Regulation of Na⁺-K⁺-2Cl⁻ cotransport by protein phosphorylation in ferret erythrocytes

Abstract

1. Na⁺-K⁺-2Cl⁻ cotransport in ferret erythrocytes was measured as the bumetanide-sensitive uptake of ⁸⁶Rb.

2. The resting cotransport rate was high but could be increased threefold by treating erythrocytes with calyculin A, a potent inhibitor of serine/threonine phosphatases. Twenty nanomolar was sufficient to maximally and rapidly (within 4 min) stimulate transport.

3. The effects of several kinase inhibitors were tested. High concentrations of K-252a, K-252b, calphostin C and hypericin caused less than 20 % inhibition. Staurosporine (IC₅₀, 0.06 μm) and 4-amino-5-(4-methylphenyl)-7-(t-butyl)pyrazolo[3,4-d]pyrimidine (PP1; IC₅₀, 2.5 μm) were more potent but still only partially (40–50 %) inhibited transport, an effect mimicked by reducing ionized intracellular Mg²⁺ concentration to submicromolar levels. Genistein may inhibit all transport at a sufficiently high dose (IC₅₀, 0.36 mM) perhaps by directly inhibiting the transporter.

4. Staurosporine, PP1 and the removal of Mg²⁺ all prevented subsequent stimulation by calyculin A, and all inhibited calyculin-stimulated transport by 20–30 %. The effects of staurosporine, PP1 and Mg²⁺ removal were not additive.

5. The phosphatase that dephosphorylates the cotransporter is probably Mg²⁺ (or possibly Ca²⁺ or Mn²⁺) sensitive and not the target for calyculin A. The data suggest that this phosphatase is inhibited by phosphorylation, and that it is the regulation of this process which is affected by calyculin A and the kinase inhibitors tested here. Phosphorylation of the phosphatase is probably regulated by members of the Src family of tyrosine kinases.

The Na⁺-K⁺-2Cl⁻ cotransport system plays a vital role in the regulation of cell K⁺ and Cl⁻ content and cell volume (Haas, 1989,1994; Haas & Forbush, 1998). In epithelia it is a major route for entry of Na⁺ and Cl⁻ into cells. Thus, in the thick ascending limb of the kidney it is the major route for the reabsorption of these ions across the apical membrane, and is the site of action of loop diuretics such as furosemide (frusemide) and bumetanide. In secretory epithelia, it allows entry of Na⁺ and Cl⁻ across the basolateral membrane, which are subsequently secreted across the apical membrane (O'Grady et al. 1987). There are at least two major isoforms of the cotransporter: NKCC1 (or BSC2), a housekeeping form found in many cell types and responsible for ion uptake on the basolateral borders of epithelia, and NKCC2 (or BSC1), the reabsorptive form found exclusively in the apical membranes of the kidney (Payne & Forbush, 1995; Mount et al. 1998; Haas & Forbush, 1998). In addition to responding to changes in the concentration of its substrate ions, the cotransporter is also affected by a variety of physiological stimuli including cell shrinkage and several hormones (Haas & Forbush, 1998). Many of these stimuli affect the cotransporter by altering protein phosphorylation (Alper et al. 1980), and these signal transduction pathways can be disrupted by treating cells with inhibitors of protein kinases and phosphatases (Pewitt et al. 1990; Klein et al. 1993; Palfrey & Pewitt, 1993). However, because these agents are non-specific, it was not clear whether the cotransporter itself (rather than an accessory protein) was the target for phosphorylation. Recent work has now shown that the cotransporter is indeed phosphorylated when cells are shrunk or treated with a variety of pharmacological agents (Lytle & Forbush, 1992; Tanimura et al. 1995; Lytle, 1997). It seems that phosphorylation activates the transporter and is also necessary for binding of inhibitors like bumetanide. A tacit assumption appears to be that the potent phosphatase inhibitor calyculin A inhibits the phosphatase that dephosphorylates the cotransporter.

There is also evidence that protein-protein interactions may influence activity independently of the cotransporter's phosphorylation state. Interference with the cytoskeleton, for instance by loading cells with phalloidin, inhibits the cotransporter without affecting ³H-bumetanide binding, which is thought to reflect the level of cotransporter phosphorylation (Matthews et al. 1994, 1998; D'Andrea et al. 1996).

Ferret erythrocytes have proved a useful simple mammalian system in which to study the properties of the cotransporter. More than 95 % of K⁺ uptake by these cells is through the cotransporter and can be inhibited by low doses of bumetanide. In addition, they have little or no sodium pump activity, thus simplifying analysis of experimental data (Flatman, 1983). Work on ferret erythrocytes has revealed that the cotransporter is stimulated by intracellular Mg²⁺ by mechanisms dependent on the presence of ATP, supporting the notion that phosphorylation activates the cotransporter either directly, or indirectly through accessory proteins (Flatman, 1988,1991). However, reduction of intracellular ionized Mg²⁺ concentration ([Mg²⁺]_i) to very low levels by treating cells with the ionophore A23187 and 2 mM EDTA inhibited less than half the ⁸⁶Rb uptake. The remaining level of uptake could be maintained for more than 1 h (Flatman, 1988). Assuming that Mg²⁺ removal inhibits all kinase activity, the finding suggests either that the phosphatases are also Mg²⁺ dependent or that the cotransporter can partially operate in a dephosphorylated state. We extend this work by examining the effects of several inhibitors of kinases and phosphatases on ⁸⁶Rb (used as a tracer for K⁺) uptake into ferret erythrocytes. We suggest the cotransporter is dephosphorylated by a Mg²⁺-sensitive phosphatase, which itself is regulated by phosphorylation. The kinases and phosphatases which regulate this process are probably the site of action for the inhibitors tested here, including calyculin A. A kinase inhibitor which potently and specifically inhibits phosphorylation of the cotransporter has still to be identified.

