ineffective on the lumen-side. Ouabain (strophanthin-G) was also ineffective when applied in the lumen-side Ringer, but led to a pronounced and irreversible potential decline on the haemolymph side (Fig. 4). Here, 10−5 mole/1. of this inhibitor caused a gradual decline of p.d., and 10−3 mole/1. a far more rapid one. 10−3 mole/1. phlorizin, an inhibitor of Na-linked glucose transport in mammalian intestine (Schultz & Zalu-sky, 1964b), was without effect on either side of the preparation.
(c) Effects of altered ionic composition of Ringer solutions
(i) Sulphate Ringer (RSO4)
The substitution of ions for Cl- ions in the Ringer on both sides of the midgut led to an increase of 75 % or more in the basic p.d., which was maintained until RCl was replaced (Fig. 5). The mean p.d. value of 80 preparations in RSO4 was 21 ± mV.
(ii) Choline Ringer (RCh)
When the midgut lumen was perfused by Ringer in which all but 2·2 m-equiv./l. Na+ was replaced with choline, the p.d. immediately fell to less than 20 % of its initial value (Fig. 6). This low level was maintained until-RC1 (Na+ =157 m-equiv./l.) was put into contact with the mucosa, when the p.d quickly rose again. RCh on the haemolymph-side induced a brief rise in p.d. followed by a steady and partly reversible decline.
(iii) Low Na-RSO4
On the lumen-side, RSO4 in which magnesium ions substituted for all but 2·2 m-equiv./l. Na+ caused an abrupt fall in p.d. (Fig. 7). This effect was immediately reversed by replacement of the original Ringer, in this case RSO4(Na+ = 157 m-equiv./l. As in the case of RCh (above) bathing the haemolymphside in low Na-RSO4 caused a rapid increase in p.d. followed by a steady decline, and an almost complete reversal of this on changing back to the initial Ringer.
(iv) Low K-RSO4
A low K-RSO4(magnesium substituting) bathing the lumen-side of the preparation had scarcely any effect on the p.d. (Fig. 8). On the haemolymphside, however, this Ringer induced a change in p.d. somewhat similar to that observed (above) with low Na-Ringer i.e. sudden initial increase of potential preceding a gradual decline.
(v) High K-RSO4
If the haemolymph-side was exposed to a RSO4 containing 123m-equiv./l. K+ the p.d. plunged dramatically to a lumen-positive value, from which it immediately rose again in normal RSO4(Fig. 9). High K-RSO4 bathing the lumen-side of the midgut caused a small (15−20%) but sudden reversible increase in p.d.
Fluxes of 22Na and 42K
The variation in the flux values obtained under each condition is thought to reflect small differences in the physiological state (indicated by the electrical potential) and in the sizes of the midgut ventriculi employed.
Normal fluxes are shown in Table 2. In every case the mean influx of an ion represented about one third of its efflux and took place against the chemical and electrical gradients. For sodium the net efflux ranged from 0·14 to 0·45 μ-equiv./hr./midgut, while for potassium it was 30% of this quantity, i.e. 0·041-0·139 μ-equiv./hr./midgut, 10−6mole/1. ouabain caused all transepithelial potential differences to fall to zero and reduced the sodium efflux by a mean amount of 0·2 μ-equiv./hr./midgut to 46% of normal (Table 3). It affected the movement of potassium differently, however, reducing mean influx of this ion by 0·014 /μ-equiv./hr./midgut to 61 % of normal, i.e; from being equivalent to 30% of normal efflux to only 19%.
DISCUSSION
Electrical potentials
From the results given above it is clear that a fairly small bioelectric potential exists across the cockroach ventriculus in vitro. There are two main reasons why this p.d. is unlikely to result from passive physical forces (e.g. diffusion, Donnan equilibrium) alone. Firstly, it can be maintained for very long periods of time in the absence of any chemical gradients between the midgut lumen and the haemolymph ; and secondly, it declines in the presence of various metabolic inhibitors, e.g. low temperature, N2, iodoacetamide or 2,4-dinitrophenol. The chemical inhibitions were generally relatively slow and irreversible, and one possible explanation is the difficulty of penetration by these molecules into the tissue. Certainly some poisons, e.g. cyanide and iodoacetic acid which had no effect on the p.d. (surprisingly, since DNP and iodoacetamide did) apparently failed to reach the appropriate sites (O’Riordan, 1968).
