Real-time imaging of axonal membrane protein life cycles

Abstract

The construction of neuronal membranes is a dynamic process involving the biogenesis, vesicular packaging, transport, insertion, and recycling of membrane proteins. Optical imaging is well suited for the study of protein spatial organization and transport. However, various shortcomings of existing imaging techniques have prevented the study of specific types of proteins and cellular processes. Here, we describe strategies for protein tagging and labeling, cell culture, and microscopy that enable the real-time imaging of axonal membrane protein trafficking and subcellular distribution as they progress through some stages of their life cycle. First, we describe a process for engineering membrane proteins with extracellular self-labeling tags (either HaloTag or SNAPTag), which can be labelled with fluorescent ligands of various colors and cell-permeability, providing flexibility for investigating the trafficking and spatiotemporal regulation of multiple membrane proteins in neuronal compartments. Next, we detail the dissection, transfection, and culture of dorsal root ganglion sensory neurons in microfluidic chambers, which physically compartmentalizes cell bodies and distal axons. Finally, we describe four labeling and imaging procedures which utilize these enzymatically tagged proteins, flexible fluorescent labels, and compartmentalized neuronal cultures to study axonal membrane protein anterograde and retrograde transport, the co-transport of multiple proteins, protein subcellular localization, exocytosis, and endocytosis. Additionally, we generated open-source software for analyzing the imaging data in a high throughput manner. The experimental and analysis workflows provide an approach for studying the dynamics of neuronal membrane protein homeostasis, addressing longstanding challenges in this area. The protocol requires 5–7 days and expertise in cell culture and microscopy.

Introduction:

Core neuronal functions depend on processes which mediate protein distribution in specific cellular compartments and include the long distance transport and cycling of proteins in and out of the cellular membrane. These processes are particularly challenging to study in axons where the discrete localization of membrane proteins such as ion channels is critically important in determining neuronal functions¹. A clear example is provided by sensory neurons where proteins are produced in the cell body yet must be transported over vast distances along axons (up to a meter in humans) to reach their functional locations.

Membrane protein trafficking dysfunction is implicated in the pathophysiology of many neurodegenerative diseases^2,3 and disorders of excitability including epilepsy, arrhythmia, and pain^4–6. Axonal trafficking systems represent potential therapeutic targets⁷, however, our understanding of the contributions of dysregulation of axonal trafficking of membrane proteins in neuronal health and disease is limited by difficulties in visualizing the movement of scarce proteins in real-time. Distinguishing proteins embedded within the cell membrane from those within the cytoplasm, especially in axons, is particularly challenging.

We have recently developed experimental tools and workflows to overcome these challenges by incorporating cultured neurons in microfluidic chambers, membrane proteins with extracellular self-labeling enzymatic tags, and a combination of bright and photostable cell-permeable and cell-impermeable fluorescent ligands. Here, we detail several assays which enable the study of multiple aspects of axonal membrane protein trafficking in real-time – including vesicle dynamics, vesicular packaging, and protein insertion and removal within axonal membranes. Some of these techniques can be extended to study protein dynamics within the somatic membrane. These workflows are amenable to the study of pathophysiological states, such as inflammation, the effects of pharmacological treatments, and cellular processes such as post-translational modifications and protein-protein interactions.

Comparison with other methods

Immunohistochemistry and immunofluorescence assays are useful for observation of protein expression and localization in axons however they require the presence of specific epitopes, which must be extracellular if specific labeling of surface antigens is desired⁸, and depend on the specificity of the immunologic reagents (antibodies, nanobodies). Moreover, immunofluorescence protocols that involve cell fixation and permeabilization do not allow for dynamic studies or differentiation of cytoplasmic proteins from those at the cell surface. The development of fusion proteins with fluorescent markers such as green fluorescent protein (GFP) allows the tagging and localization of specific proteins in live cells^9,10. However, challenges persist with both the detection of weak signals of lowly expressed proteins and the visualization of highly expressed proteins undergoing transport due to bright background signals from the large pool of static labelled proteins^11,12. Solutions to the problem of bright background have included the use of Fluorescence Recovery After Photobleaching (FRAP) and photoactivated fluorescent tags, which respectively involve either eliminating background fluorescence or activating fluorescence of the protein of interest within a specific region^13,14. However, a disadvantage of the FRAP technique is that photobleaching generates free radicals which may cause damage and affect the biology under observation^15,16. To visualize proteins at the cell surface and monitor their insertion in the cell membrane, tools include acid-sensitive fluorescent protein tags such as superecliptic pHluorin (SEP), which display increased fluorescence at the cell surface compared to the acidic environment inside cytoplasmic vesicles¹⁷. Strategies to label proteins at the cell surface and track their movement after endocytosis include live-labeling with antibodies, nanobodies, biotinylation and fluorescent toxins^18–21, which have limitations around specificity and flexibility of labeling. These and other recent advances in protein labeling and imaging techniques have been comprehensively reviewed²². Another difficulty specific to studying trafficking of large, precisely tuned membrane proteins such as voltage-gated sodium (Na_V) channels is tagging them without disrupting their physiological function. For this reason, previous imaging studies of Na_V transport relied on tagged channel fragments containing putative trafficking motifs^23,24.

We have recently employed alternative tagging approaches to visualize ion channel localization and trafficking^25,26. Here, we describe methodological solutions to the aforementioned challenges that enable real-time imaging of multiple axonal membrane proteins as they are delivered to different neuronal compartments, incorporating strategies for protein tagging and labeling, cell culture, and microscopy (Fig 1). First, we describe a process for engineering membrane proteins with extracellular tags (either HaloTag or SNAPTag). These enzymatic tags are not intrinsically fluorescent, but specifically and covalently react with synthetic fluorescent ligands, providing advantages of minimal background fluorescence in the absence of their ligand and flexibility with regards to timing and color of labeling^27,28. These tagging strategies have been used to study proteins in multiple cellular compartments including nuclear proteins^29,30. Further, Halo- and SNAP- Tagged proteins have also been used in studies in primary neurons, enabling studies of protein dynamics at single molecule resolution^31,32.

Figure 1. Flowchart of axonal membrane protein imaging assays.

All assays share common steps of generating and validating tagged protein constructs, harvesting, and transfecting DRG neurons, and seeding neurons in microfluidic chambers (MFCs). Then, we describe 4 distinct assays that evaluate transport and behavior of membrane proteins at some stages of their life cycle. Each of these assays has their own procedure, as well as their own analysis and quantification protocols.

Development of the protocol

While developing these approaches, we tested multiple different protein tagging strategies with varying success. The first successful full-length Na_V channel construct with an extracellular epitope utilized a biotin acceptor domain (BAD) recognized by a bacterial biotin ligase (BirA) that selectively ligates biotin to the lysine residue within the BAD sequence³³. Taking advantage of the strength and specificity of biotin-streptavidin interactions, we labeled surface proteins with streptavidin-conjugated fluorophores suitable for single-particle tracking studies^20,34. Since the BAD is a small epitope, it is easy to incorporate into an extracellular loop with minimal effect on channel function. However, the disadvantage of this labeling strategy is that in live-cell experiments it can only be used for surface labeling due to lack of cell-permeable ligands. Furthermore, our attempts to label using streptavidin-conjugated quantum dots (which are extremely bright and photostable, making them superior for single-particle tracking³⁵) have been unsuccessful, likely due to steric hindrance between the large quantum dot and the bulky Na_V channel. The fluorogen-activating protein (FAP) technology offers an alternative labeling method with the advantage of employing both cell-permeable and cell-impermeable labeling strategies. Additionally, FAPs are fluorogenic, meaning they remain nonfluorescent until they bind, effectively minimizing non-specific fluorescence signals³⁶. Use of the FAP to label Na_V channels worked well in our hands, but the fluorescence intensity of individual channels was dim, preventing their use for single-particle tracking or visualizing of trafficking vesicles. SEP has been successfully utilized to investigate the surface localization and trafficking of calcium and potassium channels^37,38. However, in our experience, the SEP is dimly fluorescent within the ER lumen and the presence of a substantial pool of ER-localized Na_V channels reduced the signal/noise ratio, which precluded visualization of surface proteins with high resolution. Lastly, we attempted to utilize the ‘spaghetti monster’ fluorescent protein (smFP) labeling strategy where multiple epitope tags are incorporated within the scaffold of a fluorescent protein, thereby amplifying the signal from a single channel³⁹. We had moderate success expressing a smFP-Na_V channel in HEK293 cells where we had a small number of brightly labeled channels. However, expression/labeling in neurons was poor. Ultimately, we found the HaloTag and SNAPTag labeling strategies to provide the ideal combination of sufficient expression levels, specificity of labeling, and availability of useful tagging strategies.

The usefulness of the HaloTag and SNAPTag labeling systems is greatly enhanced by an expanding toolbox of ligands with different optical and chemical properties. For example, the ligands can be conjugated to any of a suite of newly developed “JaneliaFluor” fluorophores, which are brighter and more photostable than other commonly used fluorophores^40–42, as well as biochemical tags such as biotin. Further, we take advantage of the fact that these ligands are available in either cell-permeable or cell-impermeable varieties to label either the proteins which are present at the cell surface or within the cytoplasm⁴³. Importantly, labeling proteins at the cell surface using cell-impermeable ligands requires that the protein tag protrudes into the extracellular space. Tagging proteins in this way can be challenging, especially in the case of proteins where both termini are intracellular.

Overview of the procedures:

In this Protocol, we describe procedures to overcome challenges associated with tagging proteins with intracellular termini and large membrane proteins such as Na_V channels. We further detail a procedure for the dissection, transfection, and culture of dorsal root ganglion (DRG) sensory neurons in microfluidic chambers (MFCs; Fig 3), which results in physical compartmentalization of cell bodies and distal axons. The plating strategies we describe can be extended to other neuronal subtypes. We then provide four distinct labeling and imaging procedures which utilize these enzymatically tagged proteins, flexible fluorescent labels, and compartmentalized neuronal cultures to study various aspects of axonal biology (Fig 4).

Figure 3. Microfluidic chambers separate neuronal axons from somas.

