Rectal Absorption in the Desert Locust, Schistocerca Gregaria Forskål: II. Sodium, Potassium and Chloride^*

The need for an investigation of absorption from the insect rectum has been accentuated by the intensive investigations of Ramsay (1950–1958), which have made untenable the older hypothesis that the Malpighian tubules are alone responsible for ionic regulation. Ramsay (1956) concludes: ‘There is nothing in the response of the tubules to these variations [i.e. in the ionic composition of the bathing solutions] which might suggest that they are responsible for the maintenance of the normal composition of the haemolymph ; rather it is that the activity of the tubules alone would radically alter the composition of the haemolymph were it not for the participation of the other important organs of the excretory system, the rectal glands.’ Yet the most recent review of excretion in insects (Craig, 1960) has the following single comment to make in a section on the function of the intestine in excretion : ‘The rectal region of the hindgut has not been critically studied either histologically or physiologically since the advent of the newer methods so effectively used on the Malpighian tubes.’

Comparison of the ionic composition of hindgut and rectal fluids suggests that sodium, potassium and chloride are reabsorbed from the rectum in fresh-water insect larvae against large concentration gradients and that this process is responsible for the hypotonicity of the urine (Bone & Koch, 1942; Ramsay, 1950, 1951, 1953a; Shaw, 1955). Net absorption in all these cases rests on the assumption, supported by indirect evidence, that water is not actively secreted into the rectum as Harnisch (1934) supposed. Because electropotential measurements have not yet been made, it is not known whether cation or anion is actively transported. Shaw (1955a), on the other hand, finds no evidence that the low concentrations of sodium and chloride in the rectal fluid of Sialis lutaria larvae are due to active rectal absorption.

Amongst terrestrial insects, dependent as they are upon conservation of water by the production of concentrated excreta, less is known about solute absorption in the rectum. Patton & Craig (1939) found that radioactive sodium which was removed from the haemolymph by the Malpighian tubules of Tenebrio molitor larvae could be recovered in saline bathing the rectum isolated in a separate chamber. This has been generally interpreted in the literature as indicating net absorption in the rectum but the evidence by itself only demonstrated the permeability of the rectal wall to sodium.

More conclusive evidence for net absorption of ions from the rectum in terrestrial insects has been obtained by Ramsay (1952, 1955a). In recently fed Rhodnius the osmotic pressure and potassium concentration of Malpighian tubule fluid increased during passage through the rectum but that of sodium did not, suggesting that the latter ion was reabsorbed. In Dixippus Ramsay estimated that approximately 95 % of the sodium and 80% of the potassium secreted by the Malpighian tubules were reabsorbed in the rectum.

However, in all experiments which have so far been described, net absorption of solutes from the rectum in terrestrial insects can be explained satisfactorily by assuming passive diffusion down a concentration gradient created by active water absorption. Yet the investigations of Ramsay (1952–1958) on the function of the Malpighian tubules indicate that the regulation of haemolymph composition is due mainly to selective reabsorption in the rectum. This would pre-suppose the existence of more elaborate mechanisms than passive diffusion for the absorption of solutes from the rectum.

This paper describes experiments in which absorption of sodium, potassium and chloride from the ligated locust rectum was studied while water absorption was prevented by the use of hypertonic sugar solutions (see Phillips, 1964a). A study was first made of the ability of the rectal epithelium to absorb these ions against an electrochemical gradient and the lower limits of the absorption process were determined. Next, the rate of absorption of these ions was measured as their concentrations in the rectal fluid and haemolymph were varied. Finally, the extent of possible accumulation or storage of these ions in the rectal epithelium and of back diffusion into the lumen were investigated.

Mature male Schistocerca gregaria Forskål obtained from the Anti-Locust Research Centre, London, and reared in cages at 25–30° C. on a diet of wheat shoots and bran were used in all experiments. Details of the method used to measure solute absorption from the locust rectum are described in a previous paper (Phillips, 1964a). Briefly, insects are starved for several days to remove most of the gut contents and an operation is performed to isolate the rectum from the hindgut by ligation. The remaining rectal content is then rinsed out by alternate injection and removal of distilled water through the anus. Experimental solutions are then introduced into the rectum and sampled at various times for the estimation of solute concentration. The ionic concentration and composition of the experimental solutions may be varied over a wide range while water movement is controlled by adding sugars which are only very slowly absorbed. The net absorption of a solute from the rectum is determined by comparing changes in its concentration with that of a second solute which is not absorbed from the rectum (¹³¹I-labelled human serum albumin) and which therefore acts as an indicator of volume change. The percentage of injected solute which is absorbed can be calculated from the following equation:

where [I_i] and [I_f] represent the initial and final concentrations of the volume indicator and [X_i] and [X_f] the initial and final concentrations of solute respectively.

Samples of rectal fluid and haemolymph collected from the neck (Phillips, 1964a) were quickly transferred to a clean waxed slide and taken up in self-filling micropipettes. Determinations were carried out on aliquots of fluid and several rinsings delivered from such pipettes and true concentrations were calculated from readings obtained with similar aliquots of standard solutions.

Chloride was determined by potentiometric titration with 0·2% AgNO₃ according to the first method devised by Ramsay, Brown & Croghan (1955). In the present investigation standard deviations of less than ± 2 % were observed for several measurements on 0·3−4·0 μl. aliquots of a 25 mM/l. NaCl solution.

Sodium and potassium were determined with an E.E.L. flame photometer working at maximum sensitivities of 5 p.p.m. sodium and 10 p.p.m. potassium for full-scale deflection. Over this range there is no interference from any of the biologically important ions (Collins & Polkinhorne, 1952). Aliquots (0·3−4·0 /xl. in size) of the fluid to be measured were pipetted into 1–2 ml. of distilled water in capped Polythene vials. Samples were prepared in duplicate or triplicate on which a total of 4–6 readings were made and concentrations were read from calibration curves. Standard deviations for a series of similar samples of potassium and sodium, including pipetting error, were less than ±2% (above 30% scale deflexion). Because of the low concentrations of potassium in some of the samples many measurements of this ion were made at 10% scale deflexion or less, in which case the errors reached ± 15 %.