METHODS

All solutions were prepared in double glass-distilled water with reagents of analytical quality (AnalaR, Merck). Most experiments were carried out in ferret basic medium (FBM; composition (mM): 150 NaCl, 5 KCl, 0.05 EGTA and 10 Hepes, pH 7.5 at 38°C, adjusted with NaOH). Stock solutions prepared in ethanol were: 2 mM A23187, 10 mM forskolin, 10 mM cyclosporin A and 10 μm calyculin A. Stock solutions prepared in dimethyl sulphoxide (DMSO) were (mM): 0.5 staurosporine, 30 genistein, 5 4-amino-5-(4-methylphenyl)-7-(t-butyl)pyrazolo[3,4-d]pyrimidine (PP1), 50 tyrphostins AG1296, AG1478 or B42, 0.43 K-252a or K-252b, 2 phorbol-12-myristate-13-acetate (TPA), 0.13 calphostin C and 1 hypericin. Potassium bisperoxo(1,10-phenanthroline)oxovanadate(V) (bpV(phen)) was stored as a 10 mM stock in water. Bumetanide (10 mM; a gift from Leo Laboratories Ltd, Aylesbury, UK) was prepared in water neutralized with Tris base. Staurosporine, cAMP, dibutyryl cAMP, dibutyryl cGMP, DMSO, EDTA, EGTA, genistin, Hepes, 3-isobutyl-1-methylxanthine (IBMX) and Tris base were obtained from Sigma; A23187, forskolin, cyclosporin A, AG1296, AG1478, B42, K-252a, K-252b, TPA, calphostin C, hypericin and bpV(phen) were from Calbiochem; genistein was from Calbiochem or Sigma; PP1 was from Calbiochem or Alexis (Nottingham, UK); and calyculin A was from Alexis.

Blood was taken by cardiac puncture into EDTA from adult ferrets terminally anaesthetized with sodium pentobarbitone (Sagatal; Rhône Mérieux, Harlow, UK; 120 mg kg⁻¹i.p.). Cells were washed 4 times by centrifugation and resuspended in FBM, care being taken to completely remove the buffy coat. Cells were stored packed at about 80 % haematocrit in FBM at 4°C until use (most experiments were carried out within 3 days of cell collection). All solutions used for handling blood were sterilized by passage through a 0.2 μm filter (Millipore).

K⁺ fluxes were measured using ⁸⁶Rb (Amersham Pharmacia Biotech), which is an excellent tracer for K⁺ on transport systems in ferret erythrocytes (Flatman, 1987). Most fluxes were measured over a 3 min period following the addition of ⁸⁶Rb to well-stirred suspensions of erythrocytes (5–10 % haematocrit) in FBM at 38°C. Fluxes were measured over a 12–15 min period when 20 μm bumetanide was present to inhibit the Na⁺-K⁺-2Cl⁻ cotransporter. Under these conditions the fluxes are very small. After cell suspensions had equilibrated at 38°C for a time appropriate for a particular drug treatment (between 15 min and 1 h), ⁸⁶Rb was added. ⁸⁶Rb activity in cells (A_it) was assessed by taking 0.1 ml samples of the suspensions at appropriate times (1 min intervals for most experiments, 3 min intervals when bumetanide was present) and adding these to microcentrifuge tubes containing 1 ml of ice-cold FBM and 0.25 ml of di-n-butylphthalate. The tubes were capped, shaken to mix the contents, and spun at 14 000 g for 20 s, by which time the cells had formed a tight pellet below the oil. Supernatants and oil were removed by suction, and the walls and caps of the tubes were cleaned with cotton swabs. Cell pellets were lysed in 1 ml of distilled water and 50 μl of 55 % trichloroacetic acid (TCA) was added to precipitate protein. Tubes were then spun at 14 000 g for 3 min and 0.7 ml of the supernatants was added to 3 ml scintillation cocktail (Ultima Gold, Packard). ⁸⁶Rb activity was determined in a scintillation counter. Total ⁸⁶Rb activity in the suspension (A^tot) was assessed by adding 0.1 ml suspension to 0.9 ml water, precipitating protein with TCA and processing as described above. The rate constant for ⁸⁶Rb uptake (k, h⁻¹) is given by the equation:

where t is the time in hours since the addition of ⁸⁶Rb to the suspension, and h is the haematocrit of the suspension (expressed as a decimal fraction). k with its standard error was obtained by linear regression analysis by plotting ln(1 - A_it/hA^tot) against t.

