Sculpting excitable membranes: voltage-gated ion channel delivery and distribution

Abstract

The polarized and domain-specific distribution of membrane ion channels is essential for neuronal homeostasis, but delivery of these proteins to distal neuronal compartments (such as the axonal ends of peripheral sensory neurons) presents a logistical challenge. Recent developments have enabled the real-time imaging of single protein trafficking and the investigation of the life cycle of ion channels across neuronal compartments. These studies have revealed a highly regulated process involving post-translational modifications, vesicular sorting, motor protein-driven transport and targeted membrane insertion. Emerging evidence suggests that neuronal activity and disease states can dynamically modulate ion channel localization, directly influencing excitability. This review synthesizes current knowledge on the spatiotemporal regulation of ion channel trafficking in both central and peripheral nervous system neurons. Understanding these processes not only advances our fundamental knowledge of neuronal excitability but also reveals potential therapeutic targets for disorders involving aberrant ion channel distribution, such as chronic pain and neurodegenerative disease.

Introduction

Action potentials are the basis for fast intraneuronal communication, which underlies all forms of neurologic function. Voltage-gated ion channels are the building blocks of excitable membranes that permit action potential initiation and transmission¹. Hodgkin and Huxley initially investigated action potential generation in squid giant axons, and presented a schema in which action potentials are produced by only two voltage-dependent conductances². However, mammalian neurons are far more intricate; the human genome contains at least 400 genes that encode ion channels³ and human neurons express well over a dozen types of voltage-gated ion channels. Among these, voltage-gated channels that are selectively permeable to Na⁺ (voltage-gated sodium channels, Nav), Ca²⁺ (voltage-gated calcium channels, Cav), K⁺ (voltage-gated potassium channels, Kv), and Cl⁻ (voltage-gated chloride channels, ClC) regulate both active and passive membrane properties and define electrical signaling in excitable cells.

Neurons are exquisitely polarized cells with extreme geometries and compartment-specific functions. Many core neuronal functions depend on the delivery of voltage-gated ion channels to different neuronal compartments and their insertion and maintenance within highly precise areas (domains) of the neuronal membrane. The most striking examples of ion channel localization are the plasma membranes of the axon initial segment (AIS) [G] and nodes of Ranvier (NOR) [G] of myelinated neurons, where ion channels are densely concentrated to enable action potential firing and propagation, respectively. Though the polarized and precise distribution of voltage-gated ion channels is clearly essential for neuronal homeostasis, the mechanisms through which this distribution occurs and is maintained are incompletely understood.

Most voltage-gated ion channels are synthesized in the endoplasmic reticulum (ER) of the neuronal soma. Within the ER and the trans-Golgi network, many ion channels acquire complex carbohydrate modifications^4,5 and must also clear quality control checkpoints to ensure their proper folding⁶. Post-translational modifications and association with accessory subunits [G] are often required for proper trafficking to the cell surface⁷. Both the specific vesicle populations into which the ion channels are packaged within the Golgi and the identity of the trafficking machinery define the pathway through which they will be ultimately delivered to the membrane^8–10 (Fig. 1). Long-range vesicular delivery is primarily mediated by microtubule-based transport, which depends on motor proteins [G] — kinesins for anterograde transport and dyneins for retrograde transport. The domain-specific trafficking and localization of voltage-gated ion channels is driven by a combination of factors, including the channel’s intrinsic molecular determinants and the adaptor proteins [G], motor proteins and cytoskeletal elements with which they interact¹¹. These processes have been best studied in neurons of the CNS, particularly in those of the hippocampus^12,13, while our understanding of protein trafficking and localization in peripheral neurons is less-well understood.

Fig. 1: Packaging of ion channels into vesicles for anterograde axonal delivery.

a, The delivery and distribution of ion channels (like that of other building blocks of excitable membranes) depends on their translation in the endoplasmic reticulum (ER). Here, the small subunit of the ribosome decodes mRNA while the large subunit effects the multispanning membrane protein synthesis on rough endoplasmic reticulum. After this step, the channel is further processed in the Golgi apparatus⁵. b, In the Golgi, post-translational modifications (such as complex N-glycosylation) and association with accessory subunits (not shown) are often required for proper protein packaging and trafficking⁷. c, Once these modifications are complete, vesicles containing channels and other proteins (including transporters like sodium/calcium exchangers (NCX), cytokine receptors like tumor necrosis factor receptors (TNFR), and neuropeptides like neuropeptide Y (NPY)) bud from the Golgi membrane and associate with cytoskeletal networks for export d, After export from the Golgi, ion channels are packaged into distinct vesicle populations linked to distinct motor proteins (kinesins) that associate with the vesicle cargo through distinct adaptor proteins. e, The vesicle type, motor protein and adaptor proteins define the route of anterograde transport down the axons along microtubule networks¹⁰⁶. Rab GTPases are determinants of vesicular fate, and specific Rab isoforms are present in vesicles of varying maturity, stage, and cargo types¹⁰⁸.

In this Review, we describe what is known about the polarized and domain-specific delivery and maintenance of voltage-gated ion channels in the axons of mammalian neurons, with a focus on these processes in peripheral sensory neurons. Because their axons can be a meter or more in length, these cells serve as an informative case study to understand the sculpting of excitable membranes through polarized delivery of cargoes to specific cellular compartments. Further, we highlight the advances in technology (Box 1 and Fig. 3) that have enabled the investigation of channel life cycles [G] across neuronal compartments and then discuss the relevance of altered channel distribution and delivery in neurologic disease. In much of this Review, we will focus on Nav channels because they are particularly well studied in these contexts.

Box 1: Technology for the evaluation of ion channel life cycles in sensory neurons.

The direct visualization of ion channels in hippocampal neurons via fluorescent labeling strategies has provided key information on ion channel life cycles^{106,111,123,124,135} (Fig. 2b). The clear distinction between the somatodendritic and axonal regions of such CNS neurons has provided a model that has informed our understanding of ion channel cell biology for several decades. By contrast, evaluating ion channel life cycles in sensory neurons has been hampered by the technical challenges associated with visualizing the real-time movement of proteins in cells that extend processes over vast distances, relative to the size of the soma. Furthermore, ion channels are often present in large quantities within the cytoplasm of sensory neurons¹⁰⁶, making it difficult to distinguish the proteins present in the cytoplasm from those within the cell membrane.

Early attempts at localizing voltage-gated ion channels in CNS neurons involved light microscopic autoradiography of radiolabeled toxins²³⁰. Later, the purification and generation of fluorescent proteins enabled the tagging and localization of ion channels in live cells²³¹. These approaches faced challenges related to the requirements to visualize dim signals from proteins expressed at low levels, to distinguish between membrane-inserted and intracellular channels and to achieve sufficient signal above the high background fluorescence that arises from the large pool of static proteins within the cytoplasm. Over time, the fluorescent toolbox expanded (Table 1), and included technologies such as fluorescence recovery after photobleaching (FRAP), which greatly increased the signal-to-noise ratio of live imaging at the expense of cellular health^232–234. Modifications to the fluorescent protein tags, for example to create tags that respond to changes in the environment (such as pH) allowed for the first measurements of the cell surface dynamics of neuronal ion channels²³⁵.

Further advances in protein labeling and imaging techniques⁸⁰ have enhanced our understanding of ion channel trafficking and cell biology. Modular protein tagging systems incorporating self-labeling enzymatic tags that allow for customizable and multiplexed imaging of membrane proteins in living neurons have been developed (including HaloTag, SNAPTag and ClipTag; Table 1)^236–238. Combining these strategies with in vitro culture systems that compartmentalize neuronal somas and axons (Fig. 3a) has yielded approaches that have permitted next-generation studies of neuronal ion channel life cycles and provided information on how these proteins are dynamically regulated in excitable membranes^{106,111,113,114,124,135,239}. When combined with fluorophores with unique chemical properties (such as cell-permeability or impermeability; Table 1), these systems have enabled investigations into numerous stages of the ion channel life cycle. The use of optical pulse-chase methods that allow specific spatial and temporal labeling of surface versus intracellular populations of ion channels (Fig. 3b) has enabled the study of long-distance anterograde (Fig. 3c) or retrograde (Fig. 3d) axonal trafficking, the fate of internalized channels (Fig. 3e), the rate and the locations of membrane insertion (Fig. 3f), and ion channel transcytosis. The information gathered from these studies has expanded our knowledge of the ion channel life cycle in sensory neurons. Much of this recent information was obtained using HaloTag-Nav channel constructs, bright, photostable synthetic fluorophores^240,241 and dorsal root ganglion (DRG) neurons plated in microfluidics chambers.

