B. Fakler, Department of Physiology II, Ob dem Himmelreich 7, 72074 Tübingen, Germany. Fax: + 49 7071 87815, Tel.: + 49 7071 297 7173, E-mail: firstname.lastname@example.org
Inward-rectifier potassium (Kir) channels comprise a superfamily of potassium (K+) channels with unique structural and functional properties. Expressed in virtually all types of cells they are responsible for setting the resting membrane potential, controlling the excitation threshold and secreting K+ ions. All Kir channels present an inwardly rectifying current–voltage relation, meaning that at any given driving force the inward flow of K+ ions exceeds the outward flow for the opposite driving force. This inward-rectification is due to a voltage-dependent block of the channel pore by intracellular polyamines and magnesium. The present molecular–biophysical understanding of inward-rectification and its physiological consequences is the topic of this review. In addition to polyamines, Kir channels are gated by intracellular protons, G-proteins, ATP and phospholipids depending on the respective Kir subfamily as detailed in the following review articles.
About half a decade after cloning of the first K+ channels from the Shaker locus of Drosophila melanogaster[1–3], the first two genes encoding Kir channels were isolated in 1993 by expression cloning [4,5]. These two Kir proteins termed ROMK1 (for renal outer medulla K+ channel) and IRK1 (inward rectifier K+ channel) demonstrated the prototypic structure of this type of K+ channels: they are made up of two transmembrane domains (TM1 and TM2) flanking a well-conserved pore-loop (P-loop) and large hydrophilic N- and C-termini located on the cytoplasmic side of the membrane ( Fig. 1A). This topology, first based on hydrophobicity analysis only, has recently been confirmed by the landmark effort of MacKinnon’s group that resulted in successful crystalization of KcsA , a structural homologue of Kir channels from Streptomyces lividans ( Fig. 1B). The crystal structure showed that two-segment type K+ channels are assembled from four subunits that are symmetrically arranged around a central pore lined by TM2 and the P-loop, in good agreement with functional studies performed on cloned Kir channels [8–11]. As in all other K+ channels known, the P-loop of the two-segment type K+ channels contains the famous glycine-tyrosine-glycine (GYG) or glycine-phenylalanine-glycine (GFG) stretch  that establishes the channels’ filter for K+ selectivity ( Fig. 1B).
After isolation of the first Kir proteins, subsequent homology-based cloning procedures uncovered a series of further Kir subunits that all shared the hallmark structural properties described above and showed that Kir channels indeed form a superfamily of K+ channel proteins [13–19]. Based on sequence homology, this superfamily was divided in up to seven subfamilies (Kir1–Kir7, reviewed in [20–22]) with up to four members each ( Fig. 1C); between subfamilies the level of identity in primary sequence is ≈ 40%, while individual members within one subfamily share an identity of ≈ 60% [20–22]. Upon heterologous expression, all Kir proteins were shown to be K+ selective and to exhibit an inwardly rectifying current–voltage (I–V) relation [21,22]. The latter reflects the fact that Kir channels mediate a high K+ conductance at membrane potentials negative to the K+ equilibrium potential (EK) and at a limited voltage-range positive to EK. Further depolarization of the membrane potential leads to a rapid decrease in the Kir-mediated conductance which consequently prevents outward K+ currents [23,24].
Besides common properties, the members of the various Kir subfamilies show a number of characteristics that are either unique to certain subfamilies or are only shared by a few of them. Thus, Kir1 and Kir4 channels, which are predominantly expressed in epithelia, are exquisitely sensitive to changes in intracellular pH in the physiological range. Both channel subtypes are closed down by acidification, while alkalinization results in channel opening . The members of the Kir3 subfamily are expressed in a variety of tissues and cell types, among them cardiac myocytes and central neurons, and, uniquely among Kir channels, they are activated by GTP-binding proteins (G-proteins) via stimulation of G-protein-coupled receptors [13,16,26,27]. Kir6 channels display a ubiquitous expression pattern  and in coassembly with the sulfonylurea receptor (SUR) represent the well known ATP-sensitive K+ channels (KATP) [28–30].
The members of the Kir2 family are predominantly expressed in heart, skeletal muscle and nervous system and represent the typical ‘strong’ rectifiers found in these tissues . The term ‘strong’ refers to the high voltage-dependence of rectification in these channels [24,31–33]. The nature of this voltage-dependence as well as the mechanism and physiological implication of the strong inward-rectification is reviewed in the following sections.
