Function of K2P channels in the mammalian node of Ranvier

Abstract

In myelinated nerve fibres, action potentials are generated at nodes of Ranvier. These structures are located at interruptions of the myelin sheath, forming narrow gaps with small rings of axolemma freely exposed to the extracellular space. The mammalian node contains a high density of Na⁺ channels and K⁺-selective leakage channels. Voltage-dependent Kv1 channels are only present in the juxta-paranode. Recently, the leakage channels have been identified as K2P channels (TRAAK, TREK-1). K2P channels are K⁺-selective ‘background’ channels, characterized by outward rectification and their ability to be activated, e.g. by temperature, mechanical stretch or arachidonic acid. We are only beginning to elucidate the peculiar functions of nodal K2P channels. I will discuss two functions of the nodal K2P-mediated conductance. First, at body temperature K2P channels have a high open probability, thereby inducing a resting potential of about −85 mV. This negative resting potential reduces steady-state Na⁺ channel inactivation and ensures a large Na⁺ inward current upon a depolarizing stimulus. Second, the K2P conductance is involved in nodal action potential repolarization. The identification of nodal K2P channels is exciting since it shows that the nodal K⁺ conductance is not a fixed value but can be changed: it can be increased or decreased by a broad range of K2P modulators, thereby modulating, for example, the resting potential. The functional importance of nodal K2P channels will be exemplified by describing in more detail the function of the K2P conductance increase by raising the temperature from room temperature to 37°C.

Introduction

Information processing in the nervous system requires exact timing of the generation of action potentials and their propagation along afferent and efferent myelinated nerve fibres in the peripheral and central nervous system. In the nodes of Ranvier of mammalian myelinated nerve fibres, the upstroke of action potentials is mediated by Na⁺ channel activation, and its repolarization by Na⁺ channel inactivation and a K⁺-selective ‘leakage’ conductance. The ‘leakage’ channels have recently been identified as K2P channels, TRAAK (TWIK-related arachidonic acid-activated K⁺ channel) and TREK-1 (TWIK-related K⁺ channel) (Brohawn et al. 2019; Kanda et al. 2019). In this Topical Review, I discuss the function of these K2P channels in the context of recent progress made in axon neurobiology.

Molecular neurobiology has discovered a broad range of proteins and molecular mechanisms involved in the formation of the node of Ranvier during development and the maintenance of its structure throughout life (Rasband & Peles, 2021). New insights into the structure of the node have been provided by super-resolution microscopy, e.g. by stimulated emission depletion (STED) microscopy. The new techniques have elucidated the astonishing periodicity of the axonal scaffold and the regular organization of the densely packed nodal ion channels and cellular adhesion molecules (CAMs) in the axolemma (D'Este et al. 2017). This dramatic progress in axonal neurobiology was accompanied by exciting discoveries in axon physiology. The distinct distribution of ion channels in the node of Ranvier of mammalian myelinated nerve fibres was detected more than 40 years ago: Na⁺ channels are located in the nodal axolemma and, surprisingly, voltage-dependent Kv1 channels to the juxta-paranodal axolemma (Chiu & Ritchie, 1981). Instead of delayed-rectifying K⁺ current as in the squid giant axon and the frog node of Ranvier, time- and voltage-independent nodal ‘leakage’ current was shown to repolarize the action potential in the mammalian node (Chiu et al. 1979; Brismar, 1980; Schwarz & Eikhof, 1987). The recent identification of the K⁺-selective ‘leakage’ channels, such as TRAAK and TREK-1 channels, has opened a new field of research, since almost all research into K2P activators and inhibitors has been done on K2P channels heterologously expressed in artificial expression systems. Research on native K2P channels in nodes of Ranvier is only beginning.

In addition to K2P channels, I discuss other types of ion channels contributing to nodal excitability: nodal Kv7.2/Kv7.3 channels (Devaux et al. 2004; Schwarz et al. 2006), Kv1.1/1.2 channels in the juxta-paranode, Kv3.1 channels located predominantly in central nervous system (CNS) nodes, and T- and L-type Ca²⁺ channels in the node and internode (Zhang & David, 2016). Knowledge of the molecular identity of nodal ion channels has led to the discovery of mutations in their genes, opening a new field of research by elucidating the molecular mechanisms causing human diseases. Examples are loss-of-function mutations of KCNQ2 and KCNQ3 (coding for Kv7.2 and Kv7.3 channels, respectively) causing benign epilepsy in children and myokymia (Dedek et al. 2001), and gain-of-function mutations of KCNK4 (coding for the TRAAK channel) causing a complex developmental syndrome (Bauer et al. 2018).

Morphology of the node of Ranvier

Myelinated nerve fibres consist of specialized segments (node, paranode, juxtaparanode, internode) equipped with distinct types of ion channels, CAMs, and cytoskeletal scaffold molecules (Rasband & Peles, 2021). Nerve fibres of a diameter >1 μm are wrapped in myelin. In the peripheral nervous system (PNS) one Schwann cell enwraps one internode, i.e. the nerve fibre between two nodes of Ranvier; in the CNS one oligodendrocyte enwraps one internode of several fibres. Myelin is produced by the spiral growth of glial cells around nerve fibres; a 10 μm nerve fibre contains up to 140 layers (Berthold & Rydmark, 1995). This stack of plasma membrane layers has the effect of putting together resistors and capacitances in series thereby increasing the transverse resistance and decreasing the capacitance of the myelin, prerequisites of efficient impulse propagation.

