Oxidation induces a Cl⁻-dependent cation conductance in human red blood cells

Abstract

Oxidative stress induces complex alterations of membrane proteins in red blood cells (RBCs) eventually leading to haemolysis. To study changes of membrane ion permeability induced by oxidative stress, whole-cell patch-clamp recordings and haemolysis experiments were performed in control and oxidised human RBCs. Control RBCs exhibited a small cation-selective whole-cell conductance (236 ± 38 pS; n = 8) which was highly sensitive to the external Cl⁻ concentration: replacement of NaCl in the bath by sodium gluconate induced an increase of this cation conductance by about 85 %. Exposing RBCs to t-butylhydroxyperoxide (1 mm for 10 min) induced a twofold increase in this cation conductance which was further stimulated after replacement of extracellular Cl⁻ by gluconate, Br⁻, I⁻ or SCN⁻. In addition, lowering the ionic strength of the bath solution by isosmotic substitution of NaCl by sorbitol activated the cation conductance. The Cl⁻-sensitive and oxidation-induced cation conductance was Ca²⁺ permeable, exhibited a permselectivity of Cs⁺ > K⁺ > Na⁺ = Li⁺ >> NMDG⁺, and was partially inhibited by amiloride (1 mm) and almost completely inhibited by GdCl₃ (150 μm), but was insensitive to TEA, BaCl₂, NPPB, flufenamic acid or quinidine. DIDS (100 μm) reversibly inhibited the activation of the cation conductance by removal of external Cl⁻. Oxidation induced haemolysis in NaCl-bathed human RBCs. This haemolysis was attenuated by amiloride (1 mm) and inhibited by replacement of bath Na⁺ by the impermeant cation NMDG⁺. The Na⁺- and Ca²⁺-permeable conductance might be involved in haemolytic diseases induced by elevated oxidative stress, such as glucose-6-phosphate dehydrogenase deficiency.

The permeability properties of the red blood cell (RBC) membrane govern its acid-base status and directly control the transport of carbon dioxide through the blood. In addition to this respiratory function, several transport pathways are involved in the regulation of cell volume and intracellular ion homeostasis, but little is known about the ion channels of human RBCs.

By using single channel patch-clamp techniques, different types of cation channels have been described: the Gardos K⁺ channel, which is a Ca²⁺-activated IK channel, is thought to be involved in the regulatory volume decrease (RVD) of RBCs (Grygorczyk & Schwarz, 1983; Bennekou & Christophersen, 1990; Christophersen, 1991; Pellegrino et al. 1998). The activity of a non-selective cation (NSC) channel has also been described, but the physiological relevance of this channel is not known (Christophersen & Bennekou, 1991; Bennekou, 1993; Kaestner et al. 1999; Kaestner et al. 2000). Recently, Huber et al. (2001), using the whole-cell patch-clamp configuration, described a non-selective cation conductance regulated by cell volume and by internal Cl⁻ concentration. In addition, human RBCs infected with Plasmodium falciparum showed increased permeabilities to mono- or divalent cations (Kirk & Horner, 1995) and Desai et al. (1996) have reported a Ca²⁺-permeable cation channel in these cells.

In many diseases such as glucose-6-phosphate dehydrogenase deficiency (Mavelli et al. 1984; Turrini et al. 1985), human RBCs have to deal with elevated oxidative stress. In vitro oxidation of fresh RBCs has been shown to induce a complete change in the electrophoresis pattern, especially that of membrane proteins (Koster & Slee, 1983; Ingrosso et al. 2000).

In order to study the involvement of ion conductances in oxidation-induced haemolysis, we performed whole-cell recordings in human RBCs exposed to elevated oxidative stress.

METHODS

Preparation of human erythrocytes and oxidative treatment

For control whole-cell experiments, fresh erythrocytes from healthy donors (donors gave informed consent and procedures were performed according to the Declaration of Helsinki and with local ethical committee approval) were diluted (blood dilution ∼1/1000) in NaCl bath solution used for the patch-clamp experiments (see below). Small aliquots of this cell suspension were transferred directly to the experimental chamber of the patch-clamp set-up. All experiments were performed at room temperature.