Sterilized isotopes were obtained from the Amersham Radiochemical Centre in the following forms :

The treatment and counting of isotopes are described in the previous paper (Phillips, 1964a).

The method used to measure electropotential differences across the rectal epithelium was similar to that described by Ramsay (1953b) and Shaw (1955a). Silver-silver chloride electrodes were used with liquid junction made through glass capillaries filled with 3% agar gel made up with locust saline (Hoyle, 1953), which has an ionic composition very similar to that of the haemolymph of Schistocerca. To avoid liquid junction potentials with fluids of unknown composition in the rectum the electrode which was to be inserted into the lumen was modified as follows : just before use the agar-Ringer electrode was connected to a second tapered piece of glass tubing containing 3 % agar made up with 3 M-KCI.

Potential differences were recorded with a Pye ‘Universal’ pH meter. A backing-off potential of 500 mV. was applied in parallel to the positive lead with a 1·5 V. dry cell battery and a variable resistance. Both electrodes were first placed in the haemolymph and the asymmetry potential between the two electrodes was recorded or eliminated by adjusting the backing-off potential. The electrode connected to the KCl-agar bridge was then inserted into the lumen and the potential difference across the rectal wall was recorded. After a series of measurements the asymmetry potential was again measured and subtracted from the measured transmembrane potential difference. The range of a series of similar determinations for a single locust was generally less than 4 mV.

The ionic composition of the rectal wall of locusts was determined by the following method : the rectum was quickly dissected out in the locust’s own blood by severing the gut at both ends of the rectal pads, cutting the attached tracheae and slitting open the gut to facilitate removal of the gut contents. Haemolymph adhering to the rectum was rinsed off in 0·4 M sucrose for 15 sec. and its surface was dried by touching the rectum to filter paper. The dried rectum was placed on small pieces of platinum foil and weighed within 30 sec. to obtain the wet weight.

For the determination of sodium and potassium the rectum (on platinum foil) was placed in an oven at 90° C. for 24 hr., reweighed to obtain the dry weight, and then ashed in a muffle furnace at 460° C. for 12 hr. The ashed rectum and platinum foil were then placed in 5 ml. of distilled water in Polythene containers for sodium and potassium determinations by flame photometry. Control experiments were carried out on pieces of platinum foil to detect possible contamination during the experimental procedure.

Chloride was determined by grinding up several rinsed recta in a small homogenizer followed by addition of 0·2 ml. of 50% glacial acetic acid to precipitate the protein. After centrifugation, 0·15 ml. of acidified supernatant was pipetted off into a second tube for electrometric titration of the chloride with AgNO₃. Control experiments were carried out using 25 μl. aliquots of distilled water and 120 mM/l. NaCl. Chloride concentrations per kilogram tissue water were estimated using the previously obtained values for the water content of the rectal tissues.

The concentrations of sodium, potassium and chloride in the haemolymph of locusts supplied with tap water but no food for 4–6 days are shown in Table 1. Assuming that their activities in haemolymph are the same as in pure solutions of similar concentration these three ions account for 55% of the haemolymph osmotic pressure (Δ° C. = 0·75, Phillips, 1964a). Ten amino acids (mainly serine and glycine) and sugars (mainly trehalose) which occur in the blood of Schistocerca at concentrations of 104 and 20 mM/1. respectively (Treherne 1958a, b, 1959) would account for another 30% of the osmotic pressure.

The Na: K ratio of 10 in the haemolymph of Schistocerca is similar to that in omnivorous (Boné, 1945) and more primitive insects (Duchâteau, Florkin & Leclercq, 1953; Lockwood & Croghan, 1959) and agrees with values for other orthopteroid species (Tobias, 1948; Ramsay, 1953b;Hoyle, 1954). The most unusual feature of the haemolymph of Schistocerca is the high concentration of chloride, which almost balances the total equivalents of potassium and sodium. Equally high haemolymph chloride concentrations have been reported for other orthopteroid species (Munson & Yeager, 1949; Ramsay, 1955a).

To determine whether the rectal epithelium is capable of absorbing ions against a concentration gradient, locusts were fed only on tap water for 4–6 days to induce a salt-deficient condition which might be expected to bring about full mobilization of any ion-absorbing mechanisms. The rectum was then ligated and rinsed out several times with distilled water for use in experiments at room temperature 8–16 hr. later. The following saline was made up for injection into the rectum. It contained sodium, potassium and chloride at concentrations equivalent to those in the haemolymph but was made hypertonic by addition of xylose :

A 25 μl. aliquot of this solution was injected into the rectum; initial samples of rectal fluid (10 μl.) were removed within 1– 2 min. and final samples 3– 6 hr. later for determinations of ion concentration and ¹³¹I activity. Haemolymph was obtained for analysis at the conclusion of absorption experiments. The results are recorded in Table 2.

Without exception, a large percentage of the sodium (31%), potassium (75 %) and chloride (44%) injected into the rectum was absorbed against a concentration gradient, for in all cases the final concentration of these ions in the rectal fluid was considerably less than in the haemolymph. In all but three experiments there was a net movement of fluid into the rectum, so that the observed salt absorption was not dependent on the simultaneous absorption of water. (The rate of potassium absorption represents a minimum value because net absorption was probably completed and equilibrium attained some time before the final measurements were made.) These experiments demonstrate that sodium, potassium and chloride can be absorbed against large concentration gradients from the rectum of Schistocerca gregaria.