Nodes of Ranvier contain small rings of axolemma in direct contact with the extracellular space. The paranode is formed by the close attachment of Schwann cell loops to the axolemma thereby generating a tight glia–axon junction. The nodal and paranodal axon segments are constricted to about 50% of the internodal axon diameter of thick nerve fibres (Graphical Abstract panel B) (Berthold & Rydmark, 1995). The ensuing reduced nodal area and capacitance lowers the threshold for action potential generation and supports saltatory conduction. The nodal–paranodal axon constriction increases the longitudinal axoplasmic resistance and reduces its volume. The reduced volume exaggerates the transient changes in the axoplasmic Na⁺ and K⁺ concentrations during an action potential or during bursts of action potentials (Ona-Jodar et al. 2017). The physiological function of these concentration changes is not known.

Huxley and Stämpfli assumed that the myelinated nerve fibre consists of a low resistance nodal axolemma containing Na⁺ and K⁺ conductances and a myelinated internode with a high transverse resistance and low capacitance. By merging the internodal axolemma and myelin their model generated the ‘classical’ local circuit current (Graphical Abstract panel A) (Huxley & Stampfli, 1949). Since then it has been discovered that the internodal axolemma contains different types of ion channels and pumps. These internodal ion channels are activated by a ‘local circuit current’ flowing down the axoplasm and back to the node through the internodal axolemma, periaxonal space and paranodal junction. The single-cable model of Huxley & Stämpfli (1949) has therefore been exchanged for a two-cable model (Cohen et al. 2020).

Molecular structure of the node of Ranvier

Throughout their length, axons contain a periodic cytoskeletal scaffold composed of F-actin, spectrin and ankyrin (D'Este et al. 2017; Leterrier, 2018). Tetrameric spectrin complexes bind to two neighbouring F-actin rings thereby generating a distance of about 190 nm between two rings. This basic structural module seems to provide the axon with mechanical stability and flexibility. At the nodal axolemma, the spectrin tetramers consist of two β4- and two α2-spectrins. Ankyrin G binds to the spectrin complex and provides binding sites for different types of proteins, for Na_v1.6, Na_vβ and Kv7.2/Kv7.3, as well as for CAMs (Nf186, NrCAM). Binding to ankyrin G is a common mechanism to concentrate specific proteins at high density in a circumscribed region (Leterrier, 2018). The binding region of Na_v1.6 channels to ankyrin G consists of 22 amino acids, located in the loop between domain II and III of the pore-forming alpha-subunit. The binding of Na⁺ channels to ankyrin G increases their density severalfold as compared to that of the soma membrane. Kv7.2/Kv7.3 channels bind to ankyrin G with a peptide motif homologous to that of Na_v1.6. As revealed by STED microscopy, ion channels and CAMs are incorporated into the axolemma in a highly ordered longitudinal pattern closely related to the spectrin–actin ring periodicity (D'Este et al. 2017). STED microscopy indicated that nodal Kv7 channels exhibit a higher order of axial periodicity than Na1.6 channels. In contrast, Kv1.1/Kv1.2 channels do not bind to ankyrin; they are incorporated into the juxta-paranodal axolemma in an undefined order. This is also true for TRAAK and TREK-1; these channels are located to the nodal axolemma, but as yet, the mechanism underlying their high nodal concentration is not known (Rasband & Peles, 2021). The spectrin and ankyrin complexes change along the fibre segments; outside the nodal axolemma they contain β2-spectrin instead of β4-spectrin, and ankyrin G is exchanged for ankyrin B. The reason for this exchange of molecules is not known.

In the PNS, the nodal gap is covered by a basal lamina and filled with extracellular matrix molecules and microvilli of the Schwann cell loops. Schwann cells produce gliomedin. This protein remains attached to the microvilli and binds to neuron glial-related cell adhesion molecule (NrCAM) and neurofascin (NF186). Both CAMs are embedded in the axolemma and bind to ankyrin G. In the paranodal junction, the neurofilament NF155 is incorporated in the plasma membrane of the myelin loops and binds to casprin1 located in the paranodal axolemma, thereby stabilizing this narrow contact. In the juxtaparanode casprin2 (contactin-associated protein) and TAG-1 (Transient Axonal Glycoprotein) are clustered; they bind to each other and to Kv1 channels. Therefore, in casprin2 null mice the juxta-paranodal clustering of Kv1 channels is strongly reduced, without a reduction of the total number of channels. TAG-1 null mice have similar symptoms as casprin2 deficient mice (Rasband & Peles, 2021). The review by Rasband and Peles (2021) provides an excellent overview of the complex mechanisms underlying the formation of the node of Ranvier during development.

Methods to record membrane currents and action potentials from the node of Ranvier

Action potentials and membrane currents have successfully been recorded from the frog and rat node of Ranvier with the ‘classical’ Vaseline method (Nonner, 1969; Stämpfli & Hille, 1976). The feedback amplifier of the ‘Nonner-clamp’ has been rebuilt with integrated operational amplifier components (Brohawn et al. 2019). So far, the patch-clamp technique has been used for the recording of single-channel currents from demyelinated axons (Jonas et al. 1989). Recently, for the first time, successful patch-clamp measurements (voltage and current clamp) were obtained from the rat node of Ranvier of afferent trigeminal myelinated nerve fibres (Kanda et al. 2019). The authors used a sequence of high and low hydrostatic pressures applied to the patch pipette during its approach to the nodal axolemma (pressure-patch-clamp). For future research on the node of Ranvier the patch-clamp technique will be very promising. One advantage of this technique is that it allows measurement of the absolute value of the nodal resting potential, whereas with the Vaseline method the nodal resting potential can only be measured indirectly.

