Transport of Thallium Ions Across the Isolated Midgut of Hyalophora Cecropia

Permeation of the alkali metal ions across biological membranes has been studied extensively, especially for Na⁺ and K⁺. Movements of T1⁺ have been studied much less, although the T1⁺ ion resembles the alkali metal ions in having a charge of one and an ionic radius obtained from crystal data of 1·44 Å, between the radius for Rb⁺ of 1·48 Å and that of K⁺ of 1·33 Å.

Thallium is poisonous (and this has been well known for a long time), but if care is taken, the permeation of Tl⁺ can be measured as was done by Mullins & Moore (1960), who studied the flux of Tl⁺ into and out of frog muscle. They drew the conclusion that the muscle fibre membrane cannot distinguish between the toxic heavy metal T1⁺ and K⁺, provided that the concentrations of the former ion are kept low. Skulskii, Manninen & Jmefelt (1973) studied the uptake of T1⁺ by red blood cells and found that the cell/medium distribution was only 1·8–2·0, close to the passive distribution of small anions, but they also found that the rate constants for T1⁺ flux were much larger than for K⁺.

In Chlorella T1⁺ can be accumulated by a K pump (Solt, Paschinger & Broda, 1971). Maslova, Natochin & Skulskii (1971) found that in the frog skin T1⁺ ions will activate the K⁺-dependent step of the Na⁺, K⁺-dependent ATPase, and are significantly more effective than K⁺. But this activation does not result in active Na-transport, which, as measured by the short-circuit current, is inhibited by Tl⁺ (50% for 1 mm-Tl), an inhibition reminiscent of the action of ouabain.

Koefoed (1975) compared the transfer of Tl⁺ and K⁺ in the rectal complex of the mealworm (Tenebrio molitor). She found that the two ions are treated similarly by the tissue, possibly indicating an active transport of both ions.

We have therefore performed a comparative study of the transport of T1⁺ and K⁺ across the midgut from Hyalophora cecropia (L), which is very well characterized with respect to K transport (see review, Harvey & Zerahn, 1972).

The result of the study does not support the former views of active Tl⁺ transport. T1⁺ ions are transported across the midgut many times faster than K⁺, but the transport is not active. In the short-circuited gut the flux of T1⁺ is the same in both directions and is independent of the metabolism.

It is concluded that Tl⁺ may cross the gut wall by facilitated diffusion, but other models are also possible.

The larvae of the American silkworm, Hyalophora cecropia, were reared on either leaves of willow, Salix babylonica, or on artificial diet (Riddiford, 1968). The larvae were chilled on ice for an hour or longer before the gut was removed and put up in the apparatus described by Harvey, Haskell & Zerahn (1967) (plastic tubing, no rubber). The advantage of this arrangement is that the midgut can be kept as a sphere and maintains the same thickness throughout the experiment. The composition of the bathing solution 32-K was: 32 mm-K⁺, 2 mm-HCO₃⁻, 30 IDM-Cl⁻, and sucrose to keep the total osmolarity to 260 m-osmol (Harvey & Zerahn, 1971). Analytical reagents were used and the solution contained no Ca or Mg ions (see page 114). At low concentration thallium was present in the form of T1C1, and at high concentration either in the form of Tl₂SO₄ or T1NO₃. There was no difference in behaviour if the T1 was accompanied by different anions. When cecropia larvae were not available some other satumiid larvae were used as indicated in the legends to the tables. The contents of ions in the guts were determined as usual (Zerahn, 1973). T1⁺ was determined by absorption flame spectrometry. The radio-thallium was determined in a Packard beta spectrometer (Tricarb) usually in 1 ml samples plus 10 ml Instagel with an accuracy of + 2% or better. The experiments were made by the procedure given by Harvey & Zerahn (1972). Radio-tracers were added to one side at the start and samples taken from the other side at times given in the tables. This will allow the delay of the flux through the gut (the lag time) to be determined (see Harvey & Zerahn, 1969). The short circuit current for distended guts will be equal to the net K flux with an accuracy of 10% or less (Harvey & Zerahn, 1972); it is assumed to be similar for undistended guts, but accurate measurements are not available. The values given for the short-circuit current are mean values for the period. Washing the gut out between the different periods causes a change in potential with time, but the influence of these changes was cancelled by alternating the sequence; for example first starting with an influx period, and the next time with an efflux period.