Where calyculin A and a kinase inhibitor or Mg²⁺ removal were used in combination, the cells were treated in succession for 10–15 min with each compound. Cells were not washed between additions.

Haematocrit was determined using the cyanmethaemoglobin method. Suspension (0.1 ml) was added to 4.9 ml of Van Kampen and Zijlstra's buffered diluent (Merck) and the absorbance at 540 nm was measured after about 1 h. The packed cell absorbance of ferret erythrocytes under these conditions with a 1 cm pathlength is 230 (Flatman & Andrews, 1983).

Cell ATP content was measured with the luciferin-luciferase assay (Flatman, 1988), and cell magnesium content was controlled using A23187 (Flatman, 1988). Cells were Mg²⁺ depleted by treating them for 15 min with 10 μm A23187 and 2 mM EDTA before inhibitors were added or fluxes measured. Under these conditions [Mg²⁺]_i is reduced to less than 1 μm (Flatman, 1988). However, as A23187 transports other divalent cations which bind strongly to EDTA, this treatment will probably also deplete cells of ions such as Ca²⁺, Mn²⁺ and Zn²⁺. cAMP was extracted from erythrocytes using the same protocol as for ATP determination. cAMP concentrations in cell extracts were determined using a variety of competitive protein binding assays (including acetylation assays) from Amersham International according to the manufacturer's instructions.

Fluxes and cell contents are expressed per litre of cells at their original volume (loc).

All experiments were repeated at least twice with blood from different ferrets, and either representative rate constants are given together with the standard error of the estimate, or mean values are given with the standard error of the mean, in which case the number of repeats (n) is given in parentheses. Where necessary the significance of the difference between means was assessed with a two-tailed, unpaired t test.

RESULTS

Calyculin A and other phosphatase inhibitors

Treatment with the serine/threonine phosphatase inhibitor calyculin A caused a rapid stimulation of ⁸⁶Rb influx into ferret erythrocytes, influx rate constants rising from control values of about 3 h⁻¹ to levels as high as 9 h⁻¹. This increase in rate was due almost exclusively to the activation of the Na⁺-K⁺-2Cl⁻ cotransporter since the rate constant of fluxes in the presence of 20 μm bumetanide increased only very slightly from 0.073 ± 0.002 to 0.094 ± 0.004 h⁻¹. Cotransport thus represents over 95 % of K⁺ influx under control conditions and about 99 % in calyculin-treated cells. Influx was maximal about 4 min after adding calyculin (Fig. 1), and remained at this level for at least an hour. Twenty nanomolar was sufficient to maximally stimulate flux. However, the dose-response curve was very steep (Fig. 2) and could not be described by simple Michaelis- Menten kinetics suggesting a complex interaction between calyculin and cotransporter. The extent of stimulation was strongly dependent on the length of cell storage at 4°C. Figure 3 shows ⁸⁶Rb influx into cells taken from a single ferret, stored for a total of 11 days. The calyculin-stimulated flux remained high during the first 2 days of storage, but then fell with a half-time of about a day. This pattern was repeated when transport was examined in the erythrocytes from several animals: stimulation by 20 nm or more calyculin was 169 ± 19 % (n = 5 ferrets) on the day of bleeding, was still 159.8 ± 38.5 % (n = 4) after 2 days of storage, but had fallen to 75.4 ± 10.5 % (n = 5) after 3 days and continued to fall thereafter.

Figure 1. Time dependence of the effect of calyculin A on ⁸⁶Rb uptake in ferret erythrocytes.

A suspension of ferret erythrocytes was incubated at 38 °C. ⁸⁶Rb uptake was measured over a 3 min period in aliquots of the suspension taken so that ⁸⁶Rb could be added 2 min before the indicated time thus centring the fluxes on those times. Calyculin A (20 nm) was added at zero time. Points indicate the rate constant for ⁸⁶Rb influx with standard error.

Figure 2. Concentration dependence of stimulation of cotransport by calyculin A.

Suspensions of ferret erythrocytes were treated with the indicated concentrations of calyculin A for 10 min. ⁸⁶Rb was then added and the influx rate constant determined over a 3 min period. Results are from two experiments indicated by the different symbols. Points indicate the influx rate constants. Standard errors are smaller than point size.

Figure 3. Effect of storage duration on the response of ferret erythrocytes to calyculin A.

Washed ferret erythrocytes were stored at about 80 % haematocrit in FBM at 4 °C. At the times indicated a sample of suspension was taken and the cells were washed once with FBM. ⁸⁶Rb uptake was measured in these cells suspended in FBM at 38 °C with (▪) and without (•) 20 nm calyculin A. ▴ represents the calyculin-sensitive flux. Points indicate the rate constants with their standard errors.

Cyclosporin A (10 μm), which inhibits protein phosphatase 2B by binding to cyclophilin, reduced uptake by 30 % over a 30 min period (Table 1). The inhibition was slow to develop, being significantly greater at 30 min than at 10 min (data not shown) and required micromolar amounts of cyclosporin as 10 μm was more effective than 1 μm. The potent protein phosphotyrosine phosphatase inhibitor bpV(phen) had little effect on ⁸⁶Rb uptake (Table 1).

