Activation of ferret erythrocyte Na⁺–K⁺–2Cl⁻ cotransport by deoxygenation

Abstract

Deoxygenation of ferret erythrocytes stimulates Na⁺–K⁺–2Cl⁻ cotransport by 111% (s.d., 46) compared to controls in air. Half-maximal activation occurs at a P_O2 of 24 mmHg (s.d., 2) indicating that physiological changes in oxygen tension can influence cotransport function. Approximately 25–35% of this stimulation can be attributed to the rise of intracellular free magnesium concentration that occurs on deoxygenation (from 0.82 (s.d., 0.07) to 1.40 mm (s.d., 0.17)). Most of the stimulation is probably caused by activation of a kinase which can be prevented or reversed by treating cells with the kinase inhibitors PP1 or staurosporine, or by reducing cell magnesium content to submicromolar levels. Stimulation by deoxygenation is comparable with that caused by calyculin A or sodium arsenite, compounds that cause a 2- to 3-fold increase in threonine phosphorylation of the cotransporter which can be detected with phospho-specific antibodies. However, the same approach failed to detect significant changes in threonine phosphorylation following deoxygenation. The results suggest that deoxygenation causes activation of a kinase that either phosphorylates the transporter, but probably not on threonine, or phosphorylates another protein that in turn influences cotransporter behaviour. They also indicate that more than one kinase and phosphatase are involved in cotransporter phosphorylation.

Erythrocytes are exposed to a wide range of oxygen tensions as they pass through the circulation. It is clear that these changes in oxygen tension profoundly affect the behaviour of membrane transport systems, some requiring oxygen for activity whereas others operate faster in its absence (Gibson et al. 2000). Thus judging a transporter's contribution to erythrocyte ion and volume homeostasis on measurements made solely in air can lead to serious errors in assessing what happens in tissues at low oxygen tensions.

Work on avian erythrocytes first suggested that oxygen tension might affect Na⁺ and K⁺ transport (Ørskov, 1954; Tosteson & Robertson, 1956; McManus, 1967; Allen & McManus, 1968), and later work on fish erythrocytes firmly established the powerful effect oxygen tension has on some transport systems. Hypoxia increases the activity of the Na⁺–H⁺ antiporter in trout erythrocytes (Motais et al. 1987), an effect with important physiological consequences as the resulting cytoplasmic alkalinization increases the oxygen-carrying capacity of haemoglobin (Nikinmaa, 1997). In carp erythrocytes, β-adrenergic stimulation of Na⁺–H⁺ antiport only occurs at low oxygen tensions (Nikinmaa, 1992), whereas oxygenation is necessary for the activation of K⁺–Cl⁻ cotransport (Jensen, 1992, 1995). In turkey erythrocytes, stimulation of Na⁺–K⁺–2Cl⁻ cotransport by deoxygenation is prevented by treating cells with calyculin A with or without n-ethylmaleimide suggesting that the signal transduction pathway from oxygen sensor to transporter involves changes in protein phosphorylation (Muzyamba et al. 1999).

Recently, attention has turned to the effects of oxygen tension on transport in mammalian erythrocytes. Deoxygenation inhibits K⁺–Cl⁻ cotransport in equine erythrocytes and makes it unresponsive to changes in volume and intracellular pH (Honess et al. 1996), whereas in sheep erythrocytes deoxygenation also reduces K⁺–Cl⁻ cotransport rate but does not prevent the transporter from responding to other stimuli (Campbell & Gibson, 1998). In humans, K⁺–Cl⁻ cotransport is inactive and unresponsive to other stimuli under hypoxic conditions if the erythrocytes are from donors with normal haemoglobin (HbA), but it functions well at low oxygen tensions if the donors are homozygous for sickle trait (HbS)(Gibson et al. 1998; Joiner et al. 1998) in which case activation of the transporter can dehydrate the cell and initiate sickling. The effects of deoxygenation on Na⁺–K⁺–2Cl⁻ cotransport in mammalian erythrocytes have not been investigated.

The mechanisms by which oxygen affects transport in erythrocytes are only beginning to be understood. On a priori grounds, changes in oxygen tension might be expected to alter transport rate if either the oxyor deoxy-form of haemoglobin binds preferentially to the transporter. Alternatively, the change in transport may result (secondarily) from changes in the volume or composition (pH or ionized (free) intracellular [Mg²⁺] ([fMg²⁺]_i)) of the cytoplasm that occur when haemoglobin changes between its deoxy and oxy-conformations (Hladky & Rink, 1977; Flatman, 1980). Such changes affect a variety of transport systems. However, for the K⁺–Cl⁻ cotransporter, careful studies on sheep and equine erythrocytes (Campbell & Gibson, 1998; Campbell et al. 1999) have shown that although changes in cell volume, pH or [fMg²⁺]_i have some minor effect on transport they cannot explain the large changes in rate seen when oxygen tension is altered in these cells. Alternative explanations need to be found. Work on the effects of oxygen on K⁺–Cl cotransport in trout erythrocytes suggests that a haemoprotein distinct from bulk haemoglobin is the oxygen receptor (Berenbrink et al. 2000) and in human erythrocyte ghosts, the transporter responds to oxygen tension in pink but not white ghosts (Khan et al. 2000). Residual haemoglobin retained by pink ghosts is sufficient to endow the system with oxygen sensitivity.