Fig. 3: Methods for high-resolution studies of the ion channel life cycle.

a, Schematic representation of a microfluidic chamber, in which the neuronal somas and axons of cultured neurons can be sequestered in fluidically isolated compartments. The right panel shows a fluorescent image of dorsal root ganglion (DRG) neurons expressing green fluorescent protein (GFP) plated in a microfluidic chamber. Somas are located in the left compartment, while axons extend via microgrooves across a physical barrier to the axonal chamber on the right. b, The steady state expression of enzymatically tagged channels within neuronal membranes can be evaluated in a compartment specific fashion by cell-impermeant fluorescent labeling of the axonal and somatic membranes and subsequent quantification of fluorescent signal. c, Applying cell permeable labels to the somatic compartment of the chamber allows the anterograde trafficking of labelled ion channels to the distal axon to be visualized. d, Applying cell impermeable labels to the axonal compartment of the chamber allows the retrograde trafficking of internalized channels in the distal axon to be observed. Using cell permeable fluorescent ligands will not permit visualization and robust analysis of vesicular retrograde trafficking because of the relatively large cytoplasmic pool of channels (moving in both directions) that would be labelled. e, Transfecting neurons plated in a microfluidic chamber with constructs encoding two proteins with distinct and mutually compatible tags (such that there is no cross-reactivity between the tags) and then using the labeling strategies in panels c and d allows for imaging of co-trafficking in the distal axon. f, Saturating proteins with a cell impermeant label (Label 1) and then replacing the fluid in the compartment with fluid containing another cell impermeant label (Label 2) enables visualization of the kinetics of insertion and removal of surface proteins from excitable membranes. Figure adapted, with permission, from REF¹²³.

Polarized delivery of ion channels

Polarized ion channel delivery in CNS axons

Building an axon with conduction properties that meet functional requirements depends on the delivery of the requisite number (or density) of ion channels to each part of that axon. For example, while a very low density of Nav channels (<2/μm²) can support conduction in some premyelinated axons (where the axon diameter is very small and thus input resistance is very high), the channel density needed for action potential production in larger, fully developed axons with lower input impedance is greater¹⁴. Moreover, the spatial pattern of channel expression plays a crucial role in determining action potential conduction properties¹⁵.

The first step in the polarized delivery of ion channels in CNS neurons is their segregation into distinct types of transport vesicles^16,17 that can be preferentially trafficked to either the somatodendritic or axonal compartments^16,17. A major determinant in this sorting decision in CNS neurons is the AIS, a specialized region of the proximal axon located at the boundary of the axonal and somatodendritic compartments (Fig. 2a)¹⁸. The AIS is defined by a submembrane cytoskeletal undercoat enriched for the scaffolding protein Ankyrin-G (AnkG, encoded by the gene ANK3 ), which acts as a master regulator of the assembly of the membrane proteins in this region, including Nav channels^19,20. Action potential initiation in many neuronal types, including most CNS neurons, begins at the AIS, where it is facilitated by the high concentration of Nav channels (which is ~50-fold higher than in the somatodendritic region)^21,22. AnkG anchors and concentrates Nav channels at the surface of the AIS through interactions with the cytoskeleton and cell adhesion molecules (Fig. 2b)^23–30. AnkG-mediated localization is dependent on the presence of an AnkG-binding motif within the second intracellular loop of Nav channels^24,31,32. AnkG is also involved in polarized protein delivery to other parts of the axon: in mouse neurons, the loss of AnkG not only results in a reduction in Nav concentration at the AIS, but also causes proteins normally present only in somatodendritic compartments to invade the axon^33,34.

Fig. 2: Polarized delivery of ion channels in CNS and PNS axons.

Neurons contain somatodendritic and axonal domains. Though there are many similarities in the structures of these domains in the multipolar neurons of the CNS and pseudounipolar sensory neurons of the PNS, differences in their geometries and function present unique challenges for the construction of their excitable membranes.

a, Neurons of the CNS typically have many dendrites and a single long axon. In these neurons, the axon initial segment (AIS) marks the boundary between the somatodendritic and axonal compartments. b, The AIS of hippocampal neurons, a master regulator of vesicular fate, is illustrated. A thick mesh composed of actin rings and spectrin restricts the movement of vesicles through the region; vesicles carrying cargoes destined for the somatodendritic compartment are turned away, while those destined for the axonal compartment are permitted to pass through²⁵. AnkyrinG (AnkG) is a crucial effector of ion channel trafficking through the AIS. First, AnkG drives the clustering of Nav channels (as well as other ion channels, including Kv7²²⁸) at the AIS. Secondly, AnkG directly binds kinesin 5B, which allows for the passage of Nav-containing vesicles through the AIS¹¹⁸. Disruption of the interaction of AnkG with Ndel1 permits the pass through of dendritic cargo. c, Sensory neurons of the PNS are composed of a somatic compartment, a stem axon, and two axonal branches — one central and one peripheral. The existence of an AIS in these cells is suggested, but the zone in between these domains likely represents a sorting station⁹⁰. d, In peripheral neurons, anterogradely-moving vesicles carrying axonal-destined cargoes are moved down the long microtubule network of the axon via the action of kinesin motor proteins. These vesicles carry several types of proteins in varying stoichiometries. Specific Rab GTPase isoforms associate with different cargoes (Rab6A, for example, is found in vesicles carrying Nav and Kv channels¹¹¹). e, Membrane proteins like ion channels are carried to the distal axon where they become part of a cytoplasmic pool that can be inserted into the axonal membrane. Vesicle exocytosis and membrane incorporation is mediated by the action of SNARE proteins²²⁹.

The AIS regulates axonal identity in two principal ways. The first is through the generation of a membrane diffusion barrier that impedes the movement of membrane proteins between the somatodendritic and axonal domains^35,36. The AIS in mammalian neurons contains a high density of proteins, including periodic ring-like structures that are composed of actin [G] and are spaced by spectrin bridges²⁵. This submembranous actin–spectrin–AnkG structure is thought to physically impede the diffusion of both proteins and lipid molecules within the membrane^25,35,37,38. Work performed in a Caenorhabditis elegans sensory neuron model suggests that any somatodendritic and axonal membrane proteins that do diffuse into the AIS are internalized into lysosomes and degraded³⁹. The second contribution of the AIS to axon identity involves vesicle sorting. Vesicles carrying axonal cargoes are permitted through the AIS and into the axonal trafficking apparatus, while those destined for the somatodendritic domain pause before or in the AIS and return back to the soma^40–47. This remarkable sorting capability is imparted by molecular mechanisms that are still not fully elucidated, although several models have been proposed⁴⁸. Current data suggests that actin structures in the AIS and into the proximal axon, termed actin patches, halt plus-end-directed myosin motor proteins traveling towards the axon carrying vesicles that should be destined for the somatodendritic compartment and redirect them to the somatic region along actin filaments in which the plus-end is oriented toward the cell body⁴⁹. An actin depolymerizing protein, MICAL3, is present in the AIS in rats and may serve to regulate actin in the proximal axon, including actin patches²⁸. MICAL3 is generally auto-inhibited, but can be activated by a subset of Rab GTPases, proteins that are found in different types of transport vesicles and are involved in protein trafficking²⁸. This suggests that the actin-based selectivity filter in the AIS may be altered in response to the specific type of trafficking vesicles [G] that enter the proximal axon. Microtubule-based transport mechanisms at the AIS are also important for vesicular sorting. The dynein regulator NDEL1 has been shown to be crucial to the sorting function of the AIS in rats^50,51. NDEL1 binds to AnkG present at the AIS, and both the disruption of this complex⁵⁰ and the depletion of NDEL1 alone allow dendritic cargo to pass through the AIS into the axon⁵¹. Specific kinesins also play a role in cargo polarization; the kinesin-3 family member KIF13A is a dedicated dendrite-selective kinesin, while KIF13B can switch between dendrite- and axon- selective transport modes⁵². Thus, a complex network of sorting factors, including domain specific proteins, motor protein modulators, and determinants of vesicular identity all play a role in establishing and maintaining neuronal polarity.

