K⁺ currents underlying the action of endothelium-derived hyperpolarizing factor in guinea-pig, rat and human blood vessels

Abstract

1. Membrane currents attributed to endothelium-derived hyperpolarizing factor (EDHF) were recorded in short segments of submucosal arterioles of guinea-pigs using single microelectrode voltage clamp. The functional responses of arterioles and human subcutaneous, rat hepatic and guinea-pig coronary arteries were also assessed as changes in membrane potential recorded simultaneously with contractile activity.

2. The current-voltage (I-V) relationship for the conductance due to EDHF displayed outward rectification with little voltage dependence. Components of the current were blocked by charybdotoxin (30-60 nM) and apamin (0.25-0.50 μM), which also blocked hyperpolarization and prevented EDHF-induced relaxation.

3. The EDHF-induced current was insensitive to Ba²⁺ (20-100 μM) and/or ouabain (1 μM to 1 mM).

4. In human subcutaneous arteries and guinea-pig coronary arteries and submucosal arterioles, the EDHF-induced responses were insensitive to Ba²⁺ and/or ouabain. Increasing [K⁺]_o to 11-21 mM evoked depolarization under conditions in which EDHF evoked hyperpolarization.

5. Responses to ACh, sympathetic nerve stimulation and action potentials were indistinguishable between dye-labelled smooth muscle and endothelial cells in arterioles. Action potentials in identified endothelial cells were always associated with constriction of the arterioles.

6. 18β-Glycyrrhetinic acid (30 μM) and carbenoxolone (100 μM) depolarized endothelial cells by 31 ± 6 mV (n = 7 animals) and 33 ± 4 mV (n = 5), respectively, inhibited action potentials in smooth muscle and endothelial cells and reduced the ACh-induced hyperpolarization of endothelial cells by 56 and 58 %, respectively.

7. Thus, activation of outwardly rectifying K⁺ channels underlies the hyperpolarization and relaxation due to EDHF. These channels have properties similar to those of intermediate conductance (IK_Ca) and small conductance (SK_Ca) Ca²⁺-activated K⁺ channels. Strong electrical coupling between endothelial and smooth muscle cells implies that these two layers function as a single electrical syncytium. The non-specific effects of glycyrrhetinic acid precludes its use as an indicator of the involvement of gap junctions in EDHF-attributed responses. These conclusions are likely to apply to a variety of blood vessels including those of humans.

Of the endothelium-derived relaxing and hyperpolarizing factors, the nature of endothelium-derived hyperpolarizing factor (EDHF) and the ionic mechanisms underlying its actions are least understood (Mombouli & Vanhoutte, 1997; Edwards & Weston, 1998; Hecker, 2000). The nature of EDHF as a chemical factor has been explored in a number of studies (Campbell et al. 1996; Popp et al. 1996; Randall et al. 1996; Plane et al. 1997; Chataigneau et al. 1998; Edwards et al. 1998; Fisslthaler et al. 1999) with a lack of consensus. Some studies suggest that the EDHF-induced hyperpolarization of smooth muscle cells results from electrotonic spread of activity originating in the endothelial cells (Little et al. 1995; Bény, 1997; Chaytor et al. 1998; Yamamoto et al. 1998, 1999; Edwards et al. 1999), while other studies provide evidence against coupling (Bény et al. 1997; Welsh & Segal, 1998).

Identification of the nature of EDHF, whether as a diffusible factor or as electrotonic current spread, and the detailed study of the ionic mechanisms involved, using voltage clamp of isolated cells, has been hampered by the requirement for the endothelial cells to be in close proximity to the smooth muscle cells. Hirst and colleagues developed a preparation of submucosal arterioles of guinea-pigs, possessing a single layer of smooth muscle cells (Hirst & Neild, 1980), with the endothelial cells remaining functionally intact (Hashitani & Suzuki, 1997). Edwards & Hirst (1988) voltage clamped very short segments of these arterioles using single intracellular microelectrodes coupled to a switching voltage clamp amplifier.

The aim of the present study was to use the approach of Edwards & Hirst (1988) to examine the membrane currents underlying the actions of EDHF and, importantly, to relate these ionic mechanisms to the functional response by simultaneously recording arteriole diameter and membrane potential. In view of the possibility of electrical coupling between endothelial and smooth muscle cells, cell type was unequivocally identified using Lucifer Yellow. Furthermore, derivatives of glycyrrhetinic acid were used in an attempt to uncouple the endothelium from the smooth muscle. Experiments on larger vessels, guinea-pig coronary and human subcutaneous arteries, provide a broader applicability for our conclusions, including potential clinical relevance.

METHODS

Guinea-pigs and rats of both sexes were killed by cervical dislocation with ethics committee approval. Human subcutaneous arteries were obtained during surgery following informed written consent and hospital ethics committee approval, in accordance with the Declaration of Helsinki. Guinea-pig submucosal arterioles were prepared as described previously (Hirst & Neild, 1980; Edwards & Hirst, 1988). Briefly, the muscle layers and mucosa were gently removed from a segment of ileum. The thin connective tissue sheet which remained, containing the network of submucosal arterioles, was pinned to the silicone rubber floor of an organ bath and was continuously superfused with physiological saline solution (PSS; for composition see below). Arterioles of 20-60 μm outside diameter were chosen and EDHF-induced hyperpolarizations were recorded (see below) simultaneously with contractile activity, recorded as arteriole outside diameter, using DIAMTRAK (Neild, 1989). Larger arteries, guinea-pig coronary, rat hepatic and human subcutaneous arteries, were cut into ring segments 1-2 mm in length and mounted on a wire myograph for the simultaneous measurement of membrane potential and isometric tension (Parkington et al. 1995). Tissues were superfused at 3 ml min⁻¹ and 35°C with PSS consisting of (mM): NaCl, 120; KCl, 5; NaHCO₃, 25; KH₂PO₄, 1; MgSO₄, 1.2; CaCl₂, 2.5; glucose, 11; and bubbled with 95 % O₂-5 % CO₂. The term EDHF describes endothelium-dependent hyperpolarization of vascular smooth muscle following blockade of nitric oxide (NO) and prostanoid production (Chen et al. 1988; Parkington et al. 1995; see Mombouli & Vanhoutte, 1997; Edwards & Weston, 1998). All recordings were made in the presence of N^ω-nitro-L-arginine methylester (L-NAME; 100 μM) and indomethacin (1 μM) to inhibit NO and prostanoid production, respectively, except where indicated in Fig. 8Ad. We (not shown) and others (Hashitani & Suzuki, 1997) have found that neither NO nor prostaglandins hyperpolarize these arterioles. The endothelial cells were stimulated with acetylcholine (ACh).