In a short-circuited cecropia midgut the current is derived from the active transport of potassium (Harvey & Nedergaard, 1964). T1NO₃ was added to the bathing solution to make it 1 mm in Tl⁺. The Tl⁺ was labelled with ²⁰⁴Tl on the blood-side and the flux of T1⁺ was measured and compared to the short-circuit current. The results are given in Table 1.

Since the concentration of thallium was 1 mm, it had to be multiplied by 33 in order to compare it with the K flux. When this correction for molarity is made, the magnitude of the T1⁺ flux is then comparable to the flux of K, which is about equivalent to the short-circuit current. Thus at first sight it looks as if the gut could not discriminate between K⁺ and T1⁺, and there is a possibility that T1⁺ is actively transported in the same way as K⁺. However, the flux of T1⁺ from lumen to blood-side is not significantly different from the flux in the opposite direction (Table 2). Thus it is only the unidirectional flux of T1⁺ which resembles the K flux and not the net flux, which is almost nil. Therefore there is no indication of an active transport of Tl. However, these fluxes are from different guts and the variation is quite large, so it would be better to measure the T1⁺ flux in both directions on the same gut. Preliminary experiments showed that when the guts were labelled with ²⁰⁴T1 from one side followed by washing with un-labelled saline, the activity could be washed away sufficiently in half an hour, after which the T1+ flux in the opposite direction could be determined. The animals used for the experiments were Actia selene (Hübner [1810]) and not Hyalophora cecropia, but the relation between short-circuit current and T1⁺ flux seems to be similar for the two species. Table 3 shows the results and indicates that the mean ratio between T1⁺ fluxes in opposite directions is 1·10. The large variation of the flux ratios for different experiments may be explained as biological, the guts are alive and have a muscle layer which may change the available surface for permeation. Since the expected ratio for a passive flux is 1·00, and for active K flux the ratio is probably 10 or more, we must again conclude that the T1⁺ flux is passive.

From Tables 1, 2, and 3 we concluded that the large flux is a passive one and no active transport of T1⁺ takes place. Thus we also must assume that no energy is needed for the permeation of Tl⁺

We can stop the supply of energy by stopping the metabolism and the best way is to replace the oxygen supply by stirring the solutions with nitrogen. A sign of the efficiency of this procedure is that the transmural pd drops to almost zero.

From Table 4, it is obvious that there is no decrease in the T1⁺ flux from blood-side to lumen when the K transport is abolished by quenching the metabolism. Thus we have a further indication that the large T1⁺ flux is passive.

It was not practical to compare the behaviour of T1⁺ ions with all alkali metal ions, but K⁺, Rb⁺ and Cs⁺ were used as representatives.

As seen in Table 5 the flux of T1⁺ from lumen to blood-side is 11-14 times as high per mm as that of K⁺. The delay between addition of the tracer and its appearance in the solution on the other side of the gut (see Harvey & Zerahn, 1969) is about twice as long for 42K⁺ as for ²⁰⁴T1⁺. The reason for this delay will be discussed later.

As a control, we compared the flux of ⁸⁶Rb and ¹³⁷Cs from the lumen-side to blood-side on the same gut simultaneously. These experiments were performed on guts which were only slightly distended to minimize the effect of small leaks in the gut, which would influence the ratio between Rb and Cs fluxes; for this reason the guts were not short-circuited. Rb⁺ is expected to follow K⁺ to within 10–20% (Nedergaard & Harvey, 1968). On the other hand Cs in a concentration of 1 mm will not be actively transported. Since the measured fluxes are, however, from lumen to blood-side, where we expect an ordinary passive diffusion for these ions, it is possible to compare the fluxes of Rb⁺ and Cs⁺ directly. The results from Table 6 are derived from guts which are not short-circuited, but as seen later (Table 10) this will only change the Tl⁺ flux insignificantly. Table 6 shows the results of the measurements.