A study in cultured mouse hippocampal neurons suggested that the microtubule-associated protein TRIM46 may play an important role in organizing microtubules into fascicles (the unique bundles that are observed in proximal axons), which could be determinants of vesicular trafficking and entry into the axon⁵³. However, in vivo studies using Trim46 knockout mice demonstrated that, although TRIM46 is required for microtubule fasciculation, its presence is dispensable for AIS specification and formation⁵⁴. Future work is therefore needed to understand how the AIS works in tandem with the microtubule cytoskeleton to sustain neuronal polarity.

The composition of the AIS is relatively stable, with several AIS proteins demonstrating a half-life on the order of weeks³³. However, with advances in imaging technologies and protein labeling techniques, we are starting to appreciate the functional and structural plasticity of this region. For example, the length, position, and composition of the AIS can be modified by neuronal activity to fine-tune action potential initiaton^55,56. Moreover, there is now strong evidence against the classical view that the dense actin–spectrin mesh in the AIS limits endocytosis [G] in this region⁵⁷. Experiments in cultured mouse neurons in which green fluorescent protein (GFP) was introduced into the endogenous locus of Scn2a (the gene that encodes the Nav1.2 channel) demonstrated that activity dependent AIS shortening is achieved through clathrin-mediated endocytosis of Nav1.2 channels occurring preferentially at the distal end of the AIS⁵⁸. Furthermore, robust endocytic activity at the AIS has been proposed as a mechanism for membrane protein sorting³⁹. Building on these observations, a recent study used both super-resolution optical microscopy [G] and platinum replica electron microscopy [G] to visualize clearings within the AIS, consisting of patches devoid of the actin-spectrin mesh⁵⁹ in which clathrin-coated pits (key mediators of the endocytosis of plasma membrane proteins^60,61) were observed. However, this study also suggested that, in resting neurons, endocytosis at the AIS is a rare event. Only when AIS plasticity was induced using NMDA were clathrin-coated pits unlocked and endocytosis triggered. It thus appears that endocytosis at the AIS is regulated at multiple levels and that this regulation can be controlled by the neuron in a context- and activity-dependent fashion.

In myelinated neurons, the NOR are, like the AIS, microdomains in which, Nav and Kv channels are concentrated, permitting saltatory conduction of action potentials along the axon. NOR are similar to the AIS in their structure and function, with the high density of ion channels being localized via interactions with AnkG. However, unlike the AIS (which forms through intrinsic methods), assembly of the NOR is dependent on extrinsic mechanisms⁶². During mammalian development, the structural specialization of the membrane at NOR begins before the production of compact myelin but coincides with the arrival of the glial cells that wrap loosely around the axon⁶³. It appears that the future ‘hot-spots’ of Nav channel insertion can develop pre-myelination, but that glial cells are essential for the maintenance of the mature NOR structure^64–67. It is clear that both extrinsic cues (such as glial-derived factors, both soluble and transmembrane) and intrinsic factors (such as neuronal scaffolding proteins) drive Nav channel clustering at the NOR^68,69.

Polarized protein delivery in PNS axons

Mature sensory neurons are pseudounipolar, with cell bodies that reside in the dorsal root ganglia (DRG) and trigeminal ganglia. Each cell body gives rise to a single axon, or stem axon, that bifurcates into two separate branches: one very long branch that innervates the peripheral tissues to receive sensory inputs, and a much shorter branch that synapses with second order neurons of the spinal cord (Fig. 2c)^70–72. The unique geometry and specialized functions of sensory neurons make them an extreme example of the challenges associated with polarized neuronal ion channel delivery and the generation of excitable membranes sometimes over a meter away from the site of channel biogenesis⁷³. The crucial role of sensory neurons in human disease, particularly in pain disorders^74–76, underscores the importance of understanding how their membranes are built. Some determinants of sensory neuronal structure have been explored, such as the high density of the scaffolding protein ankyrin-2 (ANK2, also known as ankyrin-B) in somatosensory PNS terminals⁷⁷. We also know that, during NOR formation in myelinated mammalian PNS neurons, Nav and Kv channels are delivered to NOR via vesicular transport, while axonal adhesion molecules, such as Neurofascin 186 (NF186), are recruited via diffusion trapping [G] from a pre-existing surface pool⁷⁸. Indeed, the presence of these ion-channel rich microdomains has provided the basis for several decades of research on ion channel delivery and distribution in myelinated axons⁷⁹. However, well-defined and distinct microdomains like the NOR have not been described in unmyelinated sensory neurons. As a result, and despite the importance of these non-myelinated axons to sensory and autonomic function, there is a paucity of studies regarding mechanisms of ion channel trafficking and distribution in these cells. Super resolution imaging techniques, however, have enabled investigations of protein localization with unprecedented resolution and have begun to elucidate mechanisms of channel localization and distribution in the unmyelinated axons of sensory neurons⁸⁰.

In contrast to CNS neurons, which generally sort proteins into two categories (axonal or somatodendritic), sensory neurons must both determine proteins that should remain in the soma versus those that are bound for the axon and face the challenge of sorting proteins based on whether they are destined for the peripheral or central axonal branches (see REF⁸¹ for a review on the topic). There is strong evidence that adhesion molecules and secretory vesicles [G] are enriched in axons^82,83, while lysosomes and mRNAs mostly accumulate in the soma⁸⁴ suggesting that a sorting mechanism exists for selective entry into the axons of sensory neurons. Furthermore, a recent study found evidence that delayed-rectifier channel Kv2.1 is present only within the soma of sensory neurons, while Kv2.2 was reported to be localized to the soma, stem axon and peripheral axons⁸⁵.

What are the mechanisms that enable differential sorting into sensory axonal branches? Studies using electron microscopy (EM) have demonstrated two sets of microtubule networks that diverge at the bifurcation of the stem axon⁸⁶, suggesting that sorting to different axonal branches likely occurs upstream of the bifurcation. The two branches show some structural differences, including a larger diameter and increased microtubule density in the peripheral branch of unmyelinated axons⁸¹. Early studies using radiolabeled amino acids demonstrated that the velocity of fast axonal transport is similar in the two branches, but that the amount of protein transported in the peripheral branch is higher^87,88. Another study reported an increase in the accumulation of the heat-sensing channel TRPV1 in peripheral, but not central, axon endings under inflammatory conditions, suggesting that the transport of some proteins to each axon branch is specifically regulated⁸⁹. A more recent report demonstrated the existence of a pre-axonal filtering zone, defined by the localization of microtubule associated protein 2 (MAP2), that is a candidate contributor to vesicular sorting in sensory neurons⁹⁰. Within this region, MAP2 controls the pairing of molecular motors to vesicular cargoes.

While the AIS is a major site of vesicle sorting in CNS neurons, the existence of an AIS or AIS-like region in sensory neurons has been contested⁸¹. A recent study reported AIS-like structures in DRG neurons in vitro and in vivo and showed that a larger proportion of the neurons demonstrated a pseudounipolar morphology and the presence of a stem axon that may act as an AIS when they were co-cultured with glial cells⁹¹. This is consistent with findings reported in a recent preprint that demonstrated that human induced pluripotent stem cells (hiPSCs) require co-cultured satellite glial cells to form the more mature pseudounipolar structure of differentiated sensory neurons⁹². Whether the formation of an AIS in DRG neurons is an intrinsic neuronal feature as it is in many CNS neurons, or requires extrinsic signaling from glial cells is yet to be determined.