Most of the experiments were performed using pre-treated erythrocytes, prepared by incubating fresh erythrocytes for 10 min at 37 °C in the NaCl bath solution containing 1 mmt-butylhydroxyperoxide (t-BHP) as the oxidising species. At the end of the oxidative stress, cells were centrifuged (2 min), washed, resuspended in fresh NaCl medium and finally transferred to the experimental chamber.

Patch-clamp experiments

Patch pipettes made of borosilicate glass (150 TF-10, Clark Medical Instruments, UK) were pulled in three steps using a horizontal DMZ puller (Zeitz, Germany). Pipettes, with a high resistance of 8–12 MΩ were connected via an Ag-AgCl wire to the headstage of an EPC 9 patch-clamp amplifier (HEKA, Germany). Data acquisition and data analysis were controlled by a computer equipped with an ITC 16 interface (Instrutech, USA), using Pulse software (HEKA, Germany). The procedure used for whole-cell configuration has been previously described by Huber et al. (2001). Briefly, seals were achieved by applying slight suction to the patch pipette. After formation of the gigaseal, the membrane was ruptured by additional suction or a small voltage depolarisation to achieve the conventional whole-cell configuration. For current measurements, cells were held at a holding potential (V_h) of −10 mV and 400 ms pulses from −100 to +100 mV were applied in increments of +20 mV. Whole-cell experiments were performed at room temperature with morphologically intact RBCs.

Data analysis

The original whole-cell current traces are depicted without filtering (the acquisition frequency was 5 kHz). The currents were analysed by averaging the current values measured between 350 and 375 ms of each square pulse. The conductance was estimated by linear regression for outward currents from +40 to +100 mV. The applied voltages refer to the cytoplasmic face of the membrane with respect to the extracellular space. Inward currents were defined as flow of positive charges from the extracellular to the cytoplasmic membrane face and the opposite movement characterised the outward currents. Data are expressed as means ± s.e.m. of n experiments.

Liquid junction potentials

The offset potentials between both electrodes were zeroed before sealing. The liquid junction potentials, ΔE = E_S-E_P, between the electrode bridge (filled with NaCl or NMDG-Cl bath solution), the pipette and the different bath solutions (S) were estimated according to the equation:

where

and where μ, α and z represent the mobility, activity and valence of each ion species, respectively. Relative μ values of μ_K = 1, μ_Na = 0.682, μ_Li = 0.525, μ_Cs = 1.05, μ_NMDG = 0.500, μ_Ca = 0.404, μ_Cl = 1.0388 and μ_gluconate = 0.33 were assumed. I-V relationships and reversal potential calculations were corrected for the estimated ΔE values.

Solutions and chemicals

The standard bath NaCl solution used for suspending the cells or for control whole-cell recording contained (mm): 115 NaCl, 10 MgCl₂, 5 CaCl₂, 10 Hepes (pH adjusted to 7.4 with 1 m NaOH). The high concentration of Mg²⁺ and Ca²⁺ increased the probability of obtaining high resistance seals. The standard pipette solution contained (mm): 120 NaCl, 5 Hepes, 1 EGTA, 1 Mg-ATP (pH adjusted to 7.2 with 1 m NaOH). Some experiments were performed using 120 mmN-methyl-d-glucamine chloride (NMDG-Cl) instead of NaCl; the concentrations of the other components were not modified.

Cation selectivity experiments were investigated by replacing the standard NaCl bath solution by the different solutions described in Table 1.

Table 1.