After 10 min the two ions have a ratio for the passive fluxes from lumen to blood-side of around one. Even if Rb⁺ is actively transported from the blood-side to lumen, and takes more time than Cs⁺ to come to equilibrium, the two ions will behave very similarly when steady state is reached.

Therefore it is sufficient to compare one of the ions with T1⁺, and Cs was chosen. Table 7 shows the flux of Cs⁺ and Tl⁺ from lumen to blood-side simultaneously across the same midgut.

A large difference is observed, with T1⁺ showing a flux which is 9–70 times as high as the flux of Cs, when the concentration of both ions is 1 mm.

The potential difference and short-circuit current of the midgut decrease with increasing concentration of T1+. This indicates that the active K transport is decreasing and that the T1⁺ ion is an inhibitor for the K transport, possibly a rather specific one. This inhibition of the K transport is often reversible, but sometimes irreversible. The inhibition was effective on the midguts from the species Hyalophora cecropia, Samia cynthia (Drury), Antherea pernyi and Actia selene. Preliminary experiments on cecropia midguts have shown that 2 mm-Tl⁺ on the lumen-side inhibits 95% of the active K flux, but 10% only of the passive K flux. In contrast, passive Tl⁺ flux is proportional to the concentration when measured within the range between 01 and 6 mm-Tl⁺ (Table 8) independent of inhibition of the active K flux.

As seen from Table 8 the T1⁺ flux is proportional to the concentration and no saturation of the flux at high concentration was observed.

In general we would expect a passive diffusion to have a low temperature coefficient in contrast to a rather large one for an active transport. The temperature dependence of the T1⁺ flux is shown in Table 9.

We found that the flux of T1⁺ in undistended guts is reduced by a factor of 2·5 on cooling the gut by 20°C, but by 3·5 times in distended guts. The measurements show that distension may have a larger effect on the Tl penetration of the gut at room temperature than at lower temperatures.

Since we had only one isotope of Tl (²⁰⁴Tl) at our disposal, we used the ‘sandwich technique’ described by Harvey & Nedergaard (1964) in order to obtain the fluxes in both directions.

Four experiments were carried out, and samples for flux determinations were taken after 5 min, 10 min, and 20 min. The washing out of labelled T1+ took about 30 min, after which the flux in the opposite direction was measured. This was followed by a new washing after which the flux in the first direction was measured again. Two experiments started by measuring the flux from blood-side to lumen and two by measuring fluxes from lumen to blood-side. The potential difference of the gut changes from about 90 mV for the first period of measurements to about 40 mV for the last. However the influence of this pd change on the T1⁺ flux was only about 10 –20% (Table 10), so that only a small error is introduced by taking the mean value.

Thus we could calculate the flux ratios for all four experiments and compare the ratios actually found with the ratios calculated from the potential difference for each period (Table 11).

In the case of high potential differences there was a pronounced difference between the calculated and observed values, indicating that much of the T1 flux is not passing the midgut by simple passive diffusion. For the low potential differences the flux ratios were rather close to the values expected for a passive diffusion.

Because we know so little of the process of T1 permeation we could imagine that T1 penetrates the gut by dissolving in the membrane phase as a salt. This could, however, be expected to be very different for different salts like the thallous chloride and the sulphate. This was tested by measuring the flux first with one and later with the other anion, and then back to the first again. It seems clear that changing the anion has very little influence on the T1⁺ flux through the midgut (Table 12). Similarly, the change from K to Na as the main cation in the solution has no significant effect (see below).

Most of the results are obtained with K⁺ as the main cation in the solution. It is possible that T1⁺ is passing the midgut by exchanging with K⁺, in which case the rapid transport will not continue when only other cations are available. We tested this possibility by replacing K⁺ in 32-K with Na+ (32-Na). Table 13 shows that T1⁺ will penetrate the cecropia midgut equally well, whether Na⁺ or K⁺ is the main cation.