According to Andersen & Ussing (i960) ‘a substance can be regarded as actively transported only if the transfer of the substance across a membrane cannot be accounted for by the action of the forces of diffusion, electrical potential gradient, solvent drag, or these forces in any combination’. The experiments summarized in Table 2 demonstrate that absorption of potassium, sodium, and chloride from the locust rectum cannot be explained by diffusion down a concentration gradient or by solvent drag. Moreover, net absorption of either the two cations or chloride must have occurred against an electropotential gradient so that it is necessary to postulate active transport of at least one of these ions. To discover which ion was being actively transported the electropotential difference across the rectal wall of locusts which had been fed solely tap water for 3– 5 days was measured at room temperature.

Using micromanipulators the electrodes were first inserted into the haemolymph through the sixth or seventh abdominal tergum to measure or back off the asymmetry potential between the electrodes. The electrode with the connected KC1 salt bridge was then withdrawn and inserted into the lumen of the rectum through the anus. Both electrodes were then sealed into place with a beeswax-resin mixture as a precaution against short-circuiting over the external surface of the cuticle. Several readings were made at 15 sec. intervals. The potential difference across the rectal wall was determined under the following conditions :

1. Rectum not ligated and containing the normal rectal contents.

2. Immediately after the injection of locust haemolymph into the well-rinsed, ligated rectum.

3. Immediately after the injection of the previously described hypertonic xylose-saline solution into the well-rinsed, ligated rectum.

4. Several hours after the injection of hypertonic xylose-Ringer solution into the well-rinsed, ligated rectum.

Under all conditions the lumen of the rectum was distinctly positive with respect to the haemocoel (Table 3). The electropotential difference ranged between 15 and 32 mV. except when measurements were made immediately after the injection of hypertonie xylose-saline solution. Under the latter condition values as high as 62 mV. were observed, possibly associated with the initial and rapid movement of fluid into the rectum. These observations indicate that the lumen must have been positive to the haemolymph during the whole course of the absorption experiments previously described, so that it is necessary to postulate the active transport of chloride across the rectal wall. (Shaw (1955a) found that the lumen of the rectum was 8-40 mV. positive to the haemocoel in Sialis lutaria.)

A steady potential difference, negative with respect to the haemolymph, was observed by chance during attempts to measure the potential difference across the rectal wall by pushing a microelectrode through the rectal pad from haemocoel to lumen. As it was thought that this might represent the potential difference between the haemocoel and the interior of the epithelium, a series of experiments was under-taken to confirm the observation. The potential difference was measured by the method used in previous experiments except that the relatively large KCl-agar electrode was replaced by a fine-tipped microelectrode. These microelectrodes were identical with and prepared in the same manner as the silica glass micropipettes used by Ramsay & Brown (1955) for the determination of freezing-point depressions and had an external tip diameter of 10– 15 μ.

The locusts which were given an abundant supply of green wheat shoots and tap water were used so that the rectum, which was not ligated, was usually distended with faecal material. The insect was held on the stage of a binocular microscope and the body cavity was opened up by making an incision along the mid-dorsal line of the last four abdominal segments. The asymmetry potential was first determined and backed off. The microelectrode filled with 3M-KCI was then pushed well into a rectal pad using the micromanipulator and the potential difference was measured at 15 sec. intervals. (These microelectrodes penetrated the soft tissue, but they would bend and break rather than pierce the cuticular intima, which was exposed by stripping away the outer tissue layers.) In two experiments the rectum was then severed so that the lumen was open to the haemocoel. The experiment was then repeated to detect any effect on the observed potential difference.

In all cases a steady potential difference was observed for periods of 1 min., the interior of the rectal pad being 8-31 mV. negative to the haemolymph (Table 4 a). Severing the rectum had no effect, confirming that the observed potential difference lay between the haemocoel and the interior of the rectal pad rather than between the haemocoel and the lumen.

These experiments were repeated using micropipettes with an external tip diameter of 20– 30 μ which would pierce the cuticular intima (Table 46). On pushing these larger microelectrodes into the rectal pad a steady potential difference of 5– 15 mV. (haemolymph positive) was again instantly developed. When more force was applied, however, there was a second rapid change to a steady potential difference of 13– 25 mV. (haemolymph negative). The latter potential difference corresponded to that which had been previously observed (Table 3) between lumen and haemocoel. This was confirmed by dissecting open the rectum and observing that the tip of the micro-electrode was in the lumen.

It seems most likely that the intermediate potential difference (haemolymph positive) is that between the interior of the large columnar epithelial cells of the rectal pad and the haemocoel. Quantitatively, the observed potential difference might be considerably smaller than the true potential difference because of short-circuiting which might have occurred owing to the use of relatively large microelectrodes which were required to penetrate the outer connective tissue. The possibility that this intermediate potential difference exists between the subintimal space (Phillips, 1964a) and the haemocoel cannot be excluded, although to accept this hypothesis would imply the existence of a large potential difference (average 37 mV.) across an inert cuticular intima, which allows very rapid exchange of water and ions (Phillips, 1961).

The maximum ionic concentration gradients which can be passively maintained across the rectal wall at equilibrium by an electropotential difference can be estimated from the equation of Andersen & Ussing (1960), which in the absence of a net flux of solute and water reduces to

According to equation (1), the normal electropotential difference of 15-32 mV. across the rectal wall could support a concentration of sodium and potassium in the rectal fluid equivalent to 30– 56% of the haemolymph concentration, assuming that simple diffusion only is involved and that the activity coefficients for these cations are the same in rectal fluid and haemolymph. The potassium concentration of rectal fluid was observed to fall below this estimated value in two insects (Table 2). The following experiments were undertaken to determine the maximum concentration gradients for chloride, sodium, and potassium which develop across the rectal wall as a result of net absorption.