Table 1.

The effects of compounds which affect kinase and phosphatase activity on ⁸⁶Rb uptake by ferret erythrocytes

Compounds tested	Concentration	Exposure time (min)	86Rb influx rate constant (% control)
K-252a	4.3 μm	15	82.9 ± 2.2(3)
K-252b	4.3 μm	15	86.2 ± 1.4(3)
TPA	1 μm	15	81.5 ± 1.0(3)
Calphostin C	1.3 μm	15	87.1 ± 0.3(3)
bpV(phen)	0.13 mM	15	112.6
Genistin	0.3 mM	30	81 ± 3.0(3)
AG1296	0.1–0.33 mM	10–15	96.4 ± 7.8(3)
AG1478	0.1 mM	15	81.2
B42	0.1 mM	15	91.5
Hypericin	0.1–1 μm	30	83 ± 3.0(3)
Hypericin (illuminated)	1–5 μm	30	94 ± 9.0(3)
Cyclosporin A	10 μm	30	67.9

Cells were treated with these compounds for the times indicated before ⁸⁶Rb influx rate was measured. Rate constants are expressed as a percentage of the control flux (mean ± s.e.m.). Numbers in parentheses, number of repeats.

Kinase inhibitors (general)

We screened a range of kinase inhibitors for effects on the cotransporter. Three compounds, staurosporine, PP1 and genistein, caused significant inhibition within 10 min and the actions of these inhibitors is explored in more depth later. High concentrations of inhibitors were used as they may penetrate the membrane poorly. K-252b and K-252a (both at 4.3 μm) caused inhibition of 15–20 %⁸⁶Rb uptake in ferret erythrocytes and 1.3 μm calphostin C, the potent protein kinase C inhibitor (Tamaoki, 1991), caused about 13 % inhibition (Table 1). The phorbol ester TPA, an activator of protein kinase C, was ineffective at 0.1 μm but was inhibitory at 1 μm. Treatment of cells for 30 min with hypericin, which inhibits casein kinase II (Agostinis et al. 1995), inhibited about 17 %⁸⁶Rb uptake in 30 min. The effect of 1 μm was not significantly greater than that of 0.1 μm, and the effect was slow to develop, suggesting that transport inhibition is a slow consequence of target inhibition, rather than that hypericin entry is rate limiting. Curiously, illumination with bright fluorescent light reduced the effectiveness of hypericin rather than enhancing it (Table 1; and see Agostinis et al. 1995).

Staurosporine

This broad specificity kinase inhibitor (Tamaoki, 1991; Hanke et al. 1996) rapidly (rate about 0.2 min⁻¹) inhibited about half of the ⁸⁶Rb uptake in ferret erythrocytes. The remaining activity declined much more slowly at about 0.01 min⁻¹ (Fig. 4). Thus even after 2 h treatment, the longest time that could be tested before haemolysis became problematic, there was still significant ⁸⁶Rb uptake (rate constant, 0.63 ± 0.03 h⁻¹). As 2 μm staurosporine only caused a small increase in the bumetanide-resistant flux (rate constant increasing from 0.073 ± 0.002 to 0.117 ± 0.01 h⁻¹) it appears mainly to inhibit Na⁺-K⁺-2Cl⁻ cotransport activity. The data suggest that staurosporine may affect two targets which regulate cotransport activity. Thus by treating cells with staurosporine for between 10 and 15 min it was possible to inhibit more than 90 % of the rapidly affected target whilst inhibiting less than 10 % of the other. Figure 5 shows the relationship between staurosporine concentration and the extent of ⁸⁶Rb uptake inhibition by the component rapidly affected by staurosporine. One micromolar was sufficient to achieve maximum inhibition, and the data suggest an IC₅₀ of about 60 nm. No correlation could be found between the extent of ⁸⁶Rb uptake inhibition and the time for which cells were stored at 4°C. Similarly no correlation could be found between the percentage inhibition and the size of the flux before staurosporine was added (data not shown). Pooled data indicated that 46.8 ± 1.8 % (n = 28) ⁸⁶Rb uptake was inhibited rapidly (within 8–20 min) by staurosporine (> 1 μm).

Figure 4. Time dependence of the inhibitory effect of staurosporine on cotransport.

Fluxes were measured as described in the legend to Fig. 1 following the addition of 2 μm staurosporine at time zero. Results from two experiments, indicated by different symbols, are combined. The line is drawn assuming that there are two components which are affected at different rates by staurosporine:

where R is the ⁸⁶Rb influx rate constant t min after the addition of staurosporine. V₁, k₁, V₂ and k₂ obtained by non-linear regression analysis are 1.78 h⁻¹, 0.19 min⁻¹, 1.62 h⁻¹ and 0.008 min⁻¹, respectively.

Figure 5. Concentration dependence of the inhibition of cotransport by staurosporine.

Erythrocytes were treated with the indicated concentrations of staurosporine for 5 min and then the ⁸⁶Rb influx rate constant was measured. The line is drawn assuming that staurosporine partially inhibits ⁸⁶Rb influx according to the equation:

where R is the ⁸⁶Rb influx rate constant in the presence of cμm staurosporine. V₁, V₂ and IC₅₀ determined by non-linear regression analysis are 1.55 h⁻¹, 1.75 h⁻¹ and 0.06 μm, respectively.