The aim of the current study is 2-fold. Firstly, to examine whether Na⁺–K⁺–2Cl⁻ cotransport in mammalian erythrocytes responds to changes in oxygen tension, and if it does, to investigate the mechanism of action. This needs to be established in mammalian cells as the pattern of cotransporter phosphorylation differs in birds, fishes and mammals (Haas, 1994). Ferret erythrocytes were used for the study as these mammalian cells have a high Na⁺–K⁺–2Cl⁻ cotransport rate with over 95% of K⁺ influx occurring through the transporter (Flatman, 1983). Secondly, the study provides a test of current models that attempt to explain how phosphorylation of the cotransporter is regulated. In the simplest model, regulatory phosphorylation of the cotransporter is accomplished by a single kinase and phosphatase. The data presented here are not compatible with this idea and support the notion that several kinases and phosphatases are involved.

Methods

All solutions were prepared in double glass-distilled water with reagents of analytical quality (BDH AnalaR, VWR International, Lutterworth, UK). Other chemicals were obtained as follows: sodium (meta)-arsenite, Fluka Chemicals, Gillingham, UK; calyculin A, microcystin and 4-amino-5-(4-methylphenyl)-7-(t-butyl)pyrazolo[3,4-d]pyrimidine (PP1), Alexis Biochemicals, Calsbad, CA, USA; staurosporine, ethylenediaminetetraacetic acid (EDTA), ethyleneglycoltetraacetic acid (EGTA), N-2-hydoxyethyl-piperazine-N′-2-ethanesulphonic acid (Hepes), Triton X-100, sodium orthovanadate, Tween 20, dithiothreitol and tris(hydroxymethyl)aminomethane (Tris) base, Sigma, Poole, UK. A23187, ASB-16 (amidosulfobetaine detergent), CHAPS and the protease inhibitor cocktail were obtained from Calbiochem, VWR International.

The composition of ferret basic medium (FBM) was (mm): NaCl 145, KCl 5, EGTA 0.05, Hepes 10; pH adjusted to 7.5 with NaOH at 38°C. The lysis buffer contained (mm): Hepes 10, EDTA 2, sodium pyrophosphate 2, NaF 2, sodium β-glycerophosphate 1, sodium vanadate 1; pH 7.6 at room temperature. The immunoprecipitation buffer (IPB) contained (mm): NaCl 150, Hepes 20, EDTA 1, NaF 2, sodium β-glycerophosphate 1; pH 7.6 at room temperature. The Tris-buffered saline (TBS) contained (mm): NaCl 136, Tris base 20; pH adjusted to 7.6 with HCl at room temperature.

Blood was taken by cardiac puncture into EDTA-containing tubes from adult ferrets terminally anaesthetized with sodium pentobarbitone (Sagatal, Rhône Mérieux, Harlow, UK; 120 mg kg⁻¹ intraperitoneal) in accordance with advice from the British Home Office. Cells were washed four times by centrifugation and resuspension in FBM. Cells were stored, packed at about 40% haematocrit in FBM containing 22 mm glucose, at 4°C until used (within 2 days unless stated otherwise). All solutions used for handling blood were sterilized by passage through a 0.2 μm filter.

Flux measurements

Fluxes were determined at 38°C in well-stirred suspensions of ferret erythrocytes (5–10% haematocrit) in FBM containing 11 mm glucose. K⁺ influx was determined using ⁸⁶Rb⁺ as tracer as previously described (Flatman & Creanor, 1999a), and is expressed as rate constants (h⁻¹). Fluxes through the Na⁺–K⁺–2Cl⁻ cotransporter were defined as the components of ⁸⁶Rb⁺ uptake inhibited by 10 μm bumetanide. As none of the treatments described here caused a major change in the bumetanide-resistant ⁸⁶Rb⁺ fluxes (rate constants ranged between 0.1 and 0.3 h⁻¹ in the presence of bumetanide), the effects reported for total fluxes (rate constants between 1 and 10 h⁻¹) are almost entirely a result of changes in the Na⁺–K⁺–2Cl⁻ cotransport rate.

Cell suspensions were equilibrated at 38°C with the required gas mixtures in tonometers (Eschweiler, Kiel, Germany) for 30 min prior to flux measurements. Gases were humidified at 38°C before being passed into tonometers or flux incubation tubes. Deoxygenation was carried out with argon (BOC, Guildford, UK) or mixtures of argon and oxygen. Oxygen tensions were measured with a Clark-type oxygen electrode mounted in a temperature-controlled cell (Stathkelvin Instruments, Glasgow, UK). ⁸⁶Rb⁺ fluxes were measured in 1 ml aliquots of suspension transferred from tonometers to borosilicate glass test tubes which were being flushed with the appropriate gas mixtures. Gas flow up the long narrow tube prevented entry of air despite frequent sampling. Oxygen tensions in suspensions did not change during the flux period.

Haematocrit was determined as previously described (Flatman & Creanor, 1999a). [fMg²⁺]_i was reduced to levels below 1 μm by treating the cells with 10 μm A23187 (an ionophore that permeabilizes the membrane to Mg²⁺) and 2 mm EDTA for 15 min (Flatman, 1988). When required, [fMg²⁺]_i was clamped at values ranging from approximately 0.2–2 mm by treating cells, incubated in FBM containing 0.1–1 mm Mg²⁺, with 10 μm A23187 for 15 min (Flatman, 1988; Flatman & Lew, 1980). Total cell Mg²⁺ content ([Mg²⁺]_i) and the concentration of ionized Mg²⁺ in the medium at equilibrium in the presence of A23187 ([Mg²⁺]_o) were determined as previously described (Flatman & Lew, 1980). Total extracellular Mg²⁺ content at equilibrium was also determined by centrifuging a 0.4 ml sample of cell suspension at 14 000 g for 1 min and then carefully transferring 0.3 ml of the supernatant to a separate tube. [Mg²⁺] was measured by atomic absorption spectroscopy in samples diluted with distilled water. Cell contents are expressed as the amount found in a litre of cells at their original volume.