Figure 8. The effect of 18β-glycyrrhetinic acid on membrane potential of an endothelial cell and diameter in a submucosal arteriole of guinea-pig.

A, recording of membrane potential (top trace; continuous impalement for a-c) and contractile activity (diameter, bottom trace). The periodic transients in the membrane potential trace are due to hyperpolarizing electrotonic potentials and also EJPs, some of which reached threshold for the initiation of action potentials and resulted in contraction; responses marked i-vi are shown on an expanded scale in B. Aa, 18β-GA (30 μM) caused considerable reversible depolarization, inhibited the action potentials (ii vs. iii), increased the amplitude of the electrotonic potential (i vs. iv), and resulted in only a small EDHF hyperpolarization in response to ACh (1 μM, 1 min). Ab, after washout of 18β-GA, the membrane was depolarized with Ba²⁺ (15 μM) and ACh produced a much larger response in the same cell. Ac, carbenoxolone (CBX; 100 μM), similarly to 18β-GA, caused a large depolarization and inhibited the response to ACh in the same cell as in Aa and Ab, but decreased the amplitude of the electrotonic potential (vi). Ad, the amplitudes of hyperpolarizations evoked by ACh (1 μM, 1 min) when the membrane was depolarized to various levels with Ba²⁺ (17 tissues). ○, hyperpolarizations recorded in normal PSS; ○, hyperpolarizations recorded in the presence of indomethacin (1 μM) and L-NAME (100 μM). The continuous line is a linear regression fitted to the combined data: individual regressions revealed no differences between the hyperpolarizations recorded in normal physiological saline and those recorded in the presence of inhibitors of NO and prostaglandins.

For most experiments, especially those involving arterioles, the membrane potential was recorded with intracellular microelectrodes whose tips were filled with 2 % Lucifer Yellow CH as the dilithium salt dissolved in 1 M LiCl; the electrodes were then backfilled with 1 M KCl. Tip resistances were about 100 MΩ. In most instances the Lucifer Yellow diffused from the tip of the microelectrode into the impaled cell (also found by Welsh & Segal, 1998), which allowed unambiguous identification of the cell from which recordings were made. Dye filling was facilitated by the hyperpolarizing electrotonic current steps applied during recording (see Results). In some experiments on arterioles, but especially those involving larger arteries, the microelectrodes were filled entirely with 1 M KCl; the membrane potential and responses to ACh were indistinguishable from recordings made with Lucifer Yellow-filled microelectrodes. To record from arteriolar endothelial cells, the microelectrodes were inserted through the single cell layer of smooth muscle cells in the arteriole wall from outside the arterioles (see also Welsh & Segal, 1998).

Arterioles were cut into short, 100-300 μm long, segments for voltage clamping with single intracellular microelectrodes and an Axoclamp-2 (Axon Instruments, USA) switching amplifier (see also Edwards & Hirst, 1988). pCLAMP 6 software (Axon Instruments) was used to generate voltage ramps, and acquire and analyse the data. In some experiments voltage steps were used to construct current-voltage (I-V) relationships; these relationships were similar to those obtained from voltage ramps (data reported on but not shown). The difference between the pre-ACh data and the data obtained in the presence of ACh gave the I-V relationship for the response attributed to EDHF and these data were fitted to the Goldman-Hodgkin-Katz (GHK) equation for a K⁺ current using Prism 2 (GraphPad Corp., USA) software. Hyperpolarizations evoked by ACh were integrated graphically using SigmaScan software (Jandel Scientific, USA).

Nerves supplying the arterioles were stimulated with pulses of 0.2 ms duration and 20-80 V (dial setting) using a DS2 isolated stimulator (Digitimer, UK) through platinum electrodes.

Drugs used were: ACh, L-NAME, indomethacin, sodium nitroprusside, ouabain, apamin, iberiotoxin (IbTX), dilithium Lucifer Yellow CH, 18α-glycyrrhetinic acid (18α-GA), 18β-glycyrrhetinic acid (18β-GA) and carbenoxolone (from Sigma Chemical Co., USA), (-)-(3S,4R)-3,4-dihydro-3-hydroxy-2,2-dimethyl-4-(3-oxo-cyclopent-1-enyl-1-oxy)-2H-1-benzopyran-6-carbonitrile (PCO 400; from Biomol, USA) and U46619 (from Cayman Chemicals, USA). Charybdotoxin (ChTX) was synthesized by Auspep (Australia) and Iloprost, a stable analogue of prostaglandin I₂, was a gift from Schering (Germany). Stock solutions of the glycyrrhetinic acid derivatives were prepared immediately before use in order to minimize the cytotoxicity of these compounds (Davidson & Baumgarten, 1988). Stock solutions were made in warmed dimethyl sulphoxide (DMSO) (18α-GA, 18β-GA) or deionized water (carbenoxolone). DMSO (from BDH Chemicals, Australia) was without effect at a concentration of 10⁻⁴ M. Antibody to Factor VIII (goat anti-human F8-RAg antibody; Atlantic Antibodies, Incstar, USA) and guinea-pig complement (Gibco BRL, USA) were also used.

Data were compared using Student's t test, paired or unpaired as appropriate, using the software packages InStat 3 or Prism 2 (both from GraphPad Corp., USA). Mean values ±s.e.m. and n, referring to the number of animals (except in Fig. 8Ad), are quoted throughout. Values of P < 0.05 were considered statistically significant. Concentration- response data for each tissue were fitted to a sigmoid curve using the least-squares method implemented in Prism 2. The concentration of agonist which produced a response that was 50 % of maximum (EC₅₀) and the Hill slope were determined and the pD₂ values (-log EC₅₀) are quoted.