Generally the solution we used was that called 32-K, with 32 mm-K⁺, 30 mm-Cl⁻, 2 mm-HCO₃⁻, and sucrose to give 260 m-osmol/1. If, however, the solution also contained 5 mm-CaCl₂ and 5 mm-MgCl₂ the findings were the same. Three experiments performed with this solution + 1 mm-TlNO₃ gave a mean value of short-circuit current of 16·5 μequiv/h and a T1+ flux (× 33) of 18·8 μequiv/h. It is obvious therefore that T1⁺ still behaves in the same way, even when Ca and Mg are also present.

Although Table 1 and 2 show that T1⁺ penetrates the midgut from blood-side to lumen almost as if Tl⁺ was hardly discriminated from K⁺, the flux from lumen to blood-side is of the same magnitude. This is confirmed in Table 3 where it is shown that the net T1⁺ flux is not significantly different from zero when the sandwich technique is used and the fluxes are measured on the same midgut. Thus the conclusion is that T1⁺ is not actively transported by the cecropia midgut. This was further confirmed by Table 4 which shows that the T1⁺ transport from blood-side to lumen does not depend on the metabolism of the midgut.

These results do not, however, imply that the flux of Tl⁺ is a simple passive diffusion for the following reasons:

• The ratio of the two fluxes through the gut as seen in Table 11 does not follow the flux ratio equation given by Ussing (1949) for a simple passive diffusion.

• The flux of T1+ through the midgut as seen from Tables 5–7 is much faster than could be expected for an ion of a size similar to K or Cs or Rb passing the gut by passive diffusion.

So we must conclude that neither diffusion nor active transport are the means for the relatively fast penetration of Tl through the midgut of Hyalophora cecropia and some of the other satumiids.

But there is another form of transport that has not been considered: the process defined by Ussing (1947) as exchange diffusion, where an ion can penetrate the tissue or membrane only by exchanging with another identical ion on the other side of membrane. Experiments with different concentrations of Tl⁺ (Table 8) show that the flux is proportional to the concentration even at very low concentrations, such 0·1 mm. Therefore there would be few T1+ ions on the opposite side to exchange with at these low concentration; as a result the flux should be proportionally less which is contrary to our findings.

We therefore measured the flux of T1⁺ across the gut with and without T1⁺ on the opposite side. With T1⁺ in the solution on both sides, the flux from blood-side to lumen was 0·121 μequiv/h, but with a Tl-free solution on the lumen-side the flux from blood-side to lumen was 0·101 μequiv/h. In the opposite direction (lumen to blood-side) the flux was 0-108 μequiv/h with 0·1 mm-Tl on both sides, but 0·091 μequiv/h with a Tl-free solution on the blood-side. The mean flux value for all determinations was 0·116 μequiv/h in 0·1 mm-Tl solutions and 0·096 μequiv/h for the experiments in which the solution in the direction of the flux contained no Tl. Exchange diffusion therefore seems to play no important role in the flux of Tl if any.

This process was first proposed as an explanation for the large flux of glucose and amino acids across the membrane of the red cell where the penetration is much larger than expected for a simple diffusion (see Wilbrand, 1960). Whether the Tl⁺ permeates the gut by a carrier-mediated transport or a specific pore system like that suggested by Mullins (1959) requires a closer study. But pore systems do fit into the picture with a flux which is without saturation kinetics and also independent of the flux of other ions. A comparison with the specific permeable glasses used for glass electrodes may also be considered (Eisenmann, 1962). Studies have shown that T1⁺ can replace K⁺ in, for example, muscle (found by comparing flux or cell concentration). In most tissues where it has been studied, K⁺ can be replaced by T1⁺ for stimulating the ATPase, even if high T1⁺ concentrations inhibit the process. Whether Tl⁺ can replace K in a particular chemical or physical process is not well established.