The mechanism of inward-rectification
Since the first description of inwardly rectifying currents by Katz , the mechanism of rectification has been subject to investigation. Armstrong  suggested that inward-rectification might result from a positively charged particle that blocks the ion conducting pathway from the intracellular side of the membrane. Indeed, extensive work on native inward rectifiers (from cardiomyocytes) showed that the divalent magnesium (Mg2+) caused inward-rectification by blocking the channel pore in a voltage-dependent manner [36,37]. However, in some cases inward-rectification remained even in the absence of intracellular Mg2+. Hence, the remaining voltage-dependent rectification was ascribed to a mechanism intrinsic to the channel protein and termed ‘intrinsic gating’ to distinguish it from the ‘extrinsic’ Mg2+-induced pore block [38,39].
However, it was not recognized until cloned Kir channels were available, that neither of these suggested mechanisms could fully account for the rectification observed with these channels under cellular conditions. Thus, intrinsic gating could be washed-off from excised membrane patches similar to the soluble blocker Mg2+[31,32]; furthermore Mg2+-induced rectification was significantly less voltage-dependent than that observed under cell-attached conditions, i.e. with the channels contacting the cytoplasm [24,32]. In three independent approaches, the mystery was resolved by identifying the naturally occurring intracellular polyamines, spermine (SPM) and spermidine (SPD), as the ‘rectifier substance’ of Kir channels [24,31–33]. Both polyamines are present in the cytoplasm at micromolar concentrations  and block outward currents through Kir channels in a steeply voltage-dependent manner ( Fig. 2A). The voltage-dependence of SPM- and SPD-mediated rectification is considerably steeper than that of Mg2+, which explains why rectification observed under native conditions is steeper than predicted by Mg2+ block alone . Intriguingly, polyamines also delivered the key to understanding intrinsic gating. As illustrated in Fig. 2B, release of SPM and/or SPD from blocked channels is a time-dependent process, while Mg2+ is released instantaneously. The latter combined with the fact that the positively charged polyamines can hardly be removed from cells or patches [31,32] explains why polyamine block appeared to be an intrinsic gating process [24,33].
What determines the much higher voltage-dependence of the polyamine block with respect to pore block by Mg2+? The ‘Woodhull model’ describing block of ion channels  assumes that voltage-dependence of the block is roughly determined by the number of charges that move during the blocking process times the distance these charges move within the electrical field (electrical distance). In agreement with such a model, the blocking potency and voltage-dependence for the various polyamines and Mg2+ indeed decreases as follows: SPM4+ (fourfold positively charged) > SPD3+ > putrescine2+ (PUT) ≈ Mg2+[24,33]. However, while the electrical distance derived from the Woodhull model for Mg2+ delivers somehow realistic values (smaller than unity) [24,42], SPM block results in ‘electrical distances’ of more than 100%, i.e. the SPM molecule would absorb more energy than offered by the transmembrane electrical field [24,32,42,43]. As a solution to this paradox it was suggested that more than one SPM molecule participates in the blocking reaction and that binding of SPM to the pore is coupled to the movement of permeant and impermeant ions [11,42–45].
Taken together, there is substantial evidence that polyamines are responsible for strong inward-rectification of Kir channels by blocking the channel pore in a voltage-dependent manner. It should be added at this point, that a similar phenomenon has recently been observed in AMPA-type glutamate receptors [46–48], where polyamines were implicated in synaptic facilitation [48,49]. Differently from Kir channels, however, SPM-mediated block of the nonselective AMPA receptors is less voltage-dependent and is released at strong depolarizations, consistent with the polyamines permeating through the channel when the driving voltage is sufficiently high [46–49].
Structural determinants of inward rectification
Although all Kir channels are blocked by intracellularly applied polyamines, the voltage-dependence of block differs significantly among Kir subfamilies [20–22]. While strongly rectifying Kirs are completely blocked at ≈ 50 mV positive to EK, members of the Kir1 and Kir6 subfamily conduct significant outward current even at very depolarized potentials (e.g. + 80 mV).
Sequence comparison between the prototypic weak and strong rectifiers, Kir1.1 and Kir2.1, uncovered two structural determinants in the channel molecule that govern the degree of rectification [31,50–52]. The first was a residue in TM2 ( Fig. 3A). This position is occupied by a negatively charged amino acid in all strong rectifiers, while it presents with an uncharged residue in weak rectifiers. In Kir1.1 (ROMK1), changing this residue to a negatively charged aspartate was sufficient to convert the channel into a strong rectifier ( Fig. 3B, left panels). The reverse mutation in Kir4.1 changes rectification properties from strong to weak , almost indistinguishable from those of Kir1.1 in the mutated channel ( Fig. 3B, right panels). The ‘phenotype’ of strong rectification thus requires a negatively charged amino acid (aspartate or glutamate) at the TM2 site suggesting electrostatic interactions between this site and the positively charged blocker molecules. This hypothesis is further substantiated by the findings that a reporter cysteine at this position can be readily modified with cysteine-reactive reagents indicating exposure to the lumen of the pore . In agreement, the KcsA structure presents the homologous position as a pore lining residue  ( Fig. 3A). Thus, it seems very likely that negatively charged amino acids at the TM2 site stabilize binding of polyamines and Mg2+ in the pore via direct electrostatic interactions.