When observed, the AIS present in cultured DRG neurons has a similar molecular profile to the AIS of multipolar neurons, including AnkG expression⁹¹. Furthermore, enrichment of Nav channels (in this case Nav1.1 and Nav1.7) has been reported, although the reported relative enrichment is only 3–4 fold, compared to 50-fold often observed in the CNS^91,93–95. Like the AIS of multipolar neurons, the AIS of DRG neurons has been shown to be able to generate electrical activity⁹¹. Indeed, the AIS of DRG neurons may contribute to the generation of pathological spontaneous activity, since time-controlled disassembly of this region decreased spontaneous activity in a model of neuropathic pain⁹¹. In healthy DRG neurons, however, action potential initiation mostly begins at the distal end of the peripheral axon⁹⁶; thus, the contribution of AIS development to excitability is unclear. The generation of an excitable membrane comes at an enormous metabolic cost⁷⁶ and so it is possible that the DRG AIS, particularly given its proximity to the bifurcation of the stem axon, is necessary for boosting action potential propagation past a site of impedance mismatch where the axons branch^15,97.

Clearly, there are many more lessons to be learned about how sensory neurons regulate the delivery and distribution of voltage-gated ion channels to sculpt their excitable membranes. Trans-Golgi export in compartment-destined secretory vesicles is only one part of the story; understanding the regulation of channel insertion, internalization, and recycling in normal conditions and in disease states is also crucial to unveiling the secrets of neuronal excitability and one day leveraging them for therapy.

Anterograde transport of ion channels

The anterograde trafficking and subcellular localization of ion channels within axons involves numerous post-translational modifications, localization motifs, and interactions with accessory and regulatory proteins. These topics have been previously reviewed for both sensory neurons⁹⁸ and hippocampal neurons⁹⁹; this section will therefore focus on recent information obtained from live-cell imaging studies of Nav channels in unmyelinated DRG neurons.

Older studies provided clear evidence that Nav channels are transported to the distal axon^100–103. However, for many years, the role of long-distance axonal transport in Nav channel localization was contested, partly due to data showing that axons contain the necessary machinery for local translation of Nav channels^104,105. The use of a Nav1.7 channel construct tagged with an extracellular HaloTag [G] enabled the first direct visualization of the anterograde trafficking of these channels in cultured rat DRG neurons, though these results must be interpreted with the caveat that these studies rely on the overexpression of tagged proteins which may result in supraphysiologic levels of channels¹⁰⁶. These studies demonstrated that Nav1.7 channels, which amplify subthreshold stimuli to allow action potential generation in nociceptors^74,107, are trafficked anterogradely in vesicles along the distal axon¹⁰⁶. While this does not preclude a role for local translation, it provides compelling evidence that axonal transport plays a major role in the polarized distribution of Nav channels in sensory neurons. This anterograde transport was blocked by a microtubule inhibitor, indicating that anterograde delivery of Nav1.7 is microtubule dependent, and thus likely kinesin based. The dim fluorescence intensity of vesicles carrying HaloTag labeled Nav1.7 channels suggests that only a small number of channels are transported in each vesicle, with the majority of imaged vesicles estimated to contain only one fluorescently labeled channel¹⁰⁶.

Members of the Rab GTPase family regulate various steps in vesicular trafficking^108,109. Live-cell imaging studies of co-trafficking demonstrated that Nav1.7 channels are preferentially co-transported with Rab6A in anterogradely trafficking vesicles^106,110, indicating the selective packaging of these ion channels into specific Rab6A-containing vesicles (Fig. 2d). Another long-standing question in the field has been whether different ion channel isoforms that are destined for the same location traffic together or in distinct vesicles. Using strategies to uniquely label different Nav channel isoforms, live-cell co-trafficking studies demonstrated that the other major sodium channels expressed in DRG neurons (Nav1.6, Nav1.8, and Nav1.9) are present in the same vesicles as Nav1.7, and are also preferentially packaged into vesicles containing Rab6A¹¹¹. This is in contrast to the findings of a recent study that used CRISPR generated tags to demonstrate that, in rat hippocampal neurons, Nav1.2 and Nav1.6 in are sorted into distinct populations of transport vesicles¹¹². This independent vesicle targeting and membrane loading contributes to differences in protein distributions in the AIS and distal axons of hippocampal neurons. Whether theses divergent results are due to differences in the cell types investigated, the specific Nav channel isoforms studied or channel tagging methods is yet to be determined.

Vesicles carrying Nav channels travel at a speed of approximately 1 μm/s^{106,110,113,114}, a speed consistent with action of kinesin motors, particularly those of the kinesin-1 family¹¹⁵. One study that employed biochemical techniques has suggested that the interaction of kinesins with Nav channels may be isoform specific, demonstrating that KIF5B (a member of the kinesin-1 family) interacts with Nav1.8 and Nav1.9 channels (but not with Nav1.7 and Nav1.6) and promotes their anterograde transport¹¹⁶. This data, however, is inconsistent with the live-cell imaging studies demonstrating that these proteins are transported in the same vesicle¹¹¹. Whether the overexpression of the tagged channels in the latter experiments contributes to this discrepancy is unclear; however, overexpression of proteins per se does not lead to co-transport with Nav1.7 channels in Rab6A associated vesicles¹¹¹

An intuitive model that might allow precise tuning of neuronal excitability suggests that ion channels that contribute to membrane depolarization (such as Nav channels) may be trafficked separately from those that contribute to membrane hyperpolarization (such as Kv channels). Consistent with this hypothesis, one study in rat DRG neurons reported independent anterograde transport of Nav-β subunits (normally associated with the pro-excitatory pore-forming Nav-α subunits) and the anti-excitatory Kv7.3 channel, as well as other axonal membrane proteins¹¹⁷. However, a second study found that multiple Nav channels and the Kv7.2 channel were transported anterogradely in the same vesicles¹¹¹. Nevertheless, this later study still demonstrated some degree of selectivity in transport of axonal proteins (Fig. 2d), with some membrane proteins being sorted into independent vesicle populations (such as the sodium/calcium exchanger NCX2)¹¹¹. While there is mixed evidence on the sorting of ion channels according to physiological function, overall it appears that selective mechanisms do exist to regulate the packaging and trafficking of specific membrane proteins — suggesting a nuanced and context-dependent packaging process.

The molecular bases for the axonal co-transport of different membrane proteins in sensory neurons are unclear. In CNS neurons, the binding of Nav channels to AnkG is important for both their trafficking and subcellular membrane localization⁹⁹. AnkG binding motifs also facilitate the common vesicular sorting of NF186 and the neural cell adhesion molecule NRCAM, which are components of the NOR in mammalian myelinated axons of the PNS¹¹⁷. KIF5 binds directly to AnkG to transport Nav1.2 channels into the AIS in CNS neurons¹¹⁸. Live-cell, single-molecule imaging of an epitope-tagged Nav1.6 channel construct demonstrated that Nav1.6 channels are preferentially inserted directly within the AIS membrane of rat hippocampal neurons, where they are immediately immobilized, and that this directed trafficking is lost when the AnkG binding motif is removed³⁰. Nav1.8 channels also possess an AnkG binding motif^24,32 and Nav1.8 has been suggested to bind constitutively to AnkG¹¹⁹. With this in mind, the dependence of long-distance axonal trafficking of Nav1.8 on AnkG binding in rat sensory neurons was investigated. Surprisingly, deletion of the AnkG binding motif from Nav1.8 did not affect its co-trafficking with other channels in Rab6A-containing vesicles, indicating that the co-packaging of these proteins occurs independently of their binding to AnkG¹¹¹.

Several mammalian and viral proteins depend on short linear polybasic motifs that act as sequence-independent electrostatic interaction motifs to facilitate their transport in vesicles containing Rab GTPases^120–122. Candidate polybasic motifs found in the first intracellular loop of Nav1.7 channels might therefore have been expected to regulate the co-association of Nav1.7 with Rab6A. However, the substitution of alanine for these conserved polybasic motifs did not affect any aspects of the vesicular co-trafficking, surface expression, or channel function of Nav1.7 in rat DRG neurons¹¹⁰. Thus, the molecular determinants of vesicular co-trafficking in sensory axons have yet to be identified. However, the technology to evaluate candidate processes is available¹²³ and the identification of mechanisms for selective packaging of cargo into specific vesicles may yield novel targets for the therapy of pain disorders.