PP1

PP1 was designed as a potent specific inhibitor of the Src family of tyrosine kinases (Hanke et al. 1996). In ferret erythrocytes it had actions very similar to staurosporine, causing a rapid inhibition of 38 ± 3 % (n = 13) ⁸⁶Rb uptake. It then slowly inhibited the remaining ⁸⁶Rb uptake. An example of the time course of inhibition in one ferret is shown in Fig. 6. In another experiment, the second slow component was more prominent. Figure 7 shows the dose- response curve for PP1 following a 10–15 min exposure to the compound. Twenty micromolar was sufficient to inhibit the fast component, and the IC₅₀ was 2.5 ± 1.6 μm. The majority of the effect of PP1 on K⁺ transport was due to inhibition of the Na⁺-K⁺-2Cl⁻ cotransporter because 50 μm PP1 caused only a very slight stimulation of bumetanide-resistant ⁸⁶Rb uptake (rate constant increased from 0.14 ± 0.01 to 0.21 ± 0.02 h⁻¹).

Figure 6. Time dependence of the effect of PP1 on cotransport.

Fluxes were measured as described in the legend to Fig. 1 following the addition of 50 μm PP1 at time zero. The line was drawn assuming that there are two components (V₁, V₂) which are affected at different rates by PP1:

where R is the ⁸⁶Rb influx rate constant t min after the addition of PP1. V₁, V₂, k₁ and k₂ obtained by non-linear regression analysis are 2.13 h⁻¹, 0.11 min⁻¹, 1.54 h⁻¹ and 0.004 min⁻¹, respectively.

Figure 7. Concentration dependence of the effect of PP1 on cotransport.

Erythrocytes were treated with the indicated concentrations of PP1 for 10 min and then ⁸⁶Rb influx rate constant was measured. The line was drawn assuming that PP1 partially inhibits ⁸⁶Rb influx according to the equation:

where R is the ⁸⁶Rb influx rate constant in the presence of cμm PP1. V₁, V₂ and IC₅₀ determined by non-linear regression analysis are 1.70 h⁻¹, 1.43 h⁻¹ and 2.5 μm, respectively.

Genistein

Figure 8 shows that 0.3 mM genistein, a compound that inhibits tyrosine kinases at low doses and serine/threonine kinases at higher doses (Akiyama & Ogawara, 1991) rapidly (about 0.35 min⁻¹) inhibited about 60 %⁸⁶Rb uptake in this sample of ferret erythrocytes. In 17 experiments where cells were exposed to 0.3 mM genistein for between 10 and 15 min uptake was inhibited by 47 ± 2 %. Figure 9 shows the extent of inhibition as a function of genistein concentration. The data suggest that genistein can inhibit all transport if a high enough dose could be applied. However, it was not possible to explore concentrations above 0.6 mM due to increased haemolysis. Assuming a single site of action the IC₅₀ was 0.36 ± 0.05 mM. The effect of genistein on ⁸⁶Rb uptake seemed to be exclusively on the Na⁺-K⁺-2Cl⁻ cotransporter since 0.3 mM genistein had no effect on ⁸⁶Rb uptake in the presence of 20 μm bumetanide (rate constants: control, 0.058 ± 0.004 h⁻¹; with genistein, 0.055 ± 0.004 h⁻¹). Genistin (0.3 mM), an analogue of genistein which does not inhibit tyrosine kinases (Akiyama & Ogawara, 1991), caused a 20 % inhibition of ⁸⁶Rb uptake (Table 1). A number of tyrphostins, potent tyrosine kinase inhibitors (Gazit et al. 1989), were also tested (Table 1). Neither AG1296 nor B42 (both at 0.1 mM) had any significant effect on ⁸⁶Rb uptake. AG1478 caused a small inhibition.

Figure 8. Time dependence of the effect of genistein on cotransport.

Fluxes were measured as described in the legend to Fig. 1 following the addition of 0.3 mM genistein at time zero. The line was drawn according to the equation:

where R is the ⁸⁶Rb influx rate constant t min after the addition of genistein. V₁, V₂ and k₂ obtained by non-linear regression analysis are 1.29 h⁻¹, 1.92 h⁻¹ and 0.35 min⁻¹, respectively.

Figure 9. Concentration dependence of the effect of genistein on cotransport.

Erythrocytes were treated with the indicated concentrations of genistein for 10–15 min and then ⁸⁶Rb influx rate constant was measured. Three separate experiments are shown by the different symbols, rate constants being normalized to the control values. The line is drawn assuming that genistein inhibits ⁸⁶Rb influx according to the equation:

where R is the normalized ⁸⁶Rb influx rate constant in the presence of c mM genistein. IC₅₀ determined by non-linear regression analysis is 0.36 mM.

Combinations of kinase inhibitors

The similarity of responses to staurosporine and PP1 suggests that these agents may inhibit the same pathway (though not necessarily the same kinase). If this is so then they should be no more potent in combination than alone. This was found to be so (Fig. 10). The same was found to be true for PP1 and genistein. However, the combination of staurosporine and genistein consistently produced the greatest inhibition of transport.