Of the effective inhibitors, DNP is generally thought to act by uncoupling oxidative phosphorylation (Slater, 1963), and would thus reduce the midgut p.d. if the energyrequiring processes responsible for the p.d. were dependent on aerobic metabolism.
But there is currently some confusion as to precisely how DNP affects cellular processes (e.g. Tosteson, 1955 ; Bricker & Klahr, 1966; Racket, 1968). Also iodoacetamide, commonly believed to inhibit glycolytic ATP formation (Hoffmann-Ostenof, 1963) is suspected by Caldwell et al. (1960) of affecting overall cellular permeability by combining with-SH groups in cell membranes. Accepting the normal interpretations of these inhibitors’ actions and considering the small inhibitory effect of nitrogen on the midgut p.d., one may suggest that this potential is probably not maintained entirely by oxidative metabolism. It may be that the cells responsible have a low energy consumption, and can function partly anaerobically using stored ATP and glycogen. Alternatively, some passive diffusion step rather than the energy requiring process may be the limiting factor in the whole system.
The fact that a low ouabain concentration on the haemolymph-side caused a fall of p.d. is very significant. This cardiac glycoside is believed to be a specific inhibitor of membrane ATP-ases and particularly of linked Na-K pumps in a variety of invertebrate and vertebrate tissues (e.g. gall bladder, (Dietschy, 1964); insect nerve, (Treherne, 1966). In the mammalian intestine (Cooperstein & Brockman, 1958) and kidney (Orloff & Burg, 1958), as in the cockroach midgut, it is without action if applied at the mucosal side (lumen side) of the epithelium. On the basis of this evidence one could therefore suggest that the cockroach midgut pump is of the linked Na-K variety, and is situated in the basal region of the epithelium, probably on the basal or folded lateral plasma membrane of the columnar cells where access to the haemolymph is relatively free (O’Riordan, 1968). In this location it is most likely that sodium ions would be pumped out of the cells into the haemolymph in exchange for potassium ions, since in vivo the sodium concentration is relatively low in the epithelial cells and the midgut lumen and high in the haemolymph, while the converse is true for potassium (Table 4).
A linked cation pump of this kind would contribute nothing directly to a transepi-thelial p.d. if equal numbers of Na and K ions were exchanged by it on a 1:1 basis since no net charge difference would be generated across that cell border. Histological tests demonstrated free polyanionic groups on the apical microvilli of the epithelial cells and on the peritrophic membrane (O’Riordan, 1968) : these may contribute to a passive Donnan equilibrium across the midgut wall, but they alone cannot account for a p.d. of the magnitude observed.
The possibility that chloride ions pumped from the haemolymph contribute to the lumen negativity is discounted by the experiment using RSO4(Fig. 5). The substitution of non-penetrating sulphate ions for smaller chloride ions led not to a decrease but to an increase in p.d. of almost 100%. Thus not only was a chloride pump not responsible for the p.d., but also, in normal Ringer (RCl) chloride ions must have been leaking through the midgut wall, thereby partly short-circuiting the p.d. and disguising its size.
The involvement of Na and K in the transepithelial potential is confirmed by the experiments in which the concentrations of these ions in the Ringer solutions were altered. The luminal border of the midgut was fairly indifferent to the ambient K concentration, judging from the very small p.d. changes which were induced by an approximately 100-fold difference in K concentration in the bathing Ringers (Figs. 8, 9). In contrast, this border was very sensitive to ambient Na concentration, for a reduction in this promptly induced a reversible drop in potential (Figs. 6,7). The steady p.d. decline observed when RCh or low Na-RSO4 bathed the haemolymph-side is thought to result from leakage of choline or magnesium ions respectively into the cells, but other explanations are possible. This side of the tissue appeared to be more permeable to K ions; in high K-RSO4(Fig. 9) the rapid plunge of p.d. to a maintained low level was presumably due to a reversal of the normal concentration gradient and a large K influx into the cells which inhibited the hypothesized pump. In low K-RSO4(Fig. 8) the p.d. fell steadily after the initial rapid rise as the enhanced K concentration gradient and intracellular K concentration declined.