Preparation of membranes and immunoprecipitation

Following treatments as described, cells were centrifuged from solution, cooled on ice and then injected into 20 volumes of ice-cold lysis buffer containing 2% protease inhibitor cocktail, and stirred on ice for 10 min. The membranes were washed three times by centrifugation and resuspension in lysis buffer. The protein concentrations in the membranes were measured with Coomassie Plus protein assay reagent (Pierce Biotechnology, Rockford, IL, USA). After adding 2% protease inhibitor cocktail, the samples were frozen and stored in liquid nitrogen. Deoxygenated cells were lysed and their membranes washed in buffers that had been deoxygenated with argon.

The cotransporter was immunoprecipitated with T4, a monoclonal antibody to the human cotransporter (Lytle et al. 1995). This antibody recognizes the cotransporter in the SDS-denatured state so membranes were heated to 60°C in the presence of 1% SDS for 20 min prior to immunoprecipitation. After cooling, samples were diluted 5-fold with IPB and the following were added: 1 mm sodium vanadate, 2% protease inhibitor cocktail, 1 μm microcystin, 1% Triton X-100 and 0.5% ASB-16. The mixture was pre-cleared with protein G sepharose (Amersham Biosciences, Little Chalfont, UK) before T4 (Developmental Studies Hybridoma Bank, University of Iowa, IO, USA) was added and the tube incubated at 4°C for 60 min. The immune complex was precipitated with protein G sepharose and the pellet washed three times with IPB containing 1% Triton X-100 and 0.5% ASB-16. Cotransporter was eluted with sample buffer heated to 70°C for 20 min. Samples were run on NuPage 3–8% Tris acetate SDS-polyacrylamide gels (Invitrogen). Blots on nitrocellulose were blocked with TBS containing 0.1% Tween 20 and 5% bovine serum albumin (fraction V, ICN Biomedical, Costa Mesa, CA, USA) and were then incubated at 4°C overnight with anti-phosphothreonine antibodies (Cell Signaling Technology, Beverly, MA, USA) diluted 1 : 2000. After washing, blots were incubated with horseradish peroxidase-conjugated anti-rabbit antibodies (diluted 1 : 10,000, Cell Signaling). Antibody binding was quantified using ECL reagents and Hyperfilm (Amersham Biosciences). Blots were scanned and the images analysed with TotalLab (Non-Linear Dynamics, Newcastle, UK).

Data analysis

Experiments were repeated at least twice with blood from different ferrets, and either representative rate constants are given with the standard deviation from regression, or mean values are given with the standard deviation. The significance of difference between means was assessed with a two-tailed, unpaired t test. Curves were fitted to data according to specific models using non-linear regression analysis (Prism version 4.01, GraphPad Software, San Diego, CA, USA). Goodness of fit is indicated by R² values.

Results

Magnitude and time course of responses to changes in P_O2

Complete deoxygenation of ferret erythrocytes (P_O2 < 3 mmHg) for 30 min causes a large stimulation of ⁸⁶Rb⁺ uptake (Fig. 1A). In cells used within a day of collection, total ⁸⁶Rb⁺ influx rate constant increased by 111% (s.d., 46.2; n = 12 ferrets) compared to oxygenated controls. Rates as high as 9.2 h⁻¹ were observed compared with values of between 2.5 and 4 h⁻¹ seen in oxygenated cells (equilibrated with air). Increasing the oxygen tension to about 710 mmHg by equilibrating the cells with pure oxygen did not further reduce ⁸⁶Rb⁺ uptake compared to the rate seen in air (Fig. 2). The stimulation of ⁸⁶Rb⁺ uptake on deoxygenation was due to stimulation of Na⁺–K⁺–2Cl⁻ cotransport as only very small changes in ⁸⁶Rb⁺ influx were observed when cells were deoxygenated in the presence of 10 μm bumetanide, the potent inhibitor of Na⁺–K⁺–2Cl⁻ cotransport. Fluxes in bumetanide-treated cells (rate constants, approximately 0.1 h⁻¹) in fact fell slightly to 90% (s.d., 18; n = 5 ferrets) of the control value on deoxygenation.

Figure 1. Time course of changes in Na⁺–K⁺–2Cl⁻ cotransport rate following deoxygenation and re-oxygenation.

Ferret erythrocytes were incubated at about 6% haematocrit in FBM containing 11 mm glucose in tonometers at 38°C. In the experiment shown in A, the tonometer was flushed with humidified argon starting at time zero. Then, 5 min before the times indicated, 1 ml samples were taken with gas-tight Hamilton syringes and transferred to separate tubes for flux and P_O2 determination. These tubes were also continually flushed with argon. Samples were taken to measure haematocrit and then ⁸⁶Rb⁺ was added and influx measured over the next 3 min. B, the tonometer was flushed with humidified argon for 40 min before a sample was removed to measure ⁸⁶Rb⁺ influx (in argon) and P_O2. The gas was now changed to humidified air and the clock restarted. Samples were transferred to flux tubes flushed with air, 5 min before the times indicated, for the determination of ⁸⁶Rb⁺ influx rate. Rate constants are shown with standard deviations if these are larger than point size, and are plotted at the mid-timepoint of their determination. Lines are drawn by eye.