(a) Microfluidic chambers enable fluidic separation of distinct compartments. The somatic compartment should be marked for easy identification.

(b) Neurons should be plated into MFCs containing a growth factor gradient. There are 2x the amount of growth factors in the axonal chamber relative to the somatic chamber. This gradient promotes the growth of axons from the somatic side (left) to the axonal side (right).

(c) After 7 days in vitro, axons are visualized growing from the somatic chamber (left) to the axonal chamber (right).

(d) MFC-plated neurons transfected with Na_Vβ4-GFP visualized by excitation with a 488 nm laser.

Figure 4. Live-cell imaging assays to interrogate the trafficking of membrane proteins and their behavior at different stages of their life cycle.

We describe protocols to investigate membrane protein surface expression, vesicular dynamics, co-trafficking of proteins, and protein exo- and endocytosis.

First, we describe the use of cell-impermeable fluorescent ligands to quantify proteins located at the surface of specific membrane domains, such as the distal axon or cell body (Fig 5). Second, we describe the principles of Optical Pulse-chase Axonal Long-distance (OPAL) imaging; in short, DRG neurons expressing tagged proteins are cultured in MFCs and fluorescent, cell-permeable tag ligands are added to the soma chamber, after which any labeled channels that are trafficked to the distal axon chamber can be visualized with very low background fluorescence (Fig 6). This allows for clear visualization of individual transport vesicles carrying a small number of labeled proteins and can be adapted to study anterograde transport (Supplementary Video 1), retrograde transport, or both. We also provide an example of how these systems can be used with super-resolution microscopy techniques to visualize proteins within subcellular compartments (Fig 7). Next, we explain how to extend the OPAL method to study the protein contents of transport vesicles through co-expression of multiple tagged proteins and simultaneous multi-color imaging (Fig 8; Supplementary Video 2). Lastly, we describe a method involving sequential labeling of surface proteins which allows for the measurement of rates at which proteins are inserted in, or removed from, the surface of distal axons (Fig 9). Additionally, we have also generated open-source software that can be used to facilitate high throughput analysis of imaging data. Together, this experimental preparation, group of assays, and analysis toolbox provide a platform for the study of neuronal membrane protein delivery, including to the cell membrane at a long distance from the soma.

Figure 5. Evaluation of axonal protein surface expression using protein labeling.

(a) Labeling schematic for the distal surface expression assay. Using neurons plated in MFCs, a cell impermeable label is applied to the axonal chamber. The use of the impermeable label ensures that only proteins present at the cell surface, and not those in the cytoplasm, will be labeled. For quantitative measurements of distal surface expression, we recommend fixation before imaging.

(b) High resolution imaging of distal protein expression can be used to show how expression of a specific protein changes in response to agent exposure. Paclitaxel is a chemotherapeutic agent known to cause peripheral neuropathy and pain. We investigated if paclitaxel treatment affected the delivery and localization of Nav1.8 channels, which are obligate mediators of pain, in distal axons. Left panels show the view through a spinning-disk confocal microscope of the distal axonal terminal of a rat pup DRG neuron expressing Halo-Na_V1.8 labeled with JF635i. Right panels are pseudo-colored, max intensity projections of acquired images of the same neurons. Top panels are from untreated neurons, while bottom panels are images acquired from neurons exposed to 25 nM Paclitaxel for 24 hours.

(c) Outlining the distal-most 30 μm of the axon to quantify fluorescence intensity. These measurements can be averaged to determine the relative amount of protein expressed at the surfaces of neuronal axons.

(d) Quantification of protein expression from axons expressing Halo-Na_V1.8 demonstrates increased surface labeling in distal axons from neurons exposed to 25 nM paclitaxel relative to control. Data is collected from 3 independent cultures and transfections. Box plot elements: Whiskers: minimum, maximum. Box limits: 1^st quartile, median, 3^rd quartile. ** p < 0.01 by unpaired t-test.

(e) Protein surface expression with tagged-proteins can be assessed at the cell body as well. Left: Halo-Na_V1.7 expression assessed with the far-red JF635i dye. Right: SNAP-Na_V1.7 expression assessed with the red JF549i dye. Intracellular signal is due to endocytosis of proteins which were labeled at the cell surface.

Panels b-d are adapted from Baker et al. Front Mol Neurosci, 2023⁴⁶ with permission of the authors. Panel e is adapted from Higerd-Rusli et al. J Neurosci, 2022²⁶ with permission of the authors.

Figure 6. Visualizing vesicular trafficking of axonal membrane proteins.

(a) Labeling schematics for trafficking assays. Top panel – Using neurons plated in MFCs, a cell permeable label is applied to the somatic chamber. Proteins of interest are labeled in the somatic cytoplasm, and the vesicular trafficking of these proteins can be visualized in the axonal chamber. Middle panel – applying cell-impermeable label to the axonal chamber labels proteins present on the axonal membrane. Following protein internalization into endosomes, retrograde trafficking of these endosomes can be visualized. Bottom panel – applying cell permeable label to the somatic chamber and cell permeable label with a differently colored fluorophore to the axonal chamber labels both membrane and cytoplasmic proteins in both chambers. Anterogradely moving and retrogradely moving vesicles can then be simultaneously visualized in the grooves of the MFCs by dual color imaging.

(b) Time series of anterogradely moving vesicles visualized by OPAL imaging. Individual vesicles are marked by colored arrowheads, and corresponding kymograph tracks (right panel) are labeled with identically colored arrowheads. Based on the orientation of our MFCs (axons extend from left to right in the microscope field), anterogradely moving particles travel from the upper left to lower right.(c) Left panels: Time series of surface-labeled channels recently internalized and retrogradely moving in a distal axon (pink arrowheads). The right panel is a kymograph plot of this axon, with the marked retrogradely moving vesicle identified by the arrow. Kymographs display distance along the axon on the x-axis and elapsed time on the y-axis. Based on the orientation of our MFCs (axons extend from left to right in the microscope field), retrogradely moving particles travel from the upper right to lower left.

(d) Automated Deep Learning based identification of vesicle tracks. The KymoButler software⁷⁰ uses artificial intelligence to trace the lines in a kymograph (input: left, bright panel, output: right, dark panels) and extracts information about particle movement including vesicle velocity and signal intensity (which reflects the number of labeled proteins present in a vesicle. Vesicular flux can be assessed by counting the number of vesicle tracks that cross the midline of the analyzed kymograph. Threshold settings that are too high result in inappropriate labeling (second panel from left). Threshold settings that are too low result in vesicle tracks not being detected by the algorithm (second panel from right). An appropriate threshold setting allows precise delineation of vesicle tracks (rightmost panel).

(e) Example application of the OPAL imaging technique. Signaling cascades downstream of pro-inflammatory cytokines can modulate function of Na_V channels and increase excitability of sensory neurons, and this contributes to inflammatory pain⁷⁸. We investigated if treatment with an inflammatory mediator cocktail (in μM: 1 bradykinin, 10 prostaglandin E2, 10 histamine, 10 serotonin, 15 ATP) affected the trafficking behavior of vesicles carrying Halo-Na_v1.7; Anterograde trafficking of Halo-Na_V1.7 carrying vesicles in the absence (control; top kymograph) and presence of an inflammatory mediator cocktail (IM; bottom kymograph). Neurons exposed to inflammatory mediators had significantly enhanced flux and signal intensity of vesicles positive for Halo-Na_V1.7 without significant changes in velocity.

(f) Retrograde trafficking of Halo-Na_V1.7 in neurons was not significantly impacted by incubation with inflammatory mediators.

Data presented as mean ± SEM. * p < 0.05, **** p< 0.0001 by Student’s t-test.

Panels b-c are adapted from Higerd-Rusli et al. J Biol Chem, 2023⁴⁷. Panel d is adapted from Tyagi et al. Front Mol Neurosci⁴⁴. Panels e-f are adapted from Higerd-Rusli et al., Proc Nat Acad Sci, 2023⁴⁸.

Figure 7. Super-resolution imaging of tagged membrane proteins.

(a) Example application of super-resolution imaging of Halo-Na_V1.7 channels in subcellular organelles. Left: DRG neurons were transfected with Halo-Na_V1.7 and LAMP1-mNeonGreen (a marker of lysosomes) and cultured in MFCs. Halo-Na_V1.7 at the surface of distal axons was labeled using cell-impermeable Halo-tag ligand JF635i and the cells were incubated for 6 hours to allow axonal membrane proteins to undergo endocytosis and retrograde trafficking. Center: cells were fixed and SRRF images of DRG somata were acquired. Right: Super-resolution images at high magnification demonstrate co-localization of Halo-Na_V1.7 and LAMP1-mNeonGreen,

(b) Normalized fluorescence across the white dashed line in (a) demonstrates the presence of internalized Halo-Na_V1.7 within the lumen of lysosomes (white dashed line).

This experiment demonstrated that Na_V1.7-carrying endosomes do not fuse with lysosomes in the distal axon but channels are eventually delivered to lysosomes in neuronal cell bodies.

Adapted from Higerd-Rusli et al. J Biol Chem, 2023⁴⁷.

Figure 8. Evaluating co-trafficking of axonal membrane proteins in vesicles.

(a) Labeling schematics for co-trafficking assays. Top panel – Neurons are plated in MFCs. Neurons are expressing one construct conjugated to a Halo or SNAP tag and one construct directly linked to a fluorescent protein. A cell permeable label is applied, and neurons that express the protein of interest tagged with the complementary enzyme of the ligands will be labeled, and anterogradely trafficking vesicles carrying one or both proteins can be visualized in the axonal chamber. Bottom panel – Co-trafficking can also be evaluated with proteins tagged with different enzymes (e.g. HaloTag and SNAPTag). In this case, two cell permeable labels, each with a spectrally unique fluorophore conjugated to a different ligand (HaloTag, SNAPTag) are applied to the somatic chamber.

Various voltage-gated ion channels have distinct physiologic functions in DRG neurons(e.g., Na_V channels, promote excitability, K_V channels, dampen excitability). In these experiments, we investigated the co-transport of voltage-gated ion channels with other proteins in trafficking vesicles within the distal axon.

