The aim of this review is to provide basic information on the electrophysiological changes during acute ischemia and reperfusion from the level of ion channels up to the level of multicellular preparations. After an introduction, section ii provides a general description of the ion channels and electrogenic transporters present in the heart, more specifically in the plasma membrane, in intracellular organelles of the sarcoplasmic reticulum and mitochondria, and in the gap junctions. The description is restricted to activation and permeation characterisitics, while modulation is incorporated in section iii. This section (ischemic syndromes) describes the biochemical (lipids, radicals, hormones, neurotransmitters, metabolites) and ion concentration changes, the mechanisms involved, and the effect on channels and cells. Sectioniv (electrical changes and arrhythmias) is subdivided in two parts, with first a description of the electrical changes at the cellular and multicellular level, followed by an analysis of arrhythmias during ischemia and reperfusion. The last short section suggests possible developments in the study of ischemia-related phenomena.
A. Aim and General Outline
The aim of the present review is to provide a description of ionic channels and electrogenic transporters present in the heart (sect.ii), to describe the biochemical and ion concentration changes during acute ischemia and early reperfusion and the effects on channels (sect. iii), to translate these findings in terms of modulation of electrical properties at the cellular and multicellular level and to analyze their role in the genesis of acute cardiac arrhythmias (sect. iv).
The reader mainly interested in the electrophysiological changes during ischemia and arrhythmias may directly start with sectioniii and return later, when necessary for a better understanding of certain processes and notions, to sectionii. The reader with a main interest in the biophysics of ion channels is directed first to section ii.
Section ii provides a description of the activation-inactivation kinetics and the permeation processes of channels and carriers; their role in the genesis of electrical activity is briefly discussed in this section. Section ii does not include information on modulation or pharmacology. It is subdivided in sections describing channels and transporters in the plasma membrane, in the gap junction, and in intracellular organelles (sarcoplasmic reticulum and mitochondrion). Although information about eventual changes during ischemia is lacking for a number of channels and transporters, it is hoped that a systematic description and analysis will be helpful to understand more fully the actual state of knowledge, to put this information into perspective, and eventually to help to plan adequate experiments.
Section iii describes how channels and transporters are modulated. The modulatory processes are not described for each channel separately but are grouped into “syndromes” related to ischemia, such as changes in ion concentration ([K+]o, [Na+]i, [Ca2+]i, and [Mg2+]i, where the subscripts o and i refer to extracellular and intracellular, respectively); disturbance of lipids resulting into the accumulation of long-chain acylcarnitines, lysophosphoglycerides, fatty acids, and arachidonic acid; production of radicals; secretion of neurotransmitters, hormones, and metabolites, with concomitant stimulation of adrenergic, purinergic, and muscarinic receptors; and the genesis of stretch. The analysis of each syndrome includes a description of the changes, the mechanisms involved, the effect on channels and transporters, and the final outcome at the electrophysiological and arrhythmia levels.
Section iv contains two major subdivisions: the electrophysiological changes at the cellular and multicellular level and the genesis of arrhythmias. Section iv Acontains a description of the changes in resting and action potential, in excitability, refractoriness, and conduction and is followed by a discussion of possible mechanisms. Section iv Bstarts with an analysis of the general processes involved in the genesis of arrhythmias, followed by a description of the type of arrhythmias encountered during ischemia and reperfusion, and finally a discussion of possible mechanisms.
B. Ischemia Models
Cardiac ischemia is characterized by a deficient energetic input as well as a deficient waste removal. The result is failure of contraction, deterioration of electrical behavior, and eventual death of the cell. At the organism level, the end point may be lethal arrhythmias or mechanical pump failure. To study ischemia, different experimental models have been used. Coronary artery ligation or obstruction by a local thrombus mimics closely the clinical settings of myocardial infarction but does not allow a direct analysis of the changes in the ionic currents involved. For this purpose, voltage-clamp measurements have been applied to multicellular preparations and single cells subjected to hypoxia, uncouplers of the mitochondrial oxidative chain, inhibitors of glycolysis, superfusion with a solution containing a high concentration of K+ and H+, and deficient in glucose. In other models, the complex nature of the ischemic process has been dissected in different facets of specific biochemical changes that occur during ischemia: amphiphiles, radicals, catecholamines, adenosine, ACh, and stretch.
C. General Biochemical Changes During Ischemia and Reperfusion
Under aerobic conditions, NADH and FADH2, formed during glycolysis and in the citric acid cycle, transfer their electrons to O2 through the electron transport chain. This provides the energy to build up the chemiosmotic gradient that drives the synthesis of ATP. Oxidative phosphorylation is coupled to the demands of the cell. A feedback is generated by the breakdown products of ATP (772) and the rise in mitochondrial Ca2+ that modulates the activity of the mitochondrial dehydrogenases (692) and of the ATP synthase (199).
The turnover rate of high-energy phosphates is 30–40 mol · g wet wt−1 · min−1, while the storage is only 15 mol/g wet wt. This means that the time limit for exhaustion is short and only 15–30 s (339). When oxygen falls below a critical level in the cytoplasm, the electron transport and the process of H+ ejection in the mitochondrion stops. The energy stored in the proton electrochemical gradient becomes insufficient to synthesize ATP in appropriate quantity. Some of the energy in ATP may be used to maintain the mitochodrion membrane potential (E m) and to inhibit irreversible reactions leading to cell death (772). As a consequence, [ATP] may be expected to fall (Fig. 1) and [ADP] to increase. The cell, however, has efficient means to maintain the ATP level: 1) energy demand falls very rapidly during the first 30 s of ischemia as a consequence of contraction failure, 2) an important amount of phosphocreatine (PCr) is continuously used to restore ATP from ADP with concomitant increase in Pi, and 3) anaerobic glycolysis starts and intensifies.
The fall in the ATP/ADP stimulates the glycolytic pathway by activation of three important enzymes: 1) hexokinase responsible for the formation of glucose-6-phosphate; 2) phosphofructokinase, responsible for the transformation of fructose-6-phosphate to fructose-1,6-biphosphate; and 3) pyruvate kinase for the formation of pyruvate. The reactions leading to pyruvate are stimulated but glucose metabolism stops at the pyruvate-lactate stage, due to inhibition of pyruvate dehydrogenase. Pyruvate is transformed in l-lactate under simultaneous oxidation of NADH to NAD+. At the same time, phosphorylase a is transformed in its active form, facilitated by accompanying increases in AMP, Pi, Ca2+, and cAMP and increases the breakdown of glycogen. High glycogen content of cells postpones the development of contracture during ischemia (190). Finally, glucose transport through the cell membrane is stimulated, secondary to a translocation of glucose transporters from an intracellular pool to the plasma membrane (1155) and increases the flow down to pyruvate. Glycolytic enzymes are especially dense in the subsarcolemmal space; stimulation of anaerobic glycolysis at this level is important for the regulation of intracellular ion concentrations and channels (KATP) (1082).
Stimulation of the glycolytic pathway explains why [ATP] stays remarkably constant during the initial 10–15 min of ischemia. However, the free energy change upon hydrolysis of ATP falls immediately because of the rise in [ADP] (281). This fall has important consequences for a number of ATP-driven transporters such the Na+-K+ATPase and the Ca2+-ATPase. The anaerobic glycolytic process is self-inhibiting however. When acidosis becomes too pronounced, glycolysis in turn is inhibited, and ATP synthesis is further reduced.
Whenever the free energy of ATP hydrolysis exceeds the energy stored in the proton electrochemical gradient, net ATP hydrolysis may occur; the ATP synthase then acts as an ATP hydrolase, and the energy is used to update the electrical gradient in the mitochondrion (224) (Fig. 2). Although the arrest of electron flow is expected to cause depolarization of the mitochondrial membrane during anoxia, such depolarization only occurs after a delay (244). The delay corresponds to the time needed to block glycolysis to cause serious ATP deficiency and rise in [Ca2+]i. Blocking ATP hydrolysis by oligomycin slows the fall in [ATP] (808) and delays activation of the mitochondrial mega-channel. When [ATP] drops to levels <1 mM and at the same time the concentration of H+, [Ca2+]i (>1 μM), PO4 3−(>10 mM), and long-chain acylcarnitines (LCAC) increase (772), the mitochondrial membrane becomes abnormally leaky through activation of the mega-channel or transition pore in the inner membrane of the mitochondrion. Efflux of ATP through the plasma membrane is an additional reason for a fall of intracellular ATP. In the extracellular medium, this ATP is further broken down to adenosine and may activate purinergic P1 and P2receptors.
Block of oxidative metabolism and fall in ATP/ADP will have consequences on lipid metabolism, the generation of oxidative stress, the release of catecholamines, and ion concentrations.
Upon reperfusion after 5–30 min of ischemia, oxygen consumption rapidly recovers. The NADH/NAD+ quickly decreases, although the level may remain higher than the control for some time. After brief (5–10 min) ischemic bouts, PCr concentration quickly returns to normal, but recovery of ATP is slow. After longer periods of ischemia, PCr concentration still recovers within 5 min, but ATP, which may have dropped to 50%, stays at this low level for 30 min or more (641).
The preischemic pattern of substrate utilization is restored, i.e., oxidation of fatty acids (FA), is the main contributor to ATP synthesis; the level of glycolysis, however, remains elevated in the early period of reperfusion and is important for ATP generation. This situation may explain why block of glycolysis at this time is disadvantageous and results in aggravation of Ca2+ overload and release of intracellular enzymes. Such a block occurs when FA are present in high concentration (621). A high rate of FA oxidation generates NADH which inhibits pyruvate dehydrogenase. Glycolysis which stops at the pyruvate level is accompanied by an overproduction of protons and may negatively affect the recovery of [Na+]i and [Ca2+]i(612). High levels of lactate are deleterious for recovery.
The high level of oxygen consumption early after reperfusion stays in contrast to the continuing deficiency of the contractile machinery or “stunning.” This uncoupling between oxygen consumption and contractility is not due to impairment of the respiratory chain flux and insufficient ATP synthesis. Energy is sufficiently present but apparently not used. As a possible explanation for the high rate of oxygen consumption, the existence of futile cycles has been proposed, such as the cycling of Ca2+ between the cytoplasm and mitochondria (189). The absorption of Ca2+ by the mitochondria as well as its removal via the Na+/Ca2+ exchanger depends on the energy flux in the electron chain. The ejection of protons generates the negative matrix potential that drives the absorption of Ca2+ and creates the proton gradient that is required for removing Na+ from the mitochondrial matrix via the Na+/H+ exchanger and subsequently Ca2+ via the Na+/Ca2+ exchanger. Adenosine 5′-triphosphate, which for its synthesis is also dependent on the proton gradient, is used by the plasma membrane Na+-K+ pump to keep cytosolic [Na+] low (Fig. 2). Under normal conditions, this cycle only requires little energy, but in the case of Ca2+overload, the expenditure may become excessive. The situation is aggravated by an eventual intermittent opening of the mega-channel activated by elevated matrix [Ca2+] (91). The leak created causes breakdown of the mitochondrial electrical gradient. Activation of the mega-channel may propagate from mitochondrion to mitochondrion, creating a Ca2+ wave in the cytosol (449).
Reperfusion after more than 30 min leads still to an important but only transient recovery of oxygen consumption, while contractile function remains severely and persistently depressed. Intracellular enzymes and other substances are lost. On the microscopic level, cells appear necrotic, and later important fibrosis develops.
II. ION CHANNELS AND TRANSPORTERS
A. Ion Channels and Transporters in the Plasma Membrane
1. Na+ channels
In most excitable cells, the Na+ current is responsible for the upstroke and the conduction of the action potential. In heart, the density of the channel is low in sinoatrial node (SAN) and atrioventricular node (AVN) cells and highest in Purkinje cells. The Na+ channel is voltage operated and shows activation and inactivation. It is specifically blocked by tetrodotoxin (TTX) (for a general review, see Ref. 291); modulation of channels is described in section iii.
a) activation and inactivation. When a cardiac cell is depolarized, an inward current is generated that rises rapidly and decreases afterward on a much slower time course. The rise is sigmoidal (567) or exponential (685). The time to peak shortens with depolarization (634). The decline in macroscopic current or inactivation is best described by a sum of two exponentials. The time constant of the first component becomes shorter with depolarization.
Peak current-voltage relation can be translated into a conductance-voltage relation from which an activation curve can be constructed. Peak current is voltage dependent and increases for more negative holding potentials. The relation has been explained as a change in availability of the channel to become activated. Both relations, activation and inactivation, can be described by a Boltzmann distribution, with slopes of ∼6 mV. Midpoints of voltages in multicellular preparations are −30 mV for the activation process and −85 mV for the inactivation process.
The time course and the steady-state properties of the current have been explained in terms of two voltage-dependent processes, activation and inactivation independent from each other (425). The opening and closing of the channel depends on the movement of charged gates, and the hypothesis thus predicts the existence of a gating current. In heart cells, such a gating current has been measured for the activation process (50,350, 372, 486) but not for the inactivation process, as if inactivation is voltage independent. The activation gating current, however, becomes smaller during the development of inactivation. The gating charge becomes immobilized, suggesting some coupling between the two processes (372,486). Depending on the type of cell, coupling between activation and inactivation is variable. Coupling is strong in neuroblastoma cells (9) in which activation shows a variable onset and inactivation follows rapidly. In cardiac cells, the coupling is weak; openings always occur at the beginning of the depolarizing pulse (activation is fast) and their duration is variable (inactivation is slower and variable) (69,90, 567, 587, 856).
Recovery from inactivation is normally very fast (1–10 ms), with rates increasing upon hyperpolarization. After long depolarizations, recovery can become very slow (order of seconds). The process has been called slow inactivation (125, 160,825, 876). The existence of this phenomenon may be important to understand changes occurring during and after ischemia, where the cells are subjected to long depolarizations.
Is there more than one type of Na+ current? In heart cells, decay of the Na+ current is normally very rapid. In a limited range of potentials, where activation and inactivation overlap, a small noninactivating current or window current can be recorded, which is due to the cycling of channels between the rested, activated, and inactivated state (27, 325,755). This is not due to any abnormal behavior of the channel, but in pathological conditions, the overlapping may increase, resulting in an enhancement of the current.
A slowly inactivating Na+ current, which represents a small percentage of the total Na+ current, can be recorded in rabbit (125) and canine (325) Purkinje fibers, in rat ventricular cells (826), and in expressed human Na+ channels (801) over a broad range of potentials. In the rat, it becomes more pronounced during hypoxia (489) (Fig. 3). Part of it does not inactivate completely. Compared with the fast Na+current component, activation is shifted in the hyperpolarized direction. This may cause more pronounced overlapping of activation and inactivation, generating in this way a constant steady-state component. The current deactivates immediately upon hyperpolarization. At the single-channel level, it is characterized by a bursting behavior, i.e., clusters of repetitive short openings (5 ms) sometimes alternating with long openings (order of 200 ms) (1187). An activity characterized by short openings (<0.5 ms), becoming shorter with depolarization, has been described as “background” current because it carries current at the resting potential (1187). The current undergoes inactivation, however, and is very selective for Na+. According to Böhle and Benndorf (90), one and the same channel can show many modes of gating behavior; this channel may thus not be different from the slowly inactivating one. The slowly inactivating current has been described as being more sensitive to block by TTX (174), a finding which has been advanced in favor of the existence of a different isoform; the result, however, can also be explained by the slow kinetics of the TTX block (126).
After exposure of membrane patches to lysophosphatidylcholine (108, 1002), a persistent current can be recorded over a voltage range of −150 to 0 mV. It is probably due to a modulation of the fast component. The current does not deactivate upon hyperpolarization and is substantial at the resting potential. It may play an important role as an inward leak current, causing K+ loss during ischemia.
A Na+-dependent, TTX-insensitive background current is induced by high concentrations of ACh in guinea pig ventricular myocytes (659). The current reverses at −25 mV in normal Tyrode solution, and its single-channel conductance has been estimated by noise analysis to be ∼2 pS (897). Because of its different pharmacological and single-channel characteristics, it can be regarded as a different isoform.
b) ion permeation. In the presence of 150 mM [Na+]o, single-channel conductance is 20–25 pS; it is dependent on the concentration of external Na+ with a dissociation constant (K d) of 300–400 mM (885) and shows different substates (554, 727,856). The channel is permeable to Na+ and Li+ and much less to K+ (9.5%) and Cs+ (2%) or tetramethylammonium (<1%) (885). A small Ca2+ permeability exists (3) and may increase in the presence of ouabain and upon β-receptor activation (840). The current-voltage relation is linear in the absence of bivalent ions; deviation from this linear behavior in normal bathing solutions is due to a voltage-dependent block by [Ca2+]o and [Mg2+]o; electrical distance is 0.3–0.4 (884). The block by [Ca2+]o and [Mg2+]ois characterized by a reduction in apparent single-channel conductance, suggesting a very fast block. Other bivalent ions efficiently but more slowly block the current: Cd2+ > Mn2+ > Co2+ > Ca2+> Mg2+ > Ba2+. Of importance to note is the highly sensitive block by Zn2+ and Cd2+ (297, 1042); these ions are much less effective in neuronal and skeletal muscle cells, whereas the opposite sensitivity exists for TTX. Internal [Mg2+] blocks outward current through the channel (786). The block is slightly voltage dependent with an electrical distance of 0.18 from inside.
At the molecular level, the channel consists of an α-subunit (260 kDa) and two β-subunits (36 and 33 kDa each) (291). The α-subunit, which is sufficient for channel activity, is composed of six transmembrane segments repeated four times in a tetrameric structure. The voltage sensor for activation has been related to the S4 segment. The short intracellular segment between domains III and IV has been identified as the fast inactivation gate. Spontaneous deletion of the KPQ segment in the junctional part gives rise to the LQT3 syndrome, characterized by a slowing of the inactivation process and generation of long action potentials. The absence or weak voltage dependency of the inactivation process is consistent with the location of this inactivation gate outside the electrical field of the membrane. Slow inactivation involves conformational changes in the external pore (C-type inactivation) (1058). The presence of sialic acid at the external surface shifts the activation and inactivation in the negative direction (64). Interaction with the cytoskeleton modulates the inactivation and activation voltage dependence; F-actin disruption and microtubule stabilization accelerate the shift in the negative direction that occurs during whole cell recording (636). The two β-subunits exert a modulatory role, speed up inactivation, and decrease block by local anesthetics (633).
The permeation process is dependent on the hydrophilic part of the α-subunit between transmembrane segments 5 and 6. The selectivity for Na+ is due to specific amino acids in this region, and mutation of only two critical amino acid residues is sufficient to confer permeability properties similar to the Ca2+ channel. A cysteine in domain I is responsible for the high affinity of the channel for Zn2+ and Cd2+ and low sensitivity to TTX of the cardiac isoform.
2. Ca2+ channels
Different types of Ca2+-permeable channels have been described in the plasma membrane of heart cells: the L- and T-type channels, both voltage activated, and a background channel (see Ref.673). They can be differentiated on the basis of their electrophysiological and pharmacological characteristics. The density of the T- and L-type channels differs in different sections of the heart. The ratio of T-type over L-type channel is highest in Purkinje and sinoatrial cells where it approaches the value of 0.2–0.6, and it is less in atrial and ventricular cells where the ratio only attains 0.015–0.025 (48).
A) BACKGROUND CA2+ CHANNELS.
Calcium-permeable channels are seen following incorporation of plasma membrane protein fractions in bilayers (806). No voltage steps are required for activation, and spontaneous single-channel activity with long openings of >100 ms can be recorded at negative E m values of −90 mV. The conductance is 7 pS in isotonic Ba2+; the channels are not blocked by dihydropyridines (DHP) and show no rundown. Their selectivity is not pronounced with a permeability ratioP Ba/P Cs of 10.
A similar channel but with higher conductance (22, 45, and 78 pS in Ba2+) has been described in neonatal rat hearts (187). It is not blocked by Cd2+ or nifedipine; instead, activity is rather increased in the presence of DHP but suppressed by protamine. In rat ventricular myocytes, the channel is induced by exposure of the inside-out patches to phenothiazines (593). Activity is also increased by exposure to oxygen free radicals and metabolic inhibition (1056). The channel may be responsible for the Na+-independent Ca2+ entry pathway described for rat trabeculae (583).
B) L-TYPE CA2+ CHANNEL.
Calcium influx through the L-type Ca2+ channel is responsible for the upstroke of the action potential in the SAN and AVN and plays an important role in determining the plateau and eventual spike-dome appearance of the action potential in other cardiac cells. It is further responsible for the coupling between excitation and contraction, induces release of Ca2+ from the sarcoplasmic reticulum, and regulates intracellular Ca2+load. In this way it determines activity of a number of mitochondrial and cytoplasmatic Ca2+-sensitive enzymes.
I) Kinetics. A) Activation and inactivation. Threshold for activation is around −25 mV, and half-maximum activation is attained at about −15 mV for most cells (see Ref. 673) and at more positive potentials (−3 mV) in the AV node (373). The rise in current follows a sigmoidal time course, suggesting a multistep process as the underlying mechanism. Activation is preceded by a gating current (50, 352, 486,898). The density of the channels derived from the gating current is much larger than the ionic current density, suggesting that some of the channels although gating are not carrying current.
At the single-channel level, three modes of activity have been distinguished (137, 403). In mode 1, the channel shows repetitive short (<1 ms) openings and closures (0.2 and 2 ms), forming a burst of activity separated from other bursts by longer closures. A number of consecutive bursts may be grouped in a cluster. A variable waiting time precedes the openings; it decreases at more depolarized levels and corresponds to the faster activation and shorter time to peak values of the Ca2+current. Mode 2 occurs in the presence of DHP agonists (403) or after β-receptor stimulation (137,767) and is characterized by much longer open times.Mode 3 is characterized by the complete absence of openings or presence of only rare short openings. The frequency of this latter mode increases with preceding depolarizations and corresponds to the occurrence of steady-state inactivation.
The L-type Ca2+ current inactivates in two ways: a voltage-dependent and a current-dependent way. The existence of two types of inactivation explains the complex time course of current decay and the presence of a dip in the inactivation curve. Half-maximum steady-state inactivation occurs at −20 to −30 mV. The curve shows a minimum at ∼0 mV and increases again at more positive potentials. When intracellular Ca2+ is well buffered, this turning up is absent and the decay of the current during a pulse is much slower. These observations have led to the conclusion that inactivation is dependent on voltage as well as on Ca2+ influx. The latter or Ca2+-induced inactivation is the faster process, whereas the voltage-induced inactivation is rather slow. Activation and inactivation show a remarkable overlapping (window current) (414,900).
B) Intracellular Ca2+ or current-dependent inactivation. Evidence for intracellular Ca2+-dependent inactivation at the whole cell level is based on the change in time course of current decay with changes in [Ca2+]i (see Ref. 673) (Fig.4, A and B). The decay is faster the larger the Ca2+ current, and it is slowed in the presence of intracellular Ca2+ buffers. At the single-channel level (cell-attached patches), Ca2+permeation through the channel reduces the open probability of subsequent reopenings of the channel and shifts the gating mode toward a mode with long-lived closed states (454). In excised patches, with Ba2+ as the charge carrier, steady-state elevation of Ca2+ in the range of micromolar concentration or flash photolysis of Ca2+ reduces the open probability of the Ca2+ channels. It is especially the Ca2+originating from the sarcoplasmic reticulum (SR) that is responsible for the inactivation process during the first 50 ms of depolarization (949, 1154); at later times also Ca2+ permeating through the Ca2+ channel contributes to the inactivation (949). The increase in cytosolic Ca2+ by the release from the SR is indeed 10-fold greater than the Ca2+ entering the cell via the L-type Ca2+ channel (915). In favor of this explanation is the observation that inactivation is much slower after depletion of the SR by caffeine or in the presence of ryanodine. It is important to note that [Ca2+] seen by the channel may importantly deviate from the cytosolic [Ca2+] as measured by fluorescence techniques. Especially during the first 50 ms of a depolarizing pulse the concentration seen by the channel may be much higher. Such a difference may explain why the relative inhibition of the current estimated from the Ca2+ transient is greater during this initial period than later (611).