Composition of the bathing solutions

Bath solutions	Cation source (mm)	Acid/base	Hepes (mm)	CaCl2 (mm)	MgSO4 (mm)
Na-gluconate	140 Na-gluconate	pH adjusted with NaOH	5	1	1
K-gluconate	140 KOH	Titrated by gluconic acid (3N)	5	1	1
Cs-gluconate	140 CsOH	Titrated by gluconic acid (3N)	5	1	1
Li-gluconate	140 LiOH	Titrated by gluconic acid (3N)	5	1	1
NMDG-gluconate	140 NMDG	Titrated by gluconic acid (3N)	5	1	1
NMDG-Cl	140 NMDG	Titrated by HCl (6N)	5	1	1
NaNO3	140 NaNO3	pH adjusted with NaOH	5	1	1
NaSCN	140 NaSCN	pH adjusted with NaOH	5	1	1
Ca-(gluconate)2	90 Ca-(gluconate)2	pH adjusted with Ca(OH)2	5	1	0
CaCl2	95 CaCl2	pH adjusted with Ca(OH)2	5	0	0

Pharmacological properties were established by the use of different inhibitors; amiloride, 5-nitro-2-(3-phenylpropyl amino) benzoic acid (NPPB) and flufenamic acid were purchased from Sigma and prepared in DMSO as stock solutions with the concentrations 250, 100 and 100 mm, respectively. 4,4′-diisocyanato-stilbene-2,2′-disulfonic acid (DIDS, Sigma) and quinidine hydrochloride (Sigma) were added directly to the bath solution at final concentrations of 100 and 500 μm, respectively. GdCl₃ (Sigma) was prepared as a stock solution of 300 mm in water. DTT was prepared as a 1 m stock solution in distilled water and used at a final concentration of 1 mm.

Haemolysis experiments

Erythrocytes of volunteers were washed and resuspended in Ringer NaCl solution (mm: 140 NaCl, 10 Hepes/NaOH pH 7.4, 5 KCl, 2.5 CaCl₂, 1 MgCl₂) to a haematocrit of 5 % and stored at 8 °C for 1–7 days. Aliquots of the erythrocyte suspension (500 μl) were oxidised by adding 1 ml of NaCl solution containing t-BHP (0, 0.1, 0.3, 1 or 3 mm final concentration) and centrifuged (15 min total incubation time). Additional experiments were carried out with different incubation times (5–25 min). Pelleted control untreated or oxidised cells were then resuspended in 400 μl of isosmotic Ringer NaCl, sodium gluconate, NaNO₃, Na-SCN, CaCl₂ or NMDG-Cl solution (see Table 1 for ionic composition) at 37 °C for 3–5 h in the absence or presence of amiloride (1 mm). Incubation was stopped by centrifugation and the haemoglobin concentration of the supernatant (200 μl) was determined qualitatively by image scanning and quantitatively by photometry (absorbance at 546 nm after oxidation to cyanomet-haemoglobin).

RESULTS

Whole-cell recordings in control cells

Patch-clamp studies were performed with NaCl solution in the pipette, whilst the bath contained NaCl solution with high MgCl₂ and CaCl₂ concentrations in order to obtain high resistance seals. Under these conditions, the spontaneous conductance of the human RBCs in whole-cell configuration was very low (Fig. 1A, left panel). The steady state currents analysed between 350 and 375 ms of each voltage square pulse showed an outwardly rectifying current-voltage (I-V) relationship with a mean slope conductance of 236 ± 38 pS (from +40 to +100 mV, n = 8) and a mean reversal potential (V_rev) of 2.3 ± 1.3 mV (Fig. 1B, •). Replacing NaCl bath solution with Na-gluconate induced a large increase in both the inward and outward currents (Fig. 1A, right panel) within 30 s. The mean slope conductance reached 442 ± 89 pS (n = 6) and the I-V curve became more linear (Fig. 1B, ▪). The removal of external Cl⁻ did not greatly affect V_rev (ΔV_rev = +5.7 ± 1.8 mV), suggesting that the induced whole-cell current was carried by Na⁺.

Figure 1. Activation of whole-cell currents in control human red blood cells (RBCs) by external Cl⁻ removal.

A, original current traces recorded with NaCl solution in the pipette and NaCl solution in the bath (left panel) or after replacing NaCl in the bath with Na-gluconate (right panel). Membrane voltage was held at −10 mV and stepped to test potential values between −100 and +100 mV, in +20 mV increments. B, average current-voltage (I-V) relationships measured between 350 and 375 ms after the onset pulse. Mean I-V relationships (±s.e.m.) recorded as in A with NaCl pipette solution and NaCl- (•, n = 8) or Na-gluconate (▪, n = 6) bath solution.