Hodgkin–Huxley action potential model in the squid and in the frog node of Ranvier

In 1952, Alan Hodgkin and Andrew Huxley performed experiments in the squid giant axon to elucidate the ion conductance changes underlying the generation of the action potential (Hodgkin & Huxley, 1952). By applying the ingenious voltage clamp they were able to separate ion currents from capacitive currents. The voltage-clamp allowed them to investigate the time- and potential-dependent activation and inactivation of Na⁺ currents, and activation of the delayed outward rectifying K⁺ currents. Based on these voltage-clamp data they calculated a membrane action potential almost identical to an action potential experimentally recorded. They concluded that the resting potential is mediated by a sustained resting K⁺ conductance, the action potential upstroke by Na⁺ conductance activation, and action potential repolarization by Na⁺ conductance inactivation and activation of a delayed activating K⁺ conductance (Hodgkin & Huxley, 1952). Their model reproduced the threshold potential, amplitude and duration of the action potential as well as its afterhyperpolarization.

To answer the question of whether the Hodgkin–Huxley model is also valid to describe the action potential of the myelinated nerve fibre, Bernhard Frankenhaeuser performed voltage and current clamp experiments in the node of Ranvier on myelinated nerve fibres of the toad Xenopus laevis. He found that the Na⁺ and K⁺ currents recorded in the toad node of Ranvier are similar to those of the squid axon, and he successfully used the Hodgkin–Huxley equations to describe the voltage- and time-dependent kinetics and steady-state values of Na⁺ channel activation and inactivation, as well as of K⁺ channel activation. Similarly to the squid giant axon, the quantitative description of the nodal voltage clamp data allowed the calculation of an action potential almost identical to that recorded from an intact toad node of Ranvier (Frankenhaeuser & Huxley, 1964). At the time when these seminal papers about the excitability properties of squid and frog nerve fibres were published, ion channels and their molecular identity were unknown.

Figure 1A and B shows membrane currents and action potentials recorded from an isolated node of Ranvier of the toad Xenopus laevis. The depolarizations elicited early transient inward and outward Na⁺ currents and delayed steady-state K⁺ currents. Application of 10 mM TEA blocked a large part of the outward current without affecting Na⁺ currents. The K⁺ current blockage by TEA depolarized the resting potential by a few millivolts, reduced the threshold potential, and increased the action potential duration. This experiment shows that TEA-sensitive K⁺ channels are important for the maintenance of the resting potential, the threshold potential and repolarization of the action potential in the amphibian node. In textbooks of physiology the Hodgkin–Huxley model of the action potential has often been described as being true for all types of myelinated nerve fibres including the mammalian node of Ranvier. As will be shown below, the ionic mechanisms underlying the action potential of the mammalian node are different from those of the amphibian node.

Figure 1. Voltage and current clamp recordings from isolated nodes of frog and rat nodes of Ranvier, before and after application of 10 mM tetraethylammonium

A and C, families of membrane currents recorded with depolarizations in steps of 10 mV. Application of 10 mM tetraethylammonium (TEA) blocked a large part of the outward current in the frog node (A), and a small part in the rat node (C). B and D, in the frog node, the presence of TEA depolarized the resting potential and increased the duration of the action potential (B); in contrast, there is only a small change in the rat node of Ranvier in the presence of TEA (D). From Brohawn et al. (2019). Reprinted by permission from eLife.

Ion channels of the mammalian myelinated nerve fibre

The first successful current and voltage clamp experiments in the node of Ranvier of single mammalian myelinated nerve fibres were performed in the laboratory of Robert Stämpfli (Nonner & Stämpfli, 1969). The authors recognized the small amplitude of the outward K⁺ currents but did not pursue this observation further. About 10 years later the observation of small delayed rectifying outward K⁺ currents was confirmed by recordings from the node of Ranvier of rat (Brismar, 1980) and rabbit (Chiu et al. 1979; Chiu & Ritchie, 1981) myelinated nerve fibres. The authors discovered that in the mammalian myelinated nerve fibre Na⁺ channels (Na_v1.6) are located to the nodal axolemma, whereas delayed-rectifying voltage-dependent Kv1 channels are absent in the node, but located to the juxta-paranode (Chiu & Ritchie, 1981). It was also recognized that the leakage current in the mammalian node is larger than in the frog node by a factor of 4–5 (Brismar, 1980) and that its magnitude is sufficient to repolarize the action potential (Chiu et al. 1979).

Membrane currents and action potentials recorded from a rat node of Ranvier at room temperature are shown in Fig. 1C and D. Application of TEA blocked only a small part of the outward K⁺ current and changed the action potential shape by next to nothing. These data show that the ion channels generating the action potential differ in the rat and frog node of Ranvier. The recent discovery that the nodal leakage current is a K⁺ current and mediated by the two K2P channels TRAAK and TREK-1 (Brohawn et al. 2019; Kanda et al. 2019) answered some of the important questions about action potential generation in the mammalian node of Ranvier but also raised additional questions. In the following paragraphs, the ion channels involved in the generation of the nodal action potential and the ion channels responsible for maintaining and stabilizing the resting potential in the mammalian node are discussed.