The existence of an intact cytoskeleton is important in determining the local [Ca2+]. The cytoskeleton normally keeps the channels separated from each other and limits the rise in local [Ca2+]. In ischemia, the cytoskeleton may become disturbed. Disruption of the cytoskeleton structure by colchicine or cytochalasin, with consequent clustering of the channels, favors inactivation, whereas substances such as taxol and phalloidin that stabilize the skeleton remove inactivation and improve reopening (taxol and colchicine act on microtubules; phalloidin on F-actin) (311).
Different mechanisms have been proposed to explain [Ca2+]i-dependent inactivation (719). 1) A fall in driving force is improbable. In cell-attached patches, single-channel conductance does not change, whereas open probability is markedly reduced. 2) Dephosphorylation by phosphatase (e.g., calcineurin) and proteolysis by Ca2+-stimulated proteases (calpain) (351) are mechanisms that may be responsible for long-term changes in Ca2+ channel behavior but are too slow to explain the time course of changes during a single depolarization. 3) Calcium binds to the channel protein and induces a change in configuration. In the cloned channel, a Ca2+ binding motif (an EF hand) exists at the COOH terminal of the α1C-subunit; deletion eliminates Ca2+-induced inactivation. The site is located near the inner mouth but outside the electrical field (205,719). The reduction of Ca2+ current by [Mg2+]i also has been explained by a direct binding to this site, reducing the number of functional channels (1132). Because Ca2+ binding is a fast process, it provides an explanation for the time course of inactivation and the observation that Ca2+ oscillations are accompanied by equivalent changes in current (911). Photolysis of Ca2+ results in a rapid inactivation, within 20 ms (351) to 75 ms (46); the process is not accompanied by a change in gating current, confirming that intracellular Ca2+-induced inactivation and voltage-dependent inactivation are two distinct phenomena (273).
C) Voltage-dependent inactivation. The evidence for voltage-dependent inactivation is based on the following observations (see Ref. 673). 1) Under conditions where intracellular Ca2+-dependent inactivation is excluded (e.g., current carried by monovalent cations or current carried by Ba2+ or Sr2+), the current decays with time, and the rate of decay is faster the higher the depolarization.2) Inactivation develops when prepulses are applied that do not result in Ca2+ inward current. 3) Inactivation is present in channels incorporated in lipid bilayers with Ca2+ buffered (806). 4) Inactivation is slowed by trypsin treatment, but the intracellular Ca2+-dependent inactivation is not affected (862). No charge movement occurs on inactivation, but as inactivation of the current proceeds, the charge movement that accompanies activation becomes smaller (351,898).
D) Recovery from inactivation: Ca2+-induced facilitation. Upon hyperpolarization, the Ca2+ current recovers from inactivation induced by a previous depolarization. Because inactivation is voltage and intracellular Ca2+dependent, it is logical to expect repriming also to depend on these two parameters. For the voltage-induced inactivation, the rate and the degree of recovery is greater the more hyperpolarized the membrane, with time constants in the order of 300 ms at −50 mV and 100 ms or shorter at −80 mV. A much slower component (seconds) is present after long depolarizations, indicating the occurrence of slow inactivation (94, 868). This slow inactivation may play a role in overdrive suppression (1074). Recovery from Ca2+-induced inactivation as such is voltage independent (911, 912) but indirectly it is modulated by voltage, since the fall in [Ca2+]i is dependent in part on the Na+/Ca2+ exchange that is faster the more negative the E m.
At negative holding potentials, recovery from inactivation may show an overshoot, i.e., the Ca2+ current transiently becomes larger than in steady state. The original finding of an overshoot was made in Purkinje fibers treated with digitalis (505) but can also be observed without Ca2+overload (see references in Refs. 606, 770). The potentiated Ca2+ current is characterized by a larger peak and a slower time course of decay (991, 1159,1193) (Fig. 4).
At the single-channel level, facilitation is characterized by an increase in open probability (P o) with a larger proportion of long openings (413) (mode 2) and an increase in number of functional channels (1132). In all these approaches, it is clear that a moderate increase in Ca2+ is required (991,1159, 1193); the overshoot in recovery is inhibited by rising intracellular Ca2+ buffering, or by ryanodine or using Ba2+ as the current carrier. Excessive rises in Ca2+ lead to inhibition by Ca2+-induced inactivation (351).
How elevated Ca2+ causes facilitation remains a matter of debate. Phosphorylation of the channel protein is a possibility. Flash photolysis induces facilitation with a delay, and the effect is counteracted by inhibitors of protein kinases (PK) (18,1159). Other groups, however, did not find an effect of PK inhibitors (46, 1132), and facilitation still occurred with nonhydrolyzable ATP analogs (1132) or even improved (46). The conclusion of these authors is that a Ca2+-nucleotide complex directly potentiates Ca2+ current (I Ca) through a phosphorylation-independent mechanism. The existence of two phases in the recovery (decrease followed by an overshoot) has practical consequences in determining down- or upregulation ofI Ca as a function of frequency and diastolic E m.
II) Permeation and selectivity. The channel is 500–1,000 times more permeable to bivalent ions such as Ca2+ and Ba2+ than to monovalent ions. The exclusion of monovalent ions depends on the presence of a minimum concentration of bivalent ions. In the absence of bivalent ions, the channel becomes highly permeable to monovalent ions (651). Although restricted, the permeability for K+ is responsible for a substantial current during the action potential, the reason being that the concentration of intracellular K+ is quite high compared with the nanomolar free Ca2+ concentration. The K+ contribution also explains why the reversal potential (E rev) of the Ca2+ current is much less positive than expected for the equilibrium potential for Ca2+. The bivalent ion current through the channel increases with the concentration and shows saturation. Compared with the single-channel conductance in isotonic Ba2+ (8–10 pS and 15–25 pS, see Ref. 673), the single-channel conductance at the physiological concentration of 1 mMCa2+ was found to be surprisingly high at 7 pS (1162). The presence of negative charges at the pore mouth of the channel, which attract bivalent ions and increase their local concentration (331), is probably the reason for this behavior. In the presence of both Ca2+and Ba2+, the current across the channel is smaller than in the presence of either Ca2+ or Ba2+ alone (997). This kind of behavior has been called anomalous mole fraction behavior and suggests a multi-ion channel. Different ions thus interact in such a way that the flux of one species is hampered by the presence of the other species. Multi-ion occupancy also explains the high conductance and at the same time the high selectivity. High selectivity is conditioned by high affinity; high conductance by the presence of more than one ion in the channel and repulsion of one ion by the other. The multi-ion nature of the channel also explains the flickery block behavior of the channel in the presence of elevated proton concentration (784).
The channel can be blocked by a number of extracellular bi- and trivalent ions such as Mg2+, Mn2+, Co2+, Ni2+, Cd2+, Zn2+, and La3+ (997). The ions Cd2+, Zn2+, and La3+ block the channel in a voltage-dependent way with an apparent electrical distance of 0.15 for Cd2+ and Zn2+ and 0.60 for La3+. Intracellular Mg2+ is needed to activate enzymes that phosphorylate the channel but may, on the other hand, reduce the current that has been augmented by isoproterenol or BAY K 8644 by a direct blocking action and by activating phosphatases (1090, 1132). The cAMP-dependent phosphorylation reduces the sensitivity to [Mg2+]i block (1131).
The cardiac Ca2+ channel molecular structure consists of four subunits: two α-subunits, α1 and α2, a β-subunit, and a δ-subunit (a γ-subunit is exclusively expressed in skeletal muscle). The α-subunit is sufficient to express channel activity. It resembles the Na+ channel with four times six transmembrane segments, a highly charged S4 segment that probably acts as the voltage sensor for activation, and an intracellular link between domain II and III responsible for inactivation. Trypsin treatment removes voltage-dependent inactivation but not internal Ca2+-dependent inactivation (862). Binding of Ca2+ to the COOH terminal is a possible mechanism for Ca2+-induced inactivation (205).
The highly conserved glutamate residues located in the pore region of all four repeats are involved in high-affinity bivalent binding (1146). A mutation from E to Q in domain III has shown that this group is the strongest determinant of Ca2+binding (331).
The function of the α-subunit is markedly modulated by the β-subunit. Coexpression of the two results in a fourfold increase of peak current, which is not due to a change in single-channel conductance (331) but to a marked increase in density of functional channels (488, 718,759). The gating current remains unchanged, but coupling between conductance and gating is improved (759).
The L-type Ca2+ channels is highly regulated; this aspect is analyzed in section iii.
C) T-TYPE CA2+ CHANNEL.
A Ca2+ current of short duration is activated at potentials more negative than the threshold for the L-type Ca2+current (48, 726; see review in Ref. 1026). The current is well represented in SAN cells, atrial cells, Purkinje cells, and nodal cells. In the embryonic chick ventricle, it is the major Ca2+ current. The current is not found in human atrium (258, 606) or human ventricle (76). The channel has been proposed to play a role in pacemaking. It may also interfere with steroidgenesis, cell proliferation, and cardiac growth (400).
I) Activation, inactivation, and repriming. Threshold for activation is around −70 to −50 mV and maximum activation is seen at −30 to −10 mV (see Ref. 1026). Inactivation is rapid and complete, with time constants of 30 ms at −50 mV, becoming shorter at more depolarized levels. This behavior is opposite to the L-type channel, where inactivation is decelerated at positive potentials. Steady-state inactivation extends from −85 to −40 mV with half maximum around −60 mV and slope of 5.5 mV (14,36, 412). An increase in [Ca2+]i does not induce inactivation but rather facilitates the T-type current (13,994). At the single-channel level, this increase in current is characterized by a shift to long openings (mode 2behavior).
Repriming is voltage dependent and becomes faster with hyperpolarization: 250 ms at −70 mV and 100 ms at −90 mV (145). It is slower the longer the preceding depolarization, suggesting the existence of slow inactivation (412).
II) Permeation. In 100 mM [Ca2+], the single-channel conductance is 8 pS, compared with 20 pS for the L type. Contrary to the behavior of the L-type channel, permeability for Ca2+ and Ba2+ ion is the same (48, 900). The channel is permeable to Sr2+, blocked by Ni2+, but much less sensitive to Cd2+. Extracellular protons inhibit the channel with greater efficiency than the L-type current, whereas intracellular protons have no effect (1000). Extracellular Mg2+ reduce the current and shift the activation and inactivation curves in the positive direction (1113).
3. K+ channels
Because the equilibrium potential of K+ is rather negative, all cardiac K+ channels when activated will carry outward current, repolarize the membrane during the action potential, or stabilize the membrane at a hyperpolarized level. Among the many K+ currents, distinction can be made between voltage-activated currents (I to,I Kur,I Kss,I Kr, andI Ks), ligand-activated currents (I KACh,I KATP,I KNa, andI KAA), and a current (the inward rectifier I K1) that apparently does not gate and can be called a background current. Under physiological circumstances the voltage-activated K+ currents,I KACh among the ligand-activated and I K1, play an important role in shaping the normal action potential. Under ischemic conditions, ligand-activated currents, especiallyI KATP andI KAA, become primordial, whereas some of the “physiological” currents are inhibited. Voltage-activated K+ currents show activation and inactivation upon depolarization; the rates of these two processes can vary from fast to ultra-slow. Ligands can bind to receptors, which then activate the channel via a G protein, or can interact directly with an intracellular site of the channel.
The amino acid composition of most K+ channels is known (201), and recently, the molecular structure of the pore has been elucidated from X-ray analysis of the crystallized molecule (234). The channels show a remarkable homology with the Na+ and Ca2+ channels. However, whereas Na+ and Ca2+ channels consist of tandemly linked four domains of six transmembrane segments that are connected in one long polypeptide, only one domain with six transmembrane segments is found for the K+ channel. Segment 4 shows a high density of positive charges and acts as potential sensor. Two types of inactivation have been described in expressed channels (794): N-type inactivation in which the negatively charged NH2 terminal acts as a ball and blocks the open channel and C-type inactivation (COOH terminal) in which conformational changes on the extracellular side close to the pore result in some kind of constriction. C-type inactivation is sensitive to drug binding and extracellular K+. The ligand-activated and background channels have a simpler structure and contain only two transmembrane segments. Recently, a new family of K+ channels with two pore segments in tandem and four transmembrane segments have been expressed. They act as background channels (530); some of them are activated by arachidonic acid and polyunsaturated fatty acids (280). A tetrameric structure for all K+ channels is highly likely. The pore in K+ channels is formed by a stretch of 19 amino acids in the link between S5 and S6 in the four repeats. The motif GYG (or of GFG) in the P-region is the signature of K+ selectivity, but other residues also participate in determining K+selectivity.
A) K+ OUTWARD CHANNELS WITH FAST ACTIVATION.
Upon depolarization, three different K+ currents are rapidly activated. They can be distinguished by their rate of inactivation, which is relatively fast forI to, slow to ultraslow forI Kur, and nonexistent forI Kss, also called background current. On the latter current, no detailed information is available at the present time.
I) The fast transient outward current. The fast transient outward current (I to) is a transient outward K+ current that is rapidly activated and inactivated and blocked by millimolar concentrations of 4-aminopyridine (4-AP). The criterion of 4-AP sensitivity is not exclusive. In some species, another K+ current, theI Kur, is also blocked even by micromolar concentrations of the drug (see below), and in the dog ventricle part of the I to is insensitive to millimolar concentrations (607). It should be distinguished from another transient outward current carried by Cl− and activated by [Ca2+]i; this current is also called I to2. In this review it will be indicated as I ClCa.I to is partly responsible for the initial fast repolarization or phase 1 during the action potential. The density of I to varies among species and in a particular species it varies in different parts of the heart (see Ref.117). It is more expressed in the atrium and Purkinje fibers and in the ventricle more in the epicardial than endocardial fibers. The density of I to in the heart increases after birth (259, 483, 639), although its presence is variable (191, 342), an observation which is possibly related to pathological downregulation.
A) Activation and inactivation. The current is activated upon depolarization. Values for time course and steady-state vary among species and experimental conditions. In the rabbit, time course is fast and monoexponential (269); in the ferret it is sigmoidal (118). Midpoint voltage values for steady-state activation vary between −10 mV in the rabbit (269) and +20 mV in the ferret (118).
After activation, the current decays. The time course of inactivation has been described as monoexponential or biexponential. Time constants again vary but are in the order of 25–75 ms and are voltage independent. Steady-state inactivation shows half-maximum potentials between −50 and −15 mV (see Ref. 117) (Fig.5 A). Recovery from inactivation is very sensitive to voltage, being faster the more hyperpolarized the membrane. It is also facilitated by increasing [K+]o(284); this supports the hypothesis that inactivation is of the N type (750). For C-type inactivation, the process is slowed by extracellular K+ acting as a “foot in the door.”
Actual time constants for recovery vary with species. In most species, including humans (15, 284, 342), recovery is fast with time constants in the order of 20–60 ms at −80 mV; frequency dependence is small (Fig. 5 B). In rabbit atrium and ventricle, in sheep and dog Purkinje fibers (see Ref. 117), and in human subendocardial fibers (708), recovery is slow to very slow (time constants of 1–6 s). In these latter preparations, the current is markedly reduced (272) and shortening of the action potential markedly less at elevated frequencies (512).
B) Permeation. On the basis of measurements ofE rev (25, 118,158, 416, 707, 708,1088), the I to current is considered to be mainly carried by K+, although it seems less selective than other K+ currents, such asI K1. At the single-channel level, the current-voltage relation is linear and single-channel conductance in 145 mM [K+]o is on the order of 10–30 pS (63, 158,717) with the exception of a much lower value of 3–4 pS for ferret ventricle (118). The conductance increases at elevated [K+]o, with aK d of 200 μM (284).
As molecular substrates for I to, Kv4.2 (922) and Kv4.3 (229) have been proposed. On the basis of differences in voltage dependence, kinetics of inactivation and recovery, and block by 4-AP, Kv4.2 is the better candidate for I to in the human, rat, and ferret, and Kv4.3 in the canine and human subendocardium. Kv1.4 may be the better choice for the sheep and the rabbit. In the rat, expression of the protein shifts from Kv1.4 to Kv4.3 after birth and during thyroid treatment (1094).
II) IKur . A rapidly activated K+current, with no or very slow inactivation is present in different heart preparations. The voltage dependency of the slow inactivation process and the sensitivity to 4-AP is species variable, and on this basis, it can be concluded that the current does not correspond to a unique channel. On the basis of the sensitivity to 4-AP, the currents can be subdivided into two groups. In the human atrium (15, 192, 388,1065), dog atrium (1164), cultured rat neonatal ventricle (347), and mouse ventricle (287, 1182), IKur is exceptionally sensitive to 4-AP and completely blocked by concentrations of 50 μM or less. In rat atrium (97) and ventricle (25) and human ventricle (707), the current is insensitive to 4-AP. In many publications it is described as a noninactivating component of I to.
A) Activation and inactivation. On depolarization to levels positive to −40 mV, an outward current remains after subtraction of a rapidly inactivating I to in many species: rat atrium (97, 1021), rat ventricle (25, 347, 1083), human atrium (890, 1065) and human ventricle (707, 1087), rabbit ventricle (272), guinea pig ventricle (1161), and dog atrium (1164). It is rapidly activated and shows no or only very slow decay. The current inactivates however. Inactivation has been determined using long (tens of seconds) conditioning pulses. Midpoint inactivation potential is variable: −70 mV (25) and −90 mV (1083) in the rat and much less negative values of −9 and −20 mV in the human atrium (284,890, 1065) (Fig. 5 A). In accord with the existence of slow recovery from inactivation, the current is markedly reduced at elevated frequencies in the rat ventricle (25), rabbit ventricle (272)), and human atrium (272, 284, 890) (Fig.5 B). In these preparations, a rest current (I Kss) remains, which seems different fromI Kur. It is reduced by α-receptor stimulation in rat atrium (1021) and by β-receptor stimulation in rat ventricle (849).
B) Permeation. I Kur is assumed to be carried by K+, but direct demonstration is mostly lacking. In human atrium, tail currents reverse at negative potentials, suggesting a predominant (388, 1065) but not exclusive (191) permeability to K+. In favor of the K+ nature is the observation of sensitivity to tetraethylammonium (TEA) in rat ventricle (25) and dog atrium (1164) and block by Ba2+ in guinea pig ventricle (33, 1161) (in humans the current is not sensitive to Ba2+). Single-channel conductance in 5.4 mM [K+]o is in the order of 14 pS for the guinea pig ventricle (1161) and 20 pS for the dog atrium (1164) and is sensitive to [K+]o. Fully activated current-voltage relations show outward rectification (388, 1065).
The Kv.1.5 protein is a possible molecular candidate for theI Kur current in the human atrium (271, 923, 1065). It shows a high sensitivity to 4-AP (1067) and a limited, partial inactivation at positive potentials but is insensitive to TEA. The single-channel conductance is 17 pS. The protein is present in the rat and the human atrium and ventricle, as determined by immunolocalization; it is highly concentrated in the intercalated disks (669). In the dog, Kv3.1 has been proposed as a molecular candidate (1164).
B) K+ CURRENTS WITH DELAYED ACTIVATION: DELAYED K+ CURRENTS.
On the basis of kinetics, rectification, sensitivity to blockers, and modulation by intracellular messengers, two delayed K+currents, I Kr andI Ks, can be distinguished (153, 838); I Krshows activation and inactivation, I Ks only activation. Both are present in the human atrium (1066), human ventricle (77, 556, 605,1033), guinea pig ventricle (838), guinea pig atrium (433), dog ventricle (324,613) and atrium (1165), rabbit atrium (700) and ventricle (832), mouse neonatal ventricle (1055), and rat ventricle (141,1115). Only I Kr has been clearly described for the cat ventricle (290), the ferret ventricle (617), and rabbit SAN (1032) (466). I Ks seems the only delayed current in the guinea pig SAN (22). Density of the two currents varies in different layers of the myocardial wall. In midmyocardium of the dog ventricle, expression ofI Ks is small (613); this explains the longer action potential in these cells. In the ferret ventricle, the ERG protein (Kr) is most abundant in the epicardial layers (99).
I) Rapid delayed K+ current, IKr . A) Kinetics. I Kr activates rapidly for depolarizations positive to −40 mV, with a midpoint voltage between −20 and −5 mV; this value is [K+]oindependent (889). Time constants of activation vary among species; in the guinea pig, they are in the order of 175 ms at −30 mV and shorten at more positive or more negative potentials to ∼50 ms (838). In the rabbit ventricular cell, time constants for activation are longer, 500 ms at −40 mV to less than 100 ms at 0 mV (128). Compared with I Ks, these time constants are shorter, a finding on which the distinction of the two currents has been based. Deactivation does not follow the same pattern. It is fast in the guinea pig but much slower in the rabbit (128) and in the dog (324); in these preparations, the time course is composed of at least two exponentials, of which the first one is on the order of 0.5 s and the second on the order of 5–10 s at −40 mV (129, 324).
At the single-channel level, the mean open time is on the order of 3 ms. Closed time distribution is biexponential with values of 0.6 and 22 ms (at −100 mV and 100 [K+]o) (889).
In whole cell recordings, the time course of the macroscopic current shows saturation with no indication of a secondary decrease. The tail currents on hyperpolarization, however, are preceded by a “hook,” and the current temporarily increases before it declines in an exponential way (838, 889). The initial increase in outward current has been interpreted (889) as due to recovery from inactivation, a process supposed to be faster than the deactivation process. The hypothesis implies that the current during depolarization very rapidly undergoes inactivation, before there is any substantial activation (Fig. 6). It implies that steady-state inactivation extends over a voltage range that is quite positive. The consequence is that the current rectifies in the inward direction. Inactivation preceding activation has also been demonstrated in the expressed HERG channel (836, 932, 1141), which has been shown to be responsible for the I Krcurrent. Fast recovery from inactivation is the reason for an increase in the number of openings of the Kr channel upon repolarization during early diastole in nodal cells (466, 1033).
Inactivation of the I Kr channel is of the C type (836). The evidence is based on the following observations: 1) truncation of the NH2terminal has no effect on the phenomenon; 2) intracellular TEA has no effect, but external TEA reduces the current and slows inactivation; and 3) an increase of [K+]o slows inactivation in expressed channels and in AT-1 cells (1141). It is further of interest to mention that increases in [Mg2]i or [Ca2+]i, which generate inward rectification in other K+ channels, do not change rectification of I Kr (836).
B) Permeation. Although preferentially permeable to K+, the channel’s K+ selectivity is less pronounced than that ofI K1. Especially at lower [K+]o, theE rev is quite positive to the equilibrium potential for K+ (E K). The conductance falls at lower [K+]o(850, 889), a phenomenon explained as block by external Na+ or more pronounced inward rectification. C-type inactivation and thus inward rectification is enhanced at low [K+]o (1141).
In 150 mM [K+]o, the single-channel conductance is ∼10 pS in SAN (466), AVN cells of the rabbit (889), human ventricular cells (1033), and guinea pig atrial cells (433). A value of <2 pS can be extrapolated for normal Tyrode solution.
External bivalent and trivalent cations block the channel. Especially sensitive is the block by Co2+ and La3+ (10 μM) (265, 837). External Cd2+causes a positive shift of the activation curve (290) and a reduction in inward rectification, in this way increasing the current during a depolarizing pulse (749). The block by Ca2+ and Mg2+ is reduced by elevating [K+]o but does not change inward rectification (421, 422). The HERG gene is responsible for the I Kr protein expression (836, 932).
II) The slowly activated IKs current. A) Activation. The I Ks current only shows activation and no inactivation. Activation occurs over a broad range of depolarizing potentials. In many experimental conditions it is difficult to obtain a clear-cut saturation. Half-maximum values vary considerably from −13 mV (666) to 26 mV (38). Kinetics are slow; the time course of the rise in current is sigmoidal, whereas the decay of the tails is monoexponential at voltages negative to −50 mV but biexponential at more positive potentials (605, 666). Deactivation is slow in the guinea pig but relatively fast in the dog and the rabbit (128, 324, 613). At the single-channel level, kinetics are complex with many open and closed times (38, 242).