RESULTS

Membrane currents underlying EDHF

Stimulation of the endothelium with ACh (1 μM) evoked an outward current (Fig. 1Aa and Ba) in all of 16 arterioles from 11 animals. I-V relationships obtained from voltage ramps during the outward current indicated an increase in membrane conductance. The K⁺ channel blocker ChTX (30-60 nM) reduced the outward current evoked by ACh (Fig. 1Ab), and subtraction of the currents evoked by ACh in ChTX from those evoked in control solution (PSS) revealed a large, slowly activating outward current (Fig. 1Ac), whose I-V relationship was outwardly rectifying, was well described by the GHK equation for a K⁺ conductance (Fig. 1Ad), and reversed at -83 ± 4 mV (n = 6), not different (P = 0.64) from the estimated K⁺ equilibrium potential (E_K) of -85 mV in these arterioles (see Edwards & Hirst, 1988). On its own, the K⁺ channel blocker apamin (0.25-0.50 μM) also reduced the amplitude of the outward current (Fig. 1Bb). Subtraction of the currents evoked by ACh in apamin from those evoked in PSS revealed that apamin also blocked an outward current component (Fig. 1Bc), whose I-V relationship (Fig. 1Bd) was also outwardly rectifying, was well described by the GHK equation for a K⁺ conductance, and reversed at -79 ± 3 mV (n = 5; not different from -85 mV; P = 0.12). A combination of apamin plus ChTX completely abolished the outward current (n = 5; not shown). That the currents were described by the GHK equation for a K⁺ current is consistent with their outward rectification, and suggests that the kinetics of the channels have little or no voltage dependence. The currents recorded during the voltage steps showed no time dependence over 150 ms (data not shown).

Figure 1. Effect of K⁺ channel blockers on the currents underlying EDHF in voltage clamped guinea-pig submucosal arterioles.

The arterioles were clamped at -63 mV and periodically ramped from -92 to -53 mV over 1.5 s (see voltage protocol in insets). ACh (1 μM) evoked outward currents (Aa and Ba), which were reduced by ChTX (30 nM; Ab) and apamin (0.5 μM; Bb). Subtraction of Ab from Aa, and Bb from Ba revealed the ChTX- (Ac) and apamin-sensitive (Bc) currents (ramp responses truncated). Ad and Bd show the I-V relationships for the ChTX- and apamin-sensitive currents, respectively. The effects of ChTX and apamin on the resting conductance were insignificant but were nevertheless subtracted out. The smooth curves in Ad and Bd are fitted K⁺ currents calculated from the GHK equation. The results in A were obtained from a continuous impalement in the same cell and those in B were from a different cell.

The EDHF-attributed hyperpolarization was not affected by IbTX (100 nM, n = 5, paired t test) or glibenclamide (1 μM, n = 3, paired t test), which block large-conductance Ca²⁺-activated K⁺ (BK_Ca) and ATP-sensitive K⁺ (K_ATP) channels, respectively (data not shown). The K_ATP channel opener PCO 400 had no effect on the resting membrane potential (300 nM, n = 3; data not shown).

The responses described above did not occur following destruction of the endothelium by pretreatment with antibody to factor VIII with complement (Juncos et al. 1994) (n = 6, not shown).

EDHF hyperpolarization

In normal PSS, the resting potential of the smooth muscle cells in intact submucosal arterioles was -74 ± 1 mV (n = 17). Under these conditions, ACh (1 μM) evoked only a small hyperpolarization, 1.7 ± 0.4 mV (n = 12) in amplitude. Partial inhibition of the K⁺ inward rectifier (K_IR) channels with Ba²⁺ was used as a means of depolarizing the arterioles. The smooth muscle membrane was depolarized over a range of values using Ba²⁺ (0-100 μM). The maximum amplitude of the EDHF hyperpolarization increased by 3.7 ± 0.4 mV per 10 mV depolarization and extrapolation to zero amplitude occurred at -84 ± 12 mV (n = 54 data points, see Fig. 8Ad). The level of the membrane potential attained during the peak of the hyperpolarization also depended on the initial value of the membrane potential, and became less negative by 6.3 ± 0.4 mV per 10 mV depolarization (n = 54 data points). To take into account the complex nature of the response to ACh, in terms of amplitude and duration, the response was integrated (Tare et al. 2000). The integral of the hyperpolarization increased 309 ± 35 mV s per 10 mV depolarization (n = 54 data points from 17 tissues). At 20-50 μM, Ba²⁺ depolarized the cells to -64.4 ± 1.3 mV (n = 8), conditions under which an EDHF hyperpolarization of around 10 mV occurred, with minimal activation of voltage-dependent Ca²⁺ channels. Thus, this concentration of Ba²⁺ was used in most of the subsequent studies.

In the presence of 20-50 μM Ba²⁺, the EDHF-attributed hyperpolarization evoked by ACh consisted of an initial peak component which was followed by a smaller, slower plateau-like component. The EC₅₀ of the peak component, 4 × 10⁻⁷ M, was a significant 2.4-fold lower concentration than the EC₅₀ of 10⁻⁶ M for the amplitude of the slow component measured at the end of the 1 min exposure to ACh (n = 4, paired t test; pD₂ values of 6.41 ± 0.08 and 6.02 ± 0.10 for peak and slow components, respectively; not shown). The hyperpolarization was subsequently studied in detail, in the presence of 20-50 μM Ba²⁺, and with 1 μM ACh (close to the EC₅₀) to stimulate the endothelium. Under these conditions, the initial peak hyperpolarization was to -72 ± 1 mV (7.2 ± 0.7 mV in amplitude, n = 7) and the following slower component was 5.4 ± 0.3 mV in amplitude (n = 7, paired t test) by the end of the 1 min exposure to ACh.

EDHF and contractile activity of arterioles

The functional significance of the EDHF-mediated conductance changes in arterioles was studied by recording membrane potential simultaneously with arteriole diameter. Exposure to higher concentrations of Ba²⁺ (50-100 μM) initiated spike-like active responses, each of which was accompanied by partial constriction of the arterioles (Fig. 2a). ACh induced a biphasic hyperpolarization which abolished the Ba²⁺-induced spike-like active responses and caused relaxation of the arterioles (Fig. 2Aa). Apamin (0.25 μM) plus ChTX (10-60 nM) all but abolished the integral of the hyperpolarization (to 9 ± 4 %, n = 6, paired data) and ACh now evoked depolarization and contraction of the arterioles (n = 3; Fig. 2Ab). Thus, the hyperpolarization, which is attributed to EDHF and is underpinned by the outward currents described above, can profoundly affect the contractile activity of these small resistance vessels.