Na⁺ channels

Two of the nine known members of the Na⁺ channel family, Nav1.2 and Nav1.6, are axonal Na⁺ channels. Nav1.2 channels occur in unmyelinated axons, and Nav1.6 channels are concentrated in the node of Ranvier of myelinated nerve fibres, where they are clustered at a density of 1000–1400 channels/μm², the highest density known in the brain (Neumcke & Stampfli, 1982; Chiu, 2005). The high Na⁺ channel concentration in the node of Ranvier is predominantly due to the binding of Na⁺ channels to ankyrin G. In addition, axo-glial proteins in the nodal gap may support nodal Na⁺ channel clustering and the tight paranodal axo-glial junction may serve as a diffusion barrier (Rasband & Peles, 2021). Nav1.6 channels consist of a pore-forming α-subunit and ancillary β-subunits (β1–β4); they also undergo multiple posttranslational modifications, including glycosylation and phosphorylation. These molecular changes influence their gating and the different types of Nav1.6 currents, which can be distinguished, transient (I_NaT), persistent (I_NaP), and resurgent (I_NaR) Na⁺ currents (Buffington & Rasband, 2013). So far, I_NaR has not been observed in nodes of Ranvier. I_NaT is the fast activating and inactivating Na⁺ current mediating the upstroke of the action potential, and comprises about 98% of the total Na⁺ current. I_NaP activates at 10–20 mV more negative membrane potentials than I_NaT and exhibits little inactivation. These properties characterize I_NaP as a small persistent Na⁺ inward current near the resting potential, tending to depolarize the membrane (Burke et al. 2001). Since I_NaP is so small, it is difficult to measure this current directly in the node of Ranvier. Indirect evidence for its existence in nodes has been obtained in electrotonic threshold tracking experiments in human peripheral nerve fibres (Bostock & Rothwell, 1997). I_NaP is also present in the first node of Ranvier of neocortical layer 5 axons. Here, I_NaP lowers the threshold potential of the nodal action potential as in all other nodes where it is present, but in addition, it lowers the threshold potential of the axon initial segment about 150 μm away, thereby facilitating high-frequency burst generation in L5 axons (Kole, 2011). The molecular basis of I_NaP is not known; presumably, the three different Na⁺ current types represent different gating modes of the Na1.6 channel in connection with the β4-ancillary subunit (Aman et al. 2009).

The first recordings of Na⁺ currents at 37°C were performed by Nonner and Stämpfli (1969). Figure 2B shows Na⁺ currents of a rat node of Ranvier elicited at 37°C. At body temperature, the time course of Na⁺ current activation and inactivation is fast. To calculate an action potential, the rate constants and steady-state parameters of Na⁺ channel activation and inactivation evaluated from current recordings like those of Fig. 2B were used, as well as a linear leakage current. The calculated action potential had a duration of 0.25 ms, almost identical to that experimentally recorded (Fig. 2A) (Schwarz & Eikhof, 1987).

Figure 2. Na⁺ currents and leakage currents, generating the action potential

A, action potential, recorded at 36°C from a rat node of Ranvier, elicited with a just suprathreshold depolarization (J. R. Schwarz, unpublished). B, Na⁺ currents of a rat node of Ranvier, recorded at 37°C in the presence of axoplasmic CsCl, to block K⁺ channels. From Schwarz & Eikhof (1987), reprinted by permission from Springer Nature. C, families of leakage currents of a rat node of Ranvier, recorded with negative and positive potential steps. Measurements were performed at different temperatures. Current recordings were made with the patch clamp technique. From Kanda et al. (2019), reprinted by permission from Elsevier. D, current–potential relationships of recordings like those shown in A. From Kanda et al. (2019), reprinted by permission from Elsevier.

K2P channels

The first detailed measurements of membrane currents in the mammalian node of Ranvier showed that the ‘leakage’ current amplitude is larger than in the frog node of Ranvier (Brismar, 1980). Its properties, like outward rectification, temperature sensitivity and reversal potential close to the K⁺ equilibrium potential, suggested that this current is possibly mediated by K2P channels (Chiu, 2005; Renigunta et al. 2015). However, it was only recently that K2P channels (TRAAK, TREK-1) were identified in the rat node of Ranvier (Brohawn et al. 2019; Kanda et al. 2019). TRAAK channels are exclusively located to nodes of Ranvier; they do not occur for example in the axon initial segment. This is in contrast to nodal Nav1.6, Kv7.2 and Kv7.3 channels, which are also present in the axon initial segment, due to their binding to ankyrin G (Leterrier, 2018; Rasband & Peles, 2021). TRAAK channels are present in about 80% of myelinated nerve fibres; it is not known why they are absent in the remaining fibres. TRAAK channels were not found in sensory receptors of the skin or the somato-dendritic compartment (Brohawn et al. 2019). The absence of TRAAK in nociceptors is astonishing since TRAAK channels are related to nociception. TRAAK^–/– mice display mechanical and temperature allodynia and enhanced mechanical hyperalgesia during inflammation, consistent with a role for TRAAK in thermal and mechanical nociception (Noel et al. 2009).