Figure 2. Relationship between oxygen tension and ⁸⁶Rb⁺ influx rate.

Ferret erythrocytes were incubated in FBM containing 11 mm glucose at 38°C in tonometers which were flushed with humidified argon–oxygen mixtures with the P_O2 indicated. After 30 min, 1 ml samples were taken for determination of P_O2 and ⁸⁶Rb⁺ influx. Flux incubation tubes were flushed with the same gas mixtures. Rate constants are shown with standard deviations if these are larger than point size. The line through data is the best fit to the Hill equation with half maximal activation occurring at a P_O2 of 25.9 mmHg and a Hill coefficient of −2.1 (R² = 0.999).

Stimulation of ⁸⁶Rb⁺ influx on changing to an oxygen-free atmosphere took about 20 min to develop fully (Fig. 1A), and having reached a peak fell slightly over the next 30 min. However, it was not possible to assess whether cotransporter activation occurs rapidly or after a lag following reduction of P_O2 to a low level. This is because it took about 15 min to remove sufficient oxygen from the suspension for its P_O2 to fall below 10 mmHg. It also took at least 3 min to measure the flux itself. On returning to an air atmosphere after 40 min incubation in pure argon, ⁸⁶Rb⁺ influx rate fell within 7 min to the value seen in controls maintained in air (Fig. 1B) and continued to fall over the next 50 min. Such reductions in rate are often seen when ferret erythrocytes are incubated over long periods, and is partly due to the gradual acidification of the medium. The rapid fall of transport rate on re-oxygenation (under these conditions it took about 5 min for the P_O2 in the suspension to rise above 25 mmHg) suggests that steps between the detection of a change in P_O2 and the alteration of transport rate are rapid and that the relatively slow rate of transport stimulation on deoxygenation reflects the slow rate at which oxygen is removed from the suspension.

Figure 2 shows the relationship between oxygen tension and the ⁸⁶Rb⁺ influx rate constant in a typical experiment. Suspensions were held at the indicated P_O2 for 30 min before the fluxes were measured. Reducing P_O2 from about 710 mmHg (pure oxygen) to about 80 mmHg had no detectable effect on ⁸⁶Rb⁺ uptake. Further reduction in P_O2 stimulated uptake, the level being 30% (s.d., 7; n = 3 ferrets) greater than control when the P_O2 was between 35 and 45 mmHg (similar to the values seen in mixed venous blood). The best fit to the data using the Hill equation indicated that ⁸⁶Rb⁺ influx was half maximally activated at a P_O2 of about 26 mmHg in the experiment shown in Fig. 2. In experiments using blood from three different ferrets the P_O2 giving half maximal activation was 24 mmHg (s.d., 2; Hill coefficient −2.3 (s.d., 0.4)). This compares with a P_O2 of 41 mmHg for half-maximal stimulation of cotransport in turkey erythrocytes (Muzyamba et al. 1999). K⁺–Cl⁻ cotransport is half-maximally activated at a P_O2 of about 30 mmHg in a variety of mammalian erythrocytes (Gibson et al. 2000).

Stimulation of cotransport was of a similar magnitude in fresh erythrocytes and those stored for 1 or 2 days at 4°C but then declined with longer storage. It was only 30% (s.d., 17; n = 3 ferrets) after 7 days and less than 2% (n = 2 ferrets) after 10 days.

Changes in intracellular ionized [Mg²⁺] induced by deoxygenation cannot explain the magnitude of cotransport stimulation

As Na⁺–K⁺–2Cl⁻ cotransport is sensitive to changes in [fMg²⁺]_i within the physiological range (Flatman, 1988) and [fMg²⁺]_i increases on deoxygenation, it is possible that the stimulation of cotransport observed on deoxygenation is simply due to the rise of [fMg²⁺]_i. To assess the contribution this rise in [fMg²⁺]_i makes to transport stimulation it is necessary to measure the change in [fMg²⁺]_i that occurs on deoxygenation, and to devise experiments that permit deoxygenation or oxygenation to occur without concomitant changes in [fMg²⁺]_i. Both can be achieved by treating the cells with A23187 in the presence of different external [Mg²⁺] ([Mg²⁺]_o) (Flatman, 1988). Previous experiments have shown that changes in [Mg²⁺]_o have little effect on cotransport, and that the effects seen in the presence of A23187 are due to Mg²⁺ acting at internal sites (Flatman, 1988). Figure 3A shows how oxygenated and deoxygenated ferret erythrocyte Mg²⁺ content increases with [Mg²⁺]_o in the presence of A23187. As described in the figure legend it is then possible to assess the [fMg²⁺]_i that normally occurs inside oxygenated and deoxygenated cells in the absence of A23187. Experiments using blood from three different ferrets indicate that [fMg²⁺]_i rises from 0.82 mm (s.d., 0.07) in oxygenated to 1.40 mm (s.d., 0.17) in deoxygenated cells.

Figure 3. Effect of changes of internal ionized magnesium concentration on ⁸⁶Rb⁺ influx.