B) Permeation. The channel is less selective thanI Kr. The E revis more positive than that for I K1 and changes only by 49 mV for a 10-fold change in [K+]o (666). The fully activated current-voltage relation approaches linearity, except for the current in frog atrial cells where inward rectification is present (243).
Extracellular K+ has no direct effect on the conductance but affects the current through changes in the chemical gradient; thus in zero [K+]o, the current is greatly increased, especially in Na+-free conditions (850). Cobalt (1 mM) (265) and La3+ (at 100 μM or higher) (38,837) block the current. A rise in [Na+]i or [Ca2+]i enhancesI Ks (978, 728).
C) Molecular structure. Coexpression of the minK and the Kv.LQT1 generates a current with the characteristics of the cardiacI Ks (39, 835). The KvLQT protein has the classical constitution of voltage-activated K+ channels. The minK protein consists of only 129 or 130 amino acids and a single putative transmembrane domain, with the NH2 terminal turned to the external side of the membrane (956,1023). It plays an essential role in the function of Kv.LQT1 and can be considered a regulator protein.
c) the inward rectifier. The inward rectifier current (I K1) is the current responsible for maintaining the negative resting potential in cardiac cells; it also plays an important role during the final rapid repolarization during an action potential (894). The density of theI K1 is highest in the Purkinje and ventricular system (445), less in atrium (396); in the SAN, the I K1current is absent (459). A substantial increase of the current occurs during development from the neonatal to the adult stage (487, 1049).
I) Activation-deactivation: inward rectification. Is IK1 a background or voltage-activated current? TheI K1 current, the first K+current to be characterized in cardiac cells, was considered initially to be a time-independent background current. Its pronounced inward rectification provided an explanation for the existence of a long plateau in the cardiac action potential (see Ref. 133). With the improvement of recording techniques, it became clear thatI K1 showed time-dependent changes, which were analyzed as activation, deactivation, and inactivation (569). Upon hyperpolarization from a holding potential of −50 mV to −100 mV, a quasi-instantaneous current jump is followed by an exponential increase in inward current to a steady state (385, 569, 982). On depolarization, the reverse sequence is seen. This led to the hypothesis that the I K1 channel opens and closes by an intrinsic gating process not different from other voltage-operated channels. More recently, the time-dependent changes in the current or gating have been recognized to be generated by a time-dependent block-unblock by Mg2+(656, 1009) and polyamines (622,1093). Magnesium and putrescine ions are responsible for the very rapid phase, spermidine and spermine ions for the slower phase (622, 723) (Fig.7). The difference in rate corresponds to the difference in positive charge (263,1139). The block is voltage dependent, with an electrical distance of 0.3 for [Mg2+]i; polyamines seem to penetrate deeper in the pore (622,723). With time at the depolarized level, the block shifts from fast Mg2+ to slow polyamine block. On hyperpolarization, this is seen as an increase in the slower phase of activation (463). This new concept of activation being an unblocking is in accord with the information obtained on the molecular structure of the inward rectifier family, in which a voltage sensor or S4 segment is lacking. The I K1 molecule consists of only two transmembrane segments with a H5 or pore sequence in between (563, 954). TheI K1 channel can thus be considered a background channel, and a distinction between gating and permeation becomes less obvious. Whether on top of block-unblock there still exists an intrinsic gating mechanism is not fully resolved, and recent experiments on IRK1 channels expressed in oocytes have been explained in this way (893).
At the single-channel level, activation at hyperpolarized levels is correlated with a change from a lower to a higher conductance substate eventually to the fully open state and a prolongation of the open time: 10 ms around E K and 100 ms at 60 mV negative to the E rev (569,652, 756, 990).
Activation or unblocking from Mg2+ or polyamines depends on [K+]o, [K+]i, and the time spent in the depolarized state. In K+-free medium, noI K1 current can be recorded (see references in Ref. 123); the channel seems to remain blocked. The process has been called “K+ activation” (123,155).
Also, intracellular K+ interferes with activation: the higher [K+]i, the faster the current rise during activation (735). The K+ gradient or E K furthermore determines the position of the apparent activation curve on the voltage axis (167).
At hyperpolarized levels, the current after being “activated” frequently undergoes a secondary decrease or inactivation, which is due to block by external ions such as Na+, Mg2+, and Ca2+ (86, 386). At the single-channel level, the inactivation corresponds to a fall in open probability (831).
II) Permeation and selectivity. TheI K1 channel is very selective for K+ (498, 756, 830). Reported values for single-channel conductance differ appreciably and show a cluster at ∼22–28 pS (655, 671,831) and another at ∼45 pS (145 mM [K+]o) (305,468, 498, 569). In expressed channels, the variabitlity is even larger with values varying between 9 and 20 pS (1092). No explanation for this difference is available.
In guinea pig ventricular myocytes, the channel shows a fully open conductance of 22 pS and three substates of 7-pS difference (652, 655) or a fully open state of 28 pS and four substates (671). In the analysis of Matsuda (655), the channel is supposed to form a three-barreled structure; it provides an explanation for the occurrence of three substates.
Extracellular and intracellular Cs+ and Rb+ can be regarded as blockers of the channel. The block is voltage dependent with an electrical distance of 0.6 and 0.14, respectively. The block by Cs+ results in a highly flickering mode of activity (498, 830) and in 20% the channel shift to a lower substate (653, 655). External Ba2+ is a very efficient voltage-dependent blocker; it shortens open time and decreases the number of openings (51, 498, 830).
The effect of protons is discussed in sectionii A2.
D) LIGAND-ACTIVATED K+ CURRENTS.
I) ACh-induced K+ current. Slowing of the heart beat is caused by activation of a specific K+channel, the ACh-induced K+ current (I KACh), in the SAN (see review in Ref.1095). In lower vertebrates and birds, the current is present in atrial as well as ventricular cells (381). In the mammalian species, the current is expressed in atrial cells, AV node cells, and Purkinje cells; in ventricular cells, the channel is not present in all species but has been described for the ferret (95), the rat (676, 1119), the dog (1142), and the human (558).
A) Activation. Activation of I KACh occurs upon binding of ACh to the M2 muscarinic receptor. The receptor is directly coupled to the K+ channel via a guanine nucleotide-binding protein or G protein, characteristically inhibited by pertussis toxin. Activation of the channel is due to binding of the βγ-subunits to the channel as demonstrated in isolated native (1126) and cloned channels (800). Gβγ binds to the COOH terminal of the channel (456).
Activation of the G protein connected to the K+ channel is also possible through stimulation of other receptors: adenosine (P1 receptors) (54, 926), external ATP (P2 receptors) (665), somatostatin (602), calcitonin gene-related peptide (522), endothelin (523), a serum albumin-associated factor of phospholipid nature (105), and sphingosine-1-phosphate (1019).
The presence of an agonist, however, is not absolutely necessary, and background openings of the channel have been observed in different situations. Such activity has been related to the presence of ATP, of increased [Na+]i, or of leukotrienes. Leukotrienes act on the G protein and stimulate the exchange of GDP for GTP (570). For activation by ATP, different mechanisms have been proposed; local GTP formation from GDP through activation of nucleotide diphosphate kinase (NDPK) (395,492), an unknown effect of NDPK (1122), phosphorylation (531, 930), which makes the channel sensitive to [Na+]i(943), or generation of phosphatiditylinositol 4,5-bisphosphate (PIP2) by ATP-dependent lipid kinases (440).
B) Permeation and rectification. The channel is very selective for K+. The conductance is highly sensitive to extracellular K+ concentration. In symmetric conditions (150 mM [K+]), the conductance is 40–44 pS. The open times are very short (1–3 ms), occurring in bursts (829,924). Open probability in human atria is very variable, from 0.03 to 0.3 in the presence of 10−5 M ACh (396).
At the whole cell level, the current is characterized by inward rectification. The current activates on hyperpolarization and deactivates incompletely on depolarization (731). The phenomenon resembles the time-dependent changes in theI K1 current, although they are of a slower nature and smaller in amplitude. Similar to theI K1 current, rectification is explained by block of the open channel by intracellular Mg2+ and polyamines (1127). The I KAChis blocked by classical K+ channel blockers such as Cs+ and Ba2+ (130,698, 1170).
Fade or desensitization is the process by which the current through the channel decreases, although the agonist concentration remains constant. Upon washout of the agonist, the response to a second exposure remains temporarily depressed (131). In the heart, fade is more pronounced for the KACh channel and less for other channels coupled to M2 receptors (I Ca,I f, and I Ks) (95, 131, 430).
For relatively short exposures of ACh, fade occurs in two phases: a rapid phase (up to 30 s) (524, 571) occurs at the channel or G protein level followed by a slower one (up to 3 min) in which the receptor is involved (1061,1171). At the single-channel level (524), the fast phase is characterized by a shortening of the open time from 4.3 to 1.1 ms; it is probably due to dephosphorylation of the channel or of the G protein (903) or inhibition of the ATP effect (531); the desensitization is heterologous and is not restricted to a single type of receptor. The slower phase is characterized by a reduction in the frequency of opening (524) and is probably related to a phosphorylation of the muscarinic receptor by a receptor kinase; this kind of desensitization is thus homologous (1171). In accordance with this explanation, the fast phase still occurs when the receptor is short-circuited by acting directly on the G protein via guanosine 5′-O-(3-thiotriphosphate) (430) or arachidonic acid metabolites (860), but the slow phase is absent.
A third phase of fade occurs when cells have been exposed to the agonist for many hours. The underlying process in that case seems to be internalization (106).
C) Molecular structure. From a structural point of view, a combination of two inward rectifier channels is proposed to be responsible forI KACh: the Kir3.1 and Kir3.4 (142, 563). Coexpression of the two proteins in combination with the M2 receptor elicits an activity that highly resembles KACh (562).
II) The ATP-inhibited K+ channel: KATP. The KATP channel, first described in heart cells (729), is present in many other cell types (465). In heart, it is expressed in ventricular, atrial, as well as nodal cells from different species, including the human heart (32). In heart cells it seems to exert a protective role during an ischemic insult. By shortening of the action potential, generation of inexcitability and shifting the E m closer to the equilibrium potential for K+, excessive loss of K+ is avoided. Activation of the channel also seems responsible for preconditioning or protection against a second insult.
A) Activation. In inside-out patches, a K+ channel with a high conductance is activated when the cytoplasmic [ATP] is decreased below a critical concentration. ATP normally decreases theP o. The K ddetermined in inside-out patches is ∼0.1 mM (540,725, 729). This value should be considered a mean value. For individual channels, the K dmay differ by as much as three orders of magnitude and ranges from 9 to 580 μM (279). Hill coefficients vary between 1.0 and 5.0. Anionic ATP as well as MgATP are able to inhibit the cardiac channel. The sensitivity to ATP is higher in the presence of free Mg2+ (278). Kinetics of the single-channel activity are complex with multiexponential distributions for open and closed times and bursts (788). The burst duration is inversely related to [ATP], and intraburst kinetics depend on Mg2+ concentration (1186).
The K d value is much lower than what is considered the normal [ATP] of 5 mM or more. How is it then possible to activate the channel? Two considerations should be made: the local [ATP] seen by the channel can be lower than the bulk concentration, and the sensitivity to block by ATP is variable.
1) The subsarcolemmal [ATP] that is seen by the channel is much lower than the bulk value when the Na+-K+-ATPase (781,998) or the adenylate cyclase (435,858) is activated, or when anaerobic glycolysis is blocked (1081).
2) The sensitivity to block by ATP can be reduced by acidosis, dinucleotides, an increase in lactate or decrease in taurine, congestive heart failure, an increase in [Mg2+]i or [Ca2+]i and polycations, receptor activation, changes in the cytoskeleton, and, most important, changes in PIP2 concentration at the intracellular side of the membrane.
Intracellular acidosis, although reducing single-channel conductance (266), increasesP o as long as ATP is not totally depleted (193, 590). The concentration-response curve is shifted to higher [ATP] (560). The presence of Mg2+ is required and suggests a change in phosphorylation of the channel (193). When acidosis is too pronounced and the pH falls below 6.0, P o is markedly reduced (560).
Dinucleotide and MgADP shift the inhibitory curve to higher [ATP]; the mechanism is not of a competitive nature, since both substances bind at different sites (494, 729,970). Adenosine 5′-diphosphate can also reactivate the channel after rundown due to the absence of ATP or presence of high [Ca2+]i. Dinucleotides thus exert effects similar to K+ channel openers.
Higher concentrations of ATP are needed to block the channel when the concentration of Mg2+ or Ca2+ and polycations such as protamine, polylysine, and poly-arginine are small. These cations are supposed to neutralize negative charges at the mouth of the channel; in their absence, access of ATP to its binding site is antagonized (212).
Activation is further facilitated by phosphorylation and G protein interaction. Protein kinase C-induced phosphorylation (439, 610) occurs after stimulation of α1, P1, P2, M1, and M2 receptors (608, 619, 1062). A direct coupling to the channel via a G protein is present for P1 and M2 receptors (469, 532, 540,969).
The functional availability of the channel is dependent on the presence of an intact cytoskeleton. Actin filament-depolymerizing agents (cytochalasin and desoxyribonuclease I) accelerate rundown in excised patches (968), whereas actin stabilizers (phalloidin) or PIP2 (inhibitor of F-actin severing protein) inhibits rundown (306). Disruption of actin filaments also impairs the sulfonylurea inhibition of the channel (98,1151).
Recently, a dramatic decrease in sensitivity to ATP of Kir6.2 expressed channels has been described after addition of micromolar concentrations of PIP2. The hypothesis is formulated that the COOH terminal of the inward rectifier binds to PIP2 in the membrane and keeps the channel activated unless [ATP] is raised to the millimolar range (47, 904).
B) Permeation, conductance, and rectification. The ATP-dependent channel shows a high conductance of 80 pS in symmetrical [K+] conditions (494,729). It is very selective for K+, with a Na+ permeability of only 0.01 relative to K+. The conductance depends on [K+]o(724). In physiological [K+]o, the conductance falls to 25 pS (277).
In the presence of high [K+]o, the channel shows inward rectification. At normal physiological [K+]o, however, the current shows no rectification or even outward rectification up to 0 mV (722, 934) (Fig.8 A). Outward rectification is also characteristic for the current generated during metabolic inhibition (462), hypoxia (61), and in the presence of potassium channel openers. Interpretation of rectification in those cases should take into account the simultaneous inhibition of the I K1 current. At positive voltages, a weak inward rectification is seen, which is caused by Mg2+block and not by spermidine or spermine ions (1127).
External cations like Ca2+, Sr2+, Ba2+, Na+, and H+ block the channel, but efficiency is small. Zn2+, Cd2+, and Co2+ with K d values of, respectively, 0.46, 28, and 200 μM are the most efficient (573). The block is voltage independent and reduced in ATP depletion. The channel is blocked by intracellular Cs+(490).
C) Molecular structure. The cardiac channel is of heteromultimeric nature (455), composed of a protein from the “inward rectifying” family, the Kir6.2 and the sulfonylurea receptor (SUR-2A), a protein from the ATP-binding cassette superfamily (4); to this family also belong the cystic fibrosis transmembrane conductance regulator (CFTR; see sect.ii A4), the P-glycoproteins, and the multiple drug resistance receptor (MDR).
The cardiac type is rather insensitive to spermine and spermidine ions (1127), and its block by Mg2+ is weak. TheK d is 2.0 mM at 0 mV and 0.29 mM at −40 mV; the Hill coefficient is 1. The SUR molecules are characterized by multiple membrane-spanning domains and two nucleotide-binding domains (NBD). They act as receptors for sulfonylurea drugs and K+ channel openers. Activation by MgADP also occurs at the SUR, probably at one of the NBD. Both ATP (338) and GTP (989), on the other hand, inhibit the channel by binding to the Kir6.2 subunit. The expression of SUR-1 facilitates the incorporation of Kir6.2 in the plasma membrane of HEK cells (484).
III) The Na+-activated K+channel. The Na+-activated K+ channel belongs to the ligand-activated channels and is activated by [Na+]i. It was first described in cardiac cells (497).
A) Activation by [Na+]i. The channel is selectively activated by [Na+]i; other substitutes such as Li+ are inefficient (497,624). Relatively high concentrations of [Na+]i, however, are required (K d 66 mM) (497). A similar sensitivity has also been found by other authors (625,804). The sensitivity to [Na+]i is variable from one patch to another and in the same patch for different channels. TheP o is independent of voltage between −100 and +60 mV and only declines for very negative potentials (834). The P o is greater in cell-attached than in inside-out patches (804); the response curve for [Na+]i is shifted to lower concentrations, and maximum activity is increased (804).
The channel’s activity is characterized by two open states with time constants in the order of 0.5 and 10 ms and three closed states with time constants in the order of 0.5, 2, and 20 ms (497,624). Sometimes the channel may enter a quiescent state for many seconds.
B) Permeation. The KNa channel shows a very high conductance (497, 625, 834,1068) with values up to 200 pS. Its actual value depends on the ratio of intracellular to extracellular [K+], on the voltage, and on the presence of other ions, such as H+, Na+, and Mg2+. Permeation shows Goldman-type behavior with a current-voltage relation that is linear, inward rectifying, or outward rectifying depending on the ratio of intracellular to extracellular [K+] (625) (Fig. 8 B). Under physiological conditions, rectification is outward and predicts the channel to carry large currents during the action potential.
Selectivity of the KNa channel for K+ is high. The change of the E rev with different K+concentrations follows the predicted values of the Nernst equation (497, 625). Substitution of K+ by Rb+ or Cs+ results in block of the channel (625);P Na/P K has been estimated to be 0.02 (497). Increases in [Mg2+]i and [Na+]i (1068) cause the channel to shift to a lower conductance state, especially at positive potentials (625).
IV) K+ channel activated by fatty acids and amphiphiles. Arachidonic acid, unsaturated fatty acids, and phospholipids activate K+-selective channels in neonatal rat atrial cells (527) and in adult rat atrial and ventricular cells (528) (Fig.9).
A) Activation. Activation occurs upon addition of fatty acids to either side of the membrane but is more efficient if applied from the cytosolic side. Available evidence suggests that unsaturated fatty acids with two double bonds are required for efficient activation. The channel shows a slight voltage dependence with higher activity at depolarized levels. Arachidonic acid activates the channel directly and not via one of its metabolites, since activation occurs in the presence of inhibitors of cyclooxygenase, lipoxygenase, or epoxygenase. The activity of the fatty acid-induced channel is increased by stretch (525), but stretch as such can also activate the channel, and this effect is not due to stretch-induced release of fatty acids. Once activated by fatty acids, activity can be increased by low pH (525) (Fig. 9). A channel with slightly different rectification characteristics is activated by phospholipids (1050).
B) Permeation. The channel is rather selective for K+. The channel activated by arachidonic acid shows outward rectification; the channel activated by phospholipids has a linear current-voltage relation (527, 1050). Outward rectification is due to a greater single-channel conductance and a higherP o at depolarized levels. The single-channel conductance, 94 pS in 140 mM symmetrical conditions, is greater than that of the KATP channel. The channel is thus different from KATP; activity of the latter is actually inhibited by fatty acids. Openings of the channel activated by fatty acids (KFA) occur in bursts. The current is blocked by 1 mM Ba2+ but not by TEA, 4-AP, quinidine, or apamine.
As possible molecular substrate a two-pore-forming K+channel can be proposed. The channel expressed in neuronal tissue is activated by arachidonic acid (280); other representatitives of this family, however, are insensitive to arachidonic acid (530).
4. Cl− channels
Four different types of Cl− channels have been decribed in cardiac cells (444): a channel activated by PKA-dependent phosphorylation, which is probably identical to the channel activated by PKC, a [Ca2+]i-activated channel, a [ATP]o-activated channel, and a stretch- or swelling-induced channel. All channels, except the PKA-induced Cl− channel, are blocked by disulfonic stilbene compounds (DIDS and SITS); the PKA-activated channel is blocked by 4,4′-dinitrostilbene-2,2′-disulfonic acid (DNDS). They show pronounced outward rectification in symmetrical Cl− conditions; this rectification is much less present for the PKA-induced current. Two channels have been cloned and expressed: the CFTR channel, which is responsible for the PKA-stimulated channel (378,596, 710), and ClC-3, which accounts for the swelling-induced current (241).
Because the E Cl is positive to the resting potential, activation of Cl− current causes depolarization of the resting potential but accelerates repolarization early during the action potential (237). The physiological role of Cl− currents, however, cannot strictly be delineated, the reason being that the condition to activate the currents may only be present in physiopathological conditions: secretion of catecholamines for the I ClPKA, cell swelling by osmotic forces for I Clswell, secretion of ATP in the extracellular medium for I ClATP. Even the I ClCa that is activated by a physiological processes, i.e., the release of Ca2+ from the SR, does not seem to be expressed in all cells (absent in human atrial cells, Ref. 604). The density and expression of the currents is very variable; in general, I ClPKA is better represented in ventricular cells andI Clswell in atrial cells.
A) PKA- AND PKC-DEPENDENT CL− CHANNEL.
The PKA-activated Cl− current (for review, see Ref.308) is typically present in the ventricle, less in the atrium; it has been described for the ventricle of the guinea pig (34,250, 660, 1012), rabbit (especially in the subepicardium, Ref. 383), and cat (1178) and in the human atrium and simian ventricle (1071). It is not present in the rabbit SAN or atrium, the dog atrium and ventricle, or the mouse and rat ventricle (437).
I) Activation. The PKA-dependent channel is typically activated after β-receptor stimulation (308). The biochemical pathway involves phosphorylation of the regulatory domain of the CFTR protein. This protein consists of two membrane domains each of six transmembrane segments, two NBD and one regulatory domain, all cytoplasmatic. β-Receptors and histamine receptors (383,447) are positively coupled to PKA; negative coupling occurs via M2 (964), endothelin A (475), and ANG II receptors (733). The nitric oxide (NO) synthase pathway does not seem to be involved (1168). Activation via cAMP is facilitated by simultaneous inhibition of tyrosine kinase (901) and previous exposure to ACh (1167).
Phosphorylation results in a change from silent to active channels (248) and modulates the kinetic behavior (446, 448). The regulatory domain of the channel can be partially or highly phosphorylated, each of these steps corresponding to two types of activity. In a first phosphorylated stage, the channel binds ATP at NBD1, followed by hydrolysis (this step can be reversed by okadaic acid-sensitive phosphatase). With Pi leaving the protein and ADP remaining on the NBD1 site, the channel is activated and shows brief openings; the channel closes when ADP is released from the channel. A second phosphorylation (reverse reaction is okadaic acid insensitive) makes it possible for the molecule to bind a second ATP molecule at site NDB2. Binding of ATP to this second site and its hydrolysis inhibits the release of ADP fromsite 1 and stabilizes the channel in the open state. The activity is characterized by long openings (1 s) and closures of 0.2 and 10 s (250). As long as the hydrolysis products stay on the protein, the channel is locked in the open state with aP o close to 1.0. Egress of the hydrolysis products suppresses the stabilization and causes channel closure. Inhibition of phosphatase 1 and 2A by calyculin A increases markedly activation of the channel by forskolin (1138).
II) Permeation and conductance. The single-channel conctance is between 7 and 14 pS (710). In asymmetric [Cl−], the current-voltage relation is outwardly rectifying for the whole cell as well as for the single channel. Current can be carried by other anions, such as ATP, but the conductance is small (871). In ischemia, it may be responsible for the release of ATP from the cell. The permeability sequence for small anions is Br− > Cl−> I− (710, 744,1012). The channel is slightly permeable to cations with aP Cl/P Na of 10–20 (149).