TRAAK and TREK-1 channels belong to the same subfamily (TREK subgroup) of the two-pore domain K⁺-selective (K2P) ion channel family (Fink et al. 1998; Maingret et al. 1999; Enyedi & Czirjak, 2010). K2P channels are composed of two subunits and assemble to form either homo- or heterodimers. Each subunit consists of four transmembrane helices and two pore domains and lacks a voltage-sensing domain (Enyedi & Czirjak, 2010). TRAAK and TREK-1 channels are K⁺-selective, and their resting conductance provides a background conductance essential for maintaining a negative resting potential. The conductance of both K2P channels is increased by a broad range of activators, e.g. by mechanical stretch, increase in temperature and arachidonic acid. Intracellular acidification activates TREK-1 channels and extracellular acidification inhibits them. In contrast, TRAAK channels are activated by intracellular alkalinisation (Enyedi & Czirják, 2010). So far, in the node of Ranvier only the physiological function of an increase in temperature on the K2P-mediated current has been studied (see below).

The TRAAK-sensitive blocker RU2 (Su et al. 2016) inhibits about 20% of the entire outward current of the rat node of Ranvier. Since the RU2-insensitive current has similar properties as the TRAAK-mediated current (K⁺-selectivity, temperature dependence, outward rectification), other K2P channels are likely expressed in the node of Ranvier. Since TREK-1, but not TREK-2, is located together with TRAAK channels to the node of Ranvier of Aβ-afferent nerve fibres of the trigeminal nerve (Kanda et al. 2019), one can assume that the nodal leakage current is mediated by at least these two K2P channels also in other axons. However, using immunocytochemistry, Kanda et al. (2019) did not see TREK-1 in motor nerve fibres of the PNS. Therefore, as yet, the molecular identity of the channel or channels mediating the RU2-insensitive leakage current in nodes of Ranvier of sciatic myelinated nerve fibres of the rat is unknown.

K2P functions in action potential generation of the mammalian node of Ranvier

At body temperature, the nodal resting K2P leakage conductance contributes to a resting potential of about −85 mV (Kanda et al. 2019). Nodal K2P current is strongly dependent on temperature (Fig. 2C and D); upon increasing the experimental temperature from 10°C to 35°C the amplitude of the nodal ‘leakage’ current recorded from a node of an Aβ-afferent nerve fibre of the trigeminal nerve increased about 4-fold. The K2P current increase is mirrored by increasing negativity of the resting potential, from −75 mV at 15°C to −84 mV at 35°C (Kanda et al. 2019). This negative resting potential has two functions, it increases the number of Na⁺ channels ready to be activated upon depolarization and it helps to repolarize the nodal action potential.

The negative resting potential reduces steady-state Na⁺ channel inactivation and ensures that more than 70% of the nodal Na⁺ channels are available for opening. This is important since the number of Na⁺ channels activated upon a depolarization determines the rate of rise of the action potential upstroke and thereby its conduction velocity. Figure 3A shows that the TRAAK channel blocker RU2 depolarizes the resting potential by about 5 mV and decreases the amplitude of the action potential. Figure 3C and D demonstrates the steep influence of the membrane potential on steady-state Na⁺ channel inactivation. In this experiment, the holding potential was adjusted to a membrane potential at which 70% of the Na⁺ channels were ready to be activated (h_∞ = 0.7). This value is traditionally defined as V = 0 mV (Stämpfli & Hille, 1976) and corresponds to a resting potential of E = −80 mV in the rat node of Ranvier at room temperature (Schwarz et al. 2006). The voltage dependence of steady-state Na⁺ channel inactivation was determined with the double pulse method. Na⁺ inward currents were elicited with a test pulse to V = 60 mV preceded by 50 ms conditioning pulses varying in amplitude between −50 mV and +50 mV. Hyperpolarizing the membrane by −10 mV and depolarizing the membrane by +10 mV induced large changes in the Na⁺ current amplitude (Fig. 3C). This is the reason why the K2P blocker-induced depolarization strongly reduced the rate of rise of the action potential (Fig. 3B).

Figure 3. TRAAK stabilizes a negative resting potential of the rat node of Ranvier, and reduces steady-state Na⁺ channel inactivation

A, action potentials, recorded before (left), in the presence of 10 mM RU2 (middle), and after wash out of the drug (right). From Brohawn et al. (2019), reprinted by permission from eLife. B, superimposed rate of rise of action potentials, shown in A (J. R. Schwarz, unpublished). C, membrane currents elicited with the pulse programme shown in the inset: a 60 mV pulse, preceded by 50 ms conditioning pulses between −50 mV and 50 mV in steps of 5 mV. D, steady-state Na⁺ inactivation, obtained by plotting the normalized Na⁺ current amplitude against conditioning pulse potential. From Brohawn et al. (2019), reprinted by permission from eLife.

Figure 4. Action potential recorded from an isolated human node of Ranvier at 20°C

Groups of ion channels are responsible for the resting potential and different action potential (a.p.) phases. Arrows indicate the resting potential changes induced by activation or inhibition of the K2P channels. Modified after Schwarz et al. (1995), reprinted by permission from Springer Nature.