In another series of experiments using NMDG-Cl pipette solution (NaCl standard bath solution; Fig. 2A, left panel) the steady state currents exhibited an inward rectification (Fig. 2B, •) with a V_rev of +58.0 ± 5.7 mV (n = 4). Replacing the NaCl in the bath solution with Na-gluconate (Fig. 2A, middle panel) induced an increase in mean slope conductance from 207 ± 28 pS (NaCl) to 320 ± 49 pS (Na-gluconate; from −100 to −40 mV; n = 4) without a large change in V_rev (Fig. 2B, ▪). Replacing the NaCl bath with NMDG-Cl solution induced a strong decrease in inward current (Fig. 2A, right panel) and a shift of V_rev towards 0 mV (n = 3; Fig. 2B, ▴). Taken together, these experiments suggested that human RBCs express a constitutively active Na⁺-permeable cation conductance. In addition, replacement of external Cl⁻ by gluconate induced activation of a Na⁺-permeable cation conductance.

Figure 2. Na⁺ selectivity of the Cl⁻ dependent currents in control RBCs.

A, original current traces recorded from an RBC, using NMDG-Cl in the pipette and NaCl (left panel), Na-gluconate (middle panel) and NMDG-Cl (right panel) bath solutions. Dotted line indicates zero current line. B, mean I-V relationships (±s.e.m.) recorded as in A with NMDG-Cl pipette solution and NaCl- (•, n = 6); Na-gluconate- (▪, n = 6) and NMDG-Cl-bath solution (▴, n = 3).

Whole-cell recordings in oxidised RBCs

Oxidised RBCs exhibited similar rectifying but larger whole-cell currents than non-oxidised controls when recorded with NaCl pipette and bath solutions (Fig. 3A and B, ○). Outward conductance was increased ∼twofold in oxidised compared with control RBCs (Fig. 3C).

Figure 3. Activation of the Cl⁻ dependent cation current by oxidation.

RBCs were exposed for 10 min to 1 mmt-BHP, centrifuged, washed, and resuspended in NaCl bath solution. A, original current traces from an oxidised RBC (NaCl pipette solution) recorded in NaCl-(far left panel), in Na-gluconate-(middle left panel), in NaCl-(middle right panel) or NMDG-Cl bath solutions (far right panel). B, corresponding mean I-V curves (±s.e.m.) measured in NaCl (○, n = 22), Na-gluconate (□, n = 14), and NMDG-Cl (▵, n = 5) bath solutions. C, comparison of the mean conductance (±s.e.m.) calculated for the outward current (+40 to +100 mV) in control (black bars) and oxidised RBCs (open bars) bathed in NaCl or Na-gluconate solution. (**P < 0.005; t test). D, time dependence of the induction of whole-cell current by oxidation: original current traces (left panel) and mean percentage of current increase (±s.e.m., n = 4, right panel) continuously recorded during oxidation with t-BHP (1 mm) starting at 0 min. E, time dependence of the decrease of the oxidation-activated cation conductance continuously recorded during an acute treatment with the reducing agent DTT (1 mm): original current traces (left panel) and percentage of current decrease (2 individual experiments, right panel). The recorded cell was pre-oxidised with t-BHP (1 mm) for 10 min. Records in D and E were obtained with NaCl pipette and bath solution; the records were analysed at +100 mV for the current-time plots.