Locusts fed on tap water only for 4– 6 days were used within 2 hr. of ligation at room temperature. A 25 μl. aliquot of the following hypertonic xylose solution which had a low ionic concentration was injected into the ligated rectum:

In initial experiments 3 μl. samples of rectal fluid were removed at intervals so that changes in volume (i.e. ¹³¹I activity), and sodium and chloride concentration could be followed (Fig. 1). After 2· 5 hr. the concentrations of these ions in the rectal fluid were constant and the rate of volume change was relatively low. Thus a state of equilibrium was closely approached. In the remaining experiments the maximum concentration gradient was determined on samples of rectal fluid and haemolymph removed 3–4 hr. after injection of fluid into the rectum. The electropotential difference across the rectal wall was then measured.

In all experiments (Table 5 a) there was a net absorption of both sodium and chloride from the rectum. The final chloride concentration in the rectal fluid fell to an average value of 4 m-equiv./l., or 4% of the haemolymph concentration, while in two locusts the final concentration was only 1 m-equiv./l. The final sodium concentration in the rectal fluid was only slightly greater (average 9·6 m-equiv./l., or 9% of the haemolymph concentration). These low concentrations were observed whether net movement of water occurred into or out of the lumen. In all cases the observed potential difference was considerably less than the value required to support the observed sodium concentration gradient.

The maximum potassium concentration gradient was determined in similar experiments at 28° C. (Table 5 b). The final concentration of potassium in the rectal fluid was less than 0·2 m-equiv./l., or 2% of the haemolymph concentration. Again, the observed potential difference was only 25 % of the value required to support the observed concentration ratio assuming that passive diffusion only is involved.

Further evidence of the very large concentration gradients of chloride, sodium and potassium which can be maintained across the rectal wall was obtained during a study of the water absorption from the locust rectum (Phillips, 1964a). In these experiments at 28° C. the chloride, sodium, and potassium concentrations in the rectal fluid remained at average values of 1, 3 and 1 m.-equiv./l. respectively for 3–6 hr. These experiments indicate that simple passive diffusion alone cannot account for the absorption of sodium and potassium from the locust rectum so that active transport of these two cations across the rectal wall must occur to some extent.

A general characteristic of active transport is that as the concentration of transported solute is increased on the absorbing side of the membrane, the rate of active transport increases less quickly and eventually becomes independent of concentration. In other words, a maximum rate of transfer is attained due, it is supposed, to the saturation of a carrier with which the solute must combine to traverse the membrane (Rosenberg, 1954). This fact is of considerable importance in osmotic and ionic regulation. Amongst the vertebrates, for example, excess solute is eliminated from the body when the rate of glomerular filtration of the solute exceeds the maximum rate of reabsorption in the kidney tubules (Smith, 1951).

Ramsay (1955-1958) has shown that in the stick insect the concentrations of inorganic ions, sugars, amino acids, and urea in Malpighian tubule fluid is directly proportional to their concentrations in the haemolymph. Moreover, in the case of potassium, sodium and phosphate the rate of urine production also increases with increasing blood concentration. The question arises whether excess solute is eliminated from the insect body due to saturation of reabsorption mechanisms in the rectum when the solute load produced by Malpighian tubule secretion is elevated.

The relationship between the sodium chloride concentration of rectal fluid and the rate of sodium and chloride absorption from the ligated rectum was studied at room temperature using locusts which had been fed only on tap water for 3– 5 days. A series of experimental solutions was used which had sodium chloride concentrations of 20– 650 mM/1. but approximately the same total osmolal concentration (0·9−1·2). These solutions was prepared by combining suitable volumes of a 0·9 molal xylose or trehalose solution and a 0·65 molar sodium chloride solution. One part each of 0·05 M amaranth solution and ¹³¹I-albumin in Ringer were added to 50 parts of xylose-saline solution. A 25 μl. aliquot of experimental solution was injected into the rectum within 1– 2 hr. of ligation, and 10 μl. samples of rectal fluid were removed after 1– 2 min. (initial) and after 1· 5– 3 hr. (final) for determination of sodium, chloride and ¹⁸¹I-albumin concentrations.

Absorption rates were plotted against the mean sodium or chloride concentration of rectal fluid (Fig. 2). Because there was generally a large change in sodium chloride concentration during experiments due to net absorption and water movement, the absorption rates represent typical values within a concentration range which in some cases was rather large. The size of this range has been indicated in Fig. 2 by broken horizontal lines extending from the mean concentration (solid circle) to the initial sodium chloride concentration in the rectal fluid.

Clearly, amongst water-fed locusts at least, the rates of both sodium and chloride absorption from the ligated rectum are directly proportional to their concentrations in the rectal fluid. Over the whole physiological range there is no marked tendency towards a maximum rate of absorption with increasing concentration. Only at concentrations below 50 m-equiv./l. is a possible departure from linearity apparent, for an extrapolation of the best straight line through the points does not pass through the origin. Therefore ionic regulation in locusts fed on tap water is not the result of a limiting rate of reabsorption.

As the locusts used in these experiments were supplied with tap water only, the salt concentration in the haemolymph was likely to be low. A system whereby absorption is proportional to concentration of solute in the rectum might allow the complete reabsorption of sodium chloride from the rectum of a salt-depleted locust when the salt concentration of the rectal fluid becomes high, owing to active absorption of water (Phillips, 1964 a) which occurs under all conditions.

Experiments with water-fed locusts indicate that regulation of the sodium chloride concentration in the haemolymph is not due simply to a limiting rate of reabsorption in the rectum. The possibility remains, however, that the rate of salt reabsorption in the rectum is governed by the ionic concentration of the haemolymph.

To test this hypothesis the net rates of absorption of sodium, potassium and chloride from the ligated rectum of locusts fed either with hypertonic saline (see Phillips, 1964 a, for composition) to raise the ionic concentration of the haemolymph or with tap water were compared over the whole physiological range of rectal fluid concentrations. Sexually mature male locusts were supplied with the appropriate fluid for 4-5 days. They were then placed in an incubator at 28 ± 1° C. and at 100% (water-fed) or below 70% (saline-fed) relative humidity for 24 hr. prior to and during experiments. After ligation the rectum was rinsed out with either 0·5 molal (water-fed) or 1·5 molal (saline-fed) trehalose solution.