(b) Rab GTPases are central regulators of vesicular trafficking, and Rab6a has been shown to travel together in vesicles carrying voltage-gated sodium channels such as Na_V1.7^25,44. We investigated whether Rab6a was present in anterograde vesicles carrying K_V7.2, as well. Time series of anterogradely moving vesicles positive for Halo-K_V7.2 and eGFP-Rab6a. Halo-K_V7.2 and eGFP-Rab6a co-localize in anterogradely trafficking vesicles. Vesicles positive for both proteins move together in time and space. The Venn diagram demonstrates that the majority of observed vesicles contain both of the tagged proteins.

(c) Neuropeptides such as neuropeptide Y are trafficked to axon terminals. To see if canonical neuropeptide-containing vesicles also carry Na_V isoform Na_V1.8, we carried out the co-trafficking assay. Time series of anterogradely moving vesicles positive for Halo-Na_V1.8 or NPY-td-Orange2. Halo-Na_V1.8 and NPY-td-Orange2 travel with distinct trajectories and do not frequently colocalize in anterogradely moving vesicles. The Venn diagram demonstrates that the two proteins are transported independently.

(d) The impact of effector molecules on protein co-trafficking can be assessed. We investigated how treatment of neurons with an inflammatory mediator cocktail (in μM: 1 bradykinin, 10 prostaglandin E2, 10 histamine, 10 serotonin, 15 ATP) affects the trafficking of vesicles carrying depolarizing Na_V channels, hyperpolarizing K_V channels, or both. Kymographs of anterogradely moving vesicles positive for K_V7.2-Halo or SNAP-Na_V1.7 after incubation with normal media (top kymographs) or media containing an inflammatory mediator cocktail (bottom kymographs). Compared to control conditions, treatment with inflammatory mediators increases the proportion of Na_V1.7 carrying vesicles. ****p < 0.0001 by chi-squared test.

Panels b and c are adapted from Higerd-Rusli et al. J Neurosci, 2022²⁶ with permission of the authors. Panel d is adapted from Higerd-Rusli et al. Proc Natl Acad Sci, 2023⁴⁸ with permission of the authors.

Figure 9. Visualizing protein insertion and removal at the axonal membrane.

(a) Labeling schematics for insertion/removal assays. Using neurons plated in MFCs, application of a cell-impermeable label (Label 1) to the axonal chamber saturates all tagged proteins present at the axonal surface. The axonal chamber is then thoroughly washed, and a second, spectrally distinct cell-impermeable fluorescent probe (Label 2) is applied to the chamber at tenfold lower concentration. When new, unlabeled proteins are inserted into the membrane, they will bind the free Label 2 present in the bath solution and fluorescence intensity will increase over time as protein insertion proceeds. Simultaneously, proteins labeled with Label 1 are internalized and thus the fluorescence intensity of Label 1 will decrease over time, reflecting the rate of protein removal from the membrane.

We used this assay to evaluate how treatment with inflammatory mediators (as discussed in Fig 6 and Fig 8) affects the insertion and removal of Na_V1.7 channels, which are effector proteins in inflammatory pain states.

(b) Treatment with inflammatory mediators causes increased insertion of Halo-Na_V1.7 channels in distal axonal terminals beginning at 5 hours.

(c) Treatment with inflammatory mediators does not affect the rate of removal of Halo-Na_V1.7 channels from the distal axonal membrane.

Panels b-c are adapted from Higerd-Rusli et al. 2023, Proc Natl Acad Sci⁴⁸ with permission of the authors.

Applications and limitations of the method:

Our applications of these methods thus far have focused on studying the trafficking of voltage-gated ion channels in peripheral sensory neurons. First, we used the co-trafficking OPAL assay to identify proteins which are transported in vesicles together with Na_V channels²⁵ and investigated putative molecular determinants of this co-trafficking⁴⁴. Interestingly, we next found that Na_V channels are transported together in the same vesicles with voltage-gated potassium (K_V) channels, despite the fact that these two classes of channels have opposite physiological functions²⁶. We have also taken advantage of the system’s amenability to pharmacological intervention, using it to study the effects of chemical inflammation and chemotherapy on the transport and expression of ion channels^25,45,46. Further, we have used these methods to study the lifecycle of ion channels in neurons, from anterograde trafficking and membrane insertion to internalization, retrograde trafficking, and degradation⁴⁷. Most recently, we showed that the trafficking of Na_V channels and K_V channels are differentially regulated by inflammation in ways that influence cellular excitability and function⁴⁸, and that these changes are neuronal compartment specific⁴⁹.

In principle, these assays could be applied to study trafficking processes in multiple neuronal or non-neuronal cell types. First, the OPAL assays are dependent on the separation of neuronal processes from their cell bodies in MFCs and therefore can be applied to any cell types which have processes long enough to grow through the inter-chamber barrier of the MFCs. Thus, it should be possible to apply these methods to dendritic membrane proteins as well as axonal ones⁵⁰. In contrast, the assays involving the measurement of membrane protein levels and turnover at the cell surface could be applied to cells in uncompartmentalized culture, including non-neuronal cells. Further, Halo and SNAPTag systems can be used in super-resolution imaging applications ^47,51,52. Finally, the assays involving labeling of membrane proteins at the cell surface require addition of an enzymatic tag to an extracellular portion of the protein, which can be difficult to achieve in some cases. On the other hand, the methods for observation of intracellular vesicle dynamics can be applied to any proteins trafficking within the axonal or dendritic lumen, and in these cases the location of the enzymatic tag is less critical.

These methods rely on the expression of engineered proteins fused with artificial enzymatic tags. In our studies thus far, we have accomplished this by exogenous plasmid DNA electroporation, which may cause supraphysiological levels of protein expression and consequent alterations in protein trafficking and localization. Thus, if transfection methods are used, it is important to use the minimal amount of DNA which is necessary for protein detection and to conduct control experiments to rule out the possibility of overexpression artifacts. Viral transduction has been successful and effective in the delivery of genetic material to neurons⁵³. However, there are limitations depending on viral serotype; for example, Adeno-Associated viruses have finite DNA capacity which can preclude packaging of plasmids encoding large proteins, such as voltage-gated sodium and calcium channels. Lentiviral vectors solve the payload problem, but require BSL-2 facilities. Expression of constructs at physiological levels can be achieved with knock-in techniques to incorporate tags into endogenous genetic loci⁸. However, this approach will be labor intensive when mutations need to be introduced to identify molecular determinants that regulate the trafficking of the channel.

A limitation of the purified neuronal culture preparation we describe here is that they lack the complexity of the in vivo environment, which may influence protein trafficking. This may be especially relevant for proteins at the cell surface which might interact with extracellular cues. It is possible to reintroduce complexity to the system by co-culturing neurons together with other cell types such as keratinocytes or oligodendrocytes^54,55. It is possible to study axonal protein trafficking in vivo, including in drosophila⁵⁶, zebrafish⁵⁷, mouse sciatic nerve⁵⁸, and mouse brain⁵⁹. While the methods described here provide high sensitivity, experimental control, and ability to specifically study multiple different trafficking processes, the physiological completeness of future studies will require incorporation of in vivo investigation when more sensitive methodologies for detecting dim signals are developed.

Experimental design, expertise, and equipment

Generation and validation of HaloTag or SNAPTag constructs (Steps 1–10)

The first design step is the choice of fluorescent labeling system to use. HaloTag labeled with silicone rhodamine derivatives was shown to exhibit up to 4-fold brighter signal compared to SNAPTag labeled with the same fluorophore⁵¹. In our experience, SNAPTag ligand conjugates display more rapid photobleaching and higher non-specific labeling compared to the same HaloTag ligand conjugates, which is consistent with the findings of a published evaluation of various SNAPTag ligands⁶⁰. While neither HaloTag nor SNAPTag enzymes react with their cognate ligands with perfect efficiency and precise values are not known, estimates suggest that the efficiency of HaloTag may be higher than that of SNAPTag (estimates range from 80% to 33% for HaloTag and 65% to 16% for SNAPTag)^61–63. Additionally, there is evidence that SNAPTag ligands bind non-specifically to DNA, at least when the SNAPTag is linked to transcription factors^32,64, though in our control experiments neither SNAPTag nor HaloTag ligands exhibited significant non-specific binding (Extended Data Fig. 1). Still, we prefer HaloTag for experiments involving only one tagged protein. Because HaloTag and SNAPTag labels have no cross-reactivity, SNAPTag can be used to label a second protein of interest in dual labeling experiments. In this case, we recommend taking advantage of the greater sensitivity of the HaloTag by using it to tag the less abundant of the two proteins. If resources permit, performing complementary experiments in which the proteins of interest are fused with the opposite tag can serve as a useful control. Lastly, while we have used the HaloTag7 variant for our studies thus far, newer HaloTag variants may provide advantages due to their ability to modify ligands for greater brightness and distinct fluorescence-lifetime properties which allows for more flexibility for multiplex imaging⁶⁵.

Some of the methods for tracking membrane proteins described here require the tag to be localized to the extracellular face of the membrane protein. This can present an additional challenge if both termini of the protein are intracellular. While some proteins (such as K_V2.1⁶⁶ and K_V7.2²⁶) can accommodate the HaloTag within an extracellular loop, our attempts to insert it in extracellular loops of Na_V channels failed to produce channels with normal function. It is advisable to avoid inserting tags near functionally critical domains, for example, the pore domain in the S5–6 linker of Na_V channels. If attempts to tag extracellular loops are unsuccessful, another strategy is to attach the enzymatic tag together with a synthetic transmembrane domain to one of the termini, which will reorient said terminus to the extracellular side of the protein (Fig. 2). We utilized this strategy to extracellularly tag multiple Na_V isoforms, using the transmembrane domain from Na_V-β4 subunit, which is naturally bound to the Na_V channel through disulfide bonds^25,26.

Figure 2. Example of an extracellular tagging strategy for a protein with two intracellular termini.