This gives a value of 32 mV/10-fold K concentration change which approaches the theoretical value of 59 mV. for a K-electrode. At the lumen-side border there is at best only a 3·5 mV. change/10-fold Na concentration change. One may assume then that neither cell border is permeable to one species of ion alone, and that the total transepithelial p.d. is not made up simply of a lumen-side Na potential and a haemolymph-side K potential.
Fluxes of 22Na and 42K
Any interpretation of the working of the cockroach midgut based purely on transepithelial p.d. measurements must be hypothetical, since voltage measurements are insufficient evidence for a linked cation pump unless confirmed by short-circuit current measurements or quantitative chemical data. Technical problems defeated this author’s attempts to measure short-circuit currents, but the flux results given above are a satisfactory demonstration that the bulk of the lumen-side negative p.d. is identifiable with net cation transport from the midgut lumen into the haemolymph. Both the p.d. and net efflux of Na ions decline very slowly during normal in vitro perfusion, and more dramatically in the presence of ouabain. The ionic concentrations of all ions were initially equal on both sides of the midgut ; therefore since the net Na efflux takes place ‘uphill’ against the electrochemical potential gradient of the sodium ion it is an energy-requiring process, and, ignoring the effects of solvent drag, it is ‘active’ by the definition of Andersen & Ussing (1960).
In a few trial experiments in which 10−4 mole/1. 2,4-DNP was used as inhibitor the p.d. fell to zero but Na efflux was only reduced by 39% i.e. somewhat less than in the presence of ouabain. It appears that a greater proportion of the Na-transport is ouabain-sensitive than is DNP-sensitive. If DNP does indeed act by uncoupling oxidative metabolism (Slater, 1963), then the ouabain-sensitive pump must utilize at least some stored or anaerobically-derived high-energy compounds. A difference between DNP-inhibition and ouabain-inhibition of Na transport has also been observed in the isolated rabbit ileum (Schultz & Zalusky, 1964a).
It is interesting that the level of ouabain-inhibited Na efflux is not significantly different from the normal Na influx or K efflux. One may therefore suppose that under normal conditions the latter fluxes are passive, and that the balance of normal Na efflux above this quantity, i.e. 0 · 20-0 ·25 μ-equiv./hr. is moved by active forces.
The extent to which such effects as solute or solvent drag (Franz & van Bruggen, 1967; Franz, Galey & van Bruggen, 1968) affect the fluxes in these experiments is believed to be small, particularly in the case of sodium. Potassium ions were 13 times less numerous than sodium ions, but their mobility is 47% more (Katz, 1966), and it is possible that the very small net K efflux may be attributable to passive drag effects. Perhaps, too, there may be preferential absorption of Na ions by the fixed negative groups present in the lumen or muscularis (O’Riordan, 1968).
These levels are similar to the fluxes recorded in the rabbit intestine (Schultz & Zalusky, 1964 a), but are somewhat higher than Ussing’s (1949 a) figures for frog skin, and lower than the K-fluxes in the potassium-pumping isolated midgut of H. cecropia(Harvey et al. 1968).