Figure 10. Combined effects of kinase inhibitors on cotransport in ferret erythrocytes.

⁸⁶Rb uptake into ferret erythrocytes was determined in control cells (Con) and cells treated with kinase inhibitors either alone or in combination for 30 min: 50 μm PP1; 2 μm staurosporine (Sta); 0.4 mM genistein (Gen).

Combination of kinase inhibitors and calyculin A

Staurosporine (2–4 μm) added before calyculin A almost completely prevented the stimulation of cotransport by calyculin. Activity following the addition of calyculin (20–100 nm) was only 5 ± 4 % (n = 6) of the activity seen when calyculin was present alone. Staurosporine added after calyculin inhibited cotransport by 31 ± 7 % (n = 5). In all cases where these agents were used in combination, cotransport activity was much greater (182 ± 72 %, n = 5) when calyculin was added first.

Addition of PP1 (50 μm) prior to calyculin A reduced the subsequent stimulation of cotransport by calyculin to 9 ± 11 % (n = 4) of the value seen with calyculin alone. Addition of PP1 after calyculin reduced cotransport seen in the presence of calyculin by 22 ± 7 % (n = 5). In all cases cotransport activity seen when PP1 was added after calyculin was greater (88 ± 15 %) than when PP1 was added first.

Genistein (0.3 mM) added prior to calyculin reduced the stimulation produced by 20–50 nm calyculin to 77 ± 11 % (n = 4) seen with calyculin alone. Genistein added after calyculin reduced transport seen in the presence of calyculin by 35 ± 2 % (n = 4). In this case, the levels of transport seen in the presence of both calyculin and genistein were independent of the order in which the compounds were added (rate constants were 4.2 ± 0.6 h⁻¹ when genistein was added first, 4.3 ± 0.5 h⁻¹ when calyculin was added first, n = 4).

Reduction of cell magnesium content combined with kinase and phosphatase inhibitors

Treatment of ferret erythrocytes for 15 min with A23187 and EDTA reduced ⁸⁶Rb uptake by 33.6 ± 2.9 % (n = 11). Pretreatment of cells with EDTA and A23187 completely prevented any subsequent stimulation of cotransport by calyculin A, and in one experiment, a significant inhibition of ⁸⁶Rb uptake was observed, the rate constant falling from 2.82 ± 0.05 to 1.07 ± 0.08 h⁻¹ on addition of calyculin. In cells already treated with calyculin to stimulate cotransport, the addition of EDTA and A23187 inhibited ⁸⁶Rb uptake by 28 ± 10 % (n = 4). The addition of 50 μm PP1 or 2 μm staurosporine after magnesium depletion had no significant effect on ⁸⁶Rb uptake (data not shown), whereas the addition of 0.3 mM genistein appeared to cause further inhibition (rate constant fell from 2.07 ± 0.04 to 1.43 ± 0.25 h⁻¹).

Role of cAMP

Treatment of ferret erythrocytes with 1 mM dibutyryl cAMP for 20 min to increase cell cAMP level inhibited ⁸⁶Rb uptake by 35 % whereas treatment with dibutyryl cGMP had no effect. We also attempted to increase cAMP levels by treating cells with 0.1 mM forskolin and 1 mM IBMX. This inhibited bumetanide-sensitive ⁸⁶Rb uptake by about 38 % (rate constant fell from 3.72 ± 0.05 to 2.29 ± 0.05 h⁻¹) while having little effect on bumetanide-resistant ⁸⁶Rb uptake (rate constant increased from 0.077 ± 0.005 to 0.109 ± 0.007 h⁻¹). IBMX alone had no effect on ⁸⁶Rb uptake. Curiously, cAMP levels fell by 20 ± 4 % (n = 3) (e.g. in one experiment it fell from 0.80 to 0.62 μmol loc⁻¹) when ferret erythrocytes were treated with IBMX and forskolin. We believe that the assay and extraction procedures were effective, as cAMP levels quadrupled when the same experiment was carried out with human erythrocytes.

Cell ATP content

Ferret erythrocyte ATP content was assessed after cells had been treated with inhibitors to determine whether any of the effects observed could be attributed to changes in ATP level resulting from interference with metabolism. Ferret erythrocyte ATP content was not altered from the normal level of about 0.6 mmol loc⁻¹ by treatment with 20 nm calyculin A, 2 μm staurosporine, 0.3 mM genistein or 50 μm PP1 (data not shown). As found previously (Flatman, 1988), treatment of cells with EDTA and A23187 caused a fall in ATP content to 0.4 mmol loc⁻¹. However, this level of ATP is more than sufficient to maintain the cotransport rate (Flatman, 1991).