The mean concentrations of sodium, potassium and chloride in the haemolymph of saline-fed locusts (Table 6) were respectively 46, 73 and 42 % higher than in water-fed locusts (Table 1). (These high concentrations are not unusual; Hoyle (1954) found that the potassium concentration in the haemolymph of grass-fed Locusta migratoria was 70% higher than that in starved animals.)

Sodium and chloride absorption from the rectum were measured using experimental solutions prepared by combining differing proportions of a 0·9M-NaCl and a I-25M trehalose solution. Potassium absorption was measured from experimental solutions prepared by combining appropriate volumes of 0·9M-KCI, 0·9M-NaCl and 1·25M trehalose solutions so that the sodium concentration was always five times the potassium concentration. One part each of 0-05 M amaranth solution and ¹³¹I-albumin in Ringer were added to 30 parts of saline-trehalose solution. A 25 μl. aliquot of the experimental solution was injected into the ligated rectum and 10 μl. were removed for the initial analysis.

Potassium absorption, which was observed to occur very rapidly, was measured over a 10– 20 min. period. Sodium and chloride absorption were then measured in a second experiment of 1–2 hr. duration using the same locust. Only two experiments were carried out on each locust. Individual absorption rates have been plotted against mean rectal fluid concentrations in Fig. 3.

The absorption of sodium and chloride will be considered first. In confirmation of the previous experiments, the rate of absorption of these two ions increased with increasing rectal fluid concentration in water-fed locusts.

The situation is quite different in saline-fed locusts. At low rectal fluid concentrations the rates of absorption of sodium and chloride are not significantly different from those in water-fed locusts ; that is, active transport of sodium chloride from the rectum still occurs even when the haemolymph concentration of this salt is increased by 40%. In saline-fed locusts, however, a maximum rate of absorption, typical of active transport, is clearly attained at a rectal fluid concentration below that of the haemolymph. Consequently the reduction in the rate of sodium chloride absorption from the rectum of the saline-fed locust compared with that in the water-fed locust increases as the rectal fluid concentration is elevated. Qualitatively, potassium absorption exhibits the same pattern.

The absorptions of these three ions from the rectum of the water-fed locust are compared in Fig. 4 a. The most striking fact is that potassium absorption is approximately ten times more rapid than sodium absorption at the same rectal fluid concentration. Assuming that these two cations traverse the rectal wall mainly by diffusion, the more rapid potassium absorption can be partly explained by the much larger concentration gradient for this ion. But the inequality of electrochemical gradient can be taken into account by comparing the absorption rates when the rectal fluid concentrations of these two cations are equal to those of the haemolymph (i.e. 11 m-equiv. K/l. and 108 m-equiv. Na/1.). The electropotential gradient is the same for both ions so that the net passive transfer of each cation should be proportional to its concentration (or rather activity) and its mobility, provided the rectal wall is not selective for potassium; i.e.

On the basis of non-selective diffusion, then, sodium absorption should be 6·7 times more rapid than potassium absorption in the absence of a concentration gradient. This is not the case. Potassium absorption (0·7 μ-equiv./hr.) at a rectal fluid concentration of 11 m-equiv./l. is approximately twice as great as sodium absorption (0·38 μ-equiv./hr.) at a rectal fluid concentration of 108 m-equiv./l., which indicates that the rectal wall is on the average twelve times more permeable to potassium than to sodium. Selective permeability to potassium with respect to sodium has, of course, been commonly observed for other biological membranes such as the plasma membranes of nerves and muscles (reviewed by Ussing, 1960). The significance of selective permeability of the rectal wall to potassium with respect to the excretory process in the locust is considered in another paper (Phillips, 1964b).

The rectal absorption of sodium, potassium, and chloride in saline-fed locusts remains to be compared (Fig. 4 b). Potassium absorption reaches a maximum rate (0·72 μ-equiv./hr.) at a rectal fluid concentration below that of the blood. This maximum rate is approximately twice that for sodium chloride (0·38 μ-equiv./l.). Thus, not only does feeding of hypertonic saline cause a general reduction in salt reabsorption at rectal fluid concentrations above those in the haemolymph but also causes a decrease in the ratio of potassium to sodium absorption. This regulation is possibly the result of the larger (73 %) increase in haemolymph potassium concentration following the feeding of saline as compared with the increase (46 %) in sodium concentration. It should be noted that the K:Na ratio in the hypertonic saline fed to locusts was 1 : 2, whereas the ratio in the haemolymph is 1:10. A reduction in the ratio of potassium to sodium absorption would tend to maintain the cation ratio in the blood by increasing sodium reabsorption relative to potassium absorption.

As a further comparison of the relative ability of water-fed and saline-fed locusts to absorb ions from the rectum against concentration gradients, the study of the lower limits for salt absorption was repeated using saline-fed locusts. The following modifications in experimental procedure were made: (1) Locusts were fed solely on hypertonic saline for 4–6 days and were incubated at 28 ± 1° C. for 24 hr. prior to and during experiments. (2) The maximum concentration gradients of sodium, potassium, and chloride were determined simultaneously from samples of rectal fluid and haemolymph taken 3 hr. after injection of the experimental solution (1 molal xylose replaced by 1·25 molar trehalose) into the ligated rectum. The results are given in Table 7.

A slight net absorption of water and a decrease in concentrations of sodium, potassium, and chloride in the rectal fluid indicated that the final concentration gradients were attained as a result of net salt absorption. The final concentration ratios and the electropotential gradient across the rectal wall of saline-fed locusts (25 ± 5 mV.) were not significantly different from those in water-fed locusts (Table 8). The observed potential difference across the rectal wall of saline-fed locusts was only 30–50% of that required to support the observed cation concentration gradients assuming simple diffusion.