To characterise the oxidation-induced conductance, NaCl in the bath solution was replaced by Na-gluconate. As in the non-oxidised controls, removal of Cl⁻ induced large increases in inward and outward currents (Fig. 3A) but only a minor shift of V_rev from 3.8 ± 0.9 to 6.9 ± 0.8 mV (n = 22 and 14, respectively; Fig. 3B, ○ and □). Re-exposing the cells to NaCl rapidly reversed the increase in whole-cell current induced by Cl⁻ removal (Fig. 3A). Replacing bath NaCl with NMDG-Cl decreased inward current and shifted V_rev to negative potentials (−42.9 ± 6.8 mV, n = 4; Fig. 3A, far right and Fig. 3B, ▵). These data indicate activation of the Cl⁻-dependent Na⁺-permeable conductance by oxidative stress as summarised in Fig. 3C. To directly demonstrate the oxidation-induced current activation, cells (n = 4) were continuously recorded (NaCl bath and pipette solutions) during bath application of t-BHP (1 mm). During acute application of t-BHP, whole-cell currents started to increase within 5 min of oxidation and equilibrated within 10 min (Fig. 3D, right panel). The induced current resembled that of the pre-oxidised cells in outward rectification and conductance (Fig. 3D, original tracing at 10 min). To test whether the oxidation induced current activation could be reversed by reduction, pre-oxidized RBCs (1 mmt-BHP, 10 min) were recorded during bath incubation with the reducing agent: DTT (1 mm; Fig. 3E, left panel). During acute application of DTT whole-cell currents decreased rapidly within 2 min of exposure and equilibrated within 8 min (Fig. 3E, right panel).

To confirm that the conductance is activated by Cl⁻ removal and not by addition of gluconate, bath NaCl was replaced by NaNO₃, NaBr or NaSCN. Substitution with NaNO₃ induced an increase of 123 ± 18 % and 118 ± 23 % in inward and outward conductances, respectively (n = 4). Similar to this manoeuvre, replacing the NaCl bath solution by NaBr or NaSCN (n = 3) increased inward and outward conductances by 112 ± 21 % and 114 ± 17 % (NaBr; n = 3) or by 127 ± 21 % and 124 ± 22 % (NaSCN; n = 3), respectively. In addition, reduction of the NaCl concentration in the bath by dilution with isosmotic sorbitol solution (25 % NaCl and 75 % sorbitol) evoked an increase in outward conductance of 102 ± 19 % and a shift of V_rev to a negative potential (−22.4 ± 4.0 mV, n = 3).

To more closely characterise the Cl⁻ dependence of this conductance, Cl⁻ in the bath was increasingly replaced by gluconate. With the decrease of the bath Cl⁻ concentration (from 145 mm to 85, 45, 25, and 5 mm; n = 5–7) whole-cell current (at +100 mV) increased with a half-maximal effective Cl⁻ concentration of 27 mm (Fig. 4).

Figure 4.

Relationship between induced current measured at + 100 mV (percentage increase, means ±s.e.m.; n = 5–7) and the external Cl⁻ concentration. Currents were recorded with NaCl pipette solution and NaCl/Na-gluconate bath solutispan

Cation selectivity

To identify a possible Ca²⁺-permeability of the oxidation-induced, Cl⁻-dependent conductance, whole-cell currents of oxidised RBCs recorded in Ca-gluconate bath solution (NaCl pipette solution; Fig. 5A) were compared with those in recorded Na-gluconate and NMDG-gluconate (Fig. 5A). The I-V relation in Ca-gluconate (Fig. 5B, ⋄) showed lower inward current (approx. −25 pA versus −50 pA at −100 mV) and higher outward rectification when compared to that recorded in Na-gluconate (Fig. 5B, ▪). The V_rev in Ca-gluconate (3.8 ± 3.1 mV; n = 5) did not differ from that in Na-gluconate but shifted to −41.6 ± 3.6 mV (n = 11) when applying NMDG-gluconate bath solution (Fig. 5B, ▵) suggesting Ca²⁺ permeability for the oxidation-induced, Cl⁻-dependent cation conductance.

Figure 5. Ca²⁺ permeability of the induced cation conductance.

A, whole-cell current traces of an oxidised RBC recorded using NaCl pipette solution combined with calcium gluconate-(left panel), NMDG-gluconate-(middle panel), and Na-gluconate-(right panel) bath solutions. Dotted line indicates zero current. B, corresponding mean I-V relationships (±s.e.m.) obtained with Ca-gluconate- (⋄, n = 5), NMDG-gluconate- (▵, n = 10) or with Na-gluconate- (▪, n = 14, already shown in Fig. 3B) bath solutions.