For proteins which don’t naturally have an extracellular terminus or extracellular loops or domains in which a tag can be inserted, an additional transmembrane domain can be added to an intracellular terminus to achieve an extracellular terminal tag.

Schematic modified from Tyagi et al. J Gen Physiol. 2019⁷⁷ with permission of the authors.

After creating a tagged construct and validating its sequence, it is critical to verify the integrity and functionality of the fusion protein. Measuring the size of the fusion protein by western blot can confirm that the enzymatic tag is not cleaved from the protein of interest. Next, trafficking of extracellularly tagged membrane proteins to the cell surface can be confirmed by labeling transfected cells with cell-impermeable fluorescent ligands. If the tagged protein is appropriately translated and trafficked, the next step is to validate functionality, which will require the use of specific assays depending on the protein. In the case of voltage-gated ion channels, this entails carrying out electrophysiological characterization to ensure the channel is still functional and has properties similar to its untagged counterpart.

Dissection, Transfection, and Culture of DRG Neurons in MFCs (Steps 11–22)

We use a previously published protocol for DRG dissection, neuron isolation and transfection by electroporation, and culture with only minor adjustments⁶⁷. We recommend that readers attempting these protocols refer to the previously published article for guidance on DRG neuron electroporation. An important note is that the procedures were optimized for Sprague-Dawley rat pup (P0-P5) DRG neurons, which are relatively easier to electroporate compared to adult rodent DRG neurons, however, we have also successfully transfected DRG neurons from adult mice and rats, and the appropriate age that meets the experimental design should be used. Our procedures comply with the 3Rs measures for the replacement, refinement, and reduction of animals in research. Careful experimental planning and a priori power analysis should take place to ensure the minimum number of animals are sacrificed for use in these experiments.

The sex of the animals used in these procedures may be relevant to the biological hypotheses that are tested. In our experiments, we have used equal numbers of male and female animals to produce each MFC culture, thus randomizing the sex of each analyzed neuron. An alternative approach is to conduct experiments segregated by sex, then analyzing between group differences to determine if data can be pooled or not.

Producing consistent, high-quality DRG cultures in MFCs is critical to obtaining reliable experimental results. MFCs are typically silicon-based devices which must be attached to microscope-compatible glass before cells are cultured within them. They can be purchased commercially or made in the lab using published protocols¹⁹. Our experiments have utilized two-chamber MFCs, which consist of two open chambers separated by a barrier with microgrooves that create passages between the chambers. Three-chamber MFCs are also available and can be used to provide additional compartmentalization for more complex experiments. MFCs are available with microgrooves of various lengths, but in our experience this variable is not critical for most experiments involving peripheral neurons since the length of the axons is relatively much greater than the barrier. Using MFCs with shorter groove lengths may be necessary in experiments with neuronal cell types with shorter axons. Special care must be taken during attachment of the MFC to the glass substrate to ensure that the entire device is fully adhered, or else cells and fluorescent labels will not remain in the desired compartment according to the experimental design. MFCs are also available in “standard” or “open” configurations which differ in the surface area in which cells can grow and the proximity of cells to the microgrooves. We prefer the “open” configuration because it allows more uniform distribution of neurons within the chamber and positions cells closer to the microgroove barrier. Additionally, the “open” configuration allows for easier addition of treatment reagents and labels and is convenient for the repeated washing steps involved in these techniques.

There is a relatively narrow window of time during which experiments can be performed on neuronal MFC cultures. After the cell bodies are dissected and cultured, sufficient time must be given for neuronal axons to sprout and grow across the microgroove barrier into the distal compartment, but as time progresses, the axons become overcrowded within the microgroove barriers, which impacts their health and interferes with protein transport. The speed with which this occurs can be modified by changing the density of cells which are seeded in the cell chamber. Experimental timing is also dependent on the expression levels of transfected proteins, which decreases over time. In our experience, culturing neurons for 5 to 7 days before performing experiments is sufficient to allow for axonal growth, but not axonal crowding or loss of transgene expression.

Labeling and Imaging (Steps 23–37; 46–50; 61–69; 81–91)

These protocols require a microscope capable of sensitive fluorescence imaging of live cells with controlled conditions while minimizing photodamage. The greatest precision of protein localization will be achieved with an optical sectioning method; our protocols describe the use of a spinning-disk confocal microscope, but two-photon, total internal reflection fluorescence (TIRF), or light-sheet microscopy modalities could also be employed. The microscope needs to be equipped with a stage-top incubator to maintain precise control of temperature, humidity, and gas composition. Temperature control is especially important during measurements of transport vesicle velocity, since motor proteins are highly temperature dependent⁶⁸. If using a bicarbonate buffered cell culture medium, the incubator should be capable of gas mixing.

There are many fluorescent HaloTag ligands and SNAPTag ligands from which to choose (see summary in Table 1). In our experiments we have utilized JaneliaFluor conjugated HaloTag or SNAPTag ligands, which have been generously provided by Dr. Luke Lavis and Dr. Jonathan Grimm and are available either commercially or by request from the Janelia Research Campus. We have used JF549, JFX554, JF585, JF635, JF646, and JFX650, all of which are very bright and highly photostable. In our experience JF549 is slightly brighter but also slightly less photostable than JF646 or JF635. JFX650 is similarly bright to JF549 but significantly more photostable. These red and far-red fluorophores are spectrally distinct enough to allow dual labeling without bleed through into neighboring spectral channels. While most tags are cell-permeable by default, cell-impermeable variants (such as JF549i and JF635i, where the “i” indicates cell-impermeability) can be used to label proteins specifically at the cell surface. Like the cell-permeable variants, JF549i seems to be initially brighter while JF635i is more photostable. Note that while cell-impermeable HaloTag and SNAPTag ligands are unable to penetrate the cell membrane while in solution, once they bind to the protein of interest the fluorophore can undergo endocytosis together with the protein. Thus, in instances where quantitative measurement of surface expression is desired, we recommend immediate fixation after labeling to prevent translocation of the surface-labeled proteins. Another property that varies between labels is fluorogenicity, which refers to the increase in fluorescence that occurs upon the ligand binding to its substrate. Some HaloTag ligand conjugates, such as JF585, JF635, JF646, and JF635i exhibit substantial fluorogenicity upon binding to the HaloTag protein, increasing in fluorescence 80, 113, 21, and 6 times, respectively⁴¹. Thus, these labels produce less background fluorescence when in solution, which is ideal for experiments which require the label to be present in excess during imaging, such as labeling channels as they are inserted in the membrane. Self-labeling reactions are not instantaneous, and sufficient time must pass for all enzymes in a population of tagged proteins to be labeled by their cognate ligand. Though many fluorophore-ligand pairings are possible, the key rate-defining feature of the labeling reaction is whether they are bound to a HaloTag or SNAPTag ligand, which have different optimal labeling periods. Thus, the various fluorophore conjugates of a particular ligand can be used in an experiment under identical labeling conditions.

Table 1.

JaneliaFluor conjugated ligands for membrane protein life cycle assays.

Fluorophore	Cell permeability	Ligand Conjugation	λEx (nm)	λEm (nm)	Fluorogenicitya	Photostabilityb	Relative Brightnessc	Application in assaysd
JF54936,38	Cell Permeable	HaloTag, SNAPTag	549	571	Low	Moderate	High	OPAL, Co-trafficking
JF549i36,38,†	Cell Impermeable	HaloTag	549	571	Low	Moderate	High	Surface Expression, Insertion and Removal (Label 1)
JFX55438,†	Cell Permeable	HaloTag, SNAPTag	554	576	Low	Very High	Very High	OPAL, Co-trafficking
JF58537	Cell Permeable	HaloTag, SNAPTag	585	609	80x	Moderate	High	OPAL
JF63537	Cell Permeable	HaloTag, SNAPTag	635	652	113x	High	Moderate	OPAL, Co-trafficking
JF635i37,†	Cell Impermeable	HaloTag	635	652	6x	High	Low	Surface Expression, Insertion and Removal (Label 2)
JF64637	Cell Permeable	HaloTag	646	664	21x	High	Moderate	OPAL, Co-trafficking
JFX65038,†	Cell Permeable	HaloTag, SNAPTag	650	667	Low	Very High	High	OPAL, Co-trafficking

^†Indicates most effective dye for the described application, in our hands

^aFluorogenicity defined as fold-increase in absorbance following conjugation of Dye-HaloTag-ligand to Halo Enzyme

^bRelative photostability expressed as a subjective measure based on our experiments. Photobleaching rates for photostable dyes can be found in the references for each dye.

^cRelative brightness based on our experience with HaloTag or SNAPTag conjugated ligands. All listed fluorophores are very bright relative to most other fluorophore families.

^dSuggested applications based on previous experiments in our hands

Additionally, the labeling of Halo and SNAP-tagged proteins with fluorescently conjugated cognate ligands is kinetically dependent on temperature. At the same time, membrane protein dynamics are also regulated by temperature. Therefore, a balance must be struck between ensuring that protein turnover during labeling steps is reduced and also ensuring that an efficient labeling reaction takes place. Thus, for assays where capturing the total protein expression at the membrane surface is important (the surface expression assay and the baseline condition of the exo/endocytosis assays), we recommend labeling at room temperature (22 ° C) rather than at 37 °C or under refrigeration. Dynamic assays of membrane trafficking, however, should be performed at physiologic temperature (OPAL, co-trafficking, subsequent time points of exo/endocytosis). A summary of recommended labeling conditions and times can be found in Table 2. Further, given that the manipulations required for these experimental preparations may influence the processes under observation, any measurements must be compared to suitable control conditions, such as vehicle-treated neurons labeled and imaged in the same manner.

Table 2.

Summary of ligand incubation conditions for imaging assays.