Oxygenated and deoxygenated (30 min in argon) ferret erythrocytes were incubated at 38°C in FBM containing 11 mm glucose and 0.1–1.0 mm Mg²⁺ under an atmosphere of air or argon, respectively, and samples were taken to measure the initial Mg²⁺ content of the cells ([Mg²⁺]_T). A23187 (10 μm) was added and 12 min later, when Mg²⁺ had reached equilibrium across the membrane, samples were taken to measure cell and medium Mg²⁺ content ([Mg²⁺]_i and [Mg²⁺]_o). ⁸⁶Rb⁺ was now added to the suspension and ⁸⁶Rb⁺ uptake into the cells was determined. A, Mg²⁺ binding curves. The physiological values for [fMg²⁺]_i in oxygenated or deoxygenated cells can be estimated by finding the Mg²⁺ concentrations that must be present in the medium to prevent a change in [Mg²⁺]_i when A23187 is added to the suspension (null point method; Flatman, 1980, 1988). These values are most conveniently found as the points where the binding curves intersect with a line representing the Mg²⁺ content of cells in the absence of A23187 ([Mg²⁺]_T), a value that is not affected by deoxygenation over a 30-min period (oxygenated cells, 2.33 mmol (l cells)⁻¹ (s.d., 0.10); deoxygenated cells, 2.32 mmol (l cells)⁻¹ (s.d., 0.07), n = 3). These points occurred when [Mg²⁺]_o was 0.36 and 0.63 mm under oxygenated or deoxygenated conditions, respectively. In the presence of A23187, [fMg²⁺]_i is given by r²[Mg²⁺]_o (where r is [Cl]_o/[Cl]_I; r² is about 2.16 in these experiments; Flatman, 1988). This suggests that deoxygenation causes [fMg²⁺]_i to increase from 0.78 to 1.36 mm in these cells. B, effect of Mg²⁺ on ⁸⁶Rb⁺ influx. ⁸⁶Rb⁺ influx rate constants are plotted against [fMg²⁺]_i for oxygenated and deoxygenated erythrocytes in the presence of A23187. Vertical dashed lines represent the estimated [fMg²⁺]_i in the same cells in the absence of A23187. Point X represents uptake in cells with [fMg²⁺]_i clamped at the level found normally in oxygenated cells, and point Y represents cells with [fMg²⁺]_i clamped at the level normally found in deoxygenated cells. ⁸⁶Rb⁺ influx rate constants in oxygenated and deoxygenated cells incubated in the absence of A23187 were 3.80 (s.d., 0.22) and 7.44 h⁻¹ (s.d., 0.57), respectively, indicating that A23187 slightly inhibits transport as has been reported previously (Flatman, 1988). Points are given with standard deviations if larger than point size. Lines through the points were fitted to the equations: [Mg²⁺]_i =[Mg²⁺]_imax[Mg²⁺]_o/(k_0.5+[Mg²⁺]_o), and Rate = Rate_max[fMg²⁺]_i/(k_0.5+[fMg²⁺]_i). R² > 0.96 in all cases.

Figure 3B shows that if cells are deoxygenated whilst [fMg²⁺]_i is clamped at the level normally found in oxygenated cells, 0.8 mm, then the transport is still substantially stimulated from 3.7 to 5.8 h⁻¹. Likewise if cells are oxygenated whilst [fMg²⁺]_i is clamped at the level normally found in deoxygenated cells, 1.4 mm, transport rate still falls from 6.5 to 4.2 h⁻¹. Analysis of the data in Fig. 3B indicates that the rise of [fMg²⁺]_i that occurs on deoxygenation can only explain about 25% of the stimulation of cotransport. In a separate experiment, using blood from a different ferret, the rise in [fMg²⁺]_i accounted for 35% of the stimulation of ⁸⁶Rb⁺ uptake on deoxygenation.

Inhibition of protein kinases both prevents and reverses the deoxygenation-induced stimulation of ⁸⁶Rb⁺ uptake

The possible involvement of protein phosphorylation in the signal transduction pathway was investigated by treating erythrocytes with agents that interfere with either protein kinases or phosphatases. Figure 4 shows the results of an experiment where protein kinases in erythrocytes were inhibited either by treating the cells with 50 μm PP1 or 2 μm staurosporine, or by reducing [fMg²⁺]_i to submicromolar levels with A23187 and EDTA (Flatman & Creanor, 1999a). These treatments, made after the cotransporter had been stimulated by deoxygenation for 30 min, reduced ⁸⁶Rb⁺ uptake to levels below those seen in oxygenated controls despite the cells being maintained under an argon atmosphere. In addition, treatment of cells with PP1 or EDTA and A23187 for 15 min prevented stimulation of ⁸⁶Rb⁺ uptake by subsequent deoxygenation. Again the rates of ⁸⁶Rb⁺ uptake in these treated and then deoxygenated cells were lower than those seen in oxygenated controls. In separate experiments, 2 μm staurosporine was shown to prevent activation of ⁸⁶Rb⁺ influx by deoxygenation (rate constants (h⁻¹) were: control, 3.77 (s.d., 0.09); deoxygenated, 7.54 (s.d., 0.01); staurosporine, 1.74 (s.d., 0.04); staurosporine followed by deoxygenation, 1.82 (s.d., 0.17)). The data suggest that prevention of protein phosphorylation stops the cotransporter responding to deoxygenation.

Figure 4. Effect of protein kinase inhibition and deoxygenation on ⁸⁶Rb⁺ influx.