Most of the anterogradely-moving vesicles carrying cargoes critical for the construction of excitable membranes become stationary upon reaching the distal axon, where they may provide a pool of channels primed for insertion into the distal axonal membrane (Fig. 2e)¹⁰⁶. The use of cell-impermeable HaloTag ligands and time-lapse live imaging allowed visualization of HaloTag-Na_V channel insertion into the membrane of cultured rat DRG neurons^114,124. Furthermore, the membrane-insertion of this pool of vesicles in these neurons is also supported by the fact that, at steady-state, Nav channels accumulate at higher density on the surface of this region of the axon than in axonal regions closer to the cell body¹⁰⁶. The accumulation of Nav channels in distal axons in cultured neurons lines up logically with the in vivo observation that they are sites of action potential electrogenesis in sensory neurons^{69,96,125–127}. However, because the in vitro culture system in which this accumulation was observed does not include end organs (such as skin), the term ‘axonal terminal’ does not accurately describe the freely outgrowing distal axon observed in these studies, though it is conceivable that this culture system could be developed and interrogated with live imaging techniques^{124,128–130}.

Ion channel turnover in axonal membranes

The lifetime of membrane proteins at neuronal surfaces is finite, and the turnover and replacement of these proteins via endocytosis, retrograde trafficking, recycling and degradation allows neurons to react to their environment and maintain homeostasis^131,132 (Fig. 4).

Fig. 4: Turnover of ion channels in sensory neurons.

The turnover of ion channels at axonal membranes is a multistage process. Following the internalization of ion channels via endocytosis, there are multiple pathways that the channel can enter. Some channels are recycled back to the axonal membrane in Rab11-containing recycling vesicles¹³⁵. Others are destined for lysosomal degradation, a process that occurs in the soma. In this case, the vesicles containing the internalized channels become part of Rab5-containing early endosomes, which then merge with other Rab5-containing endosomes and form Rab7-containing late endosomes¹³⁵. These late endosomes are transported retrogradely by dynein motor proteins to the soma, and fuse with lysosomes where their cargo is digested¹³⁵. The mechanism of somatic channel turnover has not been fully elucidated but likely follows a similar process, but without the need for long-distance axonal transport.

The Nav channels present in the AIS of mammalian hippocampal neurons, which provide the depolarizing stimulus for action potential initiation²⁰, are remarkably stable over the course of minutes to hours as investigated via fluorescence recovery after photobleaching (FRAP) [G] experiments^30,133, demonstrating slow membrane turnover. These characteristics appear to be modulated by protein partners, as mutations to the binding site for the microtubule associated protein MAP1B in the N-terminus of one of these channels, Nav1.6, accelerated its endocytosis^133,134, suggesting that the stability of Nav channel surface expression at the AIS is the result of a delicate balance between insertion into the plasma membrane and channel endocytosis – both processes influenced by protein-protein interactions³⁰.

Since action potentials in sensory neurons are initiated near the endings of the peripheral axonal branch in response to external stimuli⁹⁶, it was postulated that Nav channels in distal axons may demonstrate an AIS-like distribution, with limited membrane mobility and turnover¹⁰⁶. However, in rat DRG neurons, labeling of Nav1.7 channels present at the cell surface with an extracellular epitope revealed only a modest increase in their levels in this region relative to the axonal shaft and showed that the channels have a relatively high turnover rate, compared to channels at the AIS in hippocampal neurons¹³⁵. Using a neuronal culture system where neuronal cell bodies were isolated from their axons using microfluidic chambers, it was shown that Nav1.7 endocytosis in distal axons is constitutive, suggesting that much of the regulation of the levels of the channels at the axonal surface reflects a dynamic process of insertion and removal¹³⁵. However, these studies were performed in cultured neurons, so it remains to be seen whether these observations will be confirmed in vivo.

Following internalization, there are multiple pathways that an ion channel can enter. With regard to Nav1.7 channels, one subpopulation of these channels appears to be recycled back to the axonal membrane, but not to the somatic membrane, providing another mechanism by which neurons can maintain their polarity¹³⁵. Though studies of other ion channels have not investigated the relationship between their recycling and neuronal polarity, the recycling of these excitable membrane building blocks does appear to be important for neuronal excitability. This is evidenced by the action of the anti-allodynic drug gabapentin, which interferes with the Rab11-mediated recycling of a Cav channel subunit and thus prevents neurotransmitter release in response to neuronal activation¹³⁶.

Another fate for internalized ion channels is retrograde transport and ultimate degradation. By investigating if Nav1.7 channels appeared in vesicles with markers of early (Rab5) and late (Rab7) endosomes [G], these channels were shown to be transported retrogradely in vesicles that arise from the maturation of early endosomes into late endosomes and that also contain multiple other membrane proteins (Fig. 4)¹¹¹. This is consistent with reports that late endosomes arise from the fusion of several early endosomes¹³⁷. Following retrograde transport, the late endosomes carrying Nav1.7 fuse with the lysosome in the soma, where the cargo that they carry will be degraded¹³⁵.

The post-internalization pathway that a channel enters again represents a complex sorting decision for the neuron, and is influenced by multiple factors¹³⁸. For example, the ubiquitination of an internalized ion channel promotes its degradation. In this process, protein mono-ubiquitination (or in some cases polyubiquitination¹³⁹) is recognized by a series of multiprotein complexes belonging to the endosomal sorting complexes required for transport (ESCRT) family^140–142. Within endosomes, ESCRTs sort cargoes destined for degradation into intraluminal vesicles that are transported to the lysosome for degradation. On the other hand, the presence of recycling signal sequences within a protein sequence can drive active sequence-dependent retrieval from the endosome, facilitated by the retromer and retriever complexes. This process is thought to be involved in the recycling of internalized cargoes including membrane proteins^143–146. Modifications of these ubiquitination sites and recycling signal sequences on neuronal ion channels may be a mechanism for the regulation of the balance between channel recycling and channel degradation. Importantly, these processes can also be regulated by changes in neuronal activity. For example, the recycling of TRPV1 in rodent DRG neurons is activity dependent and scales with the strength and frequency of the activating stimulus¹⁴⁷.

The ubiquitin ligase NEDD4L is the classic regulator of ion channel endocytosis in sensory neurons, and all Nav channels except Nav1.9 contain a PXY motif in the C-terminus (where X can be any amino acid) that has been shown to be required for their NEDD4L-mediated endocytosis¹⁴⁸. NEDD4L-mediated regulation of Nav channels is regulated directly by intrinsic determinants in the channel as well as through the regulation of their interacting partners. For example, phosphorylation of the PGSP motif in the intracellular loop of Nav1.6 by activated p38 mitogen-activated protein kinases (MAPKs) generates a novel binding site for NEDD4L that, together with the PXY motif in the C-terminus, is required to induce endocytosis in rat DRG neurons¹⁴⁹. Similarly, the addition of small ubiquitin-like modifiers (SUMOylation) of collapsing response mediator protein 2 (CRMP2)¹⁵⁰ has been shown to prevent the formation of a CRMP2–NEDD4L complex that causes Nav1.7 accumulation in endosomes^151,152; blockade of this SUMOylation increases internalization of Nav1.7 and can reduce pain signaling in rat sensory neurons^153,154.

Another option for an internalized membrane protein is transcytosis, in which a protein is first inserted into the membrane of a somatodendritic compartment, internalized and then trafficked anterogradely to the axon. While the transcytosis of some membrane proteins, including the growth receptor NTRK1¹⁵⁵ and vesicle-associated membrane protein 2 (VAMP2)¹⁵⁶ contributes to axonal polarity, Nav1.7 channels do not undergo transcytosis¹³⁵. This implies that distinct mechanisms must exist for the maintenance of the polarized distribution of Nav1.7.