Assay	Optimal HaloTag ligand Incubation time	Optimal SNAPTag ligand incubation time	Incubation Temperature
Surface Expression	20 minutes	30 minutes	22 ºC
OPAL	15 minutes	25 minutes	37 ºC
Co-trafficking	15 minutes	25 minutes	37 ºC
Insertion/Removal	20 minutes	30 minutes	22 ºC

Labeling of proteins within the MFC is a key step with several important considerations. Axons of sensory neurons adhere weakly to glass and can detach easily. Thus, labeling and washing steps must be carried out gently, ideally using an automated solution exchange system. Another key consideration is the media used for labeling and imaging. Most cell media contain phenol red, which can cause high background fluorescence. Thus, for experiments lasting less than two hours, we replace the media with clear neuronal imaging saline (NIS). However, for longer experiments we use complete media without phenol red to ensure long-term cell health. For bicarbonate buffered medium, this requires an incubator capable of mixing carbon dioxide gas. While it is conceivable that Halo- or SNAPTag ligands could impact cell health, we have not observed any obvious impacts on neuronal viability across a range of ligand concentrations. Others have exposed cells to ligand for up to 60 hours without noted impact on cell health ⁴³. Riboflavin is a common auto fluorescent component of cell media that should be excluded from imaging media if possible.

Image Analysis (Steps 38–45; 51–60; 70–80; 92–100)

The protocols described here result in several types of image data that must be processed prior to analysis. Time-lapse imaging of vesicle motion along axons can yield measurements of vesicle velocity, flux (number of vesicles flowing through the axon per unit time), fluorescence intensity (which reflects the abundance of tagged proteins within a given vesicle), and protein contents (in the case of double- or triple-labeling experiments). In order to extract this data from time-lapse movies, we first use a published tool to generate kymographs, which plot vesicle movement along the axon or other trajectory on one axis and elapsed time on the other⁶⁹. In the resulting images, moving objects appear as slanting lines while stationary objects appear as vertical lines. We use an automated deep learning algorithm trained on kymograph data to trace the trajectories of individual vesicles within kymographs and extract velocity and intensity measurements for each vesicle⁷⁰. Several high-quality open source software options for single-molecule imaging analysis exist, including TrackIt, TrackMate, and uTrack^71–73. These tools can be used in the analysis of data generated from assays that we describe here. However, an advantage of using the strategy of extracting data from kymographs over tracking vesicles directly within time-lapse images is that this approach allows us to obtain data from distinct axons, more accurately track vesicles with different velocities whose tracks intersect, and clearly distinguish anterograde from retrograde motion. We measure protein abundance at the surface of cells, by obtaining three-dimensional z-stack images containing the entire axon or cell body of interest. Since all segments of an axon may not be captured by a single z-slice at a given focal plane, we collapse the z-stacks into two-dimensional images by maximal intensity projection to ensure that data from the entire axon is captured.

Regulatory Approvals

The experiments using animals were performed in accordance with the NIH Guide for the Care and Use of Laboratory Animals and were approved by the IACUC of the Veterans Administration Connecticut Healthcare System

Materials

Reagents

• Sodium Chloride (NaCl) (MilliporeSigma, cat no. 7647–14-5)

• Potassium Chloride (KCl) (MilliporeSigma, cat no. 7447–40-7)

• Calcium Chloride (CaCl2) (MilliporeSigma, cat no. 10043–52-4)

• Magnesium Sulfate (MgSO4) (MilliporeSigma, cat no. 7487–88-9)

• Monosodium Phosphate (NaH2PO4) (MilliporeSigma, cat no. 13472–35-0)

• Ascorbic Acid (MilliporeSigma, cat no. 50–81-7)

• D-Glucose (MilliporeSigma, cat no. 50–99-7)

• HEPES (MilliporeSigma, cat no. 7365–45-9)

• Sodium Hydroxide (NaOH) (MilliporeSigma, cat no. 1310–73-2)

• Neurobasal Medium (ThermoFisher, cat no. 21103049)

• Neurobasal Medium, minus phenol red (ThermoFisher, cat no. 12348017)

• Poly-L-Lysine (ScienCell, cat no. 413)

• Laminin (MilliporeSigma, cat no. L2020)

• Uridine/5-fluoro-2-deoxyuridine (MilliporeSigma, cat no. 50–91-9)

• Ca2+-free DMEM (ThermoFisher, cat no. 21068028)

• DMEM/F-12 (ThermoFisher, cat no. 11320033)

• Bovine Serum Albumin (MilliporeSigma, cat no. A7030)

• Trypsin Inhibitor (MilliporeSigma, cat no. T0256)

• Penicillin-Streptomycin (ThermoFisher, cat no. 15070063)

• 70% Ethanol (Thomas Scientific, cat no. C755G60)

• Janelia Fluor 549-HaloTag ligand (HHMI Janelia Research Campus, cat no. JF549)

• Janelia Fluor 549i-HaloTag ligand (HHMI Janelia Research Campus, cat no. JF549i)

• Janelia Fluor X 554-HaloTag ligand (HHMI Janelia Research Campus, cat no. JFX554)

• Janelia Fluor 635-HaloTag ligand (HHMI Janelia Research Campus, cat no. JF635)

• Janelia Fluor 635i-HaloTag ligand (HHMI Janelia Research Campus, cat no. JF635i)

• Janelia Fluor 646-HaloTag ligand (HHMI Janelia Research Campus, cat no. JF646)

• Janelia Fluor X 650-HaloTag ligand (HHMI Janelia Research Campus, cat no. JFX650)

• Janelia Fluor 549-cpSNAPTag ligand (HHMI Janelia Research Campus, cat no. JF549)

• Janelia Fluor 635-SNAPTag ligand (HHMI Janelia Research Campus, cat no. JF635)

• Janelia Fluor 646-SNAPTag ligand (HHMI Janelia Research Campus, cat no. JF646)

• Janelia Fluor X 554-cpSNAPTag ligand (HHMI Janelia Research Campus, cat no. JFX554)

• Dimethyl Sulfoxide (MilliporeSigma, cat no. D8418)

Animals

• Sprague-Dawley Rat Pups (Envigo)

Equipment

• 50mm Glass bottom Dish (30 mm Glass diameter; No 1.5 glass) (Mattek, cat no. P50G-1.5–30-F)

• Microfluidic Chambers, 450 mm barrier (Xona Microfluidics, cat no. DOC450)

• Bead Bath – Precision general purpose water bath with beads (ThermoFisher, cat no. TSGP28PMO5)

• Peristaltic Pump, Very Low Flow (Instech, cat no. P720/10K)

• P1000 micropipette (Rainin, cat no. 17014382)

• 40×/1.3NA oil-immersion objective (CFI Plan Fluor) (Nikon, cat no. MRD77410)

• 60×/1.4NA oil-immersion objective (CFI Plan Fluor) (Nikon, cat no. MRD01605)

• Ti Eclipse inverted microscope (Nikon, cat no. MEA53100)

• Motorized stage (Nikon, cat no. MEC56100)

• iXon Ultra 888 EMCCD camera (Andor, cat no. DU-888U3)

• Dragonfly 502 spinning disk confocal system (Andor, cat no. CR-DFLY-502)

• 405 nm laser, 100 mW (Andor, cat no. 757-LM-405–100)

• 488 nm laser, 150 mW (Andor, cat no. 757-LM-488–150)

• 561 nm laser, 150 mW (Andor, cat no. 757-LM-561–150)

• 637 nm laser, 140 mW (Andor, cat no. 757-LM-637–140)

• 450/50 bandpass filter (Andor)

• 525/50 bandpass filter (Andor)

• 600/50 bandpass filter (Andor)

• 700/75 bandpass filter (Andor)

• Incubator System for Neuroscience (Tokai, cat no. T1ZW)

• Perfusion Pencil 4 channel manifold (Automate Scientific, cat no. 04–04-xxx)

• Vacuum Pump, Edwards RV8 Rotary Vane (Edwards, cat no. A65401903)

Software

• Computer with the Fiji distribution of ImageJ (https://imagej.net/software/fiji/downloads)

• KymographClear toolkit for Fiji (https://sites.google.com/site/kymographanalysis/)

• Software analysis tools for imaging assays (https://github.com/ycnrr/NatureProtocols_2023)

• Computer with Python 3.7+ installed (https://www.python.org/downloads/)

Reagent Setup

Neuronal Imaging Saline (NIS)

Combine 136 mM NaCl, 3 mM KCl, 2.5 mM CaCl₂, 1 mM MgSO₄, 0.15 mM NaH₂PO₄, 0.1 mM Ascorbic Acid, 8 mM _D-Glucose, 20 mM HEPES in Millipore water. Adjust the pH to 7.4 with NaOH. Adjust the osmolarity to 320 mOsm by addition of H₂O or _D-Glucose. Filter-sterilize with a 0.2 μm filter. After filtering, this solution can be stored at 4° C for several months.

DRG medium

Mix (in % vol/vol): 89 DMEM/F-12, 10 fetal bovine serum, 1 penicillin/streptomycin.

Somatic Medium

Combine DRG medium with 50 ng/mL nerve growth factor (NGF) and 50 ng/mL glial cell line-derived neurotrophic factor (GDNF).

Axonal Medium

Combine DRG medium with 100 ng/mL nerve growth factor (NGF) and 100 ng/mL glial cell line-derived neurotrophic factor (GDNF).

Serum-free Somatic Medium

Combine Neurobasal medium with 2% B27, 1% penicillin/streptomycin, 50 ng/mL nerve growth factor (NGF) and 50 ng/mL glial cell line-derived neurotrophic factor (GDNF). Add 1 μM uridine/5-fluoro-2-deoxyuridine.

Serum-free Axonal Medium

Combine Neurobasal medium with 2% B27, 1% penicillin/streptomycin, 100 ng/mL nerve growth factor (NGF) and 100 ng/mL glial cell line-derived neurotrophic factor (GDNF). Add 1 μM uridine/5-fluoro-2-deoxyuridine.

BT DRG Media

Add 7.5 mg of Bovine Serum Albumin (BSA) and 7.5 mg of Trypsin inhibitor to 5 mL of DRG medium. Vortex, warm at 37° C and sterile filter (0.2 μm filter). Store at 4 C for up to 5 days.

Equipment Setup

Coating glass bottom dishes

Dilute Poly-L-Lysine in ddH₂O to working concentration of 0.5 mg/mL (1 mL of solution is required for each dish). Filter sterilize. Add 1 mL of solution to each glass-bottom dish, and gently swirl each dish so that the solution covers the entire glass surface.