Figure 2. Effect of K⁺ channel blockers and ouabain on the functional EDHF response in guinea-pig submucosal arterioles.

Aa, in arterioles depolarized and constricted with Ba²⁺ (100 μM), ACh (1 μM) evoked a biphasic hyperpolarization (top trace) and relaxation (bottom trace). Ab, in the presence of apamin (0.25 μM) plus ChTX (60 nM), ACh evoked depolarization and considerable constriction. Ba, ACh evoked hyperpolarization and relaxation in the presence of Ba²⁺ (35 μM) alone. Bb, addition of ouabain (1 μM) evoked further depolarization but ACh still hyperpolarized and relaxed the arteriole. Results in A and B are continuous recordings in the same cell.

Effects of Ba²⁺ and ouabain on EDHF-attributed membrane currents

It has been suggested that ACh stimulates K⁺ efflux from endothelial cells, which hyperpolarizes the smooth muscle cells of rat blood vessels by activating K_IR channels and the Na⁺-K⁺ pump (Edwards et al. 1998). The possible contribution of K_IR channels and the Na⁺-K⁺ pump to the hyperpolarization in arterioles was tested by recording the EDHF-attributed membrane currents in the presence of Ba²⁺ and ouabain, which block K_IR channels and the Na⁺-K⁺ pump, respectively. The current activated by ACh (Fig. 3a) was well described by the GHK equation for a K⁺ current (Fig. 3D) and neither Ba²⁺ (Fig. 3E) nor Ba²⁺ plus ouabain (Fig. 3F) had any effect on this relationship. The maximum amplitude of the current in Ba²⁺ was 89 ± 5 % of that in control solution (not significantly different, P = 0.09, n = 6, paired t test). In ouabain, the current was 118 ± 14 % of its control (P = 0.27, n = 5). Moreover, the combination of Ba²⁺ plus ouabain did not prevent EDHF-mediated hyperpolarization or relaxation (Fig. 2Bb).

Figure 3. The effects of ouabain and Ba²⁺ on EDHF currents in guinea-pig submucosal arterioles.

ACh (1 μM) evoked an outward current in control PSS (A), in the presence of Ba²⁺ (30 μM; B) and in Ba²⁺ plus ouabain (200 μM; C). Subtraction of the current transients before application of ACh (i) from those which occurred during the maximum of the outward current evoked by ACh (ii) yielded the control difference I-V relationship (D). Similar procedures provided the difference I-V relationship in the presence of Ba²⁺ (E), and the difference I-V relationship in the presence of Ba²⁺ plus ouabain (F). The smooth curves give the best fit to the GHK equation for a pure K⁺ current (D-F). The I-V relationships of the EDHF-attributed outward currents were fundamentally unchanged by Ba²⁺ or Ba²⁺ plus ouabain. Ba²⁺ inhibited an outward component of the resting current and its I-V relationship (B, trace iii minus A, trace i) was characteristic of K_IR currents (G). Ouabain inhibited a component of outward resting current whose I-V relationship (ramp responses recorded after B and before C; not shown) had a linearly extrapolated zero current potential of -141 mV, characteristic of the Na⁺-K⁺ pump (see Quinn et al. 2000).

Ba²⁺ resulted in a more negative holding current (Fig. 3Bvs. A). The I-V relationship of the inhibited current (Fig. 3G) was determined by subtraction of the current responses in control (trace i in Fig. 3a) from the current responses in Ba²⁺ (trace iii in Fig. 3B). The I-V relationship was inwardly rectifying with a shape typical of a K_IR current, and strikingly different to the outwardly rectifying I-V relationship of EDHF (Fig. 3D).

Ouabain also resulted in a more negative holding current (see Fig. 3Cvs. B). By subtraction of the I-V relationships, the ouabain-inhibited current, attributed to the Na⁺-K⁺ pump, had a shallow I-V relationship displaying little voltage dependence, a linearly extrapolated zero current potential of -134 ± 6 mV (n = 5; consistent with previous observations in arterioles, see Quinn et al. 2000), which was considerably more negative than E_K, and no obvious rectification over the voltage range studied (Fig. 3H). Thus, a significant involvement of the Na⁺-K⁺ pump in the ACh-induced hyperpolarization is difficult to reconcile with the I-V relationships for the current underlying the actions of EDHF (Fig. 3D), and the lack of effect of ouabain on the outward current (Fig. 3F) further supports this.

Comparison of EDHF- and KCl-induced membrane currents

K_IR channels are very prominent in these guinea-pig submucosal arterioles (Edwards & Hirst, 1988), which suggests that these arterioles may be ideally suited to testing the effects of raised KCl. The application of ACh to arterioles evoked hyperpolarization, whereas the addition of 5 or 10 mM KCl evoked depolarization (6 arterioles from 3 animals; data not shown). The membrane currents underlying the actions of ACh and raised KCl (an additional 5 or 10 mM) were recorded under voltage clamp in tissues from seven animals (9 arterioles) and an example from one of these for the addition of 10 mM KCl is shown in Fig. 4A. ACh evoked an outward current that was resistant to the presence of Ba²⁺ (30 μM). This is in contrast to the Ba²⁺-sensitive inward current evoked by 10 mM KCl.

Figure 4. Comparison between the effects of ACh and raised KCl on currents in guinea-pig submucosal arterioles.

A, an arteriole segment was clamped at a holding potential of -66 mV and periodic ramps (-96 to -62 mV, see inset) were applied. ACh (1 μM) evoked an outward current which was unaffected by Ba²⁺ (30 μM). Addition of 10 mM KCl induced an inward current that was markedly attenuated in the presence of Ba²⁺. Ba²⁺ (30 μM) decreased the outward holding current. B, in another arteriole in which the membrane potential was clamped at -66 mV and periodically ramped (-98 to -61 mV, see inset), the detailed response to ACh is compared with the response to the addition of 5 mM KCl. ACh evoked the usual outward current (Ba). I-V relationships before (Bb, trace i) and during (Bb, trace ii) ACh application show that the EDHF-attributed current was associated with an increased membrane conductance and that the reversal potential for the holding current was more negative in ACh (Bb, trace ii) than in control (Bb, trace i). Subtraction of the current transients (ii minus i) demonstrated that EDHF was underpinned by an outwardly rectifying current with a reversal potential consistent with E_K (Bc). Application of 5 mM KCl resulted in an inward current (Bd), which had an I-V relationship showing an increased membrane conductance and an extrapolated reversal potential (Be, trace iv) more positive than that in control solution (Be, trace iii). The difference I-V relationship shows that the KCl-activated current (Bf, iv minus iii) had a different shape and reversal potential (linearly extrapolated to -56 mV) to the EDHF current (Bc). Several current responses to voltage ramps were averaged to produce traces i and iii.