The second determinant for rectification was localized to the C-terminus of the Kir protein [44,53]. In contrast to Kir1 and Kir4, SPM block observed with the strong rectifier subfamilies Kir2 and Kir3 shows more complex kinetics and steady-state voltage-dependence [21,42,44]. When the TM2 residue was neutralized in Kir2.1, the resulting channels did not exhibit weak rectification as expected [21,44,52], although voltage-dependence was reduced and kinetics of block were changed. Using a chimeric approach, Yang et al. and Taglialatela et al. identified another negatively charged residue in the C-terminus of Kir2.1 (glutamate at position 224 [44,53]), which contributes to rectification. Neutralization of this amino acid in the TM2-neutralized Kir2.1 mutant results in a further decrease of SPM block, although channels are still significantly blocked by polyamines. Introduction of a negatively charged residue at this position into Kir1.1, however, did not induce any change in rectification properties with respect to wild-type, indicating a unique role of the C-terminal rectification site in the structural context of the Kir2 proteins [44,53].
A recent report by Baukrowitz et al. then added another twist to the C-terminal determinants of rectification . In Kir6 channels, a C-terminal histidine was identified that interacts with SPM and controls rectification in a pH-dependent manner. Around neutral intracellular pH, the histidine is protonated and KATP channels present with weak inward-rectification, while alkalinization that deprotonates the residue results in strongly rectifying channels .
Taken together, the TM2 site seems to be a common structural determinant for block of the Kir channel pore by intracellular cations. In contrast, the C-terminal contribution to pore block and rectification varies between the different Kir subunits. Furthermore, for Kir2 channels additional sites must exist as channels lacking the TM2 and C-terminal sites still display significant rectification.
Physiological role of rectification
The major physiological role of strongly rectifying Kir channels is their impact on the resting membrane potential in excitable and nonexcitable cells, as well as the establishment of the threshold that must be exceeded by excitatory stimuli to trigger generation and propagation of action potentials [23,24].
The ability of a strongly rectifying Kir channel to set a sharp excitation threshold is illustrated with a model system consisting of a cell-attached patch containing Kir2.1 channels using the current-clamp mode of the patch clamp amplifier ( Fig. 4A). In this mode, current is applied to the patch to simulate a depolarizing current in an excitable cell while the membrane potential is measured. Injection of 10, 20 and 30 pA depolarizing currents result in only small ohmic increases of the membrane potential because up to this point the membrane potential is mostly determined by the K+ conductance in the patch. However, when the applied current is large enough to depolarize the membrane potentials (VM) to values where polyamine block becomes prominent, Kir2.1 channels close. This allows VM to depolarize further and thus promotes further channel closure. This positive feedback leads to the fast all-or-nothing breakdown of the K+ conductance and thus to the observed large depolarization ( Fig. 4A). Intriguingly, this experiment demonstrates that an effective excitation threshold can be set by Kir channels in the absence of voltage activated Na+ channels.
Figure 4B,C depicts the function of Kir channels in a more native setting by correlating the K+ conductance mediated by Kir channels with the time-course of a typical neuronal action potential. At rest, the membrane potential is stabilized at values close to EK, due to the high K+ conductance mediated by the unblocked, i.e. open Kir channels (phase I in Fig. 4B). Excitatory stimuli trying to depolarize the membrane are mostly extinguished by instantaneous outward K+ currents resulting in a rather slow (and variable) prethreshold depolarization (phase II). This stabilizing effect is maintained as long as the depolarizing currents do not exceed the maximal outward current that can be mediated by the Kir channels. If this current maximum reflected by the inversion point at the steady-state I–V (≈ 30 mV positive to EK, Fig. 4C) is exceeded, Kir channels will rapidly be blocked by polyamines and thus allow for fast depolarization of the membrane, which at this point is driven primarily by the opening of voltage activated Na+ channels and the resulting large Na+ inward currents (INa). Subsequently the action potential is terminated by Na+ channel inactivation (not shown) and opening of voltage gated K+ channels (phase IV).
Thus, Kir channels via their pore block by SPM establish an effective threshold for excitation and therefore exert a major impact on excitability in many cell types. This view is supported by recent experiments by Bianchi et al. who found that a reduction in intracellular SPM levels reduced rectification of strong rectifier Kirs and consequently led to a shift of the excitation threshold to more positive potentials . Taken together, Kir channels represent an effective tool to control cellular excitation and might therefore serve as a general target for treatment of disorders in excitability. Furthermore, many Kir channels are regulated by extracellular and intracellular factors to allow adjustment of cell excitability in response to a multitude of biological stimuli. These regulatory mechanisms will be discussed in detail in the following articles.