Action potential repolarization of the mammalian node of Ranvier is determined by Na⁺ channel inactivation and the K2P K⁺ conductance. The time course of Na⁺ channel inactivation determines the time course of repolarization. As soon as Na⁺ channels inactivate the K⁺-selective ‘leakage’ conductance instantaneously repolarizes the membrane potential. If one assumes that the K2P ‘leakage’ conductance is voltage-independent, it does not accelerate repolarization, and therefore it does not shorten the action potential and does not generate an afterhyperpolarization. This is in contrast to the frog node of Ranvier; the delayed rectifying voltage-dependent Kv1 conductance is activated during the action potential and decreases its duration; blockage of Kv1 channels by TEA increases its duration (Fig. 1B and D).

Traditionally, the leakage conductance is defined as an unspecific time- and voltage-independent conductance mediated by unspecified ions. The K2P-mediated conductance is different. Recently, the current mediated by all 15 known K2P channels, including TRAAK and TREK-1, has been studied in a heterologous expression system (Schewe et al. 2016). Upon a depolarizing potential step, an outward current is mediated by the K2P channels consisting of an initial instantaneous current followed by a time-dependent increase to a sustained steady-state current. Both current components are potential-dependent. For unknown reasons, only TWIK-1 channels do not exhibit any time and voltage dependence. The instantaneous current component is due to ion flow through already open K2P channels, and – in physiological solutions – exhibits Goldman–Hodgkin–Katz rectification. The time-dependent component is due to opening of and ion flux through the selectivity filter of the K2P channels (Schewe et al. 2016). In the rat node of Ranvier, the K2P-mediated currents are characterized by the same properties; they consist of an initial instantaneous current increase and a subsequent time-dependent current increase (Brohawn et al. 2019; Kanda et al. 2019). The instantaneous current increase may represent the ‘pure’ leakage component.

How do these properties of K2P conductance contribute to nodal action potential repolarization? Action potential calculations support the assumption that the instantaneous leakage current component of K2P channels is involved in repolarization. So far, the action potential models of mammalian nodes of Ranvier include a linear leakage current, and the results show that the calculated action potentials are astonishingly similar to action potentials experimentally measured at room temperature in a rabbit (Chiu et al. 1979) and a human node (Schwarz et al. 1995), and at 37°C in a rat node (Schwarz & Eikhof, 1987). The addition of a small outward rectification does not change the action potential shape (Schwarz et al. 1995). Future experiments will investigate the effect of larger amplitudes of outward rectifying currents on the time course of action potential repolarization, e.g. those generated by increasing the K2P conductance by arachidonic acid. Important insights into the effects of K2P conductance on the resting potential and action potential have been obtained from experiments with recombinant TREK-1 or TASK-3 channels combined with Na⁺ conductance simulations. The results showed that the magnitude of the K2P conductance determines the negativity of the resting potential and that the K2P outward rectification is important for action potential repolarization (MacKenzie et al. 2015). In hypothalamic neurons of mice with a knockout of TASK1 and TASK3 channels, the resting potential was depolarized and the excitability was decreased (Gonzalez et al. 2009). In cerebellar granule neurons, deletion of TASK-3 channels induced a depolarization of about 10 mV. The increased action potential accommodation could be rescued by injecting into these neurons a non-linear leakage conductance with a dynamic current clamp (Brickley et al. 2007). Experiments in the rat node of Ranvier confirmed the functional importance of K2P conductance for repetitive firing. For example, recovery from Na⁺ channel inactivation at the resting potential of −85 mV is complete within 4–5 ms at 37°C, allowing action potential firing at frequencies up to 200 Hz in the presence of K2P conductance. A decrease in the K2P conductance led to action potential failures at much lower frequencies (Kanda et al. 2019).

K2P channels are not only activated by increasing the temperature but also by mechanical stretch of the nodal membrane. What is the function of mechanosensitive K2P channels in the node? It is known that activation of TRAAK or TREK channels by mechanical stretch or by increasing the body temperature increases their open probability (Enyedi & Czirjak, 2010). Independent of the origin of the K2P activation, the increase in K⁺ conductance shifts the resting potential to more negative values, closer to the K⁺ equilibrium potential. So far, the mechano-sensitivity of TRAAK and TREK-1 has been investigated in detail in heterologous expression systems, but preliminary experiments show that both channels are mechanosensitive in nodes of trigeminal fibres (Kanda et al. 2019) and TRAAK channels in nodes of nerve fibres of the sciatic nerve (J. R. Schwarz, personal observation). The physiological function of the K2P mechano-sensitivity is elusive. One possible function could be that ectopic action potentials generated by nerve stretch or slight mechanical damage could be counteracted by the shift of the resting potential to more negative values due to the activation and mechanical sensitivity of nodal TRAAK and TREK-1 channels (Brohawn et al. 2019). TRAAK and TREK-1 channels could also support action potential repolarization by providing an increase in K2P current by mechanical stretch due to nodal swelling during an action potential. Swelling of excitable membranes associated with action potentials has been demonstrated (Iwasa et al. 1980). In crab nerve fibres the ‘electrical’ action potential is accompanied by a ‘mechanical’ action potential due to a transient swelling of the nerve fibre (Iwasa et al. 1980). Swelling could occur by the influx of Na⁺ and water through Na⁺ channels. The additional K2P current could support action potential repolarization. As yet, the nodal function of K2P activation by stretch or change in pH has not been investigated experimentally.