The permselectivity for monovalent cations was determined with NaCl in the pipette by the shift in V_rev following replacement of bath Na-gluconate by the gluconate salts of caesium, potassium, lithium and NMDG (Fig. 6A). Under these bi-ionic conditions, the relative permeability was in the sequence of Cs⁺ > K⁺ > Na⁺ = Li⁺ >> NMDG⁺ (Fig. 6B). The conductance sequence for the inward current carried by the applied cations was Na⁺ ≥ K⁺ ≥ Cs⁺ ≥ Li⁺ >NMDG⁺ (Fig. 6C). In conclusion, this cation conductance did not discriminate between monovalent cations, indicating non-selective cation selectivity.

Figure 6. Permselectivity of the induced cation conductance.

A, whole-cell current traces of an oxidised RBC in Na-gluconate- (left panel), Cs-gluconate-(middle panel), and K-gluconate-(right panel) bath solutions. The pipette was filled with NaCl solution. B, mean shift in reversal potential (±s.e.m.) induced by replacement of bath Na-gluconate by different cations (gluconate salt) as indicated (n = number of experiments). C, corresponding mean inward conductance (calculated from −100 to −40 mV) as recorded in B.

Effect of inhibitors

All experiments with inhibitors were performed using NaCl pipette solution. Addition of NPPB (100 μm), TEA (1 mm) or quinidine chloride (500 μm) to Na-gluconate solution failed to modify the amplitude of the oxidation-activated Cl⁻-dependent non-selective cation conductance (n = 3–5; not shown). In contrast to these inhibitors, amiloride (1 mm) reversibly blocked 40 ± 4 % of the inward current with only a slight effect on outward current when added to the bath solution (n = 3, Fig. 7A). Bath application of GdCl₃ (150 μm) decreased inward and outward currents (by 59 ± 4 % and 57 ± 6 %, respectively; n = 4; Fig. 7B). In contrast to that of amiloride, the GdCl₃ effect was not reversible by washing out.

Figure 7. Inhibitors of the oxidation-induced non-selective cation conductance.

A and B, original current traces recorded in an oxidised RBC with Na-gluconate-bath solution (NaCl pipette solution) before (left panel), during (middle panel) and after (right panel) application of amiloride (1 mm; A) and GdCl₃ (150 μm; B) in the bath. C, original current traces recorded from an oxidised RBC (NaCl pipette solution) in the presence (left and middle panels) and absence (right panel) of DIDS (100 μm). Currents were obtained in NaCl- (left) and Na-gluconate-(middle and right) bath solutions, respectively. DIDS reversibly inhibited the current activation in Na-gluconate-bath solution. Application of DIDS (100 μm) after current activation in the Na-gluconate bath had no effect (not shown). D, paired increase in whole-cell current induced by Cl⁻ removal during application (black bars) and after wash-out of DIDS (100 μm; open bars). The increases in current at −100 and +100 mV are depicted as percentages of the initial current recorded in the NaCl bath in the presence of DIDS (means ±s.e.m.; n = 6).

DIDS (100 μm), while having no effect on the activated current (not shown), prevented current activation in oxidised cells by Cl⁻ removal when pre-incubated (for 2 min) before replacing bath NaCl/DIDS with Na-gluconate/DIDS (Fig. 7C, left and middle panels). DIDS removal by washing out with Na-gluconate led to current activation indistinguishable from the current activation under control conditions (Fig. 7C, middle and right panels) indicating reversibility of the DIDS blockade. The DIDS effect on activation of inward and outward currents is summarised in Fig. 7D.

Haemolysis experiments

To study the physiological relevance of the oxidation-induced cation conductance, control and oxidised RBCs were bathed in different salt solutions and haemolysis was determined photometrically. Some 6 % of the oxidised RBCs haemolysed in NaCl solution after 3–4 h of incubation whereas almost no haemolysis was apparent in the non-oxidised controls (Fig. 8A). This oxidation-induced haemolysis in NaCl was mimicked in CaCl₂ but attenuated and prevented by replacing either Na⁺/Ca²⁺ or Cl⁻ by the impermeant ions NMDG and gluconate, respectively (Fig. 8A and B) indicating that haemolysis resulted from net uptake of NaCl/CaCl₂ and water. The induced haemolysis in NaCl and CaCl₂ was inhibited by amiloride (Fig. 8A and C). Moreover, replacement of NaCl by NaNO₃ or NaSCN markedly increased the oxidation-induced haemolysis (Fig. 8D).