The molecular determinants of ion channel degradation and recycling are still incompletely understood. Along with live-imaging approaches, the incorporation of pH sensitive ligands may help us to visualize how proteins behave once internalized in endosomes¹⁵⁷. Further, tools to induce the degradation of ion channels may help us to study of membrane protein turnover. For example, a functionalized nanobody that ubiquitinates Cav channels was recently developed, and shown in cardiomyocytes to draw Cav channels into Rab7-containing late endosomes¹⁵⁸. A recent preprint has reported the development of a heterobifunctional peptide that triggers the ubiquitination of Na_v1.8 channels and showed that its expression reduced the number of Nav1.8 channels at sensory neuronal surfaces, an effect thought to be the result of reduced vesicular delivery of Nav1.8 to the membrane in response to ubiquitination¹⁵⁹. Another preprint has reported that ion channel ubiquitination occurs throughout the channel life cycle, and is regulated by distinct polyubiquitin chains present at different stages¹⁶⁰. Further explication of the mechanisms that regulate excitable membrane organization in sensory neurons may benefit from similar experimental approaches.

Mobility of ion channels

The lack of mobility of both proteins and lipids within the AIS of CNS neurons has been a point of interest for decades³⁵ and is thought to confer stable electrical properties that enable reliable action potential initiation. This limited mobility appears to be dependent on the ankyrin–spectrin–actin cytoskeleton, and on the binding of multiple membrane proteins to AnkG (including neurofascin, Kv7 channels, and Nav channels¹⁶¹). By contrast, in somas of rat CNS neurons, Nav1.6 channels show distinct behaviors, with a population of channels that are highly mobile and another that are localized to stable nanoclusters ~230 nm in diameter¹⁶². The generation of these nanoclusters is independent of the classic drivers of Nav concentration, such as ankyrin, actin and clathrin.

Since sensory neurons initiate action potentials at distal axon endings⁹⁶, it was thought that Nav channels may also show limited mobility in these endings. However, single-particle tracking of Nav1.7 tagged with an extracellular epitope revealed that there are two populations of these channels in distal axons of cultured rat DRG neurons, one population of channels that is mobile and another population of significantly less mobile channels localized to nanoclusters¹⁰⁶, an organization similar to that seen for Nav1.6 at the soma of hippocampal neurons¹⁶². The lack of evidence for a structured organization of ion channels within the membrane at the distal axon suggests that mobile Nav channels may participate in action potential generation in this neuronal compartment.

A similar relationship between the spatial regulation of ion channel localization and the mobility of the channels in the neuronal plasma membrane is demonstrated by Cav2.1 at the synapses of mammalian hippocampal neurons. Here, precise nanometer spacing between the Cav channels and synaptic release machinery determines release probability, and yet single-particle tracking demonstrates that most Cav channels at synapses are mobile and undergo transient confinement within nanodomains¹⁶³.

It is possible that accumulation of Nav channels in nanoclusters provides sites of action potential generation and that the mobile channels outside of these nanoclusters boost the response to depolarizations across a larger membrane area. This idea is supported by cytochemical and freeze-fracture electron microscopy results that suggest that Nav channels cluster along unmyelinated and pre-myelinated axons^66,67,164, computer modeling studies that suggest that this clustering enables micro-saltatory conduction¹⁶⁵ and electrophysiological investigations that demonstrate that saltatory conduction precedes remyelination in demyelinated axons¹⁶⁶. Whether the organization of the Nav channels in nanoclusters is cell-autonomous or regulated by the presence of other cell types in vivo remains to be determined.

Changes in excitable membranes in disease

Deficits in the neuronal trafficking machinery contribute to numerous neurodegenerative diseases including amyotrophic lateral sclerosis (ALS)¹⁶⁷, Parkinson disease (PD)¹⁶⁸, Huntington disease (HD)¹⁶⁹ and the peripheral neuropathy Charcot-Marie-Tooth disease¹⁷⁰. These neuropathies can be induced by mutations that affect the function of motor proteins (kinesins or dynein), the microtubule network or adaptor proteins¹⁷¹. Moreover, while the underlying mechanisms are not understood, altered patterns of ion channel distribution along demyelinated axons contribute to the pathophysiology of multiple sclerosis: these include an altered distribution of Nav channels^172,173 and disruption of the paranode and Kv channel domain^174,175.

Alterations in ion channel type, distribution or density at the plasma membrane can also contribute to neurological diseases. For example, mutations in Nav1.1 that alter its trafficking contribute to generalized epilepsy with febrile seizures¹⁷⁶, and a diffuse distribution of Nav1.6 channels along larger-than-normal regions of axons following demyelination can contribute to axonal injury in multiple sclerosis¹⁷³. Sensory neuronal hyperexcitability underlies the increase in pain sensation in a variety of disease states^177–180. Abnormal accumulations of Nav1.3, Nav1.7 and Nav1.8 have been reported within the distal ends of injured axons in human neuromas, which are sites of ectopic impulse activity that leads to chronic pain¹⁸¹. Neuronal hyperexcitability can be driven by a variety of factors, including direct modulation of single ion channel activity by disease effectors^{148,182–184}. However, here we will focus on alterations to voltage-gated ion channel distribution and delivery in sensory neurons that have been shown in recent studies to make mechanistic contributions to disease states (Fig. 5).

Fig 5: Changes in excitable membrane construction in disease.

Neurologic disease and pathology can alter the distribution and delivery of excitable membrane building blocks like ion channels. These changes result in modulation of neuronal excitability which can be defining features of certain disease states. The figure highlights points in which ion channel distribution and delivery can be modulated and how certain pathophysiological states have been shown to modulate these processes in peripheral sensory neurons. (pro-excitatory changes in membrane composition are highlighted in boxes with a red outline). Sodium flux through Na_V channels is responsible for the rise in membrane potential that forms the action potential upstroke, and potassium exit through K_V channels allows for the repolarization of the membrane potential a, b, The membrane insertion of Nav, but not Kv, channels at the distal axon is increased in response to chronic exposure to inflammatory mediators, while the rate of removal is unchanged¹¹⁴. In the somatic membrane, increased membrane insertion of Nav is observed in response to acute exposure to inflammatory mediators (20 min); however, acute inflamamtion does not alter membrane insertion in the distal axon¹²⁴. c, The relative density of Nav channels present within the plasma membrane of distal axons is increased in response to chronic inflammation^106,114 and chemotherapeutic agents^113,201, but not acute inflammation¹²⁴. d,Trafficking vesicles contain more Na_V channels in response to chronic and acute inflammation^106,114,124, as well as chemotherapeutic agents^113,201. e, The number of vesicles involved in active anterograde transport (vesicle flux) is increased in response to chronic and acute inflammation^106,114,124, as well as chemotherapeutic agents^113,201. f, Retrograde transport is not affected by chronic inflammation¹¹⁴. Additional experiments are needed to more completely understand the role of retrograde transport as well as channel turnover within the somatic compartment.

Inflammatory pain results from the sensitization of nociceptors following tissue injury and is mediated by a variety of cytokines, kinins, growth factors and eicosanoids, acting on multiple targets¹⁷⁹. These inflammatory mediators trigger a variety of signaling cascades in DRG neurons and can result in significant changes to the electrogenisome (the set of molecules that confer electrical excitability)^74,185–188, including changes in Nav channel expression. Animal models of inflammatory pain show an upregulation of Nav channels, including Nav1.3, Nav1.7, Nav1.8, and Nav1.9, that can last for days to weeks after an injury^189,190. Studies using complete Freund’s adjuvant-induced inflammatory pain in animal models have suggested that there is increased anterograde trafficking of Nav1.7 and Nav1.8 that likely involves KIF5B^116,191. However, the time course and enhanced trafficking that supports increased channel expression could only be inferred from these studies. Real-time imaging studies of Halo-tagged Nav1.7 channels showed that applying a cocktail of inflammatory mediators to rat DRG neurons in vitro causes a two-fold increase in Nav1.7 surface levels at the distal axonal membrane, as well as concomitant increases in both the loading of vesicles with Nav1.7 and the flux of these Nav1.7-carrying vesicles in axons¹⁰⁶. This increase in trafficking and membrane insertion resulted in increased neuronal excitability, as demonstrated by the fact that the spontaneous electrical activity of distal axons (as measured by calcium imaging) increased and this activity was significantly attenuated by the specific block of Nav1.7 channels¹¹⁴.