III) Are PKA- and PKC-induced currents different currents? Activation of PKC opens a Cl− channel in guinea pig (168) and feline ventricular myocytes (1178), which shows single-channel conductance (9 pS) and kinetics identical to the PKA-activated channel (168, 1178). The currents are not additive and are blocked by the same substances, i.e., 9-AC and DNDS but not by DIDS. The question remains whether the permeability sequence is different (1052). The regulatory domain of the cloned molecule contains consensus phosphorylation sites for both kinases. It is interesting to note that in some preparations in which PKA is ineffective in opening a Cl− channel, activation occurs via external ATP and PKC (500, 597,664). In guinea pig ventricular myocytes, PKC activation alone did not activate the channel, but it potentiated the channel supramaximally stimulated by isoproterenol (683). Activation of PKC may also be the mechanism underlying the occurrence of a Cl− current after stimulation of AT1receptors by ANG II (693), since intracellular Ca2+ is required to elicit the current.
B) SWELLING-INDUCED OR STRETCH-ACTIVATED CL−CHANNEL.
A volume-activated Cl− channel is present in atrial cells of the rabbit (357), guinea pig (902,1012), dog (929, 992), and human (745, 827) and in cultured cells of the chick embryo (1176) and possibly in neonatal rat cells (186).
I) Activation. Activation occurs by exposure of cells to hypotonic solutions, by pressure through the pipette, or after incorporation of an anionic amphipath in the outer leaflet of the membrane that increases the membrane curvature (992). Although stretch may be accompanied by changes in the cytoskeleton, activation does not require the integrity of the cytoskeleton. Application of colchicine, which disrupts microtubules, or cytochalasin, which disrupts F-actin, did not change activation by swelling. Activation is not immediate but takes minutes to develop (929). A phosphorylation by tyrosine kinase has been proposed as the underlying process (745,974), since the process is blocked by inhibition of this enzyme. It is known that cell stretch can lead to activation of tyrosine phosphorylation (820). Under isotonic conditions, the channel is active in a small percentage of cells (240). Activation does not require stimulation of PKA or a rise in [Ca2+]i (357,992). Once activated by volume, forskolin amplifies the current (745), probably by a direct effect of cAMP. Protein kinase C stimulation does not activate but rather inhibits the current (238, 241). Some of these channels are voltage sensitive and show inactivation on depolarization (rat, Ref. 186; guinea pig ventricle, Ref. 902).
II) Permeation. The volume-activated channel is outward rectifying with a conductance of 28 pS at theE rev and 50–60 pS at positive potentials. The P o is elevated (0.8); DIDS, SITS, and tamoxifen block the channel (239, 240,1012). The channel is permeable to anions with a permeability sequence of I− > NO3 −> Br− > Cl− > aspartate (1012). Taurine and inositol efflux occurs upon cell swelling probably via this pathway (473). In the ClC-3 clone, one amino acid mutation (N579K) is sufficient to change the permeability preference from I− to Cl−(241). In neonatal rat, a channel has been described (186) with a very large channel conductance of 400 pS and a P Cl/P Na of 25; this channel may be different from the other described above.
C) INTRACELLULAR CA2+-ACTIVATED CL−CURRENT.
A transient and intracellular Ca2+-dependent outward current, carried by Cl−, has been demonstrated in rabbit myocytes (1192) and is calledi to2. The current exists in rabbit Purkinje cells (910), rabbit atrium (604), canine ventricular (1191) and canine atrial cells (1165), ferret ventricular myocytes (985), and cultured chick heart cells (618). It has not been found in the human atrium (604).
I) Activation. During the cardiac action potential, a Cl− current is activated by Ca2+ released from the SR, subsequent to entry of Ca2+ via L-type Ca2+ channel. Evidence supporting this hypothesis is based on the observation that the current is prevented by block of the L-type Ca2+ channel, block of the SR Ca2+release channel, and intensive buffering of intracellular Ca2+, whereas stimulation of Ca2+ influx by isoproterenol has the opposite effect. Exposure to caffeine, which releases Ca2+ from a preloaded SR store, elicits a current even at negative E m values, suggesting that the essential phenomenon leading to activation is the increase in local [Ca2+] and not the depolarization (912). Activation and inactivation curves have been described (513); they should not be regarded as the expression of voltage dependence of the Cl− channel but generated mainly by the voltage dependence of the Ca2+ current.
In Purkinje cells of the rabbit (910) and in ferret ventricular myocytes (985), the Cl− current is much shorter than the intracellular Ca2+ transient and already declines before the Ca2+ transient reaches its peak value (Fig. 10). The rapid fall in current is not due to some type of accomodation or “inactivation.” Steady activation with no time-dependent decline in activity is obtained in canine permeabilized ventricular myocytes exposed to variable [Ca2+] (513); in inside-out patches, single-channel currents did not show any decrease with time (169). A possible explanation would be that the local subsarcolemmal Ca2+ transient seen by the channel is much shorter than the cytoplasmic Ca2+ transient. Subsarcolemmal [Ca2+] as seen by the Na+/Ca2+exchanger, however, does follow the time course of the intracellular Ca2+ transient (780). At the present time, the simplest explanation is to assume that activation of the channel requires an elevated [Ca2+] (high threshold), such that the time course of the current is only determined by the short peak of the transient. In Ca2+ overload, the Cl−current is prolonged and shows much slower kinetics, although the duration remains shorter than the intracellular Ca2+transient (747). It is important to note that the current can be activated at negative potentials (912,1191) and thus may play a role in the genesis of delayed afterdepolarizations and arrhythmias.
II) Permeation. A permeability study (513) shows the following sequence: SCN− > I−> Br− > Cl−; the channel is, however, impermeable to large organic anions such as aspartate (used as substitutes to change the equilibrium potential). The current-voltage relation has been described as outward rectifying (1192) or linear in symmetric conditions (910). In canine ventricular cells (169), the single channel has a small conductance of 1.5–2 pS and shows long openings. The apparent K d is rather high, 150 μM; it should be mentioned, however, that [Ca2+] seen by the channel can be very different from the bulk concentration. The current is blocked by SITS and DIDS (169,910, 1192), by DIDS and niflumic acid (594), and by glibenclamide (1133). No information is available on the structure.
D) EXTRACELLULAR ATP-ACTIVATED CL− CURRENT.
I) Activation. Purinergic activation of a Cl−current has been demonstrated in atrial cells of the guinea pig heart (664) and in ventricular cells of the mouse (597) and the rat (500). Activation occurs with ATP, adenosine 5′-O-(3-thiotriphosphate), nonhydrolyzable ATP analogs, ADP, and AMP. Adenosine activates the channel in guinea pig but not in the mouse or rat. With respect to the receptors, no clear distinction is possible between P1and P2, since ATP, adenosine, and AMP were equally active in the guinea pig (664). In rat ventricle, the receptor seems to be P2 (500). The channel is insensitive to internal cAMP or [Ca2+]i. Activation is slow and takes seconds, suggesting the intervention of an intracellular messenger. Stimulation of phosphoinositol metabolism, with secondary activation of PKC by diacylglycerol (1125) is a possibility.
II) Permeation. The current shows outward rectification in normal solutions; in symmetrical conditions, the current-voltage relation is linear. Block occurs with DIDS (500); DIDS however also blocks P2 receptors (597).
5. Nonselective cation channels and the If channel
In the heart, a number of nonselective cation channels (NSC) are activated by stimuli that also activate Cl− currents, such as a rise in [Ca2+]i, an increase in [ATP]o, and stretch. Others are activated by oxygen radicals and amphiphiles, by depolarization (sustained inward current or I st), and by hyperpolarization (I f current). Finally, a NSC current has been described with no known activation mechanism; it is spontaneously active and can thus be regarded as a background current. They all differ by permeability and block characteristics.
Both I st (only described in the sinus) andI f (present in nodal, Purkinje, and plain atrial and ventricular cells) play an important role in pacemaking under physiological and pathological conditions. The other NSC currents seem only to be activated in pathological conditions (increase in [Ca2+]i, [ATP]o, stretch, radicals, and amphiphiles). Because these currents carry inward current at negative E m values, they cause the resting potential to be more positive thanE K and thus favor K+ loss through K+ channels, of which the conductance is increased in ischemic conditions.
A) INTRACELLULAR CA2+-ACTIVATED CATION CHANNEL.
Two types can be distinguished: one is activated by [Ca2+]i alone, and the other, in addition, requires depolarization.
A) Activation. Activation of the current occurs after exposure of the cell to Na+-free medium (251), high [Ca2+]o, low [K+]o solution (368), following intracellular injection of Ca2+(650), application of caffeine, cell dialysis with solutions containing micromolar Ca2+ concentrations, and exposure to intracellular oxygen free radicals (472). TheK d for Ca2+ in inside-out patches is 1.2 μM with a Hill coefficient of 3. The channel activity is insensitive to large voltage changes (172,251).
B) Permeation. The conductance of the single channel is 30–40 pS in cultured neonatal rat cells (172) and 15 pS in the guinea pig (251). The current-voltage relation is linear and shows a reversal at zero or slightly negativeE m values. The channel is equally permeable to Na+, K+, Li+, and Cs+ but excludes anions. Kinetics of the channel are voltage independent but largely determined by [Ca2+]i. At 10 μM, mean open times are 3.8 and 140 ms, and mean closed times are 1.8 and 15 ms (251). The long open time increases with [Ca2+]i while closed times decrease; sometimes openings of many seconds are seen.
II) Second type. A second type of Ca2+-activated cation channel, characterized by completely different activation and conductance parameters, has been found by incorporation of sarcolemmal vesicles isolated from adult canine ventricle in lipid bilayers (410). The channel has not been found in cell-attached or inside-out patches, a result that could be due to the channel being situated in the clefts of the sarcolemma.
The channel is activated by a combination of depolarization and rise in [Ca2+]i. In the presence of Ca2+ below 100 nM, the channel is activated for potentials at the cis-side (cytoplasmic side) positive to −60 mV. TheK d for Ca2+ is <1 μM, and the Hill coefficient is 2. Kinetics can be described by two open times (5.6 and 22 ms) and two closed times (9 and 29 ms). The current-voltage relation is linear, and the channel is equally permeable to Na+ and K+. TheP Ca/P Na is 0.2.
B) EXTRACELLULAR ATP-ACTIVATED TRANSIENT CATION CHANNEL.
I) Activation. A transient, rapidly desensitizing inward current is activated by [ATP]o in frog atrial myocytes (301), in rabbit SAN (899), in rabbit atrial cells and guinea pig atrial and ventricular cells (411, 752), and in rat ventricular myocytes (1180). The K d for [ATP]o is 56 μM (301). Adenosine, AMP, and ADP are ineffective; channel activity is blocked by theophylline. On the basis of the action of agonists and antagonists, the receptor involved can be classified as P2x. The channel is not sensitive to [Ca2+]i(1180).
II) Permeation. The current-voltage relation is linear (49, 301) or slightly inwardly rectifying (411); the E rev is close to 0 mV, suggesting a nonselective permeability. Activation of the channel leads to elevation of [Ca2+]i(415). The current can be blocked by external Cd2+ (1180) but is insensitive to external Co2+ or Ni2+. A P2x receptor channel has been cloned from rat vas deferens (1008). Functionally, it resembles the cardiac channel activated by [ATP]o. The structure has similarities with the inward rectifier channel, and its single-channel conductance is 15 pS at −100 mV. It is permeable to Na+, K+, and Ca2+(P Ca/P Na is 4.8), Tris, and N-methyl-d-glucamine (NMDG) (1008). Inward rectification of the expressed channel is due to voltage-dependent gating and reduction in channel conductance (1183).
C) STRETCH-SENSITIVE CATION CHANNEL.
I) Activation. Stretch- or swelling-activated NSC currents have been described in rat atrial cells (526), neonatal rat cells (188), cultured chick hearts (815), guinea pig ventricular cells (841), and dog ventricular myocytes (162).
II) Permeation. The channel in rat atrial cells is permeable to monovalent cations and Ca2+(P Ca/P Kestimated to be 0.9) (526) and potentially plays a role in generating Ca2+ overload and triggering atrial natriuretic factor secretion. The current-voltage relation is linear with a single-channel conductance of 21 pS (526).
Gadolinium ions block the cardiac channel and the accompanying volume changes in dog ventricular myocytes (162) and inhibit swelling induced by osmotic gradients in rabbit ventricular cells (946). Negative results were obtained in the rat atrial channel (526). Stretch-induced increase in [Ca2+]i, probably due to activation of the stretch-induced channel, is blocked by streptomycin (313).
A channel with a greater single-channel conductance of 120 pS, a preference of K+ over Na+ of 3.4, and impermeability to Ca2+ has been described in neonatal rats (188). The channel is insensitive to TEA (20 mM), 4-AP (1 mM), TTX (100 μM), or Cd2+ (1 mM).
D) NSC ACTIVATED BY AMPHIPHILES.
In guinea pig ventricular cells (631) and in rabbit ventricular cells (114), a NSC channel is activated by amphiphiles. In the guinea pig, it shows a permeability sequence of Cs+ > K+ > NMDG+> Na+ > Ca2+. In the rabbit, the current shows some inward rectification and reverses at −18.5 mV in a solution containing a high Na+ concentration. It is insensitive to TTX (10 μM) or Cd2+ (100 μM) but blocked by La3+. Gadolinium blocks the channel in the rabbit but not in the guinea pig.
E) NSC ACTIVATED BY OXIDATIVE STRESS.
A NSC is activated by exposure of guinea pig ventricular myocytes to extracellular radicals; an increase in [Ca2+]i is not required (472). A similar current is activated by other oxidizing agents such as thimerosal and diamide (667,881) and singlet oxygen (965). The channel activated by internal free radicals is blocked by Ni2+ and Gd3+ (472).
F) NSC BACKGROUND CHANNELS.
An inward background current carried by cations has been described in SAN cells of the rabbit (356, 1038), in atrial and ventricular myocytes of the guinea pig heart (543), and in human atrial cells (191). In feline atrial myocytes, such a current is seen after addition of ACh (1062). In the SAN, the permeability sequence is K+ >> Rb+ > Cs+ > Na+ > Li+. The NMG ion is not permeant, and 10 μM Gd3+ partially blocks the channel (356). In atrial and ventricular cells, the presence of a background current causes the resting potential to be less negative than E K; in nodal cells it plays a role in pacemaker activity.
G) SUSTAINED INWARD CURRENT.
I) Activation, inactivation. In SAN cells of the rabbit (346) and the guinea pig (345), an inward current is activated on depolarizations from −80 to −60 mV and more positive levels. In the potential range where spontaneous activity is seen, the current appears as a sustained inward current. Inactivation occurs slowly for relatively large depolarizations.
II) Permeation. The channel is permeable to monovalent cations; the current is nearly suppressed by depletion of extracellular Na+. In contrast, it is not decreased by reducing [Ca2+] from 1.8 to 0.1 mM. Except for its permeation characteristics, the channel resembles the L-type Ca2+channel; it is largely increased by isoproterenol and is blocked by DHP. The current is supposed to play an important role as inward current in pacemaking.
H) PACEMAKER CURRENT.
The heartbeat in normal conditions finds its origin in the SAN. The electrical activity in the node is characterized by the existence of diastolic depolarization. Many currents participate in this process (730). One of the important currents, the pacemaker current I f, is activated during the repolarization phase of the action potential and generates an inward current. The I f current has been described primarily in SAN, AVN, and Purkinje cells (220) but is also present in atrial (124, 246,779, 973) and ventricular cells (139, 1158). In the rabbit, the density decreases and the voltage dependence shifts in the negative direction from newborn to the adult stage (2, 803). The reverse shift occurs in primary cultures of rat ventricular cells and is accompanied by dedifferentiation (267). Recently, the channel protein has been cloned; the structure resembles that of other nucleotide-activated channels (318, 623).
I) Activation. Upon hyperpolarization of the SAN membrane from −40 mV to more negative levels, an inward current is activated that slowly increases to a steady level (see Ref. 730) (Fig.11). On return to depolarized levels, the current is deactivated. Time constants are of the order of seconds at depolarized levels but become shorter with hyperpolarization. In SAN cells, the activation curve extends from −40 mV up to −100 mV, with a half-maximum value at −60 mV and a slope of 10 mV. It is not changed by [K+]o (293). In Purkinje (115) and ventricular cells (1156), the activation curve lies more negative than in the SAN. The difference in voltage dependence is translated in the diastolic depolarization occurring at more hyperpolarized levels. The difference in activation voltages may be due to differences in sympathetic tone and adenylate cyclase activity. In ventricular cells, the current may become important in pathological conditions as a consequence of the shift in the activation curve, secondary to catecholamine secretion or to dedifferentiation (267). Important modulatory processes by neurotransmitters are described in section iii,A, D, and E.
II) Permeation. The fully activated current-voltage relation is linear and reverses at −10 to −20 mV, suggesting a nonselective channel, which passes K+ and Na+ions (420). Permeability to other cations such as Li+, Rb+, and Cs+ is small or nonexistent (217). In the range of diastolic potential, the current is inward and carried by Na+. At the single-channel level, small currents have been recorded revealing a conductance of only 1 pS (218) (Fig. 11). Because the patches contained many small channels, information on the kinetics of single-channel activity could not be determined.
The amplitude of the fully activated current is sensitive to [K+]o (217,646); an increase of [K+]oenhances the inward current through the channel (293). In K+-free medium, the inward current is practically reduced to zero. The channel still undergoes activation, and upon depolarization a large outward current occurs. The voltage dependence of the activation and the kinetics of the macroscopic current, however, are not changed with [K+]o, suggesting no change in P o but a modulation of the single-channel conductance. Selectivity for Na+ is increased in low [K+]o (reversal shifts in the depolarized direction) (420).
A rise of [Ca2+]i from 10−10 to 10−7 M increases the amplitude of the I f current and shifts the activation curve in the positive direction by more than 10 mV (354). From a simple screening effect, the opposite result is expected; the mechanism therefore is probably indirect (1172).
The permeation process is affected by the anion species in the extracellular medium. Substitution of Cl− by organic anions results in a decrease of the macroscopic conductance without change in reversal and without shift of the activation curve (293). The channel is thus not permeable to anions, but its conductance is modulated by the anion species.
The current is inhibited by Cs+ in a voltage-dependent way. Block by 2–5 mM is efficient and complete at hyperpolarized levels but incomplete close to the reversal (217). The effect of Cs+ is asymmetric; inward current is specifically blocked by external Cs+ but not by internal Cs+(when [Na+]o is elevated ) (293, 420).
6. Electrogenic exchangers
A) NA+/CA2+ EXCHANGE.
The Na+/Ca2+ exchange (for review, see Ref. 765) plays an important role in the regulation of [Ca2+]i, in the excitation-contraction coupling process (increase of tension as well as relaxation), and in determining the time course of the action potential (56, 247) and of the electrical restitution following a stimulus (482).
The density of the exchanger in the heart differs with species (875). Its density is larger in the newborn (26), and with development, it shifts to the t tubules (294) or the plasma membrane overlying the junctional sarcoplasmic reticulum. This conclusion has been challenged (517). The expression is enhanced in failing hearts (387) and in conditions of increased [Na+]i after ouabain or thyroid hormone exposure (426).
I) Activation and inactivation. Both Ca2+ and Na+ are substrates of the exchange mechanism. In heart cells, the K d for [Na+]o in the forward mode is ∼80 mM and for [Ca2+]i is 0.6 μM; in the reverse mode, K d for [Na+]i is 21 mM and 1.4 mM for [Ca2+]o (686). A higher value for [Ca2+]i (20–30 μM) is obtained in macropatches (409). TheK d value for [Na+]i is furthermore dependent on [Ca2+]i, suggesting the existence of competition (661). Affinity for Na+ and even more for Ca2+ are thus highly asymmetric.
Intracellular Ca2+ not only serve as a substrate but exert a modulatory effect. A minimum of Ca2+ for instance is required for the function of the exchanger in the reverse mode. The apparent K d for the modulatory site is 2 nM, a concentration much smaller than theK d for Ca2+ as substrate (491, 534, 535).
The concentration of Ca2+ close to the transporter is modulated by the presence of negatively charged phospholipids, e.g., phosphatidylserine (408, 585,1035) and phosphatidic acid, produced by phospholipase D action on plasmologens (360). The addition of negatively charged phospholipids or detergents stimulates the activity of the transporter, whereas positively charged detergents cause inhibition (766). The concentration of negatively charged phospholipids is regulated via a “flippase” reaction (408). Flippase or ATP-dependent aminophospholipid translocase transports phosphatidylserine or phosphatidylethanolamine from the outer to the inner side of the membrane, whereas floppase is responsible for the reverse transport.
Intracellular Na+ stimulates the exchanger, but the effect shows inactivation (662) (Fig.12). The process is comparable to the inactivation of a channel and depends on the binding of Na+. Part of the exchanger is supposed to be sequestered and become inactive (389). In ATP-depleted, internal Na+-loaded rat ventricular cells, maximum velocity (V max) of the exchanger is drastically reduced, without change in the K d for Na+ or Ca2+ (389). The exchanger is not directly ATP driven, but ATP hydrolysis is required to keep [Na+]i low via the Na+-K+ pump, to hold the negatively charged phospholipids in a critical concentration in the membrane, and/or to eliminate the inactivation process caused by Na+(67, 406, 662). In this respect, the role of PIP2 has been emphasized. Inactivation by [Na+]i can be removed by addition of PIP2. The restituting effect of ATP is absent when synthesis of PIP2 from phosphoinositol is inhibited (407).
II) Permeation: electrogenic character and reversal. In extruding Ca2+ from the cell against the electrochemical gradient, energy is spent. Energy is provided by the Na+distribution that allows a large passive Na+ influx. To bring [Ca2+]i down to submicromolar concentrations, it is necessary to transport about three Na+ for one Ca2+. This stoichiometry has been verified experimentally, using efflux and voltage-clamp measurements (249, 309, 535,661), and implies that the exchange mechanism is electrogenic and can be characterized by theE rev, E Na,Ca = (nE Na − 2E Ca)/n − 2, or in the case of three Na+ for one Ca2+ asE Na,Ca = 3E Na− 2E Ca.
Under normal resting conditions, the E revcan be calculated to be approximately −30 mV (249). Negative to this potential, Na+ is moving in and Ca2+ out (forward mode) and inward current is generated; positive to this potential, Ca2+ is moving into the cell and Na+ is extruded (reverse mode), and outward current is seen (409). The current-voltage relation shows outward rectification. During the initial part of the action potential, the carrier moves Ca2+ inward and favors repolarization. As Ca2+ is released from the SR, theE rev of the carrier changes (see Ref. 75) with the result that the current reverses and slows repolarization. The exchanger is blocked by Ni2+ and Cd2+(423, 535).
III) Molecular structure. The amino acid sequence of the transporter protein is known (see Ref. 765). The glycosylated NH2 terminal is supposed to be in the extracellular compartment. The large hydrophilic linker between transmembrane segment 5 and 6 represents the regulatory part; it contains an EF Ca2+ binding site and the sites for Na+binding. A section of this regulatory link acts as an inhibitor when applied as a separate peptide (exchanger inhibitory peptide, XIP) (152, 663).
B) NA+-K+ PUMP.
The function of the pump is to transport Na+ to the outside and K+ to the inside of the cell, which are directions opposite to the passive movements of these ions. The pump, therefore, has to consume energy to make this movement possible. Experiments have shown that for each ATP molecule consumed three Na+ and two K+ are transported. This means that the mechanism generates an electrical potential but also that it is sensitive to theE m (for review, see Ref. 292).
Activation of the pump hyperpolarizes the restingE m or exerts a repolarizing effect during the action potential. These aspects are especially pronounced during high-frequency stimulation, because of the increase in [Na+]i that activates the pump. In the sinus node and Purkinje fibers, this hyperpolarizing effect counteracts the spontaneous diastolic depolarization and is known as overdrive suppression (see references in Ref. 1105).
The density of the pump sites per square micrometer varies from 1,200 in the guinea pig ventricle (714) to 2,500 in the rat ventricle (941). The density is larger in the ventricle than in the atria (1054), and in the ventricle it is greater in the subepicardial than in subendocardial cells. The turnover rate is ∼75–100/s (714, 941).