Figure 8. Spontaneous and oxidation-induced haemolysis of RBCs in isosmotic salt solutions.

A, imaged supernatants of untreated (non-oxidised) and oxidised erythrocytes (1 mmt-BHP, 15 min) after incubation in NaCl, Na-gluconate, Ca-gluconate or NMDG-Cl solutions in the absence (oxidised) or presence (oxidised amiloride) of amiloride (1 mm; results of an individual experiment). First column shows the haemoglobin of an increasing number of cells (percentage) haemolysed in water. Increasing red values indicate increasing haemoglobin concentration in the supernatant. B, substrate dependence of the mean oxidation-induced haemolysis as determined in A. C, amiloride (1 mm) sensitivity of the oxidation-induced haemolysis in NaCl and CaCl₂. D, anion dependence of the oxidation-induced haemolysis. Oxidised and control erythrocytes were incubated in NaCl, NaNO₃ or NaSCN solutions and percentages of haemolysed cells were determined as in A. Data from B-D are means ±s.e.m.; n = 6–22 experiments. *P≤0.05, **P≤ 0.01 and ***P≤ 0.001; two-tailed t test.

Figure 9 shows the concentration and time dependence of the oxidation-induced haemolysis. Starting with a concentration of 1 mm and an incubation period of 15 min, t-BHP induces haemolysis in NaCl. Higher concentrations of oxidants (Fig. 9A) or extension of the incubation time induced strong haemolysis in NMDG-Cl, suggesting formation of non-specific holes in the membrane.

Figure 9. Concentration and time dependence of the oxidation-induced haemolysis.

A, imaged supernatants of erythrocytes incubated in NaCl or NMDG-Cl solution after oxidation by increasing concentrations of t-BHP (0.1, 0.3, 1, 3 mm; left panel). Increasing red values indicate increasing haemoglobin concentration in the supernatant. Last row shows the haemoglobin of an increasing number of cells (percentage) haemolysed in water. Dose-response curve of the mean (±s.e.m.; n = 6–9) oxidation-induced haemolysis (right panel). B, time-dependence of the oxidation induced haemolysis. Imaged supernatants of erythrocytes incubated in NaCl after increasing periods of oxidation (5, 10, 15, 20, 25 min) with t-BHP (1 mm, left panel). Mean (±s.e.m.; n = 6) time-dependent appearance of the oxidation-induced haemolysis (right panel).

Taken together, these haemolysis data suggest an oxidation-induced amiloride-sensitive Na⁺ and Ca²⁺ permeability. This permeability is further induced by replacement of Cl⁻ by permeant anions such as NO₃⁻ or SCN⁻.

DISCUSSION

The present study determined the membrane conductance of human erythrocytes by patch-clamp whole-cell recording. Unstimulated RBCs exhibited low whole-cell membrane currents under the chosen experimental conditions which were mainly carried by cations. Oxidation by t-BHP induced an increase in this conductance which was further activated by removal of Cl⁻ from the bath solution. This oxidation-induced and Cl⁻-dependent conductance was amiloride- and Gd³⁺-sensitive and permeable for Ca²⁺ and alkali cations with a permselectivity of Cs⁺ > K⁺ > Na⁺ ≥ Li⁺. None of the anions tested (gluconate⁻, Br⁻, SCN⁻ and NO₃⁻) mimicked the observed inhibitory effect of extracellular Cl⁻.

Most probably the inhibitory effect was mediated by direct interaction of Cl⁻ with a Cl⁻ sensor site at the channel or an associated regulator protein. This assumption is supported by the observed instantaneous onset of current increase upon decrease of external Cl⁻ concentration. Inhibition of amiloride-sensitive epithelial Na⁺ channels by high cytosolic concentrations of Cl⁻ has been demonstrated to be mediated via a G protein (Dinudom et al. 1995). A dependence on cytosolic Cl⁻ concentration has also been reported for the NSC conductance of human RBCs (Huber et al. 2001) suggesting that the NSC conductance in human RBCs is regulated by both, extracellular and intracellular Cl⁻.