Note: Glass bottom dishes should be Number 1.5 glass for optimal use with oil immersion microscope objectives.

Incubate dishes at 37° for 24 hours. Next, suction off poly-L-lysine from each dish and wash twice with 1 mL sterile ddH₂O. Prepare working concentration of laminin (10–20 μg/mL in sterile ddH₂O). Add 1 mL of laminin to each dish and incubate at 37° for 2 hours. Remove laminin and allow the dishes to dry. With lids removed, UV sterilize dishes for 5 minutes. The glass bottom dishes are now coated and sterilized for use with MFCs. We recommend preparing dishes no more than 3 days ahead of time to ensure proper adherence of plated neurons.

Installing KymographClear

KymographClear⁶⁹ can be downloaded from the following domain: https://sites.google.com/site/kymographanalysis/. Follow the steps in the manual to install into Fiji.

Installing Fiji Macros

Custom analysis software for the assays described in this protocol can be downloaded at https://github.com/ycnrr/NatureProtocols_2023 To install these macros, navigate to Plugins > Macros > Install… within Fiji and select the appropriate macro.

Spinning Disk Microscope

Spinning disk systems will vary and many different configurations are possible. The basic components required are: an inverted spinning disk microscope with a focus drift correction system and associated imaging software, a stage top incubator, and appropriate microscope objectives. Several of these assays can be conducted before-and-after application of biological or chemical compounds. If such before-and-after capability is desired, the microscope set-up can be fitted with a perfusion and vacuum apparatus. We use a perfusion pencil (Automate Scientific) and a mobile vacuum generator unit (Edwards).

Biological Materials

Animals

Rats or mice. We use 2–4d old Sprague-Dawley rat pups from Envigo for cultures and transfections. Cultures with other rat or mouse DRG neurons are possible but have not been tested.

CAUTION: All experiments with vertebrate animals should be performed in accordance with relevant guidelines and regulations.

Procedure

Generation and validation of HaloTag or SNAPTag constructs

Timing: (Steps 1–6: 1–2 d, Step 7: 4–8 weeks, Step 8: 1–5 days; Steps 9–10: 1 week)

CRITICAL: Many HaloTag and SNAPTag fusion protein constructs are available on DNA repositories such as Addgene.

Construct design: Timing: (Steps 1–6: 1–2 d)

• 1Obtain an mRNA nucleotide sequence for your protein of interest (take note of species, isoform, etc.). The NIH Nucleotide database is a useful resource for this (https://www.ncbi.nlm.nih.gov/nucleotide/). It can be helpful to translate and annotate the sequence to identify key parts of the protein.
• 2Identify important functional domains to avoid when inserting a tag. This can be done by consulting the literature and by running a sequence alignment for similar proteins (isoforms, other species). Regions with high conservation between isoforms and/or species are likely functionally important.
• 3Determine where to put the enzymatic tag. For an extracellular tag, insertion within an existing extracellular loop or terminus would be preferable if it can be done without affecting protein trafficking or function. Engineer the nucleotide sequence of the tag (e.g., HaloTag) in frame with the nucleotide sequence of the protein of interest. Additional considerations (e.g. fusion protein placement, inclusion of a Kozak sequence) in designing the fusion protein have been detailed elsewhere^74,75.

  CRITICAL STEP: The tag can also be fused to an intracellular protein segment; however, the fused enzyme will be inaccessible to cell-impermeant fluorescent ligands.

• 4For integral membrane proteins that require an extracellular tag but have both termini intracellularly and do not have any existing extracellular loops or domains that can be used, an additional transmembrane domain can be engineered to link an intracellular terminus of the protein of interest. For the Na_V1.7 construct, we chose to add the transmembrane segment from the Na_V-β4 subunit, resulting in a protein with 25 transmembrane segments.
• 5For any transmembrane protein with an extracellular N-terminus, ensure that the signal peptide is maintained. For example, when we added the β4 transmembrane domain, its corresponding signal sequence was also included.
• 6The addition of linkers between fusion proteins is a common strategy to attempt to avoid issues with steric hindrance. These linkers are often rich in small amino acids (glycine, serine, alanine) that should allow flexibility of movement. For example, a linker used in our construct was SGSGGAV. For additional considerations in designing linkers see Chen et al. 2013⁷⁶. An example topology for a protein with an N-terminal extracellular tag may be: Signal peptide, Halo-Tag, linker, added transmembrane segment, linker, protein of interest (e.g., Na_V1.7; Fig 2.).

Plasmid Construction: Timing: (Step 7: 4–8 weeks)

• 7After designing the full sequence, the construct can be created using molecular biology techniques²⁵. Our labs’ protocol for generating these constructs is provided in the supplementary material (Supplementary Methods). Alternatively, the sequence can be synthesized by a number of companies.

Construct validation: Timing: (Steps 8–10: 1–2 weeks)

• 8Validate the amino acid sequence through sequencing.
• 9Validate the functionality of the protein as appropriate. For example, we used electrophysiology to characterize the properties of the modified Na_V1.7 construct in relation to the untagged channel.
• 10Validate the localization of the protein, if possible. For example, use antibodies to localize endogenous proteins and compare to the localization of transfected proteins. If constructs of the same protein fused with different tags or tagged in different protein domains traffic similarly to each other there is a high probability that the insertion of the tags did not alter the compartmental distribution of the protein.

Preparation of MFC, Culture and Transfection of DRG Neurons

Preparation of MFC (Timing: Steps 11–16: 2.5 h; Step 17: 3h (can be done concurrently with steps 11–16))

• 11For each MFC, choose a somatic compartment. Mark the top corner of this side with a razor blade or thin marker to keep the somatic side identified for the remainder of the protocol (Fig 3A).
• 12Place MFCs in 70% ethanol for 15 minutes. Move to sterile cabinet. Place MFCs on their narrow side and allow to dry for 1–2 hrs.
• 13Place MFCs in previously prepared glass bottom dishes coated with laminin and poly-L-lysine (see equipment set-up) and allow to adhere. Press on MFCs gently with tweezers to ensure complete adherence to the coated glass. Take extra care when applying pressure to the central groove-containing region. Verify proper adhesion under a tissue culture microscope.
• 14Add ~400 μL of somatic medium in the somatic chamber of each dish. (note: each MFC is slightly variable in terms of volume. Put enough media in such that the surface of the fluid is slightly convex and domed shaped, but not enough that there is a risk of spillage into the opposite chamber).
• 15Add ~400 μL of axonal medium in the axonal chamber of each dish.
• 16Label all dishes and incubate at 37° C until neurons are ready for seeding.
• 17Dissociate and transfect neurons via electroporation according to previously described protocols. We use DRG neurons in our experiments. Culture protocols can be found in Dib-Hajj et al. 2009⁶⁷.

  TROUBLESHOOTING

DRG neuron culturing (Timing: Steps 18–21: 24 h; Step 22: 5–7 d).

• 18Mix dissociated neurons in Ca²⁺-free DMEM (see Dib-Hajj et al. 2009⁶⁷) 1:1 with BT DRG solution (50 μL per MFC).
• 19Add 50 μL of the neuronal suspension to the somatic chamber of each MFC (Fig 3b).
• 20Incubate at 37° C in 5% CO₂ atmosphere
• 2124 hours later, carefully exchange the media in both somatic and axonal chambers with serum free somatic and axonal media, respectively. Add serum free somatic media to the somatic chamber and serum free axonal media to the axonal chamber. The precise volume of medium required will vary depending on the specific MFC used. In our experiments, we fill each chamber with 300–400 μL of serum-free medium.
• 22Replace half of the media every 3 days. 5–7 days after culture (4–6 days after switching to serum free media), neurons in MFCs are ready to be used in imaging experiments.

  TROUBLESHOOTING

Washing, Labeling, Imaging, and Analysis

CRITICAL: Many steps are shared between the various assays described in this protocol (Fig. 1). Common steps are described here, and assay-specific instructions are included in each subsection.

CRITICAL: The steps below describe washing with NIS which we employ for most of our assays. For long imaging protocols (~>2 hours), we recommend washing and keeping neurons in serum free neuronal media minus phenol red to preserve neuronal health, along with appropriate gas mixing.

Washing (Timing: Steps 23–26: 30 min – 1 h, depending on the assay.)

• 23Wash MFCs with NIS to remove neuronal media from both somatic and axonal chambers.

  CRITICAL STEP: there are many wash steps involved in these methods. There are multiple techniques that enable gentle washing of microfluidic chambers without washing away the delicately attached neuronal axons including making use of peristaltic pumps, perfusion pencils, or manual pipetting. Regardless of the method used, each wash should exchange the volume of the chamber solution at least 4x and do so at a rate that does not disrupt the adhesion of the neurons to the glass (e.g., for a 500 μL chamber, wash with at least 2 mL of saline total over a period of approximately 10 minutes). We usually perform washes on a bead bath set to 37 °C.

  CRITICAL STEP: Do not use an orbital shaker for washing or incubation as motion may disrupt neuronal adhesion to glass.

  TROUBLESHOOTING

• 24Apply Halo or SNAP ligands as described for the desired assay as shown below.

  CRITICAL STEP: If possible, during labelling, the fluid level in the compartment containing the fluorescent ligand should be lower than the compartment with no ligand, so that any flow between compartments will be away from the chamber without ligand, thus maintaining low background fluorescence. If the experiment requires adding fluorophore to both compartments, we recommend labeling the compartments sequentially to enable fluorophore isolation.

  TROUBLESHOOTING

• 25Place MFC in stage-top incubator set to 37° C and bring a 40–100x oil immersion objective (we have found 60x to be a good balance between field size and magnification for our applications) into contact with the bottom of the MFC-containing dish.

  CRITICAL STEP: Focus drift can be a concern when imaging multiple fields over multiple timepoints. Using a focus drift correction system (like Nikon Perfect Focus) is recommended to ensure imaging fields remain within the appropriate focal plane.