In view of the rapid absorption of potassium from potassium chloride solutions and the need to maintain electrical neutrality, the question arises whether the rate of chloride absorption is governed by the rate of absorption of the accompanying cation. Some initial determinations of potassium and chloride absorption from 0·35 M/1. potassium chloride at 28 ± 1° C. are relevant (Table 9). The experimental solution was prepared by combining 3·5 parts of 0·9M-KCI and 5·5 parts of 0·7M trehalose solution.

Chloride absorption from this solution averaged 4·6 μ-equiv./hr., or 83 % of potassium absorption (5·6 μ-equiv./hr.). The former value is more than three times the rate of chloride absorption (0·13 μ-equiv./hr.) from a sodium chloride solution of equal concentration (read from Fig. 3 a). This indicates that, at least at high rectal fluid concentrations, the rate of chloride absorption is governed by the relative concentrations of potassium and sodium.

While chloride absorption from the concentrated potassium chloride solution was accelerated, the mean percentage absorption rate for potassium (101% total rectal fluid K/hr.) was approximately 40% of that observed at lower rectal fluid concentrations. This suggests that the rate of potassium absorption in water-fed locusts is not a linear function of rectal fluid concentration, but approaches a maximum rate as indicated in Fig. 4 a.

In view of the commonly observed dependence of electropotential differences across biological membranes on the potassium concentration gradient (reviewed by Ussing, 1960) the possibility was considered that the reduced percentage absorption of potassium and accelerated chloride absorption from concentrated potassium chloride solution were due to a change in the electropotential gradient. To test this hypothesis the change in potential difference across the rectal wall with time was followed after the injection of a 10 pl. aliquot of 0·7M-KCl, 0·4M-K₂SO₄, or 0·7M-NaCl into the ligated rectum in water-fed locusts (Fig. 5).

Injection of 0·7M-KCl resulted in an immediate reversal of the transmembrane potential difference from + 20 mV. (lumen with respect to haemocoel) to as high as − 80 mV., but the potential difference drifted slowly back to the normal positive value within 10–30 min. of injection. The drift was possibly coincident with a fall in potassium concentration in the lumen as the latter ion was absorbed. This temporary reversal of potential difference, which was probably due to the relatively high permeability of the rectal wall to potassium, could account for the accelerated chloride absorption and decreased potassium absorption observed following the injection of concentrated potassium chloride into the rectum.

It has so far been assumed that net absorption of water (Phillips, 1964a), sodium, chloride and potassium from the lumen of the rectum is equivalent to net transfer across the rectal wall, and that no storage or accumulation of these substances occurs in the epithelium itself. This assumption requires verification since it has been suggested that chloride and iodine accumulate in the rectal pads of some insects (Wheeler, 1950; Vecchi, 1955).

The ionic content of the rectal wall was determined under the following conditions :

1. In locusts fed solely on tap water for 3–5 days.

2. In locusts fed on tap water for 3–5 days, 3 hr. after the injection of 20 μl. of concentrated saline (300 mM/l. NaCl and 300 mM/l. KC1) into the ligated rectum.

3. In locusts fed on hypertonic saline (Phillips, 1964a) for 4–6 days.

The results are presented in Table 10A.

There was no difference in the average wet weight of the rectum (9·6 mg., or 0·5 % of the total body weight) in water-fed locusts before and after absorption of hypertonic saline, while the wet weight of the rectum in saline-fed locusts was only slightly lower (9·0 mg.). Under all three conditions the ionic content of the rectal wall was similar, with the possible exception of a slightly higher sodium content (P < 0·05) under condition (a) and a slightly lower water content (P < 0·05) under condition (c). On the other hand there was a large variation in the content of individual ions and the ratio of sodium to potassium between individuals. These variations were probably due in part to variations in the size of the blood space, to movement of haemolymph into the cut ends of tracheae, and to exchange of intracellular potassium for extra-cellular sodium in consequence of tissue damage during dissection. The approximate quantities of sodium, potassium and chloride absorbed from the rectum in water-fed locusts immediately before analysis of the rectal wall (condition (b)) were estimated (Table IOB) from previous data (Fig. 3 a, b, c) to be one order of magnitude greater than the quantities of these ions present in the rectal wall, yet there was no significant difference between the ionic content of the rectal wall under conditions (a) and (b) Clearly there is no significant storage or accumulation of absorbed water, sodium, potassium or chloride in the rectal epithelium of the locust.

Since the volume of the intercellular space has not yet been determined, only broad generalizations can be made about the ionic composition of the epithelium responsible for active transport of water and ions from the locust rectum. Allowing for the haemolymph space, it can be concluded that the intracellular fluid is characterized by a very high potassium concentration, and low sodium and chloride concentrations relative to the haemolymph (Fig. 6). Histological sections (Phillips, 1964a) indicate that the large columnar epithelial cells of the rectal pad constitute the major portion of the rectal wall ; hence, analysis of the whole rectum is likely to reflect the composition of these cells. The total concentration of sodium, potassium, and chloride in the rectal epithelium of water-fed locusts (184 m-equiv./kg. tissue H₂O) is less than that in the blood (236 m-equiv./l.), which suggests that the osmotic pressure in the epithelium is closer to that of the blood than to that of the hypertonic rectal fluid. Utilizing the experimental data which have been obtained to date (Tables 1, 3, 4, 5, 10; Phillips, 1964a), a tentative scheme of the electropotential and concentration gradients across the rectal wall of the water-fed locust, under near-equilibrium conditions, is proposed in Fig. 6.

It seemed possible that the linear relationship between rectal fluid concentration and net absorption of sodium chloride in water-fed locusts (Fig. 2) might be due to the existence of considerable diffusion across the rectal wall. To determine the extent of such diffusion the influx of sodium and chloride ions into the lumen was estimated.