Several tracer flux studies observed a basal permeability of Na⁺ and K⁺ in human RBCs (e.g. LaCelle & Rothsteto, 1966). In addition, single channel recordings identified in human RBCs a voltage-activated non-selective cation channel which is permeable to Ca²⁺ (Christophersen & Bennekou, 1991; Bennekou, 1993; Kaestner et al. 1999; Kaestner et al. 2000). This channel type has been suggested to be involved in the generation of the basal permeability (Kaestner et al. 1999; Kaestner et al. 2000).

Incubation of human RBCs in low ionic strength (LIS) media induces a transport pathway for the monovalent cations Na⁺ and K⁺ (LaCelle & Rothsteto, 1966; Jones & Knauf, 1985; Bernhardt et al. 1991), as well as for different amino acids such as taurine, glutamine, etc. (Culliford et al. 1995). The LIS-induced permeabilities have been attributed to an electroneutral K⁺(Na⁺)-H⁺ exchange (Richter et al. 1997; Kummerow et al. 2000). The LIS-induced K⁺ efflux (Bernhardt et al. 1991) and the cation conductance of the present study exhibit similar dependence on extracellular NaCl -Cl⁻ concentration. In addition, the LIS-induced permeabilities are (partially) blocked by DIDS (Jones & Knauf, 1985; Culliford et al. 1995). In the present study, DIDS blocked the activation of the cation conductance induced by Cl⁻ removal (but not the previously activated current). These similarities suggest that the cation conductance shown in the present paper might generate in vivo Na⁺ and K⁺ fluxes under LIS conditions and therefore might contribute together with the Na⁺(K⁺)-H⁺ exchanger to the reported LIS effect as recently concluded for the non-selective cation channels of human RBCs (Kaestner et al. 1999), which most probably underlies the macroscopic current of the present study.

Huber et al. (2001) reported an amiloride-sensitive non-selective cation conductance in RBC ghosts that was activated by cell shrinkage, suggesting involvement of the cation conductance in cell volume regulation. In addition, the observed Ca²⁺ permeability of the cation conductance of the present study and the reported divalent cation permeability of the non-selective channels (Kaestner et al. 2000) suggests a contribution to Ca²⁺ signalling.

In the present study, oxidation induced the haemolysis of NaCl- or CaCl₂-bathed erythrocytes that was attenuated by amiloride and prevented by replacement of Na⁺/Ca²⁺ and Cl⁻ by an impermeant cation and anion, respectively. Moreover, haemolysis was enhanced after decreasing bath Cl⁻ concentration by substitution with permeant anions, indicating the contribution of the oxidation-induced Cl⁻-dependent cation conductance to net uptake of NaCl (Cl⁻ enters the cells by anion channels) and haemolysis.

Glucose-6-phosphate dehydrogenase (G-6-PD)-deficient RBCs have decreased redox buffer capacity and patients with G-6-PD deficiency suffer severe haemolytic anaemia upon exposure to increased oxidative stress (Mavelli et al. 1984; Bilmen et al. 2001). The data from the present study strongly suggest that the activation of the RBC cation conductance may contribute to the pathophysiology of haemolytic anaemia in G-6-PD deficiency.

In principle, oxidative stress can induce plasma membrane ion conductances either by triggering of signal transduction cascades or by direct modification (or de novo formation) of channel protein complexes. The latter has been demonstrated for the heterodimeric amino acid transporter LAT1/4F2hc: crosslinking of LAT1 and 4F2hc via a cytoplasmatic disulfide bridge by oxidative stress induces a non-selective cation conductance (Wagner et al. 2000).

In summary, oxidation evoked a non-selective cation conductance which promotes haemolysis of human erythrocytes bathed in isotonic NaCl solution. Both, conductance and haemolysis were dependent on extracellular Cl⁻ and inhibited by amiloride. Oxidation induced activation and Cl⁻ dependence suggest the involvement of the cation conductance in haemolytic anaemia in G-6-PD deficiency and LIS effect, respectively.