  CRITICAL STEP: The appropriate magnification will vary between imaging systems and depends on the objectives available, pinhole optimization, and camera resolution.

• 26Using the appropriate laser for the label used, bring axons into focus.
• 27Acquire images following the protocols described for each assay below.
• 28Convert images to TIFF, import for analysis in Fiji (ImageJ 2.0). Conduct analysis as described for each assay below.

Protein Surface Expression

Labeling (Timing: Steps 29–33: 25–35 min, depending on the ligand used.)

• 29Prepare 1mL of 100 nM cell impermeable HaloTag or SNAPTag ligand in NIS.

  CRITICAL STEP: The optimal concentration of this solution will be ligand dependent. We use JaneliaFluors at a working concentration of 100 nM, but other ligands may require a different concentration.

• 30After washing the MFCs, perfuse this solution containing 100 nM cell impermeable HaloTag or SNAPTag ligand into the axonal chamber of the MFC (Fig 5a).
• 31Keep MFCs at RT (~22 °C) while labeling. Expose cells to label for 15–25 minutes. (Note: label times may vary between the fluorescent tags used. In our hands, the optimal label time for HaloTag is 15 minutes, while SNAPTag requires slightly longer labeling times, roughly 25 minutes.)

  CRITICAL STEP: Incubation at lower temperatures would decrease the rate of endocytosis which is preferable for the surface expression assay. However, decreased temperature also decreases the rate of reaction of ligands with their respective enzymes. Thus, during labeling we keep cells at room temperature rather than 37 °C or under refrigeration.

• 32Wash the axonal chamber with NIS thoroughly. Any unbound label that is not washed away can cause high background fluorescence when imaging. Optional: Fix the cells with 4% PFA for 15 minutes at room temperature.

  CAUTION: Take appropriate protective measures (gloves, eyeglasses, respiratory protection) when working with PFA.

• 33Place MFC onto the microscope and bring axons into focus using a 40–100x oil immersion objective.

Imaging (Timing: Steps 34–37: 10 min – 1 h, depending on experiment.)

• 34Identify axonal terminals of positively transfected cells using the appropriate wavelength laser (Fig 5b).

  CRITICAL STEP: Some fluorescent labels are more prone to photobleaching than others. To minimize the effects of photobleaching, keep the amount of time the laser is illuminating a particular field to a minimum. Also, consider searching for fields of view while keeping the laser intensity low.

• 35Select multiple (5–10) fields of view for image collection.
• 36Acquire z-stack images from each field of view. Ensure that the size and number of the optical sections is appropriate for the axonal sizes involved in the application. For our work with DRG neurons, we commonly acquire 20 × 0.2 μM sections to capture the entire 3-dimensional morphology of the distal axonal terminal.
• 37Depending on the microscope system and software used, the file type of acquired images may vary. We convert all images to TIFF before analysis in Fiji (ImageJ 2.0).

  CRITICAL STEP: The steps below describe analysis of a single axon generated with the above protocol. If analyzing many axons from many fields of view, consider using our macro titled ‘Surface Expression’ located at (https://github.com/ycnrr/NatureProtocols_2023) to facilitate higher throughput analysis guided by a graphical user interface (GUI).

Analysis (Timing: Steps 38–45: 1–10 min per image analyzed, depending on manual vs. software assisted analysis.)

• 38Load z-stack image into Fiji.
• 39Convert z-stack into maximum intensity projection (Image > Stacks > Z Project… ‘Max Intensity’)
• 40Adjust windows to visualize maximum intensity projections of distal axonal terminals (shortcut: Ctrl + Shift + C, then Space bar).
• 41We use the modal gray value to represent background fluorescence. To determine the modal gray value of the image, make sure the box ‘Modal gray value’ is checked in Analyze > Set Measurements…. Select Analyze > Measure to display a table containing the modal gray value for the image.
• 42Subtract the background fluorescence from the image by navigating to Process > Math > Subtract and inputting the modal gray value of the image.
• 43Use a line segment to approximate the distal-most 30 μM of the axonal terminal (Fig 5c). Then, using the segmented line tool, trace a line along the center of this axonal segment. Set the line width to match the width of the axons in your experiment (in our cultures this is typically ~1 μM).

  CRITICAL STEP: The length defined as “the distal axonal terminal” is arbitrary. Keeping this definition consistent across all conditions and analyzed images is more important than the exact length used. We have used lengths of between 30–60 μM to define the distal axonal ending.

• 44Calculate the average fluorescence intensity of the defined ROI. Ensure the box ‘Mean gray value’ in Analyze > Set Measurements… is checked. Select Analyze > Measure to display a table containing the mean fluorescent intensity (in A.U.) of the distal axonal ending. Record this measurement, then repeat the above steps for each axonal ending (Fig 5d).
• 45OPTIONAL: To generate pseudo-colored images (Fig 5c) of fluorescence intensity, change the lookup table in Image > Lookup Tables.

  CRITICAL STEP: The steps described here can be easily extended to investigating protein surface expression at the somatic membrane (Fig 5e). To do this, simply label the neuronal somas instead of the axons. In the analysis step, define the ROI along the circumference of the cell body.

Trafficking - Anterograde, Retrograde, and Bi-directional

• 46
  Labeling (Timing: Steps 46–49: 25–35 min, depending on the assay and ligand used.)

  1. Anterograde OPAL

     1. Prepare 1 mL of 100 nM cell-permeable HaloTag or SNAPTag ligand in NIS.
     2. Perfuse this solution containing 100 nM cell-permeable HaloTag or SNAPTag ligand into the somatic chamber of the MFC (Fig 6a, top panels)
  2. Retrograde OPAL

     1. Prepare 1 mL of 100 nM cell-permeable or cell-impermeable HaloTag or SNAPTag ligand in NIS.
     2. Perfuse this solution containing 100 nM HaloTag or SNAPTag ligand into the axonal chamber of the MFC (Fig 6a, middle panel)
        CRITICAL STEP: Using cell-permeable ligand allows visualization of the entire population of tagged protein being transported retrogradely, while using cell-impermeable ligand will allow specific detection of proteins which were labeled at the cell surface and underwent endocytosis prior to retrograde trafficking.
  3. Bi-directional OPAL

     1. Prepare 1 mL of 100 nM cell-permeable HaloTag or SNAPTag ligand conjugated to a red fluorophore in NIS.
     2. Prepare 1 mL of 100 nM cell-permeable HaloTag or SNAPTag ligand conjugated to a far-red fluorophore in NIS.
     3. Perfuse the solution containing 100 nM cell-permeable red HaloTag or SNAPTag ligand into the axonal chamber of the MFC, and incubate for 15–25 minutes.
     4. Perfuse the solution containing 100 nM cell-permeable far-red HaloTag or SNAPTag ligand into the somatic chamber of the MFC (or vice versa) and incubate for 15–25 minutes. (Fig 6a, bottom panel)
        CRITICAL STEP: We have used red (λ_ex ≈ 550 nm) and far-red (λ_ex ≈ 635 or 650 nm) fluorophores for these experiments. Any spectrally distinct pair of fluorophores (such that the light source and filters used to detect one fluorophore does not result in off-target emission from the other fluorophore) can be used in this application.
• 47Keep MFCs at 37° C while labeling using either an incubator or a bead bath. Keep label on for 15–25 minutes.

  CRITICAL STEP: Label times may vary between the fluorescent tags used. In our hands, the optimal label time for HaloTag is 15 minutes, while SNAPTag requires slightly longer labeling times, roughly 25 minutes.

• 48Wash the chamber/chambers containing fluorescent label with NIS. Neglecting to wash away excess can lead to leak of fluorescent label into the chamber where its absence is desired.
• 49Place MFC in stage-top incubator set to 37 ° C and bring axons into focus using a 40–100x oil immersion objective.
• 50
  Imaging (Timing: Step 50: 1–2 h, depending on experiment.) Imaging parameters vary depending on the transport direction of interest:

  1. Anterograde OPAL

     1. Scan the axonal compartment for axons that contain anterogradely moving vesicles. (Fig 6b, top panels).
        TROUBLESHOOTING
     2. Acquire time-lapse movies of identified axons. Longer exposure times and frame averaging will allow for brighter signals and decreased noise but could obscure rapid events. With our typical magnification, camera, and biological preparations, 100ms exposures (10 Hz) collect sufficient signal and the average vesicle would move less than a single pixel in each frame. We typically acquire movies for 60–120 seconds, which allows for an adequate number of vesicles to be detected and subsequently analyzed.
        CRITICAL STEP: Framerates that are too low could potentially result in vesicle tracks that are discontinuous, which can potentially interfere with automated track detection.
        CRITICAL STEP: Sometimes, multiple axons containing anterogradely-moving vesicles can be visualized in the same field of view. Axons will be analyzed separately in subsequent steps.
  2. Retrograde OPAL

     1. Scan the axonal or soma compartments for axons that contain retrogradely moving vesicles. (Fig 6b, bottom panels).
     2. Acquire time-lapse movies of the identified axons. Since retrograde motor proteins move more slowly than anterograde ones, taking longer movies (10 minutes or more) at lower frame rates (0.25–1 Hz) can allow observation of a larger number of vesicles without causing photobleaching.
  3. Bi-directional OPAL

     1. Scan the microgrooves for axons that contain vesicles moving in either direction.
     2. Sequentially acquire two separate time-lapse movies of the identified axons in each color channel.
  4. OPTIONAL: While the live-cell imaging methods described above provide a combination of high spatial and temporal resolution, if higher spatial resolution is desired, these labeling methods are also amenable to super-resolution imaging modalities such as super-resolution radial fluctuation (SRRF) microscopy.

     1. To prevent motion of organelles during the long acquisition times required for this type of imaging, after labeling, cells should be fixed with 4% PFA for 15 minutes at room temperature. Cells can then be imaged with an appropriate super-resolution microscope (Fig 7).

CRITICAL STEP: Some fluorescent labels are more prone to photobleaching than others. To minimize the effects of photobleaching, keep the amount of time a particular field is exposed to the laser to a minimum. Also, consider searching for fields of view while keeping the laser intensity low. Since the number of frames captured in trafficking experiments is relatively large, it is important to use fluorophores that are relatively resistant to photobleaching (e.g., JFX650, JFX554, JF646, JF635i).