Diffusion coefficients (D′) for Na and K can be estimated, from the data obtained, using Schultz & Zalusky’s (1964a) formula, t = λ 2/6D′. This equation is based on the premise that the time (i) taken to reach a steady-state diffusion rate of isotope through the tissue depends on the diffusion rate D′) of the isotope and the thickness of the tissue (λ). In the cockroach midgut this value for Na is double that for K, and equals 0 · 68 x 10−6 cm.2/sec. (4% of the diffusion coefficient of NaCl in water at 37° C). Elul (1967) attributes such low diffusion rates as this to fixed charge effects and steric barriers, but, in the case of K, mixing of this ion with a larger pool in the epithelial cells must also be important.
where k exchange constant, Mi and Mo are influx and efflux respectively and [ion]i and [ion]o are ion concentrations on the inside and outside of the midgut (which in this case are the same). From these calculations the tissue appears to be three times as permeable in either direction to K as to Na.
Functioning of the midgut
It seems reasonably certain from the experimental evidence outlined in this paper that a linked Na-K pump is situated in the cockroach midgut epithelium. Na ions diffusing into the epithelial cells from the midgut lumen would mostly be pumped out again on the other side, and then diffuse through the muscularis (muscle cells and connective tissue, mainly (O’Riordan, 1968)) into the haemolymph. By creating a solute gradient this net sodium transport (perhaps with concomitant passive Cl- movement) might induce an osmotic movement of water in the same direction. Hence the pump would be entirely responsible, indirectly, for the p.d., since it would maintain a low Na concentration and a high K concentration inside the cell. These levels are essential for the Na and K diffusion gradients, (one at each of the two cell borders) the algebraic sum of which accounts for the bulk of the p.d. Conceivably a lumen side Na pump might also contribute.
This is in fact the situation with respect to Na and K in frog skin, as interpreted by Koefoed-Johnsen & Ussing (1958). This tissue has very low water permeability and almost perfect electrode borders on each side. It transports Na ions inwards with the aid of an exchange pump on the inner barrier; but its p.d. is sensitive to ADH and increases in the presence of 10−5 mole/1. Cu++ ions, unlike the p.d. of the cockroach midgut. (O’Riordan, 1968). Thus the midgut tissue as a whole may function on the ‘double-membrane’ principle as proposed by Curran (1960) and Diamond (1965) for vertebrate tissues involved in solute and solvent transport. The first ‘border’, across which the pump operates, is represented by the lateral or basal plasma membranes of the epithelial cell, and the second (retarding diffusion into the haemolymph) by the muscularis connective tissue, etc. Solvent diffusion, if it occurs, would take place by osmosis into the narrow (200 A) intercellular spaces between the basal halves of adjacent epithelial cells. The recent paper by Sauer et al. (1969) casts doubt on the role of the midgut in passive water transport, but if such transport does occur, then the cockroach has a simple but effective, single system for absorbing salt, water and valuable organic materials. Owing to the relatively very large volume of the haemolymph compared with the midgut fluid even a small amount of solute or water absorptionmight considerably alter the osmotic balance between these two compartments.
Although linked pumps such as that hypothesized in this paper are widespread in vertebrate epithelia, few have yet been identified in invertebrates. Cockroach blood, like that of vertebrates and most invertebrates, has a high Na : K ratio, but in some phytophagous insects, e.g. H. cecropia, this ratio is reversed. It is possible that the electrogenic K-pump in the midgut, first described by Harvey & Nedergaard (1964) in H. cecropia, is therefore associated with that type of haemolymph composition and a K-rich diet. Such a condition is very unusual, however, and pumps of that kind may prove to be less common than a linked cation pump such as occurs in the cockroach midgut. The pump in H. cecropia appears primarily to remove K ions from the haemolymph into the midgut lumen. But the precise role of the pump in the cockroach midgut in absorbing solvent and solute or in homeostasis remains to be further clarified, for until recently the latter role has been attributed only to the rectum (Wall & Ralph, 1965; Sauer et al. 1969).
ACKNOWLEDGEMENTS
The material contained in this paper formed part of a Ph.D. dissertation at Cambridge University. I am most grateful to my supervisor Dr J. E. Treherne and to Dr J. A. Ramsay for their helpful advice and constructive criticism. I am also indebted to Drs B. L. Gupta, W. R. Harvey and S. H. P. Maddrell for useful discussions; and to the Science Research Council for financial support.