I) Activation by extracellular K+ and intracellular Na+ . The K dvalue for [K+]o is ∼1 mM and for [Na+]i is ∼10 mM, with Hill coeficients of >1.0. In Na+-free medium, theK d for [K+]odecreases to 0.2 mM, suggesting competition between [Na+]o and [K+]o. The lowK d value for K+ means that the pump rate is maximally stimulated at physiological [K+]o values. The sensitivity to [Na+]i decreases at higher [K+]i, although the maximum rate is not affected. At physiological concentrations, the current is directly proportional to [Na+]i, which explains the sensitivity of the pump rate to frequency of stimulation. In spontaneously active cells, an increase in frequency will result in an enhanced outward pump current acting as a negative feedback on the rate of firing, a phenomenon called overdrive suppression. The increased pump activity plays a role in the shortening of the action potential and suppression of spontaneous activity with frequency of stimulation (1025).
II) Permeation, voltage dependence, and Erev . The transport through the pump is electrogenic and voltage dependent. The amount of energy consumed increases with hyperpolarization. When the energy required to move three Na+ out and two K+ in equals the energy obtained on hydrolysis of ATP, the pump will stop. At a givenE m called theE rev of the pump, these two values are equal. The E rev is thus determined by the free energy of ATP hydrolysis and the equilibrium potentials for Na+ and K+. Under normal conditions of [Na+]i, [K+]o, [ATP], [ADP], and [Pi], E rev is about −180 mV (E rev = ΔG ATP/F + 3E Na − 2E K).
The E rev shifts to less negative potentials when [ATP] decreases and [ADP] and [Pi] increase, changes that occur during metabolic inhibition. WhenE rev approaches theE m, the pump rate will drastically reduce and eventually completely stop. These predictions have been verified experimentally (328). ΔG is normally 61.5 kJ/mol, but during ischemia it can fall to values of 49 or less (190). Under those conditions, the pump will stop at moderate E m values (around −60 mV).
Pump rate as a function of E m shows a broad maximum extending from −80 to ∼0 mV; it declines for more negative potentials with an apparent reversal at very negative potentials around −180 mV. The pump rate also decreases at potentials positive to 0 mV (85). The voltage dependence with its positive and negative slope varies with [Na+]o and [K+]o. When [Na+]o is reduced, the positive slope of the current-voltage relation becomes less pronounced, i.e., pump rate increases at negative E m values. Calculation of [Na+]o dependence onE m shows that the apparent affinity of the transport protein for [Na+]o increases with hyperpolarization (K d decreases). The negative slope at positive potentials can be changed in a similar way by increasing [K+]o; in other words, the apparent affinity for [K+]o decreases with depolarization.
The voltage dependence of the pump has been explained by assuming that part of the molecule acts as a ion channel (310). In support of this hypothesis, gating currents have been measured (405). The voltage dependence is supposed to be due to cation binding sites of the pump buried in the membrane and only accessible from the extracellular side through a narrow channel. The actual concentration of Na+ and K+ at the binding sites will depend on the voltage gradient in this access channel. They are greater the more negative theE m, and the reverse occurs at more positive potentials. The higher [Na+], the more difficult is the release of Na+ to the extracellular medium. Also, [K+] will be greater at negative potentials, but this may not be effective if the concentration is already close to the saturating level in control conditions. At positive potentials, the actual [K+] will decrease and may fall below the saturating value and thus reduce the pump rate.
The pump is blocked by digitalis, but sensitivity is species dependent (84, 399); the rat is very insensitive with aK d of 2.4 × 10−3 M for dihydroouabain (DHO) compared with 1.4 × 10−5 M in the guinea pig. Of theoretical but also practical importance is the observation that the sensitivity to digitalis decreases with increase of [K+]o, and the opposite effect is seen when [Na+]i is elevated (942). DHO and K+ however do not bind at the same site. An increase in [K+]oincreases the cycling of the enzyme between different states and decreases the time that the external site is available for DHO binding; an increase in [Na+]i, on the other hand, will force the enzyme to a state where Na+ is delivered to the outside medium and thus a state where DHO can bind (84, 942). The antagonism between [K+]o and ouabain is voltage dependent; the degree of inhibition by a given concentration decreases at more hyperpolarized potentials. This observation is consistent with the hypothesis of a binding site for ouabain within the membrane electrical field where the actual [K+] changes withE m (84).
III) Molecular structure. The pump protein consists of one α- and one β-subunit and is probably present as a dimer. Both have been cloned. The α-subunit is a large protein of 110 kDa spanning the whole membrane; it carries the ATP binding and phosphorylation site at the intracellular portion and the digitalis binding site at the extracellular face. Three isoforms exist, of which the α1 shows low affinity and α2 and α3 high affinity for ouabain (1054).
B. Ion Channels in Intracellular Organelles
1. SR channels
The SR membrane is a leaky membrane with a high conductance for K+ and Cl− (see review in Ref. 677). The only important gradient is that of Ca2+, whereas K+and Cl− seem to be equally distributed across the membrane. The concentration of free [Ca2+] has been estimated to be 700 μM (877). With a cytoplasmic concentration of ∼0.1 μM, this means a gradient of 7.000. The Ca2+ E rev is thus >200 mV, whereas the SR potential is estimated to be 0 mV. This means that the large chemical gradient is not antagonized by an electrical gradient. During Ca2+ release, the SR interior becomes negative but at most a few millivolts because of the counterion movement of K+ and H+ inward and Cl− outward (315). Two release channels have been described: the ryanodine receptor channel and the inositol 1,4,5-trisphosphate (IP3) receptor channel.
A) THE CA2+ RELEASE CHANNEL OR RYANODINE RECEPTOR (RYR).
In mammalian cardiac cells, the SR represents the store from which Ca2+ is released during the action potential. In this process a Ca2+-permeable channel in the SR plays an important role. The Ca2+ released also activates the Na+/Ca2+ exchange current, a NSC current, and a Cl− current, which may all participate in the genesis of early and delayed afterdepolarizations. In conditions of Ca2+ overload, the channel thus plays an indirect but important role in generating triggered activity. The channel is specifically blocked by ryanodine.
I) Activation. In cardiac cells, the trigger for activation of the channel seems to be the local rise in [Ca2+]i consequent to Ca2+influx during the action potential (261). A small proportion of the SR Ca2+ release may be activated directly by voltage in a way similar to that in skeletal muscle (424). The condition to observe this voltage-dependent mechanism seems to be the presence of cAMP (276,424), and recent experiments have suggested that the Na+ channel under those conditions may become permeable to Ca2+ (840).
The process of Ca2+-induced Ca2+ release has been studied by measuring Ca2+ transients in vivo using fluorescence techniques, by flux studies of radioactive Ca2+ in isolated SR vesicles, and at the single-channel level after incorporation of the protein in artificial lipid bilayers. They have led to the following picture.
A) Ca2+: the major activating mechanism. The primary and most important activator of the cardiac SR channel is the cytoplasmatic Ca2+ (120, 261). Information on the role of Ca2+ as trigger for Ca2+ release, derived from tension measurements in skinned preparations (261) and from 45Ca2+ release in SR vesicles (678), is in favor of the existence of activation at low [Ca2+] and inactivation at elevated [Ca2+]. The results show a biphasic activation curve as a function of cytosolic [Ca2+]. The efficiency of Ca2+ as trigger also depends on the speed of Ca2+ application: the faster the rise in [Ca2+], the higher the tension developed (261). Experiments with flash photolysis of Ca2+ in native cells (709) and on pure proteins incorporated in lipid bilayers (154,916) initially did not confirm the existence of inactivation at high [Ca2+]. More recently, it has become clear (586) that the property of inactivation is lost during the isolation procedure of the protein or by the use of high concentrations of Cs+ in the test medium. When the necessary precautions are taken, steady-state activation (K d of 5 μM) and inactivation (K d of 9 mM) have been demonstrated in lipid bialyers (1120) (Fig.13). Similar experiments have further demonstrated the existence of adaptation: upon application of a constant step in [Ca2+], activation decreases with time, but release occurs again when [Ca2+] is increased by a second application, i.e., recovery is fast (349,861, 1006). The rate of adaptation is markedly accelerated in the presence of [Mg2+] or after phosphorylation of the channel (1006). In accord with the existence of adaptation is the observation that the rate of decay of a Ca2+ spark is faster the greater the magnitude of the spark or the amount of Ca2+ released (629). Adaptation and/or inactivation are responsible for the nonregenerative behavior of the Ca2+ release channels under normal conditions. According to the original model of Fabiato (261, 262), Ca2+ bind to an activating site with high rate but low affinity and to a second inactivating site with slower association rate but high affinity. Modifications of this model have been proposed by others (120, 961).
Release is modulated by the presence of nucleotides: addition of [ATP] shifts the concentration effect curve to lower [Ca2+] and upward; the opposite effect occurs with [ADP] (1120) (Fig. 13). Phosphorylation (754, 1006) results in a shift of theP o curve to lower [Ca2+] and [ATP] and in the upward direction; the gain is increased.
B) Influx of Ca2+ through L-type Ca2+channels is the important source of activator Ca2+. The stimulus for Ca2+ release during a contraction is provided by Ca2+ entering the cell through the plasma membrane via the L-type Ca2+ channel (120,709). Release upon Ca2+ entry via T-type channels (913) and via reversed Na+/Ca2+ exchange (589,598, 914, 1073) occurs but is less important and much slower.
L-type Ca2+ channels are geometrically coupled to the release channels. In the heart, one L-type channel is coupled to more than one release channel, and the coupling is less tight than in skeletal muscle. The initial early Ca2+ influx is important and not the total Ca2+ influx. The first openings of the Ca2+ channel are determining, whereas reopenings are of no importance except as a loading function (120,460, 1098). In cardiac hypertrophy and failure, the coupling may become deficient (333). The efficiency of Ca2+ influx in causing release or gain is variable. Gain, defined as the ratio between Ca2+ influx and Ca2+ release or as peak rate of Ca2+release, increases with depolarization up to a maximum and decreases again at more positive potentials. This can be explained by the fact that the current through individual L-type Ca2+channels is larger at more negative potentials because of the larger gradient; the effect on the release channels is great. At positive potentials close to the E rev, the number of active channels is large, but the individual current through the channel is small; the final result is less or no activation of Ca2+ release. Although the total Ca2+ current through the plasma membrane may be the same in both cases, the result is quite different. The efficiency in releasing Ca2+ from the SR is greater when K+ is used instead of Cs+ as the major intracellular cation, suggesting that influx of K+ into the SR through a K+ channel probably accompanies Ca2+ release. The fact that the release induced via the exchanger also is improved can explain why the relationship between Ca2+ transient and voltage is less bell-shaped and stays constant up to very positive potentials (599).
C) Regenerative Ca2+ release: importance of luminal [Ca2+] or load. Under normal conditions, Ca2+release is not propagated (987); the action potential is responsible for the propagation. The nonregenerative aspect of Ca2+ release has been shown by analyzing the behavior of sparks or microscopic Ca2+ release events (839). In the presence of a normal Ca2+ load, channels close rapidly, probably as a consequence of deactivation or fall in activator Ca2+ (460); the local [Ca2+] decreases quickly, due to diffusion from the small junctional space and by local pump reuptake. The time and space (width) constant of the spark are normally too short and too small to cause propagation of the release. The width of the spark can become larger by increasing the Ca2+ content of the SR and thus increasing the amplitude of the spark. At low Ca2+ load, release is linearly related to the Ca2+ present in the SR. Above given concentrations however, the P o of the release channel is increased and the amount of Ca2+released rises exponentially as a function of the preload (44). At the single-channel level, an increase in luminal [Ca2+] is accompanied upon release by longer and more frequent openings, and the conductance is greater (916). For a given Ca2+ load, the relative increase in P o of the channel is also greater at low cytosolic [Ca2+] (628). Under conditions of high luminal and low cytosolic [Ca2+], Ca2+ may be released spontaneously from the SR, and this release may propagate along the cell; the gain is drastically increased (148, 628, 847,987). Propagation from cell to cell, however, has a low probability (584). Increasing the Ca2+ load enhances the frequency of the spontaneous release events, such that the Ca2+ content of the SR and of the cell stays constant (986).
It is not clear how luminal [Ca2+] affects the threshold and the amount of Ca2+ released from the SR. Luminal [Ca2+] does not seem to activate the channel at the cytosolic side during its diffusion from the SR. This is concluded from the observation that Ca2+ release is not inhibited when Ba2+ or Sr2+ is added to the luminal side, although both ions permeate the channel from the luminal to the cytosolic side and block the channel when added to the cytosolic side (976, 999). In more recent experiments, however, a correlation has been found between activation-inactivation and Ca2+ flux through the channel (1121).
II) Permeation. The channel behaves as a high-conductance but poorly selective cation channel. It is permeable to bivalent (Ca2+, Ba2+, and even Mg2+) and monovalent cations. TheP Ca/P K is 6 in comparison with a ratio of >25 for the L-type Ca2+channel. The permeability sequence for monovalent ions is as follows: Cs+ > Na+ > K+ > Li+ > choline+ > Tris+(975). In theory, because of the high permeability, current carried by K+ can be large. The absence of an important electrical or K+ concentration gradient between lumen and cytoplasm, however, prevents this short-circuiting effect.
The conductance is large and differs with the type of permeating cation: ∼100–150 pS for 50 mM [Ca2+] (luminal side), much smaller for Mg2+ (40 pS), and 750 pS for 250 mM [K+] (see Ref. 146). Saturation has been described for bivalent ions (975, 999). The presence of negative charges at the mouth of the channel regulates the local [Ca2+] and the conductance of the channel (999). Neutralization of carboxyl groups using carbodiimides reduces conduction. Addition of the negatively charged heparin on the other hand increases the local [Ca2+] and reduces the threshold for Ca2+ release (999).
Caffeine activates the channel in a Ca2+-dependent way at low concentrations and in a Ca2+-independent way at high concentrations (811). Ryanodine at low concentrations increases the P o by forcing the channel in a substate in which it is more or less stabilized; at high concentrations it blocks the channel (813).
III) Molecular structure. From electron microscopic studies on skeletal muscle, the channel appears to form a square structure, with a fourfold symmetry, suggesting a tetrametric constitution (791). The structure displaces a central channel that branches into four radial channels. Cloning has revealed the existence of at least three different proteins: RyR1 (skeletal muscle), RyR2 (cardiac muscle), and RyR3 (other tissues) (711). The NH2 terminal and the shorter COOH terminal are supposed to protrude in the cytoplasm and form the foot structure seen in electron microscopic studies. The molecular site at which Ca2+ acts is not known (643) but is probably situated in the NH2 terminal. Two smaller modulatory proteins have recently been described: triadin, which is supposed to mediate the interaction with calsequestrin, and FKBP12 (FK binding protein), which stabilizes the RyR and reduces the occurrence of substates in the skeletal muscle channel. When FKBP12 is inhibited by FK-506, the rate of Ca2+ release from the SR and the Ca2+ transient are increased (see Ref. 643).
B) IP3 RECEPTOR.
Although the role of IP3 is well established in smooth muscle and nonexcitable cells (72), the evidence that IP3 might play a role in the heart is circumstantial. Three aspects should be mentioned: 1) IP3is released in cardiac cells, 2) IP3receptors are present, and 3) IP3 causes release of Ca2+ (see review in Ref. 643).
1) Inositol 1,4,5-trisphosphate is released from the plasma membrane after activation of different receptors (M2, M1, P2, endothelin) (394, 775, 936,1041). Release of IP3 is especially pronounced early during reperfusion, an effect mediated via activation of α-receptors (17, 1107).
2) Inositol 1,4,5-trisphosphate receptors can be localized by immunolocalization to the region of the intercalated disk (696). Expression is greater in Purkinje fibers than in other cells (335). Binding studies also found IP3 receptors enriched in fractions containing the intercalated disks with little or no binding in fractions containing the longitudinal SR (521). Compared with RyR, the density is 50-fold less. Expression of IP3 receptors is increased in failing hearts, whereas that of RyR is decreased (643).
From a structural point of view, the IP3 receptor shares considerable homologies with the RyR, although the molecular weight is much less. The large NH2 terminal lies in the cytoplasmic side with the IP3 binding site at its end. The COOH terminal probably plays a role in activation because antibodies that bind to this region can induce or inhibit channel opening. Three isoforms have been isolated (252).
3) Inositol 1,4,5-trisphosphate causes Ca2+release. Cardiac microsomes fused to planar bilayers exhibit channel activity modulated by IP3. Four substates have been recognized. The conductance is smaller (85 pS) than that of the RyR, but the permeability series of bivalent ions is the same with a permeability ratio of bivalent over monovalent cations around 6 (78).
Tension measurements have shown that caffeine contracture is amplified, and Ca2+-induced tension oscillations are enhanced in magnitude and frequency by IP3 (321,1043, 1185). The effects of IP3 vary and depend on the state of Ca2+loading of the SR and the cytoplasmic [Ca2+]. The influence of cytoplasmic [Ca2+] is biphasic; the release for a given concentration of IP3 increases with cytoplasmic [Ca2+], reaches a maximum at ∼300 nM, after which it begins to be inhibitory (79). Like the RyR, the IP3R may become regeneratively active in conditions of Ca2+ overload (374, 515). The IP3 receptor is modulated by PKC, PKA, and Ca/calmodulin (CaM) kinase II, MgATP, and pH (see Ref. 451). The positive inotropic effect (775) as well as the proarrhythmic effect (887) of α-receptor stimulation has been related to IP3 production. Aminoglycoside antibiotics block release (236) and suppress ventricular arrhythmias.
C) K+, CL−, AND H+ CHANNELS IN THE SR.
Outside the Ca2+-release channel, K+, Cl−, and possibly H+ channels constitute high-conductance pathways in the SR membrane. Because of their presence, the membrane of the SR is clamped at about zero potential. In this way, a large gradient forcing Ca2+ flowing from lumen to cytoplasm is guaranteed. The flux of K+, H+, or Cl− as counterions facilitates the passive release as well as the active uptake of Ca2+ via the Ca2+-ATPase.
I) H+ channel. Whether H+movement occurs through channels is not known, but the importance of proton flux has been demonstrated by measurements of changes inE m by pH in the presence of impermeant ions (679). Acidification occurs during the release, and alkalinization has been found during reuptake of Ca2+(601).
II) K+ channel. A K+ channel with an elevated P o at 0 mV exists in the SR membrane (810). The kinetics of the channel are very slow with open times of hundreds of milliseconds.
In isotonic [K+] solution, the single-channel conductance is 150 pS and the current-voltage relation is linear (810). The conductance is concentration dependent and saturates at 200 mM. The channel is blocked by Cs+(177) and TEA (1). The Cs+ block is only slightly voltage dependent and can be relieved by addition of extra K+. The block by Cs+ explains the lower efficiency of Ca2+ as activator for Ca2+release in Cs+ solutions (599) and inhibition of spontaneous Ca2+ release in skinned preparations (510). Acidosis reduces P o at 0 mV (576), but [Ca2+] or [Mg2+] does not seem to affect the channel (810).
III) Cl− channel. In cardiac SR, a Cl− channel stays open at 0 mV. Its voltage dependence is small (809) or nil (514). Gating of the single channel is characterized by two open times and two closed times (514).
The current-voltage relation is linear with a high conductance of 50–120 pS in cardiac cells (514, 809). The channel is permeable to Cl−, less to SO4 2−, and impermeable to gluconate (363). In skeletal muscle, the channel is also permeable to cations (Ca2+), with a sequence ofP Cl/P Tris/P Ca= 1.0/0.5/0.3 (944). The existence of channels that are permeable to anions and cations is not exceptional (see mitochondrial voltage-dependent anion channel, Ref. 171).
Phosphorylation of the channel prevents the rundown of channels incorporated in lipid bilayers (514, 809). Phospholamban acts as a modulator (204), and its phosphorylation results in activation of the channnel.
2. Mitochondrial channels
In the mitochondria, energy is transferred from different substrates to ATP by way of oxidation. Because one of the key reactions exists in building up a proton gradient, the inner membrane of the mitochondrion is supposed to be relatively impermeable and to show a permeability that is highly regulated by electroneutral carriers and uniporters. The outer mitochondrion membrane, on the other hand, has long been considered to act as a sieve that allows fast exchange of ATP, ADP, acetyl CoA, and other metabolites. This classic view has changed dramatically during recent years.
A) THE OUTER MITOCHONDRIAL MEMBRANE.
Permeability of the outer mitochondrial membrane has been studied after isolation and incorporation of protein fractions in artificial lipid bilayers. Two different types of channels have been found (637). One type is cation selective, of which the function is not clear. Because it is blocked by targeting peptides, it may play a role in protein transport (620). The second type is represented by a voltage-dependent channel permeable to anions and more specifically to ADP and ATP (Fig. 2). Together with the adenine nucleotide carrier in the inner membrane, it plays an important role in the traffic of ATP and ADP between the cytoplasm and the mitochondrial matrix.
I) Voltage-dependent anion channel. A) Activation. After incorporation in artificial lipid membranes the voltage-dependent anion channel (VDAC) shows a maximum P o at zero E m that falls off but not to zero on application of either a positive or negative voltage gradient. At the same time, the conductance shifts to a substate, and the channel changes selectivity, becoming more permeable to cations. The obvious question is whether this type of voltage sensitivity has functional implications. Although a direct measurement of a potential gradient over the outer membrane has not been made, it is generally assumed that a Donnan potential exists based on the presence of a difference in concentration of large, charged colloids between the cytoplasm and the intermembrane space. Phosphorylation and Ca2+ binding (during a contractile cycle) may modify this potential gradient (807). The intermembrane space contains metabolic enzymes such as adenylate cyclase, creatine kinase, nucleoside diphosphate kinase, and deoxyribonuclease.
Voltage gating is modulated by a modulatory protein and the NADH concentration. Sensitivity to voltage is increased in the presence of a modulatory protein (427) situated in the space between the inner and outer membrane. The increase in voltage dependence means that the channel closes for smaller changes in potential (0–5 mV) and becomes less permeable to anions (ATP, ADP). Voltage gating of the channel is further regulated by micromolar cytoplasmic concentrations of NADH that act as a ligand for the channel (1188). It is estimated that fluctuations of NADH could result in a sixfold change of the permeability to adenine nucleotides with consequent changes in the intensity of respiration (615). The VDAC should thus be considered a sensor of glycolysis, since cytoplasmic NADH is mainly determined by glycolysis. The closure of VDAC may be the mechanism by which glycolysis inhibits oxidative phosphorylation (Crabtree effect). In this connection it is of interest to note that the VDAC structure binds hexokinase and glycerolkinase.
B) Permeation of the VDAC. The VDAC is permeable for molecules up to 1 kDa. It stays open with a very large single-channel conductance of 650 pS at zero E m and is slightly more permeable to anions than to cations (2/1). It allows passage of ATP and ADP. Permeability to anions ranges from 100/10/1 for Cl−/succinate/ATP. The flux of ATP is high (106 molecules ATP/s, Ref. 807). The outer membrane behaves thus quite differently from the classic sieve but should be considered a highly regulated structure.
C) Structure of VDAC. The channel has been crystallized (972) and forms groups of six transmembrane pores, each pore consisting of a single 30- to 35-kDa VDAC polypeptide (87). The polypeptide contains numerous stretches of 10–14 amino acids that show an alternating polar and nonpolar character. Partial closure is supposed to be due to a positively charged stretch of the molecule moving out of the membrane (gating current, Ref. 925) and leaving the pore negatively charged and permeable to cations.
B) THE INNER MEMBRANE.