Cells were incubated at 38°C in FBM containing 11 mm glucose under the following conditions: no additions and incubated in air (Con); no additions but deoxygenated (Deox); deoxygenated and then treated with 50 μm PP1 in argon (Deox + PP1); deoxygenated and then treated with 2 μm staurosporine in argon (Deoxy + Sta); deoxygenated and then treated with 10 μm A23187 and 2 mm EDTA in argon (Deox + A/E); treated with 50 μm PP1 (in air) and then deoxygenated (PP1 + Deox); or treated with 10 μm A23187 and 2 mm EDTA (in air) and then deoxygenated (A/E + Deox). All deoxygenations were carried out in humidified argon for 30 min and the suspensions were treated with drugs for 15 min prior to flux determination. ⁸⁶Rb⁺ influxes were measured in air (Con, stippled bars) or argon (all others, filled bars). Bars represent the influx rate constants with standard deviations.

Comparison of the stimulation of ⁸⁶Rb⁺ uptake by deoxygenation with that caused by calyculin A or sodium arsenite

Deoxygenation stimulates Na⁺–K⁺–2Cl⁻ cotransport to a similar extent to calyculin A or sodium arsenite, agents that favour the phosphorylation and thus activation of the cotransporter (Flatman & Creanor, 1999a, b; Matskevich & Flatman, 2003). Comparison of the effects of deoxygenation with those of calyculin A or sodium arsenite may thus provide clues to how deoxygenation stimulates transport by affecting the activity of kinases and phosphatases. The response of cotransport to kinase inhibition (by PP1 or Mg²⁺ removal) after stimulation by deoxygenation, calyculin A or sodium arsenite alone, or in combination, is particularly revealing.

Initial evidence that deoxygenation affects different processes from calyculin A or sodium arsenite is the finding that combinations of deoxygenation with calyculin A or sodium arsenite produce greater stimulation of transport than these treatments alone. For instance in one set of experiments rate constants (h⁻¹) were: control, 3.47 (s.d., 0.13); deoxygenated, 6.49 (s.d., 0.5); calyculin A, 6.35 (s.d., 0.13); deoxygenation and calyculin A, 8.52 (s.d., 0.21); and in another: control, 3.28 (s.d., 0.11); deoxygenated, 6.41 (s.d., 0.17); sodium arsenite, 6.69 (s.d., 0.18); deoxygenation and sodium arsenite, 8.93 (s.d., 0.41).

Treatment of cells with 50 μm PP1, after cotransport had been stimulated with a combination of 20 nm calyculin A and deoxygenation, inhibits transport by 29% (s.d., 9; n = 3 ferrets) and an example of one such experiment is shown in Fig. 5. This inhibition of transport by PP1 is similar to the inhibition of calyculin A-stimulated cotransport by PP1 in oxygenated cells (Flatman & Creanor, 1999a). It is also worth noting that although PP1 inhibits cotransport in deoxygenated, calyculin A-treated cells, it does not reduce ⁸⁶Rb⁺ uptake back to levels seen in oxygenated controls; significant transport stimulation remains (Fig. 5).

Figure 5. Effect of calyculin A and PP1 on ⁸⁶Rb⁺ fluxes in oxygenated and deoxygenated cells.

Cells were incubated at 38°C in FBM containing 11 mm glucose. One group of cells was incubated in air (oxygenated, stippled bars), the other in humidified argon for 30 min (deoxygenated, filled bars). The following additions were then made before the measurement of ⁸⁶Rb⁺ influx: none (Control), 20 nm calyculin A for 10 min (Calyculin) or 20 nm calyculin A for 10 min followed by 50 μm PP1 for 15 min (Cal/PP1). Cells were maintained under air or argon during the drug treatments and flux measurements. Bars represent the influx rate constants with standard deviations.

PP1 completely reverses the stimulation of cotransport caused either by deoxygenation alone (see above) or by sodium arsenite alone (Flatman & Creanor, 1999b). Thus, if deoxygenation affects the same processes as sodium arsenite, PP1 should completely reverse the stimulation seen in deoxygenated, sodium arsenite-treated cells. However, as shown in Fig. 6, 50 μm PP1 inhibits this stimulation by only 13%. In experiments using cells from three different ferrets, PP1 inhibited ⁸⁶Rb⁺ uptake stimulated by this combination by 15.6% (s.d., 13). This suggests that deoxygenation or sodium arsenite have additional effects on transport apart from stimulating regulatory kinases. The simplest explanation is that one or other of these treatments also inhibits a regulatory phosphatase. The finding that following stimulation of transport by a combination of deoxygenation and sodium arsenite (9.38 h⁻¹ (s.d., 0.17)) addition of 50 μm PP1 after 10 min of reoxygenation causes transport to return almost to control levels (2.84 h⁻¹ (s.d., 0.15) compared to 1.77 h⁻¹ (s.d., 0.05)), rules out the possibility that sodium arsenite inhibits the phosphatase that returns transport to normal on re-oxygenation. Thus deoxygenation may also inhibit another of the cotransporter's putative regulatory phosphatases, the existence of which has been surmised from experiments carried out in the presence of sodium arsenite (Flatman, 2002).

Figure 6. PP1 does not inhibit the stimulation of cotransport evoked by the combined effects of deoxygenation and sodium arsenite treatment.