Importantly, the enhancement of ion channel trafficking by inflammation is not an indiscriminate process. Although the anterograde trafficking and membrane insertion of Nav1.7 are increased by inflammatory mediators, Kv7.2 channels (responsible for anti-excitatory effects) are not affected, even though the two channels are transported to distal axons in the same vesicles¹¹⁴. Strikingly, the increase in axonal electrical activity imparted by inflammatory mediator treatment was directly linked to increased Nav1.7 in the axonal membrane and not the reduction in the Kv7.2 levels. Furthermore, despite the dramatically increased anterograde trafficking of Nav1.7 channels, there was no modulation of channel internalization or retrograde transport. These findings indicate that there are mechanisms by which neurons selectively regulate the composition of their excitable membranes in pathophysiological states.

The impacts of specific immune cytokines on Nav channel distribution have also begun to be elucidated. A recent study showed that treatment of rat DRG neurons with the inflammatory cytokine tumor necrosis factor (TNF) increased their excitability and dramatically increased vesicular loading, flux and surface levels of Nav1.7 in distal axons¹²⁴. Given that an inflammatory pain phenotype can develop much faster than can be explained by the biogenesis of new channels and delivery to the axon^179,192,193, this study examined the acute effect of TNF on Nav channel behavior, revealing that the effects of TNF on Nav channel insertion are dependent on p38 MAPKs and regulated in a compartment specific manner¹²⁴: TNF application resulted in rapid Nav1.7 insertion into somatic membranes (within 20 minutes, in line with prior reports¹⁹⁴), but in the distal axon no changes in channel insertion were detected even after an hour of incubation with the cytokine. Furthermore, the N-terminus of the Nav1.7 channel was sufficient to drive the rapid membrane insertion in neuronal somata, and mutation of a specific serine residue, ostensibly the phosphorylation site of the p38 MAPKs, abolished rapid membrane insertion of the channel¹²⁴. Since TNF increases the accumulation of Nav1.7 channels in vesicles at the distal axon, the compartment-specific regulation of surface Nav1.7 levels appears to be due to differential kinetics of the membrane insertion step. The molecular determinants of this differential and compartment-specific regulation are unknown but may represent novel therapeutic targets for the treatment of pain.

Chemotherapy-induced peripheral neuropathy (CIPN) is a frequent side effect of cancer treatments, which can severely impact patient quality of life and survival¹⁹⁵. CIPN leads to allodynia, pain due to a stimulus that does not normally provoke pain, and is a common adverse effect of the application of microtubule-destabilizing chemotherapeutic agents (such as vincristine) or microtubule-stabilizing agents (such as paclitaxel; PTX) to sensory neurons¹⁹⁶. Human and rodent DRG neurons become hyperexcitable after treatment with PTX, due to the increased expression of various pro-excitatory ion channels, including Nav1.7^197–200. Paradoxically, even though agents like PTX disrupt microtubule homeostasis and are thus expected to impair the axonal transport of microtubule-dependent vesicles carrying proteins like Nav1.7 and Nav1.8, PTX enhanced vesicular delivery to and surface levels of Nav1.7 and Nav1.8 in distal axons^113,201. This may be because PTX enhances the interaction between tubulin and MAP2²⁰², which may affect the affinity of kinesins for microtubules²⁰³.

Concentrations of PTX that correspond to serum PTX measurements from people being treated with this medication²⁰⁴ appear to drive Nav1.7 and Nav1.8 channel trafficking even in the context of axonal degeneration^113,201. These effects add to the impact of other drivers of Nav channel trafficking, including inflammatory mediators, that may concomitantly be acting on DRG neurons in an in vivo context. Other evidence for enhanced trafficking of Nav channels in CIPN includes reports of increased trafficking of Nav1.6 due to enhanced interactions with KIF3A that are dependent on the palmitoylation of δ-catenin in response to treatment with oxaliplatin²⁰⁵. The molecular mechanisms that govern vesicular loading and the signal transduction cascades that enable an upregulation in ion channel trafficking in states of hyperexcitability need to be explored further.

Pathophysiological changes in channel trafficking can have substantial effects on neuronal excitability. Similarly, diseases that affect the determinants of vesicular fate, and consequently ion channel (among other protein) transport and distribution can result in neurologic disease. CMT disease is a sensory and motor neuropathy that has been linked to mutations in several of these proteins, including KIF1A²⁰⁶ and KIF1B²⁰⁷. CMT type 2B has also been linked to mutations in the late endosomal GTPase Rab7²⁰⁸. Though a wide variety of membrane proteins would be expected to be affected by altered function of the ubiquitously expressed Rab7 protein, mutations in this factor are found to preferentially impact peripheral sensory neurons and motor neurons in a length dependent manner^209,210. The impacts on the regulation of excitable membranes in sensory and motor neurons that is imparted by dose-dependent dysfunction in such a crucial regulator of vesicular fate likely contribute to distal sensory and motor loss^170,209,211. Other transport processes have also been implicated in neurological disease; for an excellent review on diseases caused by mutations in kinesins, dynein, and microtubules, see REF¹⁷¹.

Questions and future directions

Our understanding of the life-cycle of voltage-gated ion channels has been greatly advanced with the advent of new technologies and advanced imaging techniques; however, there are many questions left to answer. While the AIS was long presumed to be a static structure, with membrane and cytoskeletal components that could persist on the order of weeks, more detailed studies have uncovered dynamic processes that occur on the order of minutes to hours^49,58,59. How do these static and dynamic processes co-exist to generate a region that is important for so many functions, including electrical signaling, polarized vesicular trafficking and the creation of a membrane diffusion barrier? How does neuronal activity drive changes to these processes? These questions have relevance to our basic understanding of neuronal function under both normal conditions and in disease states.

Another question is how trafficking vesicles are recruited to the various membrane domains. How or why do vesicles containing voltage-gated ion channels insert into the plasma membrane within the AIS rather than traveling on to the NOR or distal axon? Single-molecule studies suggest that Nav channels delivered to the AIS are immediately immobilized within the plasma membrane upon insertion³⁰, suggesting that new proteins are delivered directly to where they are needed to contribute to the high membrane density in the AIS. What signals are involved in this precise directed trafficking? Does delivery to the membrane at the NOR occur with the same precision? Similarly, what accounts for the observed differences in rates of channel insertion in different neuronal compartments¹²⁴, and what are the biological implications?

State-of-the-art live imaging approaches have enabled the dissection of the life cycle of proteins that construct the excitable membranes of neurons, particularly the sensory afferents of the DRG. These approaches can be extended to a variety of neuronal and non-neuronal contexts and can enhance our understanding of the determinants of membrane protein delivery, cellular localization, and behaviors in disease states. Of course, the existing approaches have limitations that should inform attempts to build more sophisticated technologies to investigate protein dynamics, especially those that are present at low levels (such as ion channels). For one, live cell imaging approaches largely rely on the overexpression of proteins such as self-labeling enzymes or fluorescent protein fusions. This may result in supraphysiological levels of protein expression and consequent alterations in protein trafficking and localization. However, the expression of such constructs at physiological levels has been achieved with knock-in techniques that allow tags to be incorporated into endogenous genetic loci²¹²: this has been reported for Nav1.2¹¹², Nav1.6¹¹², and Nav1.7 (S.D.D-H, unpublished work), and many findings from overexpression systems have been recapitulated using these tools.

Due to current technical limitations, most live-imaging studies are performed in cell culture. An important question is whether these results will be confirmed in vivo. Most culture systems of neurons in vitro do not capture the complexity of their in vivo geometry and cell-to-cell interactions. For sensory neurons specifically, co-cultures with end organs or multi-compartmental 3D co-cultures incorporating keratinocytes and Schwann cells^213,214, for example, may better recapitulate native cell environments and thus lead to more physiologically relevant conclusions. This may allow the details about how trafficking vesicles are targeted either centrally or peripherally in pseudounipolar neurons to be investigated. Cultures of human induced pluripotent stem cells in which self-labeling enzymes have been knocked-into the endogenous loci of ion channels of interest may be used as ‘organ-on-a-chip’ models to study ion channel trafficking in physiological, pathological, or drug-screening contexts. In vivo observation of axonal protein trafficking is also possible, particularly in drosophila²¹⁵, zebrafish²¹⁶, the mouse sciatic nerve²¹⁷ and the mouse brain²¹⁸. At the moment, it is challenging to use this approach to study scarce proteins like axonal ion channels; however, as fluorescent probe and imaging technology improves, in vivo investigation of excitable membrane construction will become possible.