Because of the low specific activity of the ³⁶Cl available it was not feasible to measure influx by injecting this isotope into the haemolymph and measuring the increase in activity of the rectal fluid. An approximate estimate of influx was made, however, by injecting a 25 μl. aliquot of a ³⁶Cl-saline made hypertonic with xylose into the ligated rectum and measuring the fall in specific activity with time due to the influx of unlabelled chloride from the haemolymph. (The experimental solution was prepared by combining one part of 1M-Na³⁶Cl and seven parts of 0·8M xylose solution, so that the chloride concentration was equal to that of the haemolymph.) As a first approximation, the initial percentage decrease in specific activity in the rectal fluid (Table 11) gives an estimate of the percentage of the rectal fluid chloride which has exchanged with the haemolymph over and above net absorption, and hence the influx of chloride. This is so because only unlabelled chloride diffuses into the lumen from the blood, a fact confirmed by measurements at the end of experiments which showed that ³⁶Cl activity in 10 μl. of haemolymph was not significantly different from the background count.

It was not possible to calculate an exact influx because one variable, the net absorption rate, was not measured owing to the difficulty in separating the activities of ¹³¹I-albumin and ³⁶Cl. However, an average value for net chloride absorption (0·27 μ-equiv./hr., or 14% of the total rectal fluid chloride per hour, Fig. 3 a) was used for comparison with the rate of decrease of specific activity of the rectal fluid with time.

The mean influx of chloride was approximately 0·29 μ-equiv./hr., which is very similar to the net absorption rate. This suggests a total efflux (0·29 + 0·27) of 0·56 μ-equiv. Cl/hr. in the absence of a concentration gradient and a ratio of influx to efflux of approximately 0·5. The size of the passive and active components of the efflux can be estimated from the electropotential difference across the rectal wall. According to the Ussing equation, the observed average potential difference of 21 mV. (Table 3) should result in an influx:efflux ratio of approximately 2 in the absence of a concentration gradient and a drag effect. This suggests a passive efflux of 0·14 μ-equiv. Cl./hr. and an active efflux (0·56-0·14) of 0·42 μ-equiv. Cl/hr. Clearly there is considerable movement of chloride across the rectal wall by diffusion, a conclusion which is supported by the high rate of chloride absorption from concentrated potassium solutions, when the electropotential difference across the rectal wall is reversed (Table 9).

The back diffusion of sodium was determined by measuring the increase in ²⁴Na activity in the rectal fluid introduced into the lumen 2–3 hr. after injection of this isotope into the haemolymph. By this time the injected isotope had reached an equilibrium within the tissues and less accessible fluid-filled spaces (Fig. 7). The experimental solution consisted of a no mM/1. sodium chloride solution made hypertonic with trehalose (0·8 molal). To minimize the efflux of ²⁴Na back into the haemolymph the volume of rectal fluid was increased to 400 μl. (of which 100 μl. was removed within 1 min. for determination of the initial activity). Approximately 20 μl. of solution was forced into the rectum and was rinsed in and out every 5–10 min. to ensure mixing with the main bulk of fluid in the injection capillary. Net absorption under these conditions (0·3 μ-equiv. Na/hr., Fig. 3 b) represents only 1 % of the total rectal fluid sodium per hour, and the specific activity in the rectal fluid remained less than 1 % of that in the haemolymph during these experiments. Hence efflux of ²⁴Na was negligible and the influx of ²⁴Na was assumed to be a linear function of time, Under these conditions influx of sodium was calculated by the following equation (modified after Shaw, 1955b):

where I is the influx, A_h the ²⁴Na activity per unit volume of haemolymph, and A_r is the increase in ²⁴Na activity per unit volume in the rectal fluid in time t, V is the total volume of rectal fluid and K is the sodium concentration of the haemolymph.

Although there was a sixfold variation in influx between individuals (Table 12) the mean value (0·034 μ-equiv. Na/hr.) was small compared with net sodium absorption from the rectum (0·30 μ-equiv. Na/hr.) under the same conditions. This indicates an average influx:efflux ratio of approximately 1:10 in the absence of a concentration gradient compared with a ratio of 1:2 for chloride. The flux ratio for sodium is approximately the same as the maximum concentration ratio (0·09) which can be maintained across the rectal wall. Like the concentration ratio, the observed flux ratio for sodium is on the average 3–5 times greater than that predicted by the Ussing equation assuming that sodium crosses the rectal wall only by diffusion. The observed flux ratio is therefore consistent with the view that some active transport of sodium occurs across the rectal wall.

Failure to observe a maximum absorption rate for chloride in locusts fed on tap water does not necessarily conflict with the carrier hypothesis of active transport across the rectal wall. Two explanations are possible:

1. The affinity of the carrier for chloride is relatively low so that saturation is only very gradually attained at chloride concentrations well above the physiological range (Wilbrandt, 1954).

2. A considerable diffusion of chloride across the rectal wall is superimposed upon active transport which attains a maximum rate at low rectal fluid concentrations.

Subsequent experimental observations seem to favour the second hypothesis, (a) There is a large influx of chloride and chloride absorption is accelerated when the electropotential gradient across the rectal wall is reversed. These observations indicate that there is considerable diffusion of this anion across the rectal wall of water-fed locusts, (b) In saline-fed locusts at least, active absorption of chloride does attain a maximum rate at a very low rectal fluid concentration. It is not unreasonable to suppose, therefore, that the active component is also constant in water-fed locusts. A change in the amount of carrier in the rectal wall could not alone account for the difference in chloride absorption in water-fed and saline-fed locusts since this should not alter the concentration level at which saturation of the carrier is attained (Wilbrandt, 1954). To explain the difference in absorption under the two treatments in terms of the active component alone, it would be necessary to postulate a change in the affinity of the carrier for chloride.

Now consider how well the net absorption rates estimated according to the second hypothesis compare with experimental observations. According to this hypothesis, net absorption can be divided into three components:

Approximate values have been determined experimentally for these components when there is no concentration gradient across the rectal wall. Substituting these values,

Now consider how each of these components will change with the concentration of chloride in the rectal fluid. The first term, active efflux, is independent of rectal fluid concentration and thus constant. (At very low rectal fluid concentrations this component can be described by the Michaelis-Menten equation.)