I-V relationships comparing the effects of ACh and 5 mM KCl in a different arteriole are shown in Fig. 4B. The current activated by ACh (Fig. 4Ba and Bb, trace ii) and the I-V relationship attributed to EDHF (trace ii minus trace i; Fig. 4Bc) displayed the usual outward rectification. In contrast, application of 5 mM KCl evoked an inward current (Fig. 4Bd and Be, trace iv) and the I-V relationship (trace iv minus trace iii; Fig. 4Bf) was inward over the voltage range studied. Thus, the I-V relationship for EDHF (Fig. 4Bc) has a very different shape from that of the I-V relationship for the current induced by raised K⁺ (Fig. 4Bf). The addition of 5 mM KCl shifted the reversal potential for the resting current from -70 ± 3 to -62 ± 2 mV (n = 6, P = 0.03) and the K⁺-induced current had an extrapolated reversal potential of -54 ± 6 mV (n = 6). During application of 5 mM KCl, the I-V relationships during wash in of KCl were qualitatively similar to those observed during the maximal effect.

Larger arteries

The coronary artery of guinea-pigs has a prominent EDHF hyperpolarization which is resistant to glibenclamide, is only slightly reduced by apamin alone, and requires 5-10 mM tetraethylammonium to achieve blockade (Parkington et al. 1995). In the present study, EDHF hyperpolarization was not blocked by either Ba²⁺ (30 μM) or a combination of Ba²⁺ plus ouabain (1 mM) in guinea-pig coronary (n = 10; Fig. 5A and C) or human subcutaneous (n = 4; Fig. 5B) arteries. Ouabain alone (1 mM) significantly reduced the hyperpolarization evoked by ACh in guinea-pig coronary arteries (Fig. 5C) and markedly reduced the hyperpolarization and relaxation evoked by the stable prostacyclin analogue Iloprost in guinea-pig coronary (Fig. 5a), human subcutaneous (Fig. 5B) and rat hepatic arteries (not shown). Increasing [K⁺]_o from 5 to 6-10 mM evoked only depolarization in human subcutaneous arteries (Fig. 5B), a small (up to ∼5 mV) hyperpolarization in guinea-pig coronary arteries (Fig. 5a), and a larger hyperpolarization in rat hepatic arteries (10 ± 1 mV, n = 6; not shown), as found previously by others (Edwards et al. 1998). In guinea-pig coronary arteries, ChTX (± apamin) markedly reduced the ACh hyperpolarization and all but abolished relaxation (Fig. 5D), similar to observations in arterioles.

Figure 5. Effects of Ba²⁺ and ouabain on hyperpolarizations and relaxations in larger arteries: guinea-pig coronary and human subcutaneous arteries.

A, membrane potential (top trace) and contractile activity (bottom trace) of guinea-pig coronary artery in PSS and in the presence of Ba²⁺ (30 μM) and a combination of Ba²⁺ and ouabain (1 mM) in response to ACh (80 nM), sodium nitroprusside (SNP; 80 nM), Iloprost (Ilo; 8 nM) and K⁺ (8 mM in total), each for 10 s (see Parkington et al. 1995). B, responses to K⁺ (8 mM in total), ACh (2 μM) and Iloprost (8 nM) in human subcutaneous arteries in PSS and in the presence of ouabain and ouabain plus Ba²⁺. C, the level of membrane potential of the guinea-pig coronary artery before the application of ACh (upper edges of the bars) and at the peak of the hyperpolarization evoked by ACh (lower edges of the bars) in PSS (Control, n = 10), Ba²⁺ prior to addition of ouabain (Ba²⁺, n = 5), ouabain prior to the addition of Ba²⁺ (Ouab, n = 5) and the combination of both blockers (n = 8). The amplitude of the hyperpolarization in ouabain alone was significantly different from the amplitude in PSS (*, unpaired t test). D, ChTX (30 nM) combined with apamin (0.5 μM) all but abolished the hyperpolarization and relaxation evoked by ACh and abolished those evoked by 10 mM K_o⁺ in guinea-pig coronary artery depolarized and contracted with the thromboxane analogue U46619 (30 nM).

Comparison of endothelial and smooth muscle responses

Since the EDHF-attributed currents could originate in the endothelial cells and spread electrotonically to the smooth muscle cells, we assessed the existence and extent of electrical coupling between the two layers in the submucosal arterioles. Each cell recorded from, smooth muscle or endothelial, was identified by the presence of Lucifer Yellow. The Lucifer Yellow did not spread between smooth muscle cells (Fig. 6C), or between smooth muscle and endothelial cells (Figs 6C and 7C), but did spread between endothelial cells with a strong preference for longitudinal spread and little circumferential spread (Fig. 7C), as reported for other arterioles (Little et al. 1995; Welsh & Segal, 1998).

Figure 6. Arteriole diameter and the accompanying electrical activity recorded in a smooth muscle cell, identified with Lucifer Yellow.

A, simultaneous recordings of membrane potential (top trace) with contractile activity (arteriole diameter, bottom trace). Periodic transients are due to hyperpolarizing electrotonic potentials and also EJPs, some of which evoked action potentials. Each action potential resulted in contraction of the arteriole. Ba²⁺ (15 μM) depolarized the membrane while ACh (1 μM) evoked hyperpolarization attributed to EDHF. B, responses i and ii in A showing, on an expanded scale, an EJP (trace i) and an action potential arising from an EJP (trace ii). C, the cell from which the recordings were made had a circumferential orientation, indicating that it was a smooth muscle cell. Calibration bar, 25 μm.

Figure 7. Arteriole diameter and the accompanying electrical activity recorded in an endothelial cell, identified with Lucifer Yellow.