DISCUSSION

As in other cells (Palfrey & Pewitt, 1993; Klein et al. 1993; Lytle, 1997) the potent protein serine/threonine phosphatase inhibitor calyculin A (Cohen et al. 1990) caused a rapid and pronounced stimulation of Na⁺-K⁺-2Cl⁻ cotransport activity in ferret erythrocytes. Stimulation was greatest on the day of cell collection and remained high for the following 2 days. After that, stimulation decreased, roughly paralleling the decline in the cellular ATP content of these stored cells (Flatman, 1991). Treatment with calyculin A thus revealed on-going kinase activity which phosphorylated and activated the cotransporter and which was normally balanced by phosphatase activity. An obvious conclusion is that a protein phosphatase type 1 or 2A (high-affinity targets for calyculin; Cohen et al. 1990) is responsible for the dephosphorylation of the cotransporter, as has been previously suggested (Pewitt et al. 1990; Klein et al. 1993; Palfrey & Pewitt, 1993; Lytle, 1997). However, this appeared not to be the case as it is hard to reconcile this with the observation that treatment of cells with A23187 and excess EDTA to reduce the concentration of intracellular divalent cations (especially [Mg²⁺]_i) to sub-micromolar levels, inhibited bumetanide-sensitive ⁸⁶Rb uptake by only 38 %. All protein kinases should be inhibited at these very low [Mg²⁺]_i, so any remaining phosphatase activity would eventually completely suppress cotransport activity. It appears that the phosphatase which dephosphorylates the cotransporter (let us call this P1, see Fig. 11) is either Mg²⁺ sensitive (protein phosphatase type 2C?; Cohen, 1991) or is extremely sensitive to another divalent cation (Ca²⁺, Mn²⁺?), as it can function at the low Ca²⁺ or Mn²⁺ levels normally found in ferret erythrocytes. Our observations also imply that there is very little spontaneous dephosphorylation of the cotransporter, as this would also lead to gradual loss of activity in the presence of EDTA and A23187. A more plausible explanation for the stimulation of cotransport by calyculin is that a protein phosphatase 1 or 2A regulates the activity of P1, and that P1 is inhibited by phosphorylation. The kinase which carries out this phosphorylation (K2, Fig. 11) is the most likely target for most of the kinase inhibitors tested here.

Figure 11. A simple model of Na⁺-K⁺-2Cl⁻ cotransporter phosphorylation in ferret erythrocytes.

Cotransporter protein (COT) is phosphorylated (COT-P) by kinase K1 and dephosphorylated by a Mg²⁺-sensitive phosphatase P1. P1 is inhibited by phosphorylation catalysed by kinase K2. Phosphatase P2 dephosphorylates and activates P1.

We have yet to find a kinase inhibitor which rapidly and potently inhibits all Na⁺-K⁺-2Cl⁻ cotransport activity in ferret erythrocytes. Genistein at very high doses may inhibit all activity, but the toxicity of this compound has prevented the proper testing of the hypothesis. The very high estimated IC₅₀ (> 0.3 mM) suggests that inhibition is not due solely to its effect on tyrosine kinases which bind genistein with high affinity (Akiyama & Ogawara, 1991). The potent selective tyrosine kinase inhibitors tyrphostin B42 and AG1296 (Gazit et al. 1989) do not mimic the effect of genistein, though AG1478 has some effect. Although genistein has been shown to inhibit about 90 % of Na⁺-K⁺-2Cl⁻ cotransport in human erythrocytes with an IC₅₀ of about 20 μm, consistent with an effect via tyrosine kinases (Zaidi & Kaji, 1996), there has been a recent report that genistein inhibits Na⁺-K⁺-2Cl⁻ cotransport in T84 epithelial cells by a mechanism independent of its effects on tyrosine kinases (Sicklick et al. 1998). Our finding that genistein, unlike staurosporine or PP1, produced additional inhibition after Mg²⁺ removal with EDTA and A23187, a condition where kinases should already be inhibited, suggests genistein may directly inhibit the cotransporter to some extent. This is further supported by our findings that genistein causes additional inhibition in the presence of staurosporine, and that its analogue genistin, which does not inhibit tyrosine kinases (Akiyama & Ogawara, 1991), also inhibits ⁸⁶Rb uptake.

The minor effects of K-252b, calphostin C, dibutyryl cGMP and hypericin on ⁸⁶Rb uptake suggest that protein kinases C and G and casein kinase II probably play little part in regulating the cotransporter in ferret erythrocytes and are unlikely candidates for K1. Similarly, K1 is unlikely to be PKA, though the role of this enzyme is more difficult to ascertain. Treatment of ferret erythrocytes with forskolin had the unusual effect of reducing cAMP levels by 20 % and inhibited cotransport by 38 %, perhaps indicating that resting levels of cAMP maintain PKA activity in these cells. However, this is probably not the case, as treatment of cells with dibutyryl cAMP to augment their cAMP level inhibited ⁸⁶Rb uptake. It seems more likely that forskolin inhibits the cotransporter independently of any effect on cAMP, and that cAMP inhibits cotransport in ferret erythrocytes as has been previously suggested (Mercer & Hoffman, 1985). This is similar to the effects of cAMP found in human erythrocytes (Garay, 1982; though see also Zaidi & Kaji, 1996), but contrasts with the response in duck or turkey erythrocytes (Palfrey et al. 1980; Haas, 1989).