The electropotential difference across the rectal wall is relatively independent of sodium chloride concentration in the rectal fluid (Table 13). Therefore the second component in equation (3), the passive influx which is dependent on blood chloride concentration, is also independent of rectal fluid concentration.

The only variable is passive efflux. Since the electropotential difference is constant, passive efflux is directly proportional to the chloride concentration of the rectal fluid. The net absorption at any rectal concentration can then be estimated from the following equation:

or

where [Cl_r] and [Cl_h] represent the concentrations of chloride in the rectal fluid and haemolymph respectively.

The relationship between net absorption and rectal fluid concentration calculated according to equation (4) is compared with experimental observations in Fig. 2 a. The calculated and experimental curves are very similar in form although the calculated absorption rates at the very highest rectal fluid concentrations are 20% lower than the observed values. It is only necessary to postulate a slightly greater influx (e.g. 15%) or a smaller potential difference (18 mV.), both of which are well within the experimental error, in order for the calculated curve to match the experimental results exactly. It seems therefore that the second hypothesis will account, quantitatively as well as qualitatively, for net absorption of chloride from the rectum in water-fed locusts.

From these considerations a possible explanation of the difference between water-fed and saline-fed locusts in respect of chloride absorption is at once apparent. A large reduction in the passive permeability of the rectal wall to chloride would eliminate the last two terms in equation (3) so that net absorption would be due solely to active transport and therefore constant except at very low rectal fluid concentrations. It should be possible to test this hypothesis by comparing the back diffusion of isotopes into the lumen of water-fed and saline-fed locusts. If the proposed reduction in passive permeability were general, it might also explain the greater osmotic pressure gradient which can be maintained across the rectal wall of saline-fed locusts (Phillips, 1963a).

A reduction in passive permeability might not alone be sufficient to explain the difference in net absorption of chloride from the rectum in saline-fed and in water-fed locusts, since saline-fed locusts would be expected, according to this hypothesis, to exhibit (1) higher absorption rates than water-fed locusts at rectal fluid concentrations below those of the haemolymph, and (2) lower rectal fluid concentrations at equilibrium. Neither of these differences was observed, possibly owing to the degree of variability in the results. However, an alternate explanation is that there is some reduction in the active absorption of chloride, possibly due to a decrease in the effective concentration of a carrier molecule in the rectal wall.

The observed concentration ratios and flux ratios across the rectal wall indicate that there is some active absorption of sodium from the lumen. Other evidence supports this conclusion : the sodium concentration in the rectal epithelium is less than half that in the haemolymph, while the interior of the rectal pads is electrically negative with respect to the haemolymph. Since the rectal wall is permeable to sodium, it would appear that the low concentration of sodium in the rectal wall must be maintained by active extrusion of this ion into the haemolymph.

The linear relationship between net absorption of sodium and its concentration in the rectal fluid in water-fed locusts might be explained by a large diffusion component as was suggested for chloride. However, the influx of ²⁴Na into the lumen was found to be relatively small. This low influx of ²⁴Na is not necessarily inconsistent with a large net diffusion of sodium if the latter cation traverses the rectal wall by single-file diffusion (Hodgkin & Keynes, 1955).

The view that potassium is actively transported to some extent across the rectal wall is supported first by the demonstration of absorption against large electrochemical gradients and secondly by the maximum rate of absorption in saline-fed locusts, which is twice as rapid as the maximum rate of sodium chloride absorption. This situation indicates that potassium absorption involves a rate-limiting step, such as combination with a carrier in the rectal wall.

The fact that chloride is actively transported across the rectal wall of the locust might at first seem unusual since it is commonly accepted that most cells are capable of active sodium transport and since sodium transport has been observed across many biological membranes. There are, however, several examples of active secretion of chloride amongst the vertebrates (reviewed by Ussing, 1960). It is not clear how common active chloride transport might be amongst insects, for while there have been several investigations of active salt absorption from the gut of insects (see introduction) and from the external media through the anal papillae of insect larvae (Koch, 1938; Ramsay, 1953a; Treheme, 1954; Stobbart, 1959, 1960; Gloor & Chen, 1950) it is not known whether anion or cation is actively transported. In the Malpighian tubes of insects, active potassium secretion may be the ‘prime mover ‘in the formation of urine (Ramsay, 1953b) although some active secretion of sodium also occurs in Dixippus at least (Ramsay, 1955b). However, in two insects, Rhodnius and Locusta, Ramsay found that the lumen of the tubule was negative to the haemocoel, which suggests that an anion might also be actively secreted in these insects.

In view of the relatively high chloride concentration in the blood of Schistocerca compared with that in other insects active chloride reabsorption in the rectum may be peculiar to the locust. Whether active cation or anion absorption from the rectum is the general rule among insects is of reduced interest in the light of recent investigations which show that both chloride and sodium are actively transported across several epithelial membranes ; frog skin (Jorgensen, Levi & Zerahn, 1954), rat intestine (Curran & Solomon, 1958) and crayfish gills (Shaw, i960). There is good reason to believe that this is also the case in the rectum of Schistocerca.

In its ability actively to reabsorb ions from the lumen of the rectum against very large concentration gradients the locust (which inhabits semi-desert regions) does not appear to differ significantly from fresh-water insect larvae (Edney, 1957). The main difference between the two seems to be that the locust is forced to reabsorb water actively and to produce hypertonic excreta under all conditions (Phillips, 1963 a), whereas fresh-water larvae (except Aedes detritus, Ramsay, 1950) do not actively absorb water from the rectum.

I would like to thank Prof. V. B. Wigglesworth, F.R.S., for his encouragement and advice during the course of this work and for accommodation in the Zoology Laboratory. I am also grateful to Drs J. E. Treheme, J. W. L. Beament and A. P. M. Lockwood for much helpful discussion.