Two different de novo missense mutations in KCNK4, the gene encoding TRAAK, have been thoroughly characterized (Bauer et al. 2018). These mutations are gain-of-function mutations and have been identified in three human families. They are thought to cause the complex developmental and neurological disorder FHEIG, an abbreviation standing for its characteristic syndrome of facial dysmorphism, hypertrichosis, epilepsy, intellectual disability and gingival over growth (Bauer et al. 2018). The TRAAK channel mutations exhibit a strong increase in their open probability. The increased conductance could shift the resting potential to more negative values thereby inhibiting action potential generation and transmission.

The propagation of action potentials along myelinated nerve fibres proceeds with high reliability and precision. In the PNS, Na⁺ channels and K2P leakage channels are involved in the generation of the nodal action potential. A multitude of other types of ion channels has been found in nerve fibres (Fig. 4). The function of many of these channels has not been studied. Some of them stabilize the resting potential and dampen excitability following a single action potential or burst of action potentials. The properties of some of these ion channels are described below.

Kv7.2, Kv7.3 channels

In the rat node of Ranvier, a voltage-dependent non-inactivating K⁺ current (I_Ks) is present that slowly activates and deactivates (Roper & Schwarz, 1989). This current is an M-current and is mediated by homomeric Kv7.2 channels (Devaux et al. 2004; Schwarz et al. 2006). For unknown reasons, Kv7.2 and Kv7.3 are distributed differently among nerve fibres. In nodes of large-diameter axons homomeric Kv7.2 channels occur; Kv7.2 and Kv7.3 channels are present in fibres with small and intermediate axon diameters. The reason for this different distribution is not known. In most neurons, the M-current is mediated by hetero-multimeric Kv7.2/7.3 channels (Brown & Passmore, 2009).

Nodal M-channels have two functions, they contribute to the maintenance of the resting potential and they induce accommodation. The activation curve of the nodal M-current is shallow and starts to activate at membrane potentials as negative as −110 mV. At the resting potential of −80 mV a fraction of about 30% is activated, thereby contributing to the maintenance of the resting potential (Roper & Schwarz, 1989). This function is the same as that of K2P channels; the negative resting potential removes part of the resting steady-state Na⁺ channel inactivation and ensures that upon a depolarization a large Na⁺ inward current is activated to strengthen action potential upstroke (Battefeld et al. 2014). As in many types of neurons, the nodal M-current dampens repetitive activity (Schwarz et al. 2006; Battefeld et al. 2014). Due to the slow time course of deactivation, the amplitude of the M-current successively increases during repetitive activity, because additional M-current is activated by each action potential. This cumulative increase of M-current during repetitive activity induces frequency adaptation. In vivo application of the M-channel blocker XE991 induced sustained heavy repetitive activity in motor axons innervating the tail muscles of the rat (Schwarz et al. 2006).

Loss-of-function mutations in KCNQ2 or KCNQ3, the genes encoding Kv7.2 and Kv7.3, respectively, cause benign familial neonatal epilepsy. Patients suffer from brief generalized seizures occurring during the first 3 months after birth; thereafter the seizures disappear (Dedek et al. 2001). In some patients with a point mutation in KCNQ2, spontaneous involuntary muscle movements occur known as myokymia, induced by spontaneous repetitive firing generated in motor neuron axons (Dedek et al. 2001).

Kv1.1, Kv1.2 channels

Kv1.1/1.2 channels, as well as the ancillary subunit Kvβ2, are located to the juxta-paranodal axolemma, where they form heteromeric channel complexes (Chiu & Ritchie, 1981; Rasband & Peles, 2021). The interaction between casprin2, TAG-1 and Kv1 channels may explain Kv1 clustering in the juxtaparanode. Kv1 channels are low-threshold channels, are partially open at the resting potential, and contribute to its maintenance. It is generally assumed that juxta-paranodal Kv1 channels may be activated in the wake of an action potential, thereby stabilizing the membrane potential and preventing re-excitation of the node (Chiu, 2005).

The stabilizing function of Kv1 channels is supported by the phenotypes of mutations in KCNA1, the gene encoding Kv1.1. Dominantly inherited mutations in KCNA1 cause episodic ataxia type 1, characterized by brief attacks of ataxia and myokymia, induced by spontaneous repetitive axonal activity (Browne et al. 1994). Mice with a deletion of Kv1.1 (or Kvβ2) have seizures and die 3–5 weeks after birth. The absence of Kv1.1 channels results in the tendency of abnormal re-excitation of myelinated axons (Zhou et al. 1998).

K_v3.1 channels

Kv3.1 channels are present in CNS nodes of Ranvier (Devaux et al. 2003). Due to alternative splicing, two channel subunits exist, Kv3.1a and Kv3.1b. Kv3.1b channels are more abundantly expressed than Kv3.1a channels. The molecular mechanism underlying the clustering of Kv3.1b at nodes of Ranvier is not known; the channel does not contain an ankyrin G binding sequence. In contrast to CNS axons, Kv3.1b channels are present in only 27% of PNS nodes of Ranvier (Devaux et al. 2003). It would be interesting to know, whether Kv3.1 channels are expressed in nodes of Ranvier lacking TRAAK channels. Kv3 channels are high threshold channels; they are activated well above the threshold potential of action potentials and open with fast kinetics, thereby inducing Na⁺ channel deactivation. Deactivated Na⁺ channels shorten the refractory period; they can be re-activated instantaneously and allow nerve fibres to fire action potentials at very high frequencies (Rudy & McBain, 2001).