⁸⁶Rb⁺ influx was determined in fresh cells incubated at 38°C in FBM containing 11 mm glucose and the reagents indicated. Prior to flux measurement the cell suspensions were treated in the following ways: incubation in air with no additions (Con); 15 min incubation with 50 μm PP1 in air (PP1); or 1 h incubation with 1 mm sodium arsenite in air (Asn). Cell suspensions were also deoxygenated for 30 min in argon (Deox), or were treated with 1 mm sodium arsenite initially for 30 min in air followed by a further 30 min in argon (Asn/Deox). An aliquot of these deoxygenated, sodium arsenite-treated cells was then treated with 50 μm PP1 for 15 min under argon (Asn/Deox/PP1). Fluxes were determined in air (stippled bars) or under argon (filled bars). Bars represent the ⁸⁶Rb⁺ influx rate constant with standard deviations.

Does threonine phosphorylation of the cotransporter increase on deoxygenation?

The cotransporter is activated when key threonine residues are phosphorylated (Lytle & Forbush, 1992; Lytle, 1997; Kurihara et al. 1999; Darman & Forbush, 2002). Following isolation of the cotransporter from erythrocytes by immunoprecipitation, phosphorylation of these threonine residues can be measured using a phosphothreonine-specific antibody. Maximum activation of the cotransporter by treating ferret erythrocytes with calyculin A or sodium arsenite is associated with a 2- to 3-fold increase in phosphothreonine detected in cotransporter immunoprecipitated with the T4 antibody. Similarly a reduction in transport rate induced by PP1 is associated with a fall in phosphothreonine (Matskevich & Flatman, 2003). The same approach was used to see whether deoxygenation of erythrocytes causes a rise in phosphorylation of the cotransporter on threonine residues. As shown in Fig. 7, little if any change in phosphothreonine levels could be detected when cells were deoxygenated, whereas phosphothreonine levels changed as expected when samples of the same cells were treated with calyculin A, sodium arsenite or PP1. In an experiment (Fig. 7A) where great care was taken to ensure that the same amount of membrane protein was loaded onto gels by measuring the amount of the integral membrane protein, band 3, the amount of phosphothreonine in the cotransporter was 1.66 arbitrary units (s.d., 0.31) in oxygenated cells compared to 2.06 (s.d., 0.57) in deoxygenated cells. The cells were obtained from three separate suspensions under each condition. Although there is a slight increase (24%) in phosphothreonine, the means are not significantly different from each other (P = 0.46). This pattern was repeated with blood from several different ferrets – deoxygenation appeared to cause a slight increase in threonine phosphorylation but this was much less than the 2- to 3-fold increase produced by calyculin A or sodium arsenite (Fig. 7B).

Figure 7. Deoxygenation does not increase threonine phosphorylation of the immunoprecipitated cotransporter.

T4 immunoprecipitates were prepared from the membranes of ferret erythrocytes that had been incubated at 38°C in FBM containing 11 mm glucose and the following additions: none for 30 min (Ox, Ox1–3), 1 mm sodium arsenite for 60 min (Asn), 50 nm calyculin A for 15 min (Cal), deoxygenated under argon for 30 min (De, De1–3). Immunoprecipitates were run on 3–8% Tris-acetate gels, blotted and probed with an antibody to phosphothreonine. A, shows results of an experiment where membranes were prepared from six separate suspensions of erythrocytes (from the same ferret); three were incubated in air for 30 min and three were deoxygenated for 30 min in argon. B, shows results using blood from a different ferret where the levels of phosphothreonine detected in immunoprecipitates from deoxygenated cells are compared with those from cells treated with calyculin A or sodium arsenite.

Discussion

Deoxygenation of ferret erythrocytes results in a large stimulation of ⁸⁶Rb⁺ influx. This is due to stimulation of the Na⁺–K⁺–2Cl⁻ cotransporter as deoxygenation causes a slight fall in the already minimal ⁸⁶Rb⁺ uptake measured in the presence of the cotransporter inhibitor, bumetanide (10 μm). About 25–35% of this stimulation can be attributed to the increase in [fMg²⁺]_i that occurs on deoxygenation but the majority of the stimulation requires another explanation. Simple protein–protein interactions between oxyor deoxyhaemoglobin and the transporter cannot explain the deoxygenation-induced cotransporter stimulation. For instance, although the hypothesis that deoxyhaemoglobin binds to the cotransporter and activates it may explain the results in fresh cells it cannot explain the loss of stimulation in cells stored for long periods. The extent of Na⁺–K⁺–2Cl⁻ cotransporter stimulation is equivalent to that produced by treating erythrocytes with calyculin A or sodium arsenite, procedures which increase protein phosphorylation (Flatman & Creanor, 1999a, b; Matskevich & Flatman, 2003). Like stimulation by these agents, deoxygenation-induced stimulation declines the longer cells are stored before use, suggesting that labile compounds are involved in the signal transduction pathway. The findings that deoxygenation-induced stimulation is both prevented and reversed by protein kinase inhibitors, such as staurosporine and PP1, and by reduction of [fMg²⁺]_i, further indicate that phosphorylation of proteins is crucial in the process.