The study of the permeability of the inner membrane (see Ref. 538) has been possible by application of the patch-clamp method to mitoplasts (928). Mitoplasts are mitochondria stripped of their outer membrane. Under physiological conditions, the permeability of the inner membrane is low. In ischemic conditions, however, this permeability barrier can be weakened or completely vanish by opening of large channels. Five different channels have been described (Fig. 2):1) a multiple conductance channel (MCC) or mega-channel permeable to large cations and anions, 2) an ATP-dependent K+ channel, 3) a 107-pS slightly anion-selective channel, known as the mCS channel, acronym for mitochondrial centopicosiemens channel, and two channels activated when the matrix side is exposed to very alkaline conditions:4) a cation-selective 15-pS channel, and 5) an anion-selective 45-pS channel.
I) The MCC or mitochondrial permeability transition pore. A) Activation. Under normal physiological conditions, the channel is closed mainly because of the high [ATP] and low [Ca2+]. ATP is normally needed to maintain a low cellular [Ca2+] and a negative mitochondrial E m(908). The channel is activated by elevation of intramitochondrial [Ca2+] in the micromolar range, in combination with low [ATP] (189, 244,952). Activation is facilitated by long-chain acyl CoA (771), pro-oxidants, and dithiol oxidatives (reversed by disulfide reduction), high O2 tension in combination with low [ATP], and high [Ca2+]; it is inhibited by H+, Mg2+, and the immune suppressive drug cyclosporin (953). It is also sensitive to drugs that bind to the benzodiazepine receptor. Typical during activation is the increase of conductance in steps. A possible underlying mechanism is the assembly of subunits into a large channel by the action of Ca2+ (953).
The presence of a high conductance in the inner membrane poses a problem. Open pores spanning the inner membrane would dissipate the energy-transducing gradients and so uncouple oxidative phosphorylation (comparable to the effect of the uncoupling protein in brown fat cells). As intramitrochondrial [Ca2+] is normally low and matrix [ATP] high, the MCC is not expected to be activated under normal conditions. In ischemia or Ca2+-overload conditions, and especially upon reperfusion with high O2 tension restored, the channel may be activated and is probably responsible for the permeability transition. Activation is accompanied by depolarization (see Fig. 19) and release of the mitochondrial content, especially Ca2+ itself. This release of Ca2+ may trigger a Ca2+ wave from one mitochondrion to the other. The mitochondrion may still be operating in a low-conductance mode, and the activation of the MCC remains reversible; the mitochondrion acts as an excitable organelle (449). If activation is steady, however, the cell will die (244, 953). Pore opening disrupts mitochondrial energy transduction by providing a short circuit for protons, bypassing the chemiosmotic proton circuit that normally links the oxidation of substrates with phosporylation of ADP. Once uncoupled, mitochondria act as a drain for glycolytic ATP. The molecular nature of the mega-channel as well as all other inner membrane channels is not known.
B) Permeation. The mega-channel has a variable conductance between 40 and 1,000 pS and is unselective to cations or anions. Large anions such as ADP and ATP can permeate.
II) The ATP-dependent K+ channel. The ATP-dependent K+ channel is similar to the channel in the plasma membranes of many cells (458). It has the same conductance and the same activation pattern. It opens by reduction of [ATP] below a critical concentration. It is also blocked by glibenclamide and sensitive to K+ channel openers. The opening is facilitated by PKC-dependent phosphorylation (846). In contrast to the cardiac plasma membrane channel, it is sensitive to diazoxide and resembles more the pancreatic channel (316). Its role in mitochondria is not obvious; its activation may stabilize the mitochondrialE m and avoid activation of the mega-channel. A role in preconditioning (316,619, 846) and in volume regulation has been proposed; during ischemia, mitochondria swell, and opening of large-conductance channels may induce efflux of solutes and water and reduce volume.
III) The 107-pS anion channel or mCS channel. The 107-pS anion channel of mCS channel (928) is slightly anion selective. It is activated at positive matrix potentials, exponentially increasing with positivity. Its anion selectivity is poor (P Cl/P K of 4.5). It is not pH sensitive but inhibited by oxidative uncouplers. Under normal conditions, the channel is thus closed. For its function, different hypotheses have been offered: 1) a function as uncoupling protein analogous to that in brown fat mitochondria,2) a role in volume homeostasis, and 3) a function as import channel for proteins and mitobiogenesis. The majority of mitochondrial proteins are synthesized in the cytoplasm and hence have to be imported into the organelle; this requires a two channel assembly in tandem in the inner and outer membrane.4) Recently, it has been proposed (37) to function as a protective mechanism against a fall in mitochondrial potential. Permeability under physiological conditions is supposed to be small but large enough to allow passive equilibrium for Cl−. In a situation of metabolic inhibition when the matrix negativity would normally drop, this process might be inhibited by opening of the mCS channel clamping the mitochondrial membrane at a very negative matrix potential.
IV) Alkaline-activated channels. When the matrix side of a mitoplast preparation is exposed to very alkaline solutions, two channels can be observed: one is cation selective and shows a conductance of 15 pS and a second is anion selective and shows a conductance of 45 pS. The role of these channels is unknown.
C. Gap Junction Channels
Gap junctions are responsible for the syncitial nature of cardiac tissue. Their high conductance and permeability allow for fast conduction of the action potential (electrical coupling) and for an efficient flow from cell to cell of molecules or metabolites with a molecular mass up to 1.2 kDa (K+, Na+, Ca2+, cAMP, cGMP, IP3) (metabolic coupling). In the canine heart, each ventricular cell is connected to 11 other cells by way of gap junctions (824). Different values for the conductance between cells have been published from 250 to 2,500 nS/cell. For a normal conduction of the action potential between a pair of cells, ∼35 gap channels seem to be sufficient (1078). Redundancy is thus great. This contrasts with the recent finding that mice heterozygous for connexin (Cx)43 null mutation show a 45% reduction in conduction velocity of the action potential (344).
The distribution of gap junctions in the normal heart is nonuniform or anisotropic (761). Gap channels are found almost exclusively in the intercalated disks. Large intercalated disks exist at the end of the cells, smaller along the length. A small number is present at the SAN-atrium junction and the Purkinje-muscle junction. These observations explain why conduction in the transverse direction or between the SAN and the atrium or between the Purkinje system and the ventricular muscle can be critically reduced. In culture, the number of junctions is smaller, which translates in slower conduction of the action potential and absence of propagated Ca2+ waves.
2. Formation of gap junctions and structural considerations
Gap junctions can be induced by forcing two single cells into physical contact (1030). Precursors are present in the plasma membrane from which cell-to-cell channels can form rapidly upon pairing. Analysis with different biophysical techniques has shown that the channels responsible for the large conductance of the gap junction membrane consist of two hemichannels or connexons in two apposing cells (933). Each hemichannel is composed of six polypeptide subunits or connexins. Subunits derived from different genes differ in molecular mass varying between 26 and 50 kDa and in conductance between 40 and 160 pS. In the heart, three connexins are expressed: Cx40, Cx43, and Cx45 (197). They show a different distribution and differ in functional characteristics. Connexin43 is the most abundant. Connexin40 is preferentially present in atria, nodal tissue, and Purkinje system (501, 574). Connexin43 and Cx45 turnover quite rapidly (2–3 h) in cultured neonatal rat heart cells (197).
In the formation of gap junctions, hemichannels act as ligands for each other. Of importance for this activation is the presence of six cysteines in the extracellular loop of the connexin molecule. Docking and/or opening of channels involves disulfide exchange.
Each subunit or connexin is supposed to consist of four transmembrane segments, M1 to M4. M3 is amphiphatic and probably part of the pore structure. Amino acids at or near the NH2terminal and the M1/E1 border form part of a charged complex that may act as a voltage sensor. The COOH terminal together with the intracellular loop between M2 and M3 form the so-called proton gate (195).
The gap junction channel is gated via two important mechanisms called chemical and voltage gating.
3. Chemical and voltage gating
A) CHEMICAL GATING BY PROTONS AND CALCIUM IONS: A SLOW PROCESS.
Protons cause the gap channel to close (see Ref. 104) (Fig.14). The mechanism is proposed to be a ball-receptor inactivation process; protons increase the positive charge of the receptor site and allow the negatively charged ball of the COOH terminal to bind and to occlude the channel pore. Deletion of part of the COOH terminal causes a significant fall in pH sensitivity (695). In the cytoplasmic loop between M2 and M3, which probably acts as the receptor site of the Cx43 gap junction, histidine residues are present that can be titrated by protons (933). The presence of positive charges at critical positions in this loop leads to channel closure.
An increase in [Ca2+] also reduces the total junction conductance (109, 206, 285,732, 984). In the experiments of Noma and Tsuboi (732), the sensitivity to block by Ca2+was rather high under normal pH conditions (pKCa of 6.6 at pH 7.4), which means that concentrations of Ca2+that close the channel may be reached during a normal contraction. Because uncoupling does not occur during a contraction, the authors suggested uncoupling to be rather slow or [Ca2+] to be much smaller at the gap channels than at the myosin level. Larger increases in [Ca2+] occur during ischemia and may induce uncoupling.
Possible underlying mechanisms are direct binding of Ca2+/CaM to the channel or Ca2+/CaM-activated phosphorylation (984). Phosphorylation of the COOH terminal increases the negativity and thus the ability to bind to the positively charged histidine of the loop between M2 and M3.
Information about interaction of Ca2+ and H+ is controversial. The outcome is of importance, however, since simultaneous changes in the concentrations of these two ions occur frequently. A synergistic action of Ca2+ and H+has been demonstrated by studying dye movement between cells (109) or measuring electrical conductance (1091). A decrease of intracellular pH (pHi) to 6.0 combined with an elevation of [Ca2+]i to 425 nM fully blocked the transjunctional conductance. This conclusion is at variance to the findings of Noma and Tsuboi (732), who described a competition between Ca2+ and H+.
B) VOLTAGE GATING: A FAST PROCESS.
Gap junction channels in cardiac cells are sensitive to transjunctional voltage (691, 1053). Upon application of a voltage gradient, the instantaneous current-voltage relation is linear. With time, channels close however and the relation becomes S type, a process that can be compared with inactivation (1053) (Fig. 15). Such a process accelerates uncoupling during ischemia.
At the molecular level, amino acids at or near the NH2terminus and the M1/E1 border form part of a charged complex and thus may act as a voltage sensor (1040). The data are interpreted by supposing that each gate of the six connexins responds to the applied voltage in an independent way. By making chimeras, junctions with different time- and voltage-gating properties have been constructed; in this way, rectification may be generated, with a voltage independence on one side and closure on application of large polarities of a given sign on the other side. The transjunctional voltage dependence, together with the effects of Ca2+, may explain the sealing phenomenon that occurs when part of the heart tissue is injured. During ischemia, this process is useful in isolating ischemic cells from viable functional cells. However, at the same time, it may be arrhythmogenic.
In native cardiac gap channels, a reduction of gap junction conductance following a decrease of transmembrane voltage has not been demonstrated yet, but if present, it might explain to a certain extent the uncoupling of cells when [K+]o is increased in ischemia (see Ref. 581).
The gap junction shows a high permeability and passes substances with a molecular mass up to 1 kDa, including IP3 and cAMP (1030). The single-channel conductance decreases markedly for TEA and tetrabutylammonium. The ionic radius of these molecules is ∼12 Å. This dimension may thus be close to the limiting pore diameter. Calcium ions easily permeate, but propagation of spontaneous Ca2+ waves, occurring under conditions of Ca2+ overload, is rather rare, with a probability of 0.15 (584).
Values for the single-channel conductance vary, and this variability may be related to the relative expression of the three connexins, the state of phosphorylation, and the voltage (111, 690, 955,1031). Connexin43 has a conductance of 45–100 pS, depending on the phosphorylation state, is equally permeable to anions and cations, is highly permeable to dyes, but is insensitive to transjunctional voltage. Connexin40 shows a conductance of 120–160 pS, and its permeability to cations is about fivefold greater than to anions. Connexin45 has a smaller conductance of 30 pS, and its permeability to anions and dyes is limited; it is very sensitive to voltage (501). The current-voltage relation is linear, but rectification can be generated by expression of chimeras consisting of proteins with different voltage dependency (1040). If such asymmetry exists in native cells, it could play a role in unidirectional conductance.
For modulation by phosphorylation, see sectioniii D.
III. ISCHEMIA SYNDROMES
Block of oxidative metabolism and fall in [ATP]/[ADP] causes important changes in ion concentrations ([K+]o, [H+], [Na+]i, [Ca2+]i, [Mg2+]i); disturbs lipid metabolism with accumulation of long-chain acylcarnitines, lysophosphoglycerides, fatty acids, and arachidonic acid; leads to the production of radicals, secretion of neurotransmitters, hormones, and metabolites, with concomitant stimulation of adrenergic, purinergic, and muscarinic receptors; and causes mechanical changes and stretch in the ischemic regions of the tissue. These changes have important modulatory effect on channels and carriers. In this section, this type of information is grouped in physiopathological topics, characteristic for acute ischemia. The analysis of each syndrome includes 1) a description of the changes, 2) the mechanisms involved,3) the effect on channels and transporters, and4) the final outcome at the electrophysiological and arrhythmia level.
A. Changes in Ion Concentrations
In this section we distinguish successively accumulation of [K+]o, intra- and extracellular acidosis, accumulation of [Na+]i, accumulation of [Ca2+]i, and depletion of [Mg2+]i.
1. Increase of [K+]o
A) DESCRIPTION OF EXTRACELLULAR K+ ACCUMULATION.
Under aerobic conditions, [K+]i is high and [K+]o is low. Passive K+efflux is compensated by active K+ influx via the Na+-K+ pump. During ischemia, this dynamic equilibrium is broken, and external K+ accumulates. Potassium loss during ischemia typically occurs in three phases (Fig.16 A). Within 20 s after the occlusion of a coronary artery there is a fast accumulation of K+ in the extracellular space that reaches a plateau after 3–10 min and is followed by a third slower increase starting between 15 and 30 min (see Ref. 1099).
The time to reach the plateau and the actual levels of [K+]o differ widely among species, the models used (isolated perfused papillary or septum muscle, Langendorff-perfused heart, and in situ heart), the rate of beating, and the activation of the sympathetic nervous system. Plateau level of [K+]o reached is highest (20 mM) in the isolated guinea pig heart (544), 10–11 mM in the in situ pig (288) and the rabbit perfused septum (1079), and 8–9 mM in the Langendorff-perfused rat (1100). The perfused rabbit papillary muscle shows a broad range of [K+]o levels from 8 to 20 mM that depend on the diameter of the preparation. The difference has been correlated to a much more pronounced acidosis in the center of thick preparations (136).
Rate of stimulation is important between 0 and 60–90 min−1, but not above this value (377,418, 544, 1080). In the absence of stimulation, [K+]o accumulation starts with a delay and its rate of accumulation is clearly less rapid; an outspoken plateau is not present.
A delay is also obtained in the presence of β-receptor blockade (418, 1097); the plateau level, however, was not different. Stimulation of the sympathetic nervous system on the other hand causes a faster rise in [K+]o(1101).
During the plateau, [K+]o is constant or slowly changing in the positive or negative direction (288, 418, 1101). The preparation remains excitable or regains its excitability. When stimulation is stopped, [K+]o falls, suggesting an active Na+-K+ pump (1080). During the plateau, catecholamines are massively released from nerve endings and cells (865), glycolysis is stimulated, and lactate production is increased. The plateau level of [K+]o is dependent on glycolytic activity (1101). When the heart is depleted of glycogen during successive ischemia periods, lactate production (348) and acidification (288) are less, and [K+]o rises faster.
The changes in [K+]o are far from homogeneous (857). They are most pronounced in the center of the infarct. Higher levels are present in subepicardium compared with subendocardium.
When ischemia is continued over 20–30 min, [K+]o starts to rise again. This phase is accompanied by inexcitability, an increase of longitudinal resistance, elevation of the resting tension, and development of rigor contracture (135, 1079). Lactate production stops (281), and extracellular pH (pHo) remains at a low level. The metabolic changes suggest that anaerobic glycolysis is blocked.
Upon reperfusion, [K+]o recovers completely and rapidly during the first rising phase or the plateau (8). The [K+]o may even fall transiently below the control level. Recovery is incomplete, and [K+]o remains elevated when reperfusion is started after 30 min of ischemia when the secondary rise of [K+]o is already pronounced. Under those conditions, the resting tension of the muscle may increase and often shift into hypercontracture. At the same time, intracellular enzymes are released, evidencing an irreversible increase in membrane permeability.
B) MECHANISMS RESPONSIBLE FOR THE CHANGE IN [K+]O.
Three main factors contribute to accumulation of [K+] in the extracellular space (scheme 1): shrinkage of the extracellular space, decrease of active K+ influx, and increase of passive K+ efflux.
The multifactorial nature should be taken into account when discussing experiments in which only one factor is changed. In such situations, other factors may compensate.
1) Shift of water from the extracellular to the intracellular phase occurs as a consequence of increase in osmotically active particles in the cell, such as lactate and phosphate. Experimentally the amount has been evaluated and found to be ∼20–35 mM, a value which is lower than expected on theoretical considerations (988). Much larger changes in osmotic pressure have been deduced from the changes in the concentration of the extracellular marker tetramethylammonium (552); however, a recent reevaluation, using higher concentrations of the marker, has yielded smaller values (1134). Taken together, it seems reasonable to assume that the increase in osmotically active particles can result in a restriction of the extracellular space by 15% after 10 min of ischemia. A similar value has been estimated in single rat ventricular myocytes exposed to silicon oil (282). Although small in itself, a decrease in the extracellular space acts as an amplification mechanism for any other mechanism involved in the increase of [K+]o.
2) There is a decrease of active K+ influx by reduction of the Na+-K+ pump activity. Experimental evidence suggests partial inhibition of the pump. The block is incomplete; a transient arrest of electrical stimulation during ischemia results in a fall of [K+]o (1080), whereas block of the pump by digitalis increases K+ loss (544).
To justify the observed increase in [K+]o however, the pump does not need to be blocked to a large extent. The capacity of the pump is remarkable, and a maximally active pump (150 pA or 3 × 10−15mol/s) would totally deplete the extracellular space of K+(5 mM) in a very short time of 17 s (change in concentration is 18 mmol · l−1 · min−1; extracellular space assumed to be 50% of the intracellular space or 10 pl/cell; [K+]o is 5 mM; 1 Coul = 10−5 mol). Even if the pump is only working at half the maximal value and inhibition in early ischemia is only 50%, the change in [K+]o will still be 4.5 mmol · l−1 · min−1, which is more than needed.
Ischemic conditions make a moderate inhibition of the pump plausible: LCAC (886) and oxygen free radicals (880) have been found to reduce pump activity. Metabolic energy needed to fuel the pump becomes deficient. It is not so much the [ATP] alone, however, that should be considered (a fall to 150 μM which is theK d value of the pump is probably an unrealistic low value during early ischemia) but much more the increase in [ADP] and [Pi] which cause a substantial fall in the energy delivered from ATP hydrolysis or the phosphate potential. Although the normal value is ∼61 kJ/mol and theE rev of the pump is −180 mV, the phosphate potential may drop below 50 kJ/mol, and theE rev of the pump to −60 mV, which is about the E m of the cells during the plateau of [K+]o accumulation. The pump, in other words, will stop functioning at the E mexpected under ischemic conditions (328). In accord with the assumption of a partial block of the pump activity, recent measurements reveal substantial increases in [Na+]i (see sect.iii A3) and internalization of pump molecules by endocytosis in [ATP] depletion (M. J. Shattock, personal communication).
3) There is an increase in K+ outward movement. The first question is whether K+ loss, in the form of potassium lactate or KCl cotransport, exists? On a molar basis, the amount of lactate production is four times larger than the net loss of K+ (502). If only part of the lactate movement were linked to K+ efflux, this process could make an important contribution to [K+]oaccumulation. However, the following observations are against a tight coupling between K+ and lactate: 1) K+ loss still occurs after blockade of anaerobic metabolism, when lactate production is strongly reduced; 2) K+ loss can be reduced by drugs, e.g., glibenclamide, whereas lactate production is not changed (502); and3) there is no evidence for an electrogenic transport of lactate (892). When cells are exposed to lactate, the intracellular medium becomes acidic, a process which is not accompanied by any current under voltage-clamp conditions or is not affected by a change in E m. Acidification was saturable, partially stereospecific for l-lactate overd-lactate and inhibited by monocarboxylate carrier inhibitors, consistent with a carrier-mediated transport mechanism.
Part of K+ moves out of the cell via a KCl cotransport, which is stimulated by an increase in cell volume as occurs in ischemia. Inhibition of the KCl cotransporter by bumetanide causes a decrease, and stimulation by ethylmaleimide an increase in [K+]o accumulation (1136). The quantitative contribution of this mechanism is unknown.
The next question is whether evidence exists for electrogenic K+ outward movement.
Quantitatively electrogenic K+ movement can be expressed by the product of the conductance to K+ and the electrochemical gradient, according to the equation:I K =g K(E m −E K). It tells us that net outward movement will take place and increase when K+ conductance (g K) is elevated, but only when theE m is positive to the equilibrium potential for K+ (E K). Evidence for the existence of increased g K is given below. It should be stressed that an increase ing K alone may be self-limiting because it causes hyperpolarization and moves theE m closer toE K. This is the reason why K+loss during ischemia is less when the preparation has been pretreated with a K+ channel opener (499). As long as the preparation is excitable, K+ loss occurs during the action potential, and loss will be greater when g Kis increased. Computations have shown that, although the action potential is shortened, net K+ efflux is enhanced (134, and see Ref. 1099). This process is responsible for the initial phase of K+ loss when the cells are still excitable. Later, when the cell becomes inexcitable, the rate of K+ loss will be dependent on the existence of an inward leak current.
Evidence for an increased inward leak current as well as an increased K+ conductance during ischemia has been obtained on models simulating the ischemic condition, because measurements of ionic currents using the voltage-clamp technique cannot be done in real ischemia. These models include multicellular and single cells exposed to hypoxia, metabolic inhibition, elevated [K+]o, and increased [H+]. In preparations exposed to hypoxia or metabolic inhibitors (see Ref.134), an increase in outward K+ current has been measured. The current reverses at the expected E K, and its current-voltage relation is linear or even outward rectifying, over a broad range of negative and positive potentials. TheI K1 channel is an unlikely candidate, since the I K1 current is strongly inward rectifying, and inward rectification should even become more pronounced in the presence of an increased [Mg2+]i(652), [Na+]i, and [Ca2+]i (654,671). The I K1 current furthermore was found to be decreased in the presence of lysophosphatidylcholine (LPC) (161, 542), free oxygen radicals (542), and intracellular acidosis (468).
A better candidate is the ATP-dependent K+ channel, and a number of observations are consistent with this hypothesis.1) Single-channel activity typical for the ATP channel is seen in cell-attached patches during exposure of the cell to dinitrophenol (DNP) (990) or cyanide (CN) (729). A correlation in time is found between the shortening of the action potential and single-channel activity (694, 725); shortening of the action potential was reversed by intracellular injection of ATP (963). 2) Glibenclamide, a blocker of the KATP, reverses the outward current induced by DNP or hypoxia (300) and the shortening of the action potential duration (317). Results on K+ efflux and [K+]o accumulation, however, are less convincing (1137).
A serious objection against the KATP channel being responsible for the increased efflux is its high sensitivity to ATP block (K d 100 μM), whereas the [ATP] during the early minutes of ischemia remains at a sensibly higher level. The following remarks should be taken into consideration. 1) During ischemia, the sensitivity of the channel to ATP block is decreased by the simultaneous rise in [ADP]i(970), Pi, and lactate (366,516); acidosis (193, 266); the presence of free oxygen radicals (471) and extracellular adenosine (469, 540, 969); a decrease in taurine concentration (364); and the existence of stretch (1020). 2) The [ATP] in the subsarcolemmal space or the concentration seen by the channel may be different from the bulk concentration. Block of anaerobic glycolysis, a pathway important for local subsarcolemmal [ATP] together with continued consumption of ATP by the Na+-K+-ATPase and adenylate cyclase may decrease the local [ATP] far below the bulk concentration (435, 781, 858,998). 3) Because of the large single-channel conductance and the high density of the channels, only a small proportion of the channels need to be activated to generate a relative large increase in conductance or current (725).