A, membrane potential (top trace) of an identified endothelial cell recorded simultaneously with arteriole contractile activity (diameter, bottom trace). Periodic transients are due to hyperpolarizing electrotonic potentials and also EJPs, some of which reached threshold to initiate action potentials. Ba²⁺ (20 μM) evoked depolarization while ACh (1 μM) evoked EDHF-attributable hyperpolarization. Each action potential recorded from the endothelium was associated with contraction of the arteriole. B, the EJP (response i in A) and an action potential (response ii in A), on an expanded scale. C, the cell from which the recordings were made was a long thin cell running along the axis of the arteriole, indicating that it was an endothelial cell. Calibration bar, 25 μm.

The membrane potential responses recorded from a dye-labelled smooth muscle cell and the accompanying changes in arteriole diameter are depicted in Fig. 6. Periodic stimulation of the sympathetic nerves resulted in excitatory junction potentials (EJPs) (Fig. 6B, trace i), which occasionally gave rise to action potentials (Fig. 6A, top trace, and B, trace ii) associated with constriction of the arteriole (Fig. 6A, bottom trace). Ba²⁺ (15 μM) evoked depolarization from which ACh evoked hyperpolarization attributed to EDHF. These responses can be compared with those in Fig. 7, which shows responses recorded from a dye-labelled endothelial cell together with the accompanying changes in arteriole diameter. In endothelial cells, EJPs occurred upon sympathetic nerve stimulation (Fig. 7B, trace i) and action potentials were also recorded (Fig. 7B, trace ii), which were indistinguishable from those recorded from smooth muscle cells (Fig. 6). Furthermore, each action potential recorded in an endothelial cell (Fig. 7A, top trace) was associated with constriction of the smooth muscle (Fig. 7A, bottom trace), indicating the simultaneous occurrence of action potentials in the smooth muscle cells. Ba²⁺ and ACh evoked depolarization and hyperpolarization, respectively, in the endothelial cells (Fig. 7A), which were indistinguishable from the responses in smooth muscle cells (Fig. 6a), providing compelling evidence that the two layers function as a syncytium. Of 41 identified endothelial cells, all responded with action potentials and EJPs, indicating close and extensive electrical coupling between the endothelial and smooth muscle layers.

Glycyrrhetinic acid

In view of the extensive electrical coupling between the endothelial and smooth muscle layers in these arterioles, the effects of the putative gap junction inhibitor glycyrrhetinic acid (GA), and its derivatives, were tested on EDHF-attributed responses in the two cell types. The reputedly more specific and potent form, 18α-GA (Davidson & Baumgarten, 1988) (50 μM) was without effect on the membrane potential or on the amplitude of the EDHF-attributed hyperpolarization of endothelial cells or on the input resistance of short segments of arterioles, although the input time constant was increased and the amplitude of the EJP was reduced (Table 1).

Table 1.

The effects of glycyrrhetinic acid derivatives on the electrical properties of endothelial cells

	Membrane potential (mV)	EDHF amplitude (mV)	Rin (MΩ)	τin (ms)	EJP amplitude (mV)	EJP decay (ms)
Control	−53 ± 5(7)	13 ± 3(7)	54 ± 12(5)	147 ± 38(5)	3.8 ± 0.9(5)	391 ± 50(5)
18α–GA	−53 ± 4	13 ± 2	74 ± 22	204 ± 55*	2.6 ± 0.8*	399 ± 52
Control	−62 ± 3(7)	18 ± 2(54)†	43 ± 9(5)	97 ± 22(5)	2.7 ± 0.4(5)	257 ± 92(5)
18β–GA	−31 ± 6*	8 ± 3(4)*	94 ± 14*	132 ± 37	0.8 ± 0.4*	‡
Control	−65 ± 2(5)	19 ± 2(54)†	52 ± 35(5)	121 ± 51(5)	3.0 ± 0.8(3)	159 ± 38(3)
Carbenoxolone	−31 ± 2*	8 ± 2(5)*	25 ± 6	51 ± 15	0.6 ± 0.1	62 ± 12

^†The level of membrane potential, the amplitude of the hyperpolarization evoked by ACh (EDHF), the values of small segment input resistance (R_in) and input time constant (τ_in), and the amplitude and time constant of decay of the excitatory junction potential (EJP) immediately prior to (Control) and in the presence of the 18α(50 μM), 18β(30 μM) and carbenoxolone (100 μM) analogues of glycyrrhetinic acid(GA) are shown. The number of tissues is given in parentheses (except †). Paired t tests were used for all data since the drug effects were tested against the control in the same cell in all cases. Thus, only one n number per control and GA pair is given, except when the data were extrapolated from the data in Fig. 8Ad

^*Significantly different in GA compared with the corresponding control.

^‡The EJP amplitude was too small to determine the decay time constant.

The more commonly used, but less selective, isoform (Davidson & Baumgarten, 1988), 18β-GA (30 μM), reversibly depolarized identified endothelial cells (by 30 mV) (Fig. 8Aa; see Table 1). The amplitude of the ACh-evoked hyperpolarization was only 44 % of that expected at the same potential when Ba²⁺ was used to depolarize the cells (response in the same cell shown in Fig. 8Ab). Overall, application of ACh during membrane depolarization to -31 ± 6 mV should have evoked a hyperpolarization of 18 ± 2 mV, by extrapolation from the data in Fig. 8Ad (see also Table 1), while the hyperpolarization observed was a mere 8 ± 3 mV (n = 4). The form of the action potential was markedly altered by 18β-GA, with a decrease in the initial spike, together with a broadening/slowing of the repolarization phase (Fig. 8B, ii vs. iii). The input resistance was significantly increased while the amplitude of the EJP was decreased by 18β-GA. The effects of GA (18β) in identified smooth muscle cells (n = 3, data not shown) were indistinguishable from those recorded in identified endothelial cells (illustrated in Fig. 8), with similar effects on the resting potential, the action potential and the input resistance. The amplitude of the EJP was reduced to 60 ± 8 % by 18β-GA (n = 3, data not shown).

Carbenoxolone (100 μM), similarly to 30 μM 18β-GA, reversibly depolarized endothelial cells (Fig. 8Ac and Table 1). ACh evoked hyperpolarization which was reduced to 42 % of the value expected for tissues depolarized with Ba²⁺ to this level (see Fig. 8Ad). The mean value of the input resistance was smaller in carbenoxolone (Fig. 8B, vi vs. i), but the difference was not statistically significant (paired data; Table 1).