The actions of PP1 and staurosporine on cotransport were similar, though staurosporine was generally the more potent, despite both agents being used at concentrations 20 times greater than their IC₅₀. PP1 inhibited about 38 % and staurosporine 46 % of cotransport rapidly, and then slowly inhibited some, if not all, of the remaining activity. Both prevented subsequent activation by calyculin A and also caused partial (20–30 %) inhibition of the transport seen in the presence of calyculin A. Staurosporine has been shown to reduce ²²Na uptake by the cotransporter in ferret erythrocytes (Mairbäurl & Herth, 1996), and to prevent stimulation of cotransport by calyculin in duck erythrocytes, but in this case had no effect when added after calyculin (Lytle, 1997). The simplest explanation for the effects of staurosporine and PP1 is that they both inhibit (directly or indirectly) kinase K2 (Fig. 11), which phosphorylates and inactivates phosphatase P1. Further evidence that they target the same pathway is the finding that a combination of these agents is no more effective than either alone. Thus when staurosporine or PP1 is added to cells, phosphatase P1 is dephosphorylated and becomes more active, reducing the cotransport rate. If calyculin is added at this point, it is ineffective as all P1 is in the dephosphorylated state. This model predicts that inhibiting kinase K2 with PP1 or staurosporine after treatment with calyculin should have little effect as the initial exposure to calyculin should have resulted in most P1 being in the phosphorylated, inactive state. However, both PP1 and staurosporine cause significant inhibition of ⁸⁶Rb uptake (20–30 %). This suggests that there may be other targets for these agents (see later), or that there is residual P1 activity when kinase K2 is inhibited due to incomplete inhibition of P2, or spontaneous dephosphorylation of P1 in the presence of calyculin.

Further evidence for the model shown in Fig. 11 is the finding that high doses of potent kinase inhibitors, alone or in combination, and reduction of [Mg²⁺]_i to submicromolar concentrations only partially inhibit Na⁺-K⁺-2Cl⁻ cotransport rate. Even with kinase K2 fully inhibited, the maximal rate of P1 may be insufficient to match the rate of K1 so that the cotransporter remains partially phosphorylated and partially active. In order to inhibit the Na⁺-K⁺-2Cl⁻ cotransporter completely a kinase inhibitor would have to target kinase K1.

Kinase K2 or perhaps the kinases which regulate its activity are inhibited by staurosporine, PP1 and possibly genistein. As PP1 was designed as a specific Src family kinase inhibitor and as staurosporine also inhibits these kinases (Hanke et al. 1996) it seems likely that K2 or one of its regulatory kinases is a member of the Src family of protein tyrosine kinases. In view of this it is surprising that the powerful tyrosine phosphatase inhibitor bpV(phen) had little effect on cotransport activity, though this is in accordance with a previous finding that vanadate does not significantly stimulate cotransport in ferret erythrocytes with normal ATP content (Flatman, 1991).

The effects of combinations of genistein and calyculin are harder to explain as it was not possible to use a dose of genistein much higher than its IC₅₀. However, the rate of cotransport in the presence of both agents together was independent of the order of addition, suggesting that genistein affects a pathway which is completely independent of the calyculin-sensitive route. The fact that genistein inhibits cotransport after calyculin treatment makes it less likely that K1 is genistein's chief target, though it is consistent with the idea that genistein inhibits the cotransporter directly. Alternatively, the similarities of rate could be a coincidence, and genistein, like PP1 and staurosporine, inhibits through affecting K2.

An effective way of non-specifically inhibiting kinases is to reduce [Mg²⁺]_i to submicromolar levels by treating cells with A23187 in the presence of EDTA. This treatment produced effects which were strikingly similar to the effects of PP1. ⁸⁶Rb uptake was inhibited by 33 %, stimulation by calyculin was completely prevented and ⁸⁶Rb uptake in the presence of calyculin was inhibited by about 28 %. Mg²⁺ removal has also been shown to reduce the stimulation of cotransport by calyculin in duck erythrocytes (Palfrey & Pewitt, 1993). The most likely explanation of this effect is that transient residual activity of P1 becomes apparent when K1 is inhibited by Mg²⁺ removal. For this to be so, K1 must be more sensitive to Mg²⁺ removal than P1. Further evidence for this is the fact that Mg²⁺ removal partially inhibited cotransport rather than freezing activity at the existing level. Although treatment of cells with A23187 and EDTA reduced [Mg²⁺]_i to very low levels, cells experience intermediate [Mg²⁺]_i for several minutes (Flatman, 1988) during which time the balance of kinase and phosphatase activity will be altered.

We have focused our analysis on the effects of phosphorylation of the cotransporter. However, phosphorylation of elements of the cytoskeleton may also be important in regulating cotransport rate, through protein- protein interactions (Haas & Forbush, 1998). For instance, inhibition of myosin light chain kinase has been shown to reduce Na⁺-K⁺-2Cl⁻ cotransport in other cell types independently of effects on cotransporter phosphorylation (Klein & O'Neill, 1995). K-252a, K-252b and staurosporine are all potent inhibitors of myosin light chain kinase so part of their effect may be due to inhibition of this enzyme. Changes in cAMP levels also affect cytoskeletal structure through PKA-mediated phosphorylation, and thus cAMP or inhibitors of PKA (e.g. K-252a and b) could indirectly affect the cotransport rate through this route (Matthews et al. 1994, 1998; D'Andrea et al. 1996).