Two other candidate channels with outspoken outward rectifying properties are KNa and KAA. The KAA is a channel with a conductance slightly larger than the KATP; its P o is enhanced in acidosis (527, 528) and by stretch (525), two conditions present during ischemia. With respect to the possible role of the KNa channel, the problem is similar to that of the ATP-dependent channel, the question being whether the [Na+]i concentration can reach the high values needed to activate the channel (497,625). Recent measurements usuing the NMR technique show substantial increases in [Na+]i. It is furthermore important to realize that [Na+]i close to the membrane can be much higher than in the bulk; gradients and local accumulation exist (642, 1085; and see Refs. 127, 591). For these reasons, activation ofI KNa during ischemia and especially during reperfusion (enhanced Na+/H+ exchange) remains a real possibility.
Three pathways for an increased inward leak probably participate in keeping E m away fromE K: modified TTX-sensitive Na+ channels, NSC channels, and Cl− channels.
Sodium channels are modified by LPC in such a way that they open and show continuous activity at the resting potential level (107, 108, 1002). Inward current via NSC channels can be induced by stretch (188,529, 815), [ATP]o(301, 411), free radicals (472), long-chain fatty acids (441), and a rise in [Ca2+]i (172,251, 410). Some of the NSC channels activated by stretch are equally permeable to Ca2+(529). Calcium leak channels as such are induced by radicals (1056). An important Cl− current is activated via β-receptor stimulation (443), a rise in [Ca2+]i (910,1192), hypotonic distension (357,929, 992), and [ATP]o(500, 597, 664). BecauseE Cl in cardiac cells is positive to the resting potential, an increase of Cl− conductance generates an inward current at negative potentials. Outward movement of lactate through channels would generate an important inward current. At present, no evidence for such a mechanism exists, and the major efflux seems to occur through a proton-coupled carrier mechanism.
Finally, outward current is reduced through a partial block of the Na+-K+-ATPase by the fall in free enegy of hydrolysis, LCAC (959), and oxidative stress (880), which thus favors depolarization. In conclusion, it may be said that three K+ channels with large conductance can be activated during metabolic inhibition, whereas a plethora of inward currents carried by Na+, Ca2+, and Cl− are available to generate an electrical gradient for K+ loss.
C) EFFECTS OF INCREASED [K+]O ON CHANNELS AND CARRIERS.
The effect of the increase in [K+]o on channels, other than K+ channels, is mainly indirect via the depolarization it causes (for references, see sect.ii A). The extent of depolarization during acute ischemia is responsible for a partial or complete inactivation of the fast Na+ channel, the T-type Ca2+ channel, the I to, and theI Kur. Recovery from inactivation of these currents is slowed and is accompanied by a decrease in excitability, prolongation of refractoriness, and conduction slowing of the action potential.
For a number of K+ channels, but not for all, the conductance is increased when [K+]orises; the I K1 current (123) is the most sensitive, but voltage-activated currents such asI Kr (851) andI to (284) and ligand-gated K+ channels such as KACh, KATP, KAA, and KNa all increase their conductance and carry more current at elevated [K+]o. In the case ofI K1 and I Kr, the mechanism is less pronounced inward rectification, consequent to a smaller block by intracellular cations (I K1) or smaller inactivation (I Kr). These changes again will stabilize the E m and reduce excitability.
Worthwhile to mention is the enhanced conductance of theI f current of which the Na+-carrying capability is increased (293,419). Because this current is only activated at hyperpolarized levels, the functional implications of this change remain limited, except transiently during reperfusion.
Extracellular K+ stimulates the Na+-K+ pump, but because the normal K+ concentration is already sensibly greater than theK d of 1 mM (715), the increase in pump current via this mechanism is of no functional importance.
D) EFFECTS OF [K+]O ON ELECTROPHYSIOLOGICAL CHARACTERISTICS.
At the multicellular level, the changes by elevated [K+]o will cause the cell to depolarize and the action potential to be reduced in amplitude, rate of rise, and duration. Inactivation of the Na+ conductance, concomitant to the depolarization, is responsible for the fall in action potential amplitude and rate of rise and the decrease in excitability. Transiently, however, excitability may be increased, the reason being that the depolarization, although causing inactivation of the Na+ current, at the same time moves theE m closer to threshold with as result a reduction in the current required to reach threshold (231). At depolarized levels, the recovery from inactivation is slower; in the presence of increased [K+]o, the result is a prolongation of the postrepolarization refractoriness (see Ref. 480). Conduction velocity is depressed and slowed, especially during the relative refractory period (545). Shortening of the action potential is mainly due to an increase of theI K1 and I Krconductance and to a lesser extent to a decrease of Na+conductance. The changes in excitability, refractoriness, and conduction, together with the shortening of the action potential, favor the occurrence of reentry arrhythmias.
Potassium loss during early ischemia occurs in three phases: a fast increase results in a plateau (10–20 mM) after 3–10 min and is followed (after 15–30 min) by a secondary and irreversible increase. Upon reperfusion before the irreversible phase, [K+]o recovers rapidly. Three main factors contribute to the rise: shrinkage of the extracellular space, inhibition of active K+ influx, and increase of passive K+ efflux. The increase in K+ efflux is due to the existence of an inward leak current (I Na, I NSC andI Cl) which keeps theE m positive to theE K, concomitant with an increase in K+ conductance (activation ofI KATP, I KAA andI KNa). The effect of elevated [K+]o, apart from causing an increase in conductance of K+ channels, is indirect via the depolarization. The final result is inexcitability and block of conduction.
2. Intracellular and extracellular acidosis
A) PROTON DISTRIBUTION UNDER NORMAL AND ISCHEMIC CONDITIONS.
Under normal perfusion and oxygenation, pHi is slightly more acidic than pHo. During ischemia, CO2 retention and the net production of protons shift pHi as well as pHo in the acidic direction. External pH has been described to change rapidly and monotonically from 7.4 to values as low as 6.0 (159). Diverging results have been obtained on pHi, depending mostly on the experimental model and the technique used [ion-sensitive electrodes (ISE), distribution of weak acids, 31P-NMR spectroscopy, and fluorescence indicators]. With the use of NMR and fluorometric measurements, a pronounced fall in pHiand pHo has been found in preparations subjected to total ischemia. From a control value of 7.15–7.2, pHifell to 6.5 after ∼4 min and to 6.2–6.0 after 10–20 min (rat, Refs.116, 159, 962, 1047; ferret, Ref. 640; rabbit, Ref. 689) (Fig.16 C). A short delay of 1 min before the decline in pHi has been observed by Vandenberg et al. (1011). Smaller changes in pHi have been found using the ISE. This in part is due to a selection of superficially located cells for measurement that is an unavoidable bias with this technique. In a study on gas-superfused but blood-perfused papillary muscles of the rabbit heart subjected to ischemia (1135), internal pH was found to be rather resistant to any change when Pco 2 was held at the initial value; however, if the CO2 tension in the experimental chamber surrounding the preparation was gradually increased to values that are probably present in vivo during ischemia, pHi underwent a fall from 7.0 under control conditions to 6.55 after 18 min of ischemia. At the same time, changes in pHo were more pronounced and decreased from 7.39 to 6.13. In other words, the pH ratio reversed, an effect that has been correlated to the higher buffering capacity of the intracellular medium. The finding that pHi depends largely on Pco 2 predicts heterogeneity in pHi changes, depending on the depth of the myocardial layer. The resistance of pHi to undergo dramatic changes further demonstrates that protons continue to be transported against their electrochemical gradient even during ischemia. A more pronounced fall in pHo than pHi has been described in another study in which ischemia was simulated by bathing a papillary muscle in oil saturated with nitrogen (1018).
Changes in pH are less dramatic if the perfusion is not totally suppressed. In a comparative study where total ischemia and 10% perfusion were tested, pHi was found to drop to 6.1 after 20 min of total ischemia, whereas this value was only 6.8 in 10% flow ischemia (769).
After an ischemic period short enough not to cause irreversible injury, pH on reperfusion recovers rapidly (398,1011). Recovery is >50% in 1–1.5 min and may slow down a little during the following minutes.
B) REGULATION OF PROTON CONCENTRATION IN NORMAL, ISCHEMIC, AND REPERFUSION CONDITIONS.
Intracellular proton concentration is kept at a much lower level than predicted for a passive distribution. For a −80 mV resting potential, the pHi at equilibrium should approach 6.0. The fact that the pHi is around 7.2 means that protons are actively transported outward. In aerobic conditions, the continuous production of protons is compensated by CO2elimination, Na+/H+ exchange, and Na+-HCO3 − cotransport, the latter two processes being ultimately coupled to the energy-consuming Na+-K+ pump (493, 595a, 616).
The Na+/H+ exchanger is activated by intracellular H+, with a K d of 7.4. Although this will make it an efficient mechanism (1114), the exchanger is far from equilibrium as can be learned from the [Na+] and [H+]. The sensitivity of the exchanger to proton concentration is enhanced, and its activation curve shifted in the alkaline direction by α1-receptor stimulation and activation of PKC (298). β-Receptor activation however inhibits (580). The exchanger is also inhibited by extracellular acidosis (1114).
The second important mechanism to protect the cell against acidosis is the Na+-HCO3 − cotransporter. According to a recent publication (7), it is electrogenic (2 HCO3 − for 1 Na+ ). In the guinea pig heart, the cotransporter is responsible under physiological conditions for about the same amount of acid extrusion as the Na+/H+ exchanger (198,580). The cotransporter is also coupled to α- and β-receptors but in an opposite way to that of the Na+/H+ exchanger. α-Receptor stimulation in other words depresses whereas β-receptor stimulation enhances the activity of the cotransporter. The final effect will thus depend on the relative importance of the α versus β and cotransporter versus exchanger pathways. In the guinea pig, the α-effect is predominant over the β-effect, and the effect on the symporter is stronger than on the exchanger (580). Sympathetic stimulation therefore will decrease the rate of acid extrusion in this species. These effects are mimicked by application of extracellular ATP, whereas adenosine or ADP is without effect. In the rat, α-receptor stimulation has been reported to result in alkalinization, implying a larger effect on the exchanger (470). Lysophosphatidylcholine impairs the symporter and slows the recovery from an acid load (1129). The fact that two different mechanisms reduce the proton load explains why block of one does not necessarily result in acidosis.
In heart cells, two other processes, the Cl−/HCO3 − and Cl−/OH− exchange, act in the opposite way and may cause acid load (592, 948). They are stimulated by activation of β-receptors and [ATP]o. Whether the stimulation involves a direct modification of the exchanger or is due to an indirect effect is not known. β-Receptor stimulation in addition stimulates glycolysis, which indirectly contributes to the acidification.
During ischemia, the fall in pHi and pHo is caused by an increased production of protons and insufficient removal (210).
1) Glycolytic ATP turnover in contrast to oxidative ATP turnover is accompanied by a net H+ production. During ischemia, ATP production shifts from the mitochondrial system to glycolysis, and this shift is accompanied by an obligatory increase in protons. The finding that inhibition of glycolysis abolishes the acidosis during CN exposure is consistent with this explanation (10).
2) Net ATP hydrolysis is accompanied by net production of protons. One mole of ATP breakdown produces 0.8 mol H+. This contrasts with the absorption of 0.35 mol H+ upon breakdown of PCr which thus may shift the pH early in ischemia in the basic direction. The ratio of [PCr] to [ATP] is normally around two. Metabolic inhibition, at least in the beginning, can result in alcalosis or acidosis depending on the relative amounts of PCr or ATP that disappear (253). A delay in pHidecline has been observed for global ischemia of the ferret heart (1011).
3) Removal of protons is insufficient. Under normal conditions, a large amount of the acidic load is removed by simple CO2 diffusion. Although the production of CO2 by aerobic metabolism ceases after coronary occlusion, it is still formed from HCO3 − as a consequence of metabolic acidosis. The acidosis will thus be different across the ventricular wall with much more acidic conditions in the subendocardial regions (1077). Differences also exist between the central and border zones of the infarct. Acidosis will also greatly depend on the presence of buffer systems, e.g., blood or saline solutions (253, 390). These considerations should be taken into account when interpreting the data obtained in experimental models where the oxidative as well as the glycolytic pathways are inhibited by metabolic blockers but superfusion and elimination of CO2 persist. The two other mechanisms for acid removal, i.e., the Na+/H+ exchanger and the Na+-HCO3 − cotransporter, are infavorably affected by the increase in [Na+]i. Upon reperfusion, pH in general rapidly recovers (398, 1011). The underlying mechanisms are washout of acid metabolites (H+, lactate−, and CO2) and Na+-coupled acid extrusion (Na+-HCO3 − cotransport and Na+/H+ exchange). The pH does not recover when hearts are perfused with HEPES buffer (no Na+/HCO3 − symport), and Na+/H+ exchanger and lactate transport are simultaneously inhibited. The most important of the three mechanisms is washout of CO2 and lactate, whereas Na+-dependent extrusion (Na+/H+ and Na+/HCO3 −) accounts for 20–35% of the recovery (1011). Among these two mechanisms, the relative contribution of Na+-HCO3 − cotransport is larger. The minor contribution of the Na+/H+exchanger is evident from the small effect of blockers on the recovery of pH (398, 769, 864). Both extrusion mechanisms and the accompanying Na+ influx, however, should be taken into account to understand the possible increase in Ca2+ load of the cell upon reperfusion. This explains why improved recovery of contractile behavior occurs (398), and arrhythmias are less frequent (29,1123) when the exchanger is blocked.
C) EFFECTS OF THE PH CHANGES ON CHANNELS AND CARRIERS.
Because channel proteins act like enzymes, large changes by protons may be expected. Most plasma membrane currents, with the exception ofI KATP and I KAA, are inhibited, some by extracellular (I Na,I Ca, I to,I Kr, I Ks), others by intracellular acidosis (I KNa,I K1) (for references, see Table1) (Fig.17). The mechanism is a decrease in single-channel conductance, P o, or both and eventual voltage shifts in the activation-inactivation kinetics. The change in conductance in general is not voltage dependent; binding of protons outside the electrical field probably reduce the local concentration of permeating cations or shift the conductance to a lower substate by an allosteric effect. TheP o is diminished by shortening of the open time (I Na,I CaL), prolongation of the closed time (I K1), and increase in nulls (I CaL). External acidosis normally causes also positive shifts in the activation-inactivation curves (I Na, I CaT,I to, I Kr). In case of shift, the effect on macroscopic current for a channel with activation-inactivation kinetics will thus depend on the resting (holding) E m (938). In normal polarized cells, the current is not changed at positive test potentials but reduced for smaller depolarizations; the difference is due to the shift of the activation curve. When the resting potential is depolarized, however, the current may paradoxically be increased, an effect due the positive shift of the inactivation curve (442, 938). An exception to the rule of voltage independence can be found in the change of macroscopic conductance of the I Kr current by extracellular protons (Fig. 17), which cause a larger reduction of the current at hyperpolarized potentials (129).
Among the plasma membrane channels, the KATP and KAA channels have to be mentioned separately, because they can show an increase in activity in the presence of elevated proton concentrations. Although in acidosis single-channel conductance is reduced and multiple substates become apparent (264), changes in P oare more important and can be positive or negative depending on the presence or absence of ATP (193, 590). In the presence of ATP, the P o (number of openings and open time) is increased for moderate decreases in pH (down to 6.0). The concentration response curve is shifted to higher [ATP] (560). The presence of Mg2+ seems required, which may suggest a change in phosphorylation of the channel, although this process has not been verified (193). For a fall of pH below 6.0, P o is markedly reduced (560), an effect which may be caused by a faster rundown in inside-out patches. In the absence of [ATP], acidosis still increases mean open time but reduces overallP o (193). In trypsin-treated patches, where rundown is absent, acidosis consistently increased activity in the presence of [ATP] (266, 304) but had no effect in ATP-free medium.
The gap junction channel is very sensitive to acidosis (109, 285, 732,1091). Synergistic (109, 1091), additive (285), but also antagonistic interactions (732) have been described for simultaneous increases in [H+] and [Ca2+]. The reduction in conductance by an increase of [H+]i has been explained by a ball-receptor mechanism, comparable to the inactivation in voltage-operated channels, with the COOH terminal acting as the negatively charged ball and the M2-M3 link as the positively charged receptor (933).
Among the transporters, the Na+/Ca2+ exchanger is very sensitive to pH and strongly inhibited by acidosis (230, 764). Intracellular protons act as substrate on the Na+/H+ exchanger, but extracelluar acidosis inhibits (1029, 1114).
D) ELECTROPHYSIOLOGY AND PH CHANGES.
The effects of acidosis at the channel level are translated at the multicellular level into a fall in resting potential, a reduction of upstroke velocity, a prolongation of the action potential duration, with eventual oscillations at the plateau level and occurrence of early afterdepolarizations (EAD) (175). Refractoriness is prolonged, due to a slowing of the Na+ current recovery. Conduction of the action potential is retarded (319) as a consequence of the decrease in excitatory current and a fall in gap conductance. Of these changes, the prolongation of the action potential and the occurrence of EAD are not seen during ischemia but may be present upon reperfusion. During ischemia, the action potential is shortened as a consequence of extracellular K+ accumulation and activation of outward currents, mainly carried by KATP and KAA (see sects. iii A1 and iv A). All the acidosis-induced changes are proarrhythmic and may explain why the threshold for ventricular fibrillation is reduced in acidosis (950).
Under aerobic conditions, the continuous production of protons is compensated by CO2 elimination, Na+/H+ exchange, and Na+-HCO3 − cotransport. During ischemia, pHi and pHo fall by one pH unit or more. The fall is caused by an increased production and reduced removal of protons. Most plasma membrane channels with the exception of KATP and KAA, as well as gap junction channels, the SR Ca2+release channel, and Na+/Ca2+ exchange current are inhibited, some by extracellular and others by intracellular acidosis. The mechanism for the channels is a decrease in single-channel conductance and/or a reduction inP o. In most instances, except forI Kr, the effect is voltage independent. These changes result in depolarization, prolongation of the action potential, and occurrence of EAD.
3. Accumulation of [Na+]i
A) DESCRIPTION OF NA+ DISTRIBUTION.
Cytoplasmic free [Na+] concentration measured with ISE, NMR, and fluorescent dyes varies between 5 and 10 mM intracellular water for different animal species (see Refs. 30, 1037). Species differences exist, and [Na+]i is higher in the rat and mouse ventricle cells (15–16 mM) (1143). In single cells, actual values depend on the cell isolation procedure; the presence of taurine, for instance, improves the quality of the cells and reduces [Na+]i in rat cells from 20 to 10 mM (232, 945). The [Na+]i is significantly higher in subepicardial than in subendocardial cells (173).
It is important to realize that [Na+]iis not uniformly distributed in the cytoplasm. Recent estimations with electron-probe microanalysis (EPMA) (642,1085), concomitant with evaluation ofI KNa activation (1085), show that the [Na+] in the subsarcolemmal space may differ substantially from the bulk cytoplasmic phase. Measurements of the time course of the Na+ pump current (82) and the change in Na+/Ca2+ exchange current (303) upon reactivation of the Na+ pump are consistent with the presence of a fuzzy space in which diffusion of Na+ is restricted (see Refs. 127, 591, 872). In mitochondria, the [Na+] is only half the cytoplasmic value (fluorometry, Ref. 232; EPMA, Ref. 1085).
A substantial monotonic increase in [Na+]i occurs during ischemia or metabolic inhibition (NMR, Refs. 635, 769, 1015; fluorometry, Ref.232). An increase of two- to fivefold with actual values of 20–25 mM have been recorded after 15–20 min of ischemia. These results stay in contrast to earlier results obtained using the ISE technique in the guinea pig (544, 1018) where no rise and even a small fall have been reported. A possible reason for the divergence in results is the preferential recording from superficial and thus from less ischemic, less acidotic cells with the ISE technique. A small fall in cytoplasmic [Na+] may be due to a shift from the cytoplasm to the mitochondria. Such a shift would be difficult to detect with other techniques. Recent experiments with ISE in the rabbit, however (1136), show a doubling of [Na+]i after 15 min (Fig.16 B). Of importance was the observation that simultaneous stimulation of thrombin receptors caused a rise in [Na+]i that was to 2.5 times greater, implying a role for thrombin receptors in the activation of a Na+-permeant pathway (1136). Concomitantly, a greater accumulation of LPC in the membrane occurred, and the probability of arrhythmias increased (180,751).
Information on recovery of [Na+]i upon reperfusion is limited. If ischemia is restricted in duration to 20–30 min, [Na+]i decreases monotonically and normalizes within 5 min (635). A transient slight increase at the very start of the reperfusion has been recorded in the rat (962). In the same species, recovery was fast initially but remained incomplete (769, 1015).
B) MECHANISMS INVOLVED IN THE CHANGES OF [NA+]I DURING ISCHEMIA.
During ischemia, an increase in [Na+]iis caused by a reduced active outward movement and an increased passive inward leak.
A partial block of the pump during early ischemia is highly probable and may explain the observed changes in [Na+]i (232). Inhibition of the pump is due to the fall in free energy of ATP hydrolysis as a result of changes in [ATP], [ADP], and [Pi]; a reduction in enzymatic activity under the influence of LCAC (959) and oxidative stress (880); and a reduction in the number of Na+ pump molecules in the plasma membrane consequent to endocytosis (Shattock, personal communication). Theoretical calculations show that a reduction of the pump activity by one-fourth of its maximal activity will result in a change of [Na+]i amounting to 3.4 mmol · l−1 · min−1 (Na+influx is assumed to remain constant). Under normal conditions, the energy available from ATP hydrolysis amounts to 61.5 kJ/mol, but during ischemia, it can fall to 49 kJ/mol or less. Concomitant with this fall, the E rev of the pump shifts in the depolarized direction and approaches the resting potential (190).
Inward leak of Na+ occurs via carriers and channels. An important carrier is the Na+/H+ exchanger; extrusion of protons via the Na+/H+ exchanger causes Na+ to move inward. Block of the Na+/H+ exchanger has been found to reduce [Na+]i increase during ischemia (19, 769), although it had no effect on pHi (398). In other experiments, the beneficial effect of Na+/H+ exchanger block was restricted to reperfusion (962).
Sodium ions move inward through TTX-sensitive Na+channels and NSC channels. The fast Na+ channel is modulated by the rise in LPC to open repetitively and show continuous activity at the resting potential level (107,1002). Block of this pathway by lidocaine reduces this leak inward current (112, 1016).
Activity in NSC channels is induced by stretch (188,529, 815), [ATP]o(301), free radicals (472), and a rise in [Ca2+]i (172,251, 410), all changes which occur very early during ischemia. Some of these NSC channels activated by stretch are equally permeable to Ca2+ (526). In pacemaker cells (Purkinje fibers), an important influx of Na+ occurs via the I f channel (327).
The fast recovery of [Na+]i upon reperfusion is somewhat unexpected, since washout of protons from the extracellular space should stimulate Na+/H+exchange and lead to a net Na+ influx. Activation of the Na+-K+ pump however seems to prevail in most cases. Results may vary depending on the relative activity of the Na+/H+ exchanger and the pump. When glycogen is still available, glycolysis causes acidification and an increase in Na+/H+ exchange. In those instances, inhibition of the exchanger improves recovery of contractile activity and prevents arrhythmias (28, 29, 1123). When cells are depleted of glycogen, acidification is less, and activity of the exchanger remains limited. Under those conditions, [Na+]i does not increase upon reperfusion, and only a rapid fall is observed.
C) EFFECT OF [NA+]I INCREASE ON CHANNELS AND CARRIERS.
Intracellular [Na+] if sufficiently elevated, activates specifically the I KNa (497,625), and enhances I Ks(978) and I KACh(943). In the case of I KNa, an increase in [Na+]i also reduces the single-channel conductance (1068), but this effect is largely compensated by the increase in P o. Outward current through other K+ channels, such asI K1 (654, 848) and I KATP (494,434), is blocked by intracellular Na+; the block is voltage dependent, rapid, and seen as a decrease in overall conductance.