DISCUSSION

By voltage clamping short segments of submucosal arterioles in which the functional endothelial/smooth muscle relationship was maintained, we have recorded the membrane currents which underlie the relaxation attributable to EDHF. The current consisted of two components which could be distinguished pharmacologically and which we described further in terms of their I-V relationships. One component was blocked by ChTX but not by apamin, while the other component was blocked by apamin but not by ChTX. Neither component was affected by IbTX or glibenclamide. Both currents were outwardly rectifying and displayed little, if any, voltage-dependent gating. These characteristics describe intermediate conductance Ca²⁺-sensitive K⁺ channels (IK_Ca) under physiological K⁺ gradients, and small conductance Ca²⁺-sensitive K⁺ channels (SK_Ca), respectively (Alexander & Peters, 2000). The two current components are consistent with the actions of the blockers on EDHF-induced hyperpolarization and relaxation in guinea-pig coronary arteries (Fig. 5), and are in general agreement with results from a variety of arteries (Hashitani & Suzuki, 1997; Plane et al. 1997; Zygmunt et al. 1997; Chataigneau et al. 1998; Edwards et al. 1998; Quignard et al. 1999).

Two phases were evident in the EDHF hyperpolarization evoked by ACh, with the initial rapid component being more sensitive to ACh than the slower phase. In a simplistic model, it might be thought that one component might be due to the activity of IK_Ca channels while the other component represented the activity of SK_Ca channels. This is unlikely to be the case since the currents carried by IK_Ca and SK_Ca channels did not occur solely as an initial rapid component or solely as a slow component. Stimulation with agonists of many cell types, including endothelial and smooth muscle cells, can result in an initial peak component resulting from the increase in cytoplasmic free Ca²⁺ released from intracellular stores, followed by a smaller, plateau phase involving Ca²⁺ influx from the extracellular medium. These two processes of Ca²⁺ mobilization are likely to underlie the initial peak and slower phases of the EDHF hyperpolarization.

The prominence of the inward rectifier K⁺ current in the submucosal arterioles means that these tissues should be exquisite sensors of changes in extracellular K⁺ adjacent to the smooth muscle cells. Even though exogenous application of K⁺ (an additional 5 or 10 mM) caused appreciable activation of these channels, the outwardly rectifying nature of the EDHF-activated K⁺ channels immediately excludes the involvement of inwardly rectifying K⁺ channels (K_IR), the Na⁺-K⁺ pump, and their activation by K⁺ in the actions of EDHF. The effectiveness of ChTX and apamin, the ineffectiveness of Ba⁺ and/or ouabain against the actions of EDHF, and the failure of added K⁺ to evoke hyperpolarization and relaxation support these conclusions in the guinea-pig submucosal arterioles and guinea-pig coronary and human subcutaneous arteries (present study), guinea-pig carotid and porcine coronary arteries (Quignard et al. 1999) and rat mesenteric arteries (Vanheel & Van de Voorde, 1999; Lacy et al. 2000).

In guinea-pig submucosal arterioles the application of K⁺ resulted in considerable activation of K_IR channels. Due to the very negative nature of the membrane potential K⁺ had a depolarizing effect. It is unlikely that concentrations of K⁺ of less than 5 mM would have led to different conclusions. In the larger arteries we applied 1-7 mM additional K⁺ (7-14 mM K_o⁺), all with qualitatively similar effects. In arterioles, the washing in of 5 mM K⁺ would have resulted in initial lower concentrations. Yet early I-V relationships were not qualitatively different from those observed when the effect was stable and maximal. Since the aim of the present study was to describe the ionic currents underlying EDHF and to assess any role of K⁺ in their activation, we did not test the ability of K⁺ to hyperpolarize and relax arterioles more depolarized than about -55 mV, but 7-10 mM K_o⁺ did not significantly hyperpolarize larger guinea-pig coronary or human subcutaneous arteries.

Our results with added K⁺ recorded from the rat hepatic artery are similar, albeit of smaller amplitude, to those of Edwards and colleagues (Edwards et al. 1998). However, we found that the Iloprost-induced hyperpolarization, which is blocked by glibenclamide (Parkington et al. 1995), was also reduced by ouabain and this raises concerns regarding the mechanisms by which ouabain inhibited hyperpolarizations in these blood vessels. Ouabain-induced Ca²⁺ overload has been invoked to explain an inhibition of a K⁺ channel by ouabain in canine ventricular myocytes (Saxena et al. 1997).

In the present study, ouabain inhibited a current in arterioles, which had an extrapolated zero current potential of around -134 mV and an I-V relationship with a shallow slope, similar to the zero current potential of -160 mV reported for a ouabain-sensitive current in cerebral arterioles (Quinn et al. 2000). The current also has characteristics that are consistent with those attributed to the Na⁺-K⁺ pump in cardiac (Sakai et al. 1996) and cultured endothelial cells (Oike et al. 1993).

A strength of the present study is that, as well as recording the EDHF-activated currents under voltage clamp, we extended these results to encompass the functional response of the tissue by recording changes in arteriole diameter of the same preparations. Importantly, we recorded membrane potential simultaneously with diameter in the same tissues. In arterioles modestly depolarized with Ba²⁺, EDHF evoked hyperpolarization which was accompanied by relaxation. When the K⁺ channels, and therefore the hyperpolarization, were blocked with ChTX and apamin, the relaxation was abolished, revealing depolarization and contraction, thus demonstrating the functional importance of this K⁺ current which underlies the actions of EDHF in regulating the calibre of these arterioles.

The nature of EDHF is unresolved, with some studies suggesting that EDHF is a diffusible factor, while others suggest that the EDHF hyperpolarization results from electrotonic spread of current from the endothelial cells (see Introduction). In the present study, the membrane potential responses in identified endothelial cells included both action potentials and EJPs, which are generated exclusively in smooth muscle cells. The correlation of activity in endothelial cells recorded simultaneously with contractile activity of the smooth muscle cells further indicates that the two cell types are electrically coupled in these arterioles. That the responses in the endothelial and smooth muscle cells, including the characteristics of the action potential and EJPs, are indistinguishable indicates that the coupling is so strong that the two layers function as a single syncytium. The observations of strong electrical coupling in submucosal arterioles are in general agreement with a mathematical model of electrical coupling in these arterioles (Crane et al. 1999) and support and extend results obtained in earlier studies in a variety of blood vessels (von der Weid & Bény, 1993; Marchenko & Sage, 1994; Little et al. 1995; Bény, 1997; Chaytor et al. 1998; Emerson & Segal, 2000) and are consistent with detailed anatomical evidence of myoendothelial gap junctions obtained from serial section electron microscopy (Sandow & Hill, 2000). All of these observations support the very strong possibility that the EDHF-attributed hyperpolarization of smooth muscle cells is due to electrotonic spread from the endothelial cells.