There is strong evidence that phosphorylation of the cotransporter itself on threonine and serine, but not tyrosine residues, plays a major role in regulating transport rate in response to a wide range of stimuli (Lytle & Forbush, 1992; Haas, 1994; Tanimura et al. 1995; O'Donnell et al. 1995; Lytle, 1997; Haas & Forbush, 1998; Flemmer et al. 2002; Matskevich & Flatman, 2003) and recent work shows that phosphorylation of three threonine residues in the N-terminus of the cotransporter is particularly important in this process (Kurihara et al. 1999; Darman & Forbush, 2002). There is also evidence that phosphorylation of cytoskeletal proteins may stimulate transport independently of changes in cotransporter phosphoryl ation (Matthews et al. 1994; Klein & O'Neill, 1995). However, it has been very difficult to identify the enzymes involved (Dowd & Forbush, 2003; Piechotta et al. 2003). Phosphopeptide maps of the cotransporter activated by different stimuli are very similar and this led to the attractive idea that phosphorylation at multiple sites may be carried out by a single kinase and dephosphorylation by a single phosphatase. A strong candidate for the phosphatase is protein phosphatase 1, for which a binding site has been identified close to the key threonine residues in the N-terminus of the cotransporter, and this would explain the stimulation of cotransport by calyculin A (Lytle & Forbush, 1992; Lytle, 1997; Darman et al. 2001). The nature of the kinase or kinases remains a mystery. On the other hand, analysis of the stimulation of Na⁺–K⁺–2Cl⁻ cotransport in ferret erythrocytes by calyculin A or sodium arsenite and of how this stimulation is affected by kinase inhibitors such as PP1, staurosporine or Mg²⁺ removal suggests that the cotransporter may be phosphorylated by several kinases and dephosphorylated by several phosphatases (Flatman, 2002).

Analysis of the effects of kinase and phosphatase inhibitors in combination with deoxygenation provides clues to how changes in oxygen tension affect cotransporter operation. Deoxygenation stimulates cotransport to the same extent as calyculin A or sodium arsenite, but this almost certainly involves a different pathway. For instance PP1, staurosporine or Mg²⁺ removal completely inhibit stimulation produced by deoxygenation (Fig. 4) but only reduces that produced by calyculin A by about 30% (Flatman & Creanor, 1999a) (Fig. 5). In addition, although these agents completely prevent and reverse stimulation by deoxygenation or sodium arsenite when applied alone, PP1 only reduces the stimulation produced by a combination of sodium arsenite and deoxygenation by 16% (Fig. 6). It is important to note that the observation that PP1, staurosporine or Mg²⁺ removal both prevent and reverse the stimulation caused by deoxygenation suggests that deoxygenation primarily activates a kinase that phosphorylates the cotransporter (or an associated protein) rather than inhibits a phosphatase (in which case transport would remain activated in the presence of these agents). The responsiveness to PP1 also implicates a tyrosine kinase from the Src, c-Kit or Bcr-Abl families in the process (Hanke et al. 1996; Tatton et al. 2003).

Initial results show that threonine phosphorylation of the cotransporter in ferret erythrocytes increases 2- to 3-fold when ferret erythrocytes are treated with calyculin A or sodium arsenite paralleling the effects of these agents on transport (Matskevich & Flatman, 2003). However, using the same techniques, it was not possible to demonstrate a significant change in threonine phosphorylation of the cotransporter following deoxygenation (Fig. 7). As both calyculin A and sodium arsenite cause phosphorylation of several threonine residues it is possible that deoxygenation causes the phosphorylation of a single key threonine residue, and that this cannot be resolved by the immunoprecipitation technique. However, as we can reliably resolve a 40% change in phosphorylation (Matskevich & Flatman, 2003), these data, together with the data on the effects of kinase and phosphatase inhibitors, suggest that if the cotransporter is phosphorylated on deoxygenation, then this is not on threonine, but on another residue, for instance serine. Alternatively, stimulation of cotransport by deoxygenation may involve phosphorylation of an accessory protein that activates transport in this situation. Whatever the explanation, the results reveal that activation of cotransport by deoxygenation involves a different kinase from that (or those) which phosphorylates and activates the transporter when cells are treated with calyculin A or sodium arsenite.

The observations reported here thus provide further evidence that several kinases and phosphatases are involved in regulatory phosphorylation of the cotransporter (Krarup et al. 1998; Flatman, 2002; Zhao et al. 2004). Not only is the kinase activated by deoxygenation distinct from that, or those, revealed when cells are treated with calyculin A or sodium arsenite but the phosphatase that returns transport to normal on re-oxygenation is also different. It is not protein phosphatase 1 or 2A as it is not inhibited by calyculin A (Fig. 5) nor protein phosphatase 2C as it is not inhibited by Mg²⁺ removal (Fig. 4).

Insight into how oxygen tension may affect transport is provided by studies of haemoglobin binding to band 3, the anion exchanger of which there are more than 10⁶ copies in the membranes of each erythrocyte. The cytoplasmic domain of band 3 has a higher affinity for deoxy than for oxyhaemoglobin (Chétrite & Cassoly, 1985) so deoxygenation leads to increased binding of haemoglobin to band 3, and as haemoglobin competes with several enzymes for the binding site, the release of these enzymes into the cytoplasm (Messana et al. 1996). Amongst these enzymes may be a kinase that phosphorylates nearby Na⁺–K⁺–2Cl⁻ cotransporters. It is of particular interest that the binding of both haemoglobin and enzymes is influenced by phosphorylation of band 3 tyrosine residues (Low et al. 1987) by members of the Src family of tyrosine kinases (Harrison et al. 1994; Brunati et al. 2000; Bordin et al. 2002). Tyrosine phosphorylation of band 3 increases on deoxygenation (Barbul et al. 1999), and this facilitates the release of enzymes (Low et al. 1987). This may explain how the tyrosine kinase (Src) inhibitor PP1 prevents and reverses stimulation of transport by deoxygenation. Alternatively changes in the phosphorylation state of band 3 may directly influence cotransport through protein–protein interactions (Guizouarn et al. 2004).