How can the high rate of permeation of T1+ be explained? From the values of the ion radius given from crystal data (see Mullins & Moore, 1960) it seems clear that the differences in size of the ions should be too small for explaining the large permeation rate for Tl⁺. The passive permeation rates of K, Rb, Cs (see Tables 5, 6, 7) are rather similar. T1⁺ is the exception. The chemistry of T1⁺ seems to provide no explanation for this. T1⁺ forms many chemical compounds that are only slightly soluble, namely chloride, bromide, iodide, sulphide, etc., so one might have thought that formation of insoluble compounds might delay the permeation. But actually, T1⁺ ions pass faster than the other ions through the gut, and even if small amounts are found to be bound to the membrane this does not seem to delay the transport, which if anything is too fast. Tl⁺ seems to behave similarly in several membranes, as judged from the literature.

We have considered possible effects of different solubility of T1 compounds, but two different anions make no difference (see Table 12) and T1⁺ does not have any compounds which are especially lipoid soluble. So we return to the assumption of facilitated diffusion, which in one respect is clearly supported: T1⁺ permeation is much more rapid than expected.

Certain findings do not agree with this hypothesis: (a) the T1⁺ flux is not saturated at high concentrations and (b) we know no compound which inhibits the flux, because it has not been investigated.

But to deal with point (a), it is possible that the concentrations used were too low and that a saturation may in fact occur at higher concentrations. An inhibitor may be found, but none except T1⁺ seems to be known at the moment. At this point the only way to decrease the T1⁺ flux seems to be to decrease the surface of the gut.

In the cecropia midgut T1⁺ acts as an inhibitor for the active K flux and causes the potential across the midgut to fall. But T1⁺ is in many cases a potent inhibitor of K transport. A T1 concentration as low as 0·1 mm on the lumen-side alone would sometimes depress the potential to less than 30% of the starting value. The action of T1 as an inhibitor, although not investigated sufficiently, makes T1 an important tool for studying the midgut, because in contrast to the study of Na transport in many other membranes, there is in this case no specific inhibitor available. But even if T1 is not actively transported in the cecropia midgut it may behave quite differently in other membranes, like the frog skin, where it has been found that potassium and thallous ions are showing the same behaviour and where both are actively transported from inside to outside (private communication).

Neher (1975) reported that typical gramicidin channels in black membranes lack the high specificity for certain cations which is common in biological systems, but the thallous ion is an exception. Eisenmann, Krasne & Ciani (1974) found that the permeability ratio for Tl⁺ versus K⁺ is approximately 50 as compared to ratios between 0·1 and 2 for most other monovalent cations. Neher found that in black membranes T1⁺ blocks Na⁺ currents at concentrations which are two orders of magnitude lower than the sodium concentration even though the ion itself carries current through the channels better than Na⁺ and K⁺. In the cecropia midgut, T1⁺ penetrates the gut much faster than Cs ions (see Tables 6 and 7), but this was not found for K ions (or Rb⁺). Preliminary experiments showed that 95% inhibition of the active K flux with 2 mm-Tl⁺ on the lumen side produced an insignificant inhibition of 10% of the passive K flux.

A careful study of the movement of Tl⁺ across the membrane of frog sartorius muscle fibres in both directions was made by Mullins & Moore (1960). The flux of T1⁺ was measured directly by use of the isotope ²⁰⁴Tl. For example, with an external T1+ concentration of 2·5 mm, the T1⁺ influx was 8·3 pmol/cm² s and the influx of K was estimated to be 6pmol/cm²s. Thallous ions reach a steady-state distribution between fibre water and solution that is very close to the corresponding ratio for K⁺: Tl₁/Tl₀ was about 40, whereas the ratio for K was 38. Also during stimulation it seems clear that the permeability of the muscle membrane to T1⁺ is as large as to K⁺.