An emerging technology for elucidating protein–protein interaction utilizes proteomic assays following proximity-induced biotinylation^219–221. In these modalities, enzymes that generate reactive radicals covalently tag neighboring proteins. These tagged proteins can then be analyzed by mass spectrometry. These approaches enable comparisons of protein–protein interaction networks in both health and disease^222,223. One can imagine using these techniques to interrogate important determinants of ion channel trafficking. For example, they could be used to identify regulators of the faster insertion of Nav1.7 channels in the soma versus distal axons in response to TNF treatment by studying the channel macrocomplex in these compartments. Perhaps a similar approach could identify the cellular determinants of the exclusion of Nav1.2 channels from mature NOR. With new spatially resolved proximity-labeling approaches^224–226, these questions can potentially be answered in a compartment-specific manner.

These techniques may be instrumental for our understanding of sensory biology and disease and could be used to identify novel, druggable targets for the treatment of sensory neuronal pathology. For example, although there are a number of treatments for acute pain syndromes, including NSAIDs, opioids, and the recently approved Nav1.8 inhibitor Suzetrigine²²⁷, there are no approved therapies for chronic pain syndromes, even though these conditions involve many of the same peripheral mediators. Lessons learned from studies in peripheral sensory neurons may also be relevant for disorders related to CNS excitability because of the involvement of ion channels that are expressed in both CNS and PNS neurons. Investigating how ion channel regulatory networks change across the time course of a disease process may unlock a therapeutic arsenal in these incredibly prevalent medical conditions.

Conclusion

It is becoming increasingly clear that there are multiple mechanisms that contribute to localization of ion channels within precise neuronal subcellular domains. The diseases that arise when these processes go awry highlight the importance of understanding the intricacies of ion channel trafficking and localization. The advent of novel imaging techniques has enabled real-time imaging of membrane proteins, including voltage-gated ion channels, in sensory neurons, advancing our understanding of how the excitable membranes of these neurons are sculpted and leading to new questions about the basic mechanisms underlying dynamics of ion channel regulation in these structures. These advances will hopefully lead to novel therapeutic strategies for the management of neuropathic pain and other neuropathic diseases.

Table 1.

Comparison of fluorescent tagging and labeling strategies for the study of ion channel life cycles

Strategy	Description	Advantages	Disadvantages
Technologies for fluorescent protein labeling
Antibodies	Fluorophore-bound antibodies can bind to targets and be visualized by fluorescence microscopy.	Commercially available for many targets Fluorophore selection is diverse	Difficulties in specific antibody generation Only extracellular epitopes accessible Large molecules that can agglutinate Physiological function of target proteins may be affected
Fab fragments	Fluorophore bound antigen-binding domains isolated by removal of the Fc region of antibodies allow for target binding with a smaller protein.	Smaller size than full size antibodies	Difficulties in specific antibody generation Only extracellular epitopes accessible Physiological function of target proteins may be affected
Fluorescent proteins	Fluorescent proteins can be attached to proteins of interest	Bright fluorescent signals that can be attached to proteins of interest directly Many FP varieties to choose from	Overexpression of FP-tagged proteins may alter trafficking and localization Bright background from large pool of static labeled proteins
Self-Labeling Enzymatic tags (e.g. HaloTag, SNAPTag)	Enzymatic tags which are not intrinsically fluorescent can react covalently with synthetic fluorescent ligands to provide minimal background fluorescence and flexible labeling	Minimal background fluorescence Multiplex imaging with different colors Increased temporal resolution Can engineer tags to face both intra- and extra-cellularly depending on the application.	Overexpression of tagged proteins may alter trafficking and localization
Strategies for visualization of specific membrane protein life cycle stages
Steady state surface expression	Levels of enzymatically tagged channels at neuronal membranes can be evaluated in a compartment specific fashion by cell-impermeant labeling.	This approach labels only surface channels, eliminating fluorescent background from pool of cytoplasmic channels and enabling robust quantitation of surface protein levels.	Requires an extracellularly tagged membrane protein.
Anterograde trafficking	Forward trafficking can be studied by applying cell permeable labels to cell bodies and imaging the distal axons.	This configuration significantly increases the signal to noise ratio, allowing the imaging of vesicles with dim fluorescent signals, carrying as little as one labeled channel. Preferential visualization of vesicles undergoing fast transport is also achieved.	Excludes visualization of locally translated proteins. Cannot visualize somatic trafficking due to the size of the cell body and labeling of cytoplasmic channels in the soma.
Retrograde trafficking	Retrograde trafficking can be investigated by applying cell impermeable labels to neuronal axons and imaging internalized channels in the distal axon. Using cell permeable fluorescent ligands will not permit visualization and robust analysis of vesicular retrograde trafficking because of the relatively large cytoplasmic pool of channels that will be moving in both directions.	This approach labels only surface channels, eliminating fluorescent background from cytoplasmic channels.	Only allows visualization of proteins internalized from distal axons.
Co-trafficking	Co-trafficking of membrane proteins in vesicles can be visualized by engineering protein constructs with distinct and mutually compatible tags and using appropriate labeling strategies as described above.	This configuration significantly increases the signal to noise ratio, allowing the imaging of vesicles with dim fluorescent signals. Preferential visualization of vesicles undergoing fast transport is also achieved.	Labeling efficiency of different enzymatic tags may vary. Excludes visualization of locally translated proteins.
Insertion and removal	Insertion and removal of membrane proteins from the membrane can be detected by saturating tagged proteins with a cell impermeant label and then replacing the fluid in the compartment with another cell impermeant label. This then enables visualization of the kinetics of insertion and removal of surface proteins from excitable membranes.	This approach allows for the visualization of protein turnover and dynamics of surface populations. Both insertion and removal can be measured with this assay.	Requires imaging over long timeframes.

Glossary

Accessory subunits

Auxiliary proteins that modulate the function, localization, and trafficking of sodium channels without forming the channel pore.

Actin

A cytoskeletal protein involved in maintaining cell morphology and facilitating intracellular transport.

Adaptor proteins

Molecules that mediate interactions between two proteins.

Axon initial segment (AIS)

A specialized region of the proximal axon that is enriched in sodium channels anchored by scaffolding proteins.

Channel life cycles

The stages that sodium channels undergo including synthesis, trafficking, membrane insertion, internalization, recycling, and degradation.

Diffusion trapping

A mechanism in which proteins diffuse laterally in the membrane until captured by anchoring proteins at specific sites.

Endocytosis

A process by which cells internalize portions of the plasma membrane by vesicle formation.

Endosomes

Membrane bound compartments involved in transporting molecules within a cell.

Fluorescence recovery after photobleaching (FRAP)

A technique for measuring the movement and exchange of fluorescently labeled molecules in live cells.

HaloTag

A self-labeling enzymatic tag that forms a covalent bond with synthetic ligands, enabling specific labeling for imaging or biochemical studies.

Motor proteins

Proteins, like myosin, kinesin, and dynein, that work to transport cargo along cytoskeletal filaments

Nodes of Ranvier (NOR)

Gaps in the myelin sheath of axons, enriched in sodium channels, that enable saltatory conduction along the axon.

Platinum replica electron microscopy

An imaging technique that involves coating samples with platinum to produce high-resolution images of cellular structures.

Secretory vesicles

Membrane-bound sacs that transport and release molecules, such as peptides or neurotransmitters, from cells

Super-resolution optical microscopy

Advanced light microscopy techniques that achieve higher resolution than conventional methods, revealing structures at the nanoscale.

Trafficking vesicles

Membrane-bound carriers that transport molecules between different compartments within cells.