The existence of electrical coupling between the endothelial and smooth muscle layers raises the fundamental question of the location of the K⁺ channels which underlie the EDHF-attributed hyperpolarization. IK_Ca channels occur in vascular smooth muscle cells that have been cultured (Neylon et al. 1999), but there seems to be little, if any, evidence for their presence, or for the presence of SK_Ca channels, in non-cultured vascular smooth muscle cells. Ca²⁺-activated K⁺ channels of intermediate conductance occur in vascular smooth muscle cells, though block of the channels by apamin suggests that they are not IK_Ca channels (Gebremedhin et al. 1996). If the IK_Ca and SK_Ca channels were on the smooth muscle cells, then in the present study one might expect that, due to the Ca²⁺ dependence of the channels (Marchenko & Sage, 1996), the Ca²⁺ influx during action potentials might activate the channels resulting in ‘mini EDHFs’ after each action potential. This was not evident (e.g. Fig. 7), suggesting that the IK_Ca and SK_Ca channels are on the endothelial cells, separated from the Ca²⁺ influx of the action potential in the smooth muscle cells. Recordings of intracellular Ca²⁺ from guinea-pig mesenteric arterioles (Fukuta et al. 1999) indicate that the endothelial and smooth muscle cells behave as separate Ca²⁺ compartments. Recordings from endothelial cells reveal K⁺ currents with the properties of IK_Ca and SK_Ca channel currents (Van Renterghem et al. 1995; Marchenko & Sage, 1996), while widespread expression of human IK_Ca channels occurs in tissues rich in epithelia (Jensen et al. 1998). Furthermore, endothelial cells which are isolated and not in contact with vascular smooth muscle cells respond to ACh with hyperpolarization which can be reduced by ChTX (Chen & Cheung, 1992; Ohashi et al. 1999) and abolished by ChTX plus apamin (Ohashi et al. 1999). Consistent with these observations is the recent study by Doughty et al. (1999) in which the EDHF-induced relaxation of rat mesenteric arteries was blocked when ChTX and apamin were applied to the lumen, and thus selectively to the endothelial cells, but not when the blockers were applied to the solution superfusing the adventitial-smooth muscle surface of the arteries, indicating an endothelial location of the K⁺ channels. Overall, given that endothelial cells that are separate from smooth muscle cells can respond to ACh with a hyperpolarization which has a strikingly similar time course and pharmacology to those of the EDHF-attributed hyperpolarization of the smooth muscle cells and, in view of the strong electrical coupling between the endothelium and smooth muscle in the arterioles in the present study, the most economical interpretation of our results is that the EDHF hyperpolarization represents electrotonic spread of the endothelial hyperpolarization to the smooth muscle.

Strong evidence that EDHF reflects electrotonic spread from endothelial cells would be provided by a loss of the EDHF-attributed hyperpolarization by gap junction inhibitors. While the GA compounds, including the reputedly selective α-isoform, tended to cause an increase in input resistance and membrane time constant, the effect was small and not always statistically significant. Although they reduced the amplitude of the EJP, the failure of these agents to block EJPs in endothelial cells demonstrates that inhibition of gap junctions was, at best, limited. Thus, while the GA compounds may have caused some blockade of gap junction patency, the effect was far less than would be required to provide the definitive evidence that electrotonic spread accounted for EDHF in a tissue in which endothelial and smooth muscle cells are so well coupled. The increase in input resistance and membrane time constant in GA could also be explained by a change in membrane conductance, hinted at by the substantial changes in membrane potential, action potential and junction potentials elicited by these agents. Moreover, the third GA compound tested, carbenoxolone, tended to decrease the input resistance and membrane time constant, adding to the confusion that surrounds the interpretation of the effects of these agents. 18β-GA and carbenoxolone are notorious for non-specific effects, with actions on other ion transport processes (Davidson & Baumgarten, 1988). Thus, the actions of these triterpenoid saponins are indicative of interference with various unspecified conductances and preclude any assessment of gap junction patency from the electrotonic potential. Of pivotal significance to the present argument is the observation that the GA compounds markedly inhibited the hyperpolarization evoked by ACh in the endothelial cells themselves. Recently, Santicioli & Maggi (2000) concluded that 100 μM 18β-GA had non-specific effects on guinea-pig renal pelvis and ureter which limited the usefulness of this agent in the study of gap junction communication. Action potentials, recorded with the sucrose gap technique, were inhibited by 30 μM 18β-GA, and this is consistent with the inhibition of action potentials in the smooth muscle or endothelial cells of arterioles (present study). Thus, it would seem that GA compounds fall far short of being ‘the standard’ indicators for the involvement of electrical coupling, as widely assumed.

Irrespective of whether EDHF is a diffusible factor or represents electrotonic spread from the endothelium, a significant result from our study is that the conductances described here are those which underlie the functional effects of EDHF, that is, EDHF-induced relaxation of the arterioles. Their block by ChTX plus apamin is consistent with results from a number of blood vessels including human arteries, indicating a generality of our conclusions. Evidence suggests that EDHF is impaired in diseases such as diabetes and hypertension. Thus, therapeutic modalities aimed at EDHF will need to focus on the K⁺ channels described above.

In conclusion, our results demonstrate that stimulation of endothelial cells results in outwardly rectifying K⁺ currents attributable to EDHF which play a significant role in bringing about relaxation of arterioles. The channels underlying the actions of EDHF are likely to be IK_Ca and SK_Ca channels. The demonstrated strong electrical coupling between the endothelial cells and smooth muscle cells means that the channels may well be located on the endothelial cells. Whether an endothelial-derived diffusible factor(s) is involved in these effects remains unresolved, but the responses to ACh in resistance arterioles, and larger arteries, including human arteries, do not indicate a significant role for the activation of K_IR channels and the Na⁺-K⁺ pump. As indicated by the effects of apamin and ChTX, any factor(s), or pathology (e.g. diabetes, hypertension), which alters the effectiveness of EDHF can have a major impact on dilator/constrictor balance and hence on blood flow and tissue perfusion.