It is therefore convincing when Mullins & Moore (1960) conclude (p. 772) as follows. ‘It seems difficult to escape the conclusion that neither the mechanism that determines the selectivity of the resting membrane toward ions, nor the mechanism that determines the selectivity of the active membrane, can discriminate between K+ and Tl⁺ in spite of the obvious chemical differences in these ions. ‘

Gehring & Hammond (1967) report that the (Na + K) ATPase from rat erythrocytes is activated by T1⁺ (Na present). Skulskii et al. (1973) found that in human or rabbit erythrocytes, Tl is as effective as K in this activation at only one-tenth of the concentration of K. This was also found by Lishko, Kolchynska & Parkhomenko (1977) in human red cells. ATPase from rabbit kidney is also stimulated by T1+ (Britten & Blank, 1968), and in addition, Tl⁺ is 10 times as effective as K⁺. The same was found in beef brain by Inturrisi (1969), and by Maslova et al. (1971) for frog skin, where the affinity of the (Na + K) ATPase for Tl is also 10 times higher than for K. It seems that not only can Tl replace K at low Tl concentration with respect to the activation of (Na + K) ATPase, but does so with an even higher affinity.

Cavieres & Ellory (1974) found that T1⁺ can replace K⁺ with a higher affinity in the activation of the ouabain sensitive Na+ efflux. This is also found for squid nerve by Landowne (1975).

Skulskii et al. (1973) found that the T1+ and K+ flux rate constants for fragmented red cells were 9·36 h⁻¹ for T1⁺ and 0·153 h⁻¹ for K⁺ inward, and 4·68 h⁻¹ for T1⁺ and 0·0066 h⁻¹ for K⁺ outward.

The permeability of K channels to various cations was studied in myelinated frog nerve by Hille (1973). The ratios P_T1: P_K: P_Rb: are 2·3: 1·00:0·92:0·13. No other ions are found to be measurably permeant, including Li⁺, Na⁺, or Cs⁺.

Skulskii et al. (1973) studied the snail brain and found that the inward rate constants had the following sequence T1+ > Rb+ > Cs+. The passive permeability of T1⁺ in the snail brain was 50–100 times less than the T1⁺ permeability of red cell membrane.

For the (Na + K) ATPase it was found that low concentrations of T1⁺ may stimulate the ATPase, but high concentrations will inhibit the enzyme. The active Na transport in blood cells will also be inhibited by high concentrations of T1⁺ (Cavieres & Ellory, 1974). Natochin & Skulskii (1971) found that active transport of Na in the frog skin was inhibited by T⁺ and explained that T1⁺ inactivates the ability of the pump (and possibly this particular enzyme) to translocate ions. This may, however, be different for the cecropia midgut, because here the inhibition is often reversible, which must imply that the Tl⁺ cannot be firmly bound.

Koefoed (1975) found that T1⁺ follows K⁺ in the uptake by the rectal complex of the mealworm, Tenebrio molitor. This could be explained as an active transport of T1⁺, but other possibilities were also taken into account. Whether the transport is a kind of facilitated diffusion, as we seem to have in the Hyalophora cecropia midgut, cannot be decided, because different tissues behave differently with respect to T1⁺.

In the gut of Hyalophora cecropia T1⁺ permeates from blood-side to lumen as rapidly as if T1⁺ was a marker for K⁺. Nevertheless T1⁺ is not actively transported. The high permeability is the same in both directions when the gut is short-circuited. The T1⁺ permeation does not depend on the metabolism of the gut, but is up to 50 times as high as the passive permeation of Cs or Rb. It is suggested that some special kind of diffusion may take place such as facilitated diffusion, or a specific diffusion like the Tl⁺ diffusion through gramicidin channels in black membranes.

T1+ was found to be an inhibitor for the active K transport in the midgut even though it is not actively transported. This property may be useful in later work because no good inhibitor of active K transport in this tissue is known.

It should be noted that these conclusions have been shown to be valid only for the Hyalophora cecropia midgut. How far these properties of the T1⁺ flux apply in other membranes must be studied carefully. In other membranes, such as frog muscle and skin, T1⁺ and K⁺ both behave in a similar way and both can either be actively transported or show a simple passive diffusion.

We thank Dr V. Koefoed-Johnsen, Professor S. O. Andersen, Professor J. Ringsted, and Professor H. H. Ussing for helpful criticism in preparing the manuscript and Dr Leon Pape for correcting the language.