Intracellular Na+ is the main substrate that activates the Na+-K+ pump current. The process becomes important during reperfusion when oxygen tension is restored. On the Na+/Ca2+ exchanger, a rise in [Na+]i favors reverse-mode activity, causing extra Ca2+ load. The relation between Na+ and Ca2+ load explains the close relationship between [Na+]i and the frequency of spontaneous Ca2+ release (216). Via interaction with the endogenous XIP region, a rise in [Na+]i causes a state of inactivation of the exchanger (662, 663). The simultaneous rise in [Ca2+]i, however, will partly compensate for this inhibitory mechanism.
D) EFFECT OF [NA+]I INCREASE ON ELECTROPHYSIOLOGY.
All changes in channels and carriers, except for the reduction inI K1 and I KATP, tend to hyperpolarize the membrane, but the effect will only be visible during reperfusion. The marked hyperpolarization of the diastolicE m and the excessive shortening of the action potential, despite the low [K+]o, can be ascribed to stimulation of the pump and eventual activation of outward K+ currents. Insufficient recovery of the Na+ pump or excessive Na+ influx via the Na+/H+ exchanger may be responsible for increased Ca2+ influx via the reversed Na+/Ca2+ exchanger, causing Ca2+overload, absence of recovery, and eventual arrhythmias. Prevention of Na+ and Ca2+ overload by blocking the Na+/H+ exchanger (19,548, 769, 863) explains the inhibitory effect of such intervention on reperfusion arrhythmias.
During ischemia, [Na+]i increases from 10 to 20 mM after 20–30 min. The mechanism is reduced active outward pumping (inhibited by the fall in free energy of ATP hydrolysis by LCAC and radicals) and an increased inward leak (through the Na+/H+ exchanger,I Na, I NSC, andI Cl). The effect of increased [Na+]i is activation of the pump and eventually of I KNa. During reperfusion, these effects result in hyperpolarzation and short action potentials combined with low [K+]o. Dispersion is pronounced and is favorable to arrhythmias.
4. Changes in the distribution of [Ca2+]i
A) CA2+ DISTRIBUTION UNDER NORMAL CONDITIONS AND CHANGES DURING ISCHEMIA AND REPERFUSION.
I) Ca2+ distribution under normal conditions. Under normal conditions, concentrations of Ca2+ differ in the cytosol, the SR, the mitochondria, and the nucleus (for review, see Ref. 773). In each compartment, it plays a different role. It modulates the activity of myofilaments in the cytosol and of ionic channels in the plasma membrane. The SR acts as a store for Ca2+ and modulates contraction and relaxation. In the mitochondrion, the free [Ca2+] regulates the activity of three dehydrogenases (pyruvate dehydrogenase, α-oxoglutarate dehydrogenase, and NAD-dependent isocitrate dehydrogenase) and adapts in that way energy production to energy utilization. In the nucleus, Ca2+ modulates gene expression.
Free [Ca2+] can be estimated in the cytoplasm (909) and mitochondria (341,687) using phosphorescence, fluorometric methods, and NMR spectroscopy (640). For the measurement of [Ca2+] in mitochondria, the signal in the cytoplasm is Mn2+ quenched or the cytoplasm is unloaded from the indicator by washout.
Cytoplasmic [Ca2+] during diastole and at low stimulation frequencies is on the order of 100 nM (from 50 to 200 nM), with somewhat higher values in the subendocardial layers (116). During systole, this value may increase to between 500 and 1,000 nM or even higher values depending on the Ca2+ load in the SR; similar values are obtained on rapid application of caffeine, suggesting that most of the SR Ca2+ can be released during a contraction (see Ref. 73). From a comparison with values for total [Ca2+], it appears that only a very small fraction or <0.03% is present as the free ion at rest and most (order of 0.5 mM) is bound to cytoplasmic proteins; during activity, this fraction rises to 0.1%.
In the mitochondrion, the free [Ca2+] concentration at rest or during diastole is somewhat below the cytoplasmic [Ca2+] concentration (225, 312,687, 870) or close to it (888). Because the mitochondrial matrix is negative with respect to the cytoplasm (estimations vary between −150 and −180 mV), this means that mitochondrial [Ca2+] is far from equilibrium.
The [Ca2+] in the mitochondrion is variable depending on the cytoplasmic load. When cytoplasmic [Ca2+] is <500 nM, the gradient of mitochondrial [Ca2+] to cytoplasmic [Ca2+] is less than unity but increases in an exponential way at higher values (687). Mitochondrial [Ca2+]-to-cytoplasmic [Ca2+] ratios larger than unity have been measured during stimulation of isolated cells at 2 Hz (225). Total mitochondrial [Ca2+] measured by the electron probe transiently increases during a depolarization (461, 1084); the increase in free Ca2+ transient during metabolic inhibition has been interpreted as due to a fall in Ca2+ uptake by the mitochondria during the cardiac systole. The rise in free [Ca2+] has been confirmed using a combination of a fluorescence and confocal microscopy technique (140). In previous fluorescence measurements, such fast changes were not detected. The reason for this divergence is not clear but may be related to the buffer characteristics of the fluorescent probe (461). Mitochondria are thus no source of Ca2+for the cytoplasm during stimulation but rather act as a buffer. Because mitochondria represent a large fraction of the cell volume, they can accumulate massive amounts of Ca2+(43), a situation which occurs during pronounced and prolonged ischemia. The amount of Ca2+ stored in the mitochondrion is sufficient to lead to a contracture when the mitochondrial E m breaks down after application of an uncoupler (1106).
In the cardiac SR, total [Ca2+] expressed per liter cytosolic volume has been measured to be between 100 and 200 μM (see Refs. 73, 323, 1024), with estimations of free [Ca2+] per liter SR volume at 700 μM (877). An enormous gradient of 7.000 thus exists between the SR and cytoplasm.
II) Changes of [Ca2+]i during ischemia and reperfusion. During anoxia, metabolic inhibition or ischemia, diastolic free [Ca2+]i rises with some delay. The delay is variable, and quite important differences in time course have been reported, depending on the model. In studies of ischemia (coronary ligation), the delay attained 10–20 min and the rise in [Ca2+]i slightly preceded the development of contracture and uncoupling of the gap junctions (640, 1047) (Fig. 16 D). A much shorter delay between 2 and 5 min has been described in other studies (519, 689). A large variation in time course also exists in experiments on single cells subjected to hypoxia or metabolic inhibition (see references in Ref. 817). The reasons for these divergent results are not known but are probably related to the intensity of the metabolic blockade.
Results on systolic [Ca2+] are variable. A transient rise may occur (Fig. 18 A) (461) but is not always present [total heart (689, 1047), ventricular muscle (920), and singe cells (253,461)] and is followed by a decrease when diastolic [Ca2+]i rises (519,640). The initial rise is not accompanied by an enhancement of the contraction and is probably due to a shift of Ca2+ from the mitochondrion to the SR (461) (Fig. 18 A) and/or a release from binding sites by the increase in intracellular acidosis (312). The fall in systolic [Ca2+]i may be accompanied by oscillations during diastole, caused by release from the SR (906).
In mitochondria, an increase is seen upon metabolic inhibition when the ratio of mitochontrial-to-cytoplasmic [Ca2+] is low, a decrease when the starting level is greater than unity (225). In this last case, Ca2+ initially is taken by the SR and the Ca2+ transient is increased. Later, when the cytosolic [Ca2+] increases, the mitochondrial [Ca2+] simply follows and rises to the same extent. Release of Ca2+ from mitochondria and contracture development have been demonstrated when mitochondria are uncoupled (1106).
Changes in SR Ca2+ content during ischemia have not been measured directly. From the temporary increases in Ca2+transient that have been recorded (461), a higher content early in ischemia can be deduced. Later on, less Ca2+ will be stored in the SR as a consequence of inhibition of the SR Ca2+-ATPase activity (503).
Upon reperfusion, the change in diastolic [Ca2+] can follow three different time courses: 1) a rapid decline to the normal value (ferret, Ref. 920; rat, Refs. 641, 1047);2) an initial fall to an intermediate value followed by oscillations that can end with a return to normal or be followed by a secondary increase (773, 817,906) (Fig. 18 B); at the multicellular level oscillations may then trigger arrhythmias (703);3) no recovery but an immediate rise to elevated values in the micromolar range accompanied by contracture and irreversible injury (11, 397, 906,962). Irreversibility is often the outcome when the cytoplasmic [Ca2+] has risen above 1 μM, and this occurs when the ischemic period is extended over 20 min.
Data on changes in mitochondrial [Ca2+] upon reoxygenation or reperfusion are scarce. On theoretical grounds an increase in mitochondrial [Ca2+] may be expected when oxidative metabolism restarts the mitochondrial battery and increases the matrix negativity. No change in mitochondrial [Ca2+] was seen on reoxygenation after 50 min of hypoxia, but a dramatic 10-fold increase was seen after 80 min of hypoxia. Addition of ruthenium red that blocks Ca2+ uptake in the mitochondria exerted a protective effect (11). Large increases in mitochondrial [Ca2+] are accompanied by irreversible changes and hypercontracture (687). The mitochondrial damage is not due to Ca2+ alone but to the concomitant increase in phosphate, LCAC, proton concentration, oxidative stress, and low [ATP], changes that result in the opening of the mitochondrial mega-channel. This contention is based on the observation that large Ca2+ overload caused by digitalis is easily reversible when it is not accompanied by energy shortage (595).
Recordings of Ca2+ transients suggest that reuptake of Ca2+ in the SR is quickly restored upon reperfusion. In some cases, however, overload is the consequence with spontaneous oscillatory release and generation of EAD or DAD.
B) MECHANISMS FOR [CA2+]I CHANGES.
The increase in cytoplasmic [Ca2+] in the cells submitted to ischemia is due to a less efficient removal from the cell via the Na+/Ca2+ exchanger and a reduced Ca2+ uptake in the SR, an increased inward leak through the plasma membrane, and displacement of Ca2+ from binding sites in the cytoplasm and in mitochondria by protons. Mitochondria are probably not responsible for an increase in cytoplasmic [Ca2+] except in excessive overload; in most instances, the mitochondria absorb Ca2+ during ischemia.
In the early stage of ischemia, free [Ca2+] in the cytoplasm increases due to a displacement of Ca2+ by H+ from binding sites (312). The experimental basis for this mechanism was provided by the observation that an increase of CO2 tension from 5 to 20% caused [Ca2+] in the cytoplasm to rise from 130 to 221 nM. The effect was not changed by application of ryanodine (block of the SR Ca2+ channel) or ruthenium red (block of mitochondrion uniporter).
Calcium influx and efflux through the plasma membrane that are normally in equilibrium (986, 1057) go out of balance during ischemia with Ca2+ influx exceeding Ca2+efflux. The most important mechanism for removal of Ca2+from the cell is the Na+/Ca2+ exchanger, and it is responsible for 77% of the Ca2+ extrusion (583, 986). During ischemia, the Na+/Ca2+ exchanger is less efficient because of an increase in [Na+]i and [H+]i. An increase in [Na+]i promotes the reversed mode of the Na+/Ca2+ exchanger whereby Ca2+ is entering the cell and Na+ is removed (358,920, 962). All mechanisms responsible for an increase in [Na+]i thus also contribute to the increase in [Ca2+]i via the exchanger. The efficiency of the exchanger is furthermore reduced by the acidosis (230, 764) and the influence of radicals (163).
Whereas Ca2+ efflux is reduced, Ca2+ influx is increased during metabolic ihibition. The role of T- and L-type Ca2+ channels in this respect is negligible. Background channels carrying Ca2+ inward become more important during ischemia as they are activated by radicals (472,1056). Inward leak of Ca2+ also occurs via NSC channels (see sect. iii A5) that are activated by [ATP]o (301, 411), mechanical stretch (188, 526,841), and a rise in [Ca2+]iitself (172, 251, 650).
During ischemia, less Ca2+ is stored in the SR, an effect mainly due to block of Ca2+-ATPase activity, whereas changes in the release mechanism are less important. The decrease in Ca2+ reuptake is due to a reduction ofV max, not to a change inK d (0.5–0.7 μM) or the Hill coefficient (503). The effect is related to the reduction in free energy change of ATP hydrolysis (339, 644) and modification of the carrier by oxygen radicals (disulfide bridge formation) (565). Although the SR Ca2+ channel can become activated by exposure to oxygen radicals (429) or to arachidonic acid and its metabolites (196), the increased levels of [H+], [Mg2+], and [PO4 3−] exert a blocking effect. The overall result is probably of less importance.
Influx as well as efflux of Ca2+ in the mitochondria depend on the electrical and proton gradient (Fig. 2). Influx through the uniporter is driven by the electrical potential difference that is estimated to be −150 to −180 mV. Efflux from the mitochondria depends on the proton gradient and is continuously guaranteed by the Na+/Ca2+ exchanger and the Na+/H+ exchanger which act in concert to keep intramitochondrial [Na+] and [Ca2+] at low levels. The immediate source of energy lies in the Na+gradient, but the final energy source is respiration and the proton gradient. Block of the oxidative metabolism, resulting in the disappearance of the electrical and proton gradients, is thus expected to result in equilibration of the mitochondrial [Ca2+] and the cytoplasmic [Ca2+] with a ratio of mitochondrial-to-cytoplasmic [Ca2+] approaching unity. Because the cell volume fraction of mitochondria is 30–36%, they may absorb a large amount of Ca2+ and act as a buffer.
There are mechanisms responsible for [Ca2+]i changes during reperfusion. Depending on the severity and duration of the ischemic period, the cytosolic [Ca2+] concentration may normalize or on the contrary increase to dramatic levels leading to cell death.
The rapid initial recovery phase seems to depend on the activity of the SR Ca2+-ATPase (see Refs. 503, 697). Although the Ca2+-ATPase activity and the number of ryanodine receptors (1005) are reduced, uptake in the SR and increase of the luminal [Ca2+] may be so large that threshold for release is attained resulting in oscillations (987). If, in the mean time, removal of Ca2+ from the cell through the Na+/Ca2+ exchanger is restored, the Ca2+ level may return to normal values (773,906). In some cases, however, the oscillations are followed by a secondary increase. Recovery depends much on the restoration of the Na+ gradient and is facilitated by blockade of the Na+/H+ exchanger (578, 864) or acidosis (see Ref. 91). Protection against Ca2+ overload contracture is also provided by the natriuretic peptide that stimulates guanylate cyclase (397). On the other hand, production of IP3 secondary to α1-receptor stimulation may inhibit recovery (17, 236).
A critical situation may occur when the ischemic period has been too long and restoration of the Na+-pump activity has been retarded. In this case, the [Ca2+] remains elevated. On restoration of the oxygen supply, the electron flow through the oxidative chain generates an electrical gradient in the mitochondrion causing a massive absorption of Ca2+ from the cytoplasm. The presence of such an abnormally high [Ca2+] together with low [ATP] and oxidative stress may cause the opening of the mitochondrial mega-channel (Figs. 2 and19). The ATP, instead of being exported to the cytoplasm, may even be hydrolyzed to generate a proton gradient and limit the absorption of Ca2+ in the mitochondrion. The final result is disappearance of all mitochondrial gradients, hypercontracture, and cell death (244, 340). It is important to note that high [Ca2+] alone does not cause the permeability transition as shown by experiments with toxic concentrations of digitalis in which [Ca2+] increases to much higher levels than in ischemia and cells remain viable (595).
C) EFFECTS OF [CA2+]I ON CURRENTS AND CARRIERS.
An increase in [Ca2+]i modulates multiple currents in cardiac cells. The L-type Ca2+current itself is very sensitive to increases in [Ca2+]i and shows an increased rate of inactivation, limiting in this way the Ca2+ load of the cell (see Ref. 673). In contrast, the current may also undergo facilitation for moderate increases of the Ca2+ level (505) (see sect. ii A2).
Some currents are dependent on [Ca2+]ifor their activation: I Cl(Ca)(910, 1192), I NSC(172, 251), the Ca2+ release channel (120, 261), and the Na+/Ca2+ exchanger (686). An increase in [Ca2+]i stimulates the Na+/Ca2+ exchanger and counteracts in this way the Na+-induced inactivation (663). Other currents are modulated; an increase is seen for the T-type Ca2+ current (994),I Ks (728, 978),I Kr (851),I f (354), the Na+-K+ pump (314), and the mitochondrial MCC (189, 952). Two currents are inhibited: the I K1 (654,670, 671), a phenomenon comparable to the block by [Mg2+]i, and the gap junction channel (206, 732).
The [Ca2+]i influences furthermore a number of enzymes. Adenylyl cyclase is inhibited (1154), and others are stimulated, e.g., mitochondrial dehydrogenases (692), the ATP synthase (199), PKC, phosphatases, and proteases (causing rundown of channels) (see Ref.496).
D) ELECTROPHYSIOLOGICAL EFFECTS OF [CA2+]I.
Enhancement of Ca2+-induced inactivation of the Ca2+ current, increase of the transient Ca2+-activated Cl− current, and of the two delayed K+ currents will lead to shortening of the action potential.
When the rise in [Ca2+] is more pronounced (Ca2+ overload), EAD as well as DAD are generated, and arrhythmias may result (519, 971). One of the mechanisms underlying EAD is reactivation of L-type Ca2+ current (684), but also release of Ca2+ from the SR (782), followed by activation of the “transient inward current” can cause EAD. The transient inward current (not described as a separate current in this review) is a composite current and is carried through the Cl− channel (747, 912), the NSC channels (121, 368), and the Na+/Ca2+ exchanger (369,579). Although the primary activator for Ca2+release is the cytoplasmic Ca2+, other factors such as ATP (1120), the phosphorylation state of the channel (754, 1006), and luminal [Ca2+] (917) determine the efficiency of the release mechanism or, what has been called, the gain of the system (120). Under conditions of moderate Ca2+ load, the release is linearly related to the luminal [Ca2+] concentration and not propagated; the gain is constant. With increase of Ca2+load however, gain increases nonproportionally, and release becomes regenerative and propagated (44, 367,476, 847, 940). Under these conditions, release may be spontaneous and occur in the absence of trigger Ca2+. The result is EAD during the action potential or DAD, following repolarization. Important to note is that Ca2+ overload also increases the tendency in Purkinje fibers to become spontaneously active; the activation curve of theI f current is shifted in the positive direction (354, 1172), which favors activation of the current and increases the rate of diastolic depolarization. The fall in gap junction conductance by [Ca2+]i further enhances the probability for arrhythmias by reducing conduction (206,732). In a final stage, activation of the mitochondrial mega-channel uncouples the ATP synthase, causes hydrolysis of ATP, and leads to rigor and contracture (189,244). High [Ca2+] alone, however, is not sufficient (595); a simultaneous fall in [ATP] and the presence of oxidative stress that may occur during reperfusion creates a favorable condition for activation of the channel (189,244) (Fig. 19). Once activated, the opening of the MCC channel leads to irreversible processes and cell death.
The [Ca2+] differs in the cytoplasm, the SR, and the mitochondria. Whereas systolic [Ca2+] may transiently increase during the first minutes of ischemia, diastolic [Ca2+] only increases after a delay of 10–20 min. In the mitochondria, [Ca2+] follows the changes in cytosolic level, although there may be an initial release if the [Ca2+] in the mitochondria is higher than in the cytoplasm at the start of ischemia. The increase in cytosolic [Ca2+] is due to a less efficient removal to the extracellular space via the Na+/Ca2+ exchanger, a reduced uptake in the SR, an increased inward leak, and a displacement by protons from binding sites. An increase of cytosolic [Ca2+] activates a number of channels, carriers, and enzymes (I NSC,I Cl, andI Na,Ca) and modulates others (I CaL, I CaT,I Ks,I Kr,I f, I Na,K, and gap junction channels). The result is the occurrence of EAD, DAD, and arrhythmias. If on reperfusion, removal of Ca2+ to the extracellular space and into the SR is insufficient, while O2 tension is restored and the mitochondrionE m energized, a massive absorption of Ca2+ may occur in the mitochondrion and result in activation of the mitochondrial mega-channel, which signals irreversibility.
5. Changes in [Mg2+]i
A) DISTRIBUTION AND REGULATION OF [MG2+] UNDER AEROBIC CONDITIONS AND ISCHEMIC CONDITIONS.
Estimations of total Mg2+ in heart cells using EPMA (see Ref. 701) vary between 50 and 80 mmol/kg dry wt for the cytoplasm (A band), 20–80 mmol/kg dry wt for mitochondria, and 20–75 mmol/kg dry wt for junctional SR. From the estimated dry weight-to-wet weight ratio of 10% for the cytoplasm and 25% for mitochondria, these values are translated in 5–8 mM for the cytoplasm and 6–20 mM for mitochondria.
Measurements of free [Mg2+] in the cytoplasm, using ISE, NMR, and fluorescence techniques, have yielded data between 0.5 and 0.8 mM intracellular water (299, 701). In contrast to Ca2+, which is largely concentrated in the SR, Mg2+ does not seem to be concentrated in a particular intracellular compartment. The free [Mg2+] in mitochondria, estimated to be 0.8 mM (818), is similar to the cytoplasmic [Mg2+]. Isolated mitochondria can absorb and extrude Mg2+ in a respiration-dependent way, but in vivo this does not seem to result in any large gradient.
In the cytoplasm, only about one-tenth of the total [Mg2+] is free; the rest is bound to nucleotides (ATP) and proteins. A competition between Mg2+ and protons has been demonstrated by the fall in free [Mg2+] when the cell is subjected to a pulse of alkalosis (299).
After 10–15 min of ischemia, free [Mg2+] has been found to increase from 0.5–0.8 mM to 2–6 mM (539,702, 869). Upon reperfusion, the level of free [Mg2+]i decreases to the original level (869) but may remain transiently elevated at 1.5 mM (539, 702).
B) MECHANISMS OF [MG2+] REGULATION.
Free [Mg2+] in the cytoplasm is of the same order of magnitude as the extracellular concentration. With the fact that the intracellular medium is negative taken into account, the distribution of free [Mg2+] is thus far from equilibrium. Thermodynamically active transport is required. Two mechanisms are involved: a Mg2+-ATPase and a Na+/Mg2+ exchange process (361,907). In favor of a Na+/Mg2+exchange mechanism are the observations that [Mg2+]i increases upon application of ouabain. Reduction of [Na+]o also results in an increase of [Mg2+]i(214). The [Mg2+] content can be modulated by β-receptor stimulation, causing a release of intracellular Mg2+, and by M2 receptor stimulation, leading to the opposite effect. These changes may be secondary to mobilization of the mitochondrial [Mg2+] pool by an effect on the adenine nucleotide metabolism.
The most important mechanism for the increase in [Mg2+]i during ischemia is the net hydrolysis of ATP to which Mg2+ was bound. Block of the Mg2+-ATPase and reduction of the activity of Na+/Mg2+ exchange mechanism, however, should not be neglected.
C) ELECTROPHYSIOLOGICAL EFFECTS OF INTRACELLULAR MG2+.
Intracellular [Mg2+] is needed to activate enzymes that phosphorylate or dephosphorylate channels (Na+, Ca2+, K+, Cl−, and f channels). On the other hand, as a cation interacting with the pore structure, it blocks a number of channels, and in the case of K+ channels generates inward rectification.
An increase in [Mg2+]i blocks outward current through the fast Na+ channel. The block is slightly voltage dependent with an electrical distance of 0.18 from inside. The phenomenon has only a biophysical importance (66).
On the Ca2+ channel, [Mg2+]iexerts two effects: it is necessary for the phosphorylation of the channel, but an increase to concentration levels reached during ischemia reduces the current. The effect is direct through block and indirect via a change in phosphorylation state (1090,1132) or modulation by Mg-nucleotide complexes (743). Sensitivity for Mg2+ block is dependent on the state of phosphorylation and increases in the dephosphorylated state (1131). Conductance is not changed, but the number of functional channels and the P o are reduced (1130). At very high concentrations of 9 mM, the current is completely blocked (see Ref. 5).