Ca²⁺ channels

In mouse myelinated nerve fibres of the PNS, trains of action potentials induce a local increase in axoplasmic [Ca²⁺] around nodes of Ranvier. In intact myelinated nerve fibres, Ca²⁺ influx is mediated by T-type Ca²⁺ channels and reversed activity of the Na⁺/Ca²⁺ exchanger (Zhang & David, 2016). Reversed activity takes place when the axoplasmic [Na⁺] is too high, Na⁺ is moved outward, and Ca²⁺ inward into the axoplasma. Following paranodal demyelination, stimulation of a train of action potentials induced a still larger increase in the axoplasmic [Ca²⁺]. This increase is strongly reduced by the L-type Ca²⁺ channel blocker nimodipine (Zhang & David, 2016).

Possible functions of the increased axoplasmic [Ca²⁺] are the activation of Ca²⁺-dependent K⁺ channels or the modulation of metabolic processes in the axon. Single-channel recordings from demyelinated peripheral nerve fibres demonstrated the existence of Ca²⁺-dependent K⁺ channels (Vogel & Schwarz, 1995). In axons of cerebellar Purkinje cells, T-type Ca²⁺ channels are co-localized with Ca²⁺-activated IK channels (K_Ca3.1) (Grundemann & Clark, 2015). The physiological function of L-type Ca²⁺ channels buried below the myelin is not known. A low or moderate increase in axoplasmic [Ca²⁺] may activate enzymatic cascades leading to the modulation of processes like axonal transport. If the axoplasmic [Ca²⁺] is too high, e.g. following demyelination, proteases are activated inducing nerve fibre damage.

HCN channels

HCN channels mediate an inwardly rectifying unspecific current (I_h) carried by Na⁺ and K⁺ that depolarizes the resting potential. I_h is known as a pacemaker current in the heart and in neurons (Biel et al. 2009). HCN channels have also been discovered in peripheral myelinated nerve fibres with the method of electrotonic threshold tracking (Howells et al. 2012). In nerve fibres, the resting potential is the result of the interaction between depolarizing currents, mediated by HCN and Na_vp channels, and hyperpolarizing currents, mediated by K2P, Kv7, Kv1 channels and ion pumps. I_h counteracts the hyperpolarization of the resting potential during periods of high-frequency firing (Baker & Bostock, 1997). It is unlikely that HCN channels mediate a pacemaker current in nerve fibres.

Conclusions

The focus of this review is the function of K2P channels in the node of Ranvier. In addition to K2P channels, I also discussed other ion channels that are important for reliable action potential generation and repetitive firing. However, there are additional types of ion channels in nerve fibres. Recording of single-channel currents from acutely demyelinated axons revealed the presence of a broad range of ion channels, e.g. K_Na, K_ATP and K_Ca (Vogel & Schwarz, 1995). The function of most of these ion channels for axonal excitability is still elusive.

There are important unanswered questions about nodal physiology. Do nodes lacking TRAAK channels contain other types of K2P channels, or e.g. Kv3 channels? What differences exist between ion channels expressed in PNS and CNS nerve fibres or in nerve fibres of different diameter? How are nodal K2P channels activated by volatile anaesthetics or inhibited by substances like bipuvacaine or fluoxetine? Does their mechanism of action on K2P channels explain their anaesthetic effects? Future research on nodes of Ranvier relies on suitable methods. Unexpectedly, the patch-clamp technique has been successfully employed for recording of the resting potential and action potentials in the rat node of Ranvier (Kanda et al. 2019). Another promising technique is high-speed optical recording from CNS and PNS axons (Cohen et al. 2020).

More than 40 years ago, the distinct distribution of ion channels in the mammalian node of Ranvier was discovered. Since then we know that repolarization of the nodal action potential is brought about by K-selective ‘leakage’ channels (Chiu et al. 1979). The identification of these channels as K2P channels was the most recent exciting finding (Brohawn et al. 2019; Kanda et al. 2019). In mammals, evolution has created an effective ‘minimalistic’ system, consisting of Na⁺ channels and K2P channels, to ensure reliable generation of short action potentials (0.25 ms at 37°C) and action potential firing at frequencies as high as 200 Hz.

Biography

• 

  Jürgen R. Schwarz studied medicine. After a research position at the Institute of Physiology, Kiel University (1969–1972) he specialized in Neurology (1972–1977). Since 1977 he has been at the Institute of Physiology, and from 1991 to 2006 was chairman of the Department of Applied Physiology, Hamburg University. Since 2006 he has been guest scientist at the Institute of Neurogenetics, Center of Molecular Neurobiology Hamburg (ZMNH). His research interests are the function of ion channels in the node of Ranvier of mammalian nerve fibres, ether-à-go-go-related gene K⁺ channels and their function in neurons and synaptic plasticity.

Additional information

Competing interests

The author declares that there are no competing interests.

Author contributions

Sole author.

Funding

This work was funded by Deutsche Forschungsgemeinschaft (DFG), grant no. Schw292/17.

Acknowledgements

I want to thank Dr Christiane Bauer and Dr Vitya Vardanjan for helpful comments on the manuscript, Dr Matthias Kneussel for his hospitality at the Institute of Neurogenetics, ZMNH, and Yvonne Pechmann for unfailing help. I also want to thank the referees for their constructive comments.

Open access funding enabled and organized by Projekt DEAL.