CRITICAL STEP: The described assays (options a-c) allow for the visualization and recording of vesicle movement. The physical orientation of the MFCs containing neurons on the microscope will affect the direction in which vesicular movement is detected. We recommend orienting MFCs with the somatic chamber on the left. This way, anterograde motion in an axon extending from left to right will be from left to right, and retrograde motion will be from right to left.

Analysis (Timing: Steps 51–60: 5–30 min per image, depending on manual vs. software assisted analysis).

CRITICAL: Steps 51–60 describe analysis of a single axon generated with the above protocols. If analyzing many axons from many fields of view, consider using our macro titled ‘Trafficking’ located at (https://github.com/ycnrr/NatureProtocols_2023) to facilitate higher throughput analysis guided by GUI prompts.

• 51Ensure Fiji (ImageJ 2.0) has had the package ‘KymographClear’ installed (see equipment set-up) and loaded into the toolbar.
• 52Open a time lapse image in .tif format through the KymographClear max intensity image generation tool (keyboard shortcut: F2)
• 53
  Draw a segmented line over a region of the axon in the max intensity projection.

  1. CRITICAL STEP: The direction of this line defines the forward direction. Our convention is to draw this line in the anterograde direction.
• 54Press the kymograph button in the KymographClear toolbar. The line width used will depend on the width of axons in each user’s application. In our hands, for rat DRG neurons, a line width of 1 μm is sufficient to capture the width of the axon.
• 55The software will generate 5 total kymographs each colored according to particle tracking properties (see ref: KymoClear). The basic, unfiltered export will be titled: “kymograph1.tif”. Save the kymograph to a folder.
• 56Repeat the above analysis process for all kymographs intended to be analyzed in a group, and place kymograph .tif files in a folder. Compress the folder to a .zip file.
• 57Upload each kymograph to the KymoButler portal. The KymoButler software is available free at deepmirror.ai/kymobutler. The free version of the software allows for the analysis of one kymograph at a time.
• 58The landing page following kymograph upload allows the user to define parameters that modulate the AI’s threshold for particle track detection. Input acquisition settings such as frame duration and scale, and adjust particle detection settings such that vesicle tracks are detected completely, without segmentation, and few falsely positive tracks are detected (Fig 6c).

  TROUBLESHOOTING

• 59The results file from KymoButler includes information about vesicle velocity, signal intensity, and pause behavior for each kymograph. These measurements can be aggregated for population or condition analysis. A python script to concatenate measurements from multiple result spreadsheets can be accessed at https://github.com/ycnrr/NatureProtocols_2023.
• 60To calculate vesicular flux, count the number of vesicle tracks that cross the midline in each of the kymographs analyzed by KymoButler and normalize this measurement to the time of acquisition. By using the midpoint of the kymograph as a consistent metric to assess flux, the measurement of vesicles crossing the midline/axon/unit time remains consistent.

Co-trafficking

Evaluating vesicular co-transport of proteins of interest can be accomplished by tagging one protein with a HaloTag enzyme and one protein with a SNAPTag enzyme. Alternatively, such studies can be performed by coupling one protein with a HaloTag or SNAPTag and the other protein with a fluorescent protein (e.g., GFP). We describe steps for this latter approach below. If you are instead using the first, two enzymatic tag approach, simply include spectrally distinct fluorescent ligands for both of those tags in the initial incubation step.

Labeling (Timing: Steps 61–65: 25–35 min, depending on the assay and ligands used.)

• 61Prepare 1 mL of 100 nM cell-permeable HaloTag or SNAPTag ligand
• 62Perfuse this solution containing 100 nM cell-permeable HaloTag or SNAPTag ligand into the somatic chamber of the MFC (Fig 8a).
• 63Keep MFCs at 37 ° C while labeling using either an incubator or a bead bath. Keep label on for 15–25 minutes. (Note: label times may vary between the fluorescent tags used. In our hands, the optimal label time for HaloTag is 15 minutes, while SNAPTag requires slightly longer labeling times, roughly 25 minutes.)
• 64Wash the somatic chamber once (to prevent diffusion of any fluorophore from the somatic chamber to the axonal one).
• 65Place MFC in stage-top incubator set to 37 ° C and bring axons into focus using a 40–100x oil immersion objective.

Imaging (Timing: Steps 66–69: 1–2h, depending on experiment.)

• 66
  Search for axons that contain anterogradely moving vesicles with enzymatically tagged proteins of interest while also brightly fluorescent in the wavelength of the fluorescent-protein tagged protein of interest.

  1. TROUBLESHOOTING
• 67Maneuver the stage such that the identified axon is in the middle of the field of view, and new vesicles enter the field of view from the edges.
• 68Depending on their cellular distribution, FP-tagged proteins often produce diffuse, robust fluorescence throughout the cell. To decrease this background fluorescence, photobleach the fluorescent protein present within the field of view by exposing the field to the appropriate wavelength laser at high intensity. Effective photobleaching may take several minutes, depending on the fluorophore photostability and laser intensity. After this, new vesicles containing FP-tagged proteins which enter the field of view may be visible above the decreased background.
• 69Acquire time-lapse movies of the axon identified in the previous step. Dual color movies can be acquired by rapidly alternating laser and color filters; under such conditions the actual frame rate will depend on how quickly the filters can be switched by the system. Alternatively, incorporating a dichroic beamsplitter and a second camera can direct light from each fluorophore to a separate detector and allow for simultaneous acquisition.

Analysis (Timing: Steps 70–80: 5–30 min per image, depending on manual vs. software assisted analysis)

CRITICAL: The steps below describe analysis of a single axon generated with the above protocols. If analyzing many axons from many fields of view, consider using our macro titled ‘Co-trafficking’ located at https://github.com/ycnrr/NatureProtocols_2023 to facilitate higher throughput analysis guided by GUI prompts.

• 70Ensure Fiji (ImageJ 2.0) has had the package ‘KymographClear’ installed (see equipment set-up) and loaded into the toolbar.
• 71Open a time lapse image in .tif format. In co-trafficking experiments, the image should contain 2 channels, one in each fluorescent color.
• 72Collapse the channels of the image file into a composite image (Image > Color > Make Composite).
• 73Using the channels tool, assign a pseudocolor to each channel. We match the color of each channel to the corresponding fluorescent color of the ligand used in the experiment. (Image > Color >Channels tool… > Dropdown to select channel > More > Select desired color).
• 74Split the image into two separate images, one for each fluorescent channel, and save each image as a .tif file. (Image > Color > Split Channels)
• 75Generate a maximum intensity projection for each channel using the KymographClear toolbar. (Keyboard Shortcut: F2 > Select .tif file of the image from channel 1, then channel 2).
• 76Using the segmented line tool on one of the generated max intensity images, trace the portion of the axon you wish to analyze for vesicular co trafficking. Save the drawn segmented line as a .ROI file.
• 77Generate a kymograph for one channel using the KymographClear toolbar (“Kymograph button”) using the appropriate line width. For our analysis of sensory neurons imaged at 60x magnification, a line width of 1 μm is appropriate.
• 78Repeat step 8 for the remaining channels.

  CRITICAL STEP: several kymographs are generated using the KymographClear tool. We generally use the generated, non-colored kymograph, which follows the naming convention kymograph”N”.tif, where “N” is the number of kymographs generated.

• 79Generate a composite kymograph (Image > Color > Merge Channels…) and pseudocolor the resulting kymographs depending on the color of the originating channel.
• 80Evaluate vesicle tracks and assign vesicles as containing a single tagged protein, or co-trafficking multiple tagged proteins.

Protein Exocytosis/Endocytosis (Insertion/Removal)

Labeling (Timing: Steps 81–87: 30–40 min, depending on the assay and ligands used.)

• 81Prepare 1mL of 100 nM cell-impermeable HaloTag or SNAPTag ligand conjugated to a red fluorophore in NIS (i.e., JF549i, Label 1).
• 82Perfuse this solution containing 100 nM cell-impermeable HaloTag or SNAPTag ligand into the axonal chamber of the MFC (Fig 9a).
• 83Keep MFCs at RT (~22 °C) while labeling. Expose cells to label for 20–30 minutes. (Note: label times may vary between the fluorescent tags used. In our hands, the optimal label time for HaloTag at RT is 20 minutes, while SNAPTag requires longer labeling times, roughly 30 minutes.)

  CRITICAL STEP: Incubation at lower temperature would decrease the rate of endocytosis which is preferable for detecting the baseline surface expression in the endocytosis assay. However, decreased temperature also decreases the rate of reaction of ligands with their respective enzymes. Thus, during labeling we keep cells at room temperature rather than 37 °C or under refrigeration.

• 84Prepare 1 mL of 10 nM cell-impermeable HaloTag or SNAPTag ligand conjugated to a far-red fluorogenic fluorophore in NIS (i.e., JF635i, Label 2).

  CRITICAL STEP: Because Label 2 remains in the bath solution for the entire imaging period, it is critical to use the most fluorogenic label available at a concentration that does not produce significant background fluorescence. In our hands, including JF635i at 10 nM strikes a balance between low background fluorescence and rapid labeling of newly inserted proteins.

• 85Following the incubation with Label 1, wash the axonal chamber with NIS thoroughly. Any unbound Label 1 that is not washed away can be picked up by newly inserted proteins and contaminate measurements of protein exo- and endocytosis (Fig 9a).
• 86Perfuse the solution containing 10 nM Label 2 into the axonal chamber of the MFC (Fig 9a).
• 87Place MFC in stage-top incubator set to 37° C and secure the dish in place on the stage. Bring axons into focus using a 40–100x oil immersion objective.

  CRITICAL STEP: The remainder of this assay seeks to quantify protein turnover at the membrane which is a temperature-dependent process. For this reason, it is important to ensure that cells are imaged under normal physiological temperature (37 °C) in a stage-top incubator.

Imaging (Timing: Steps 88–91: 1–24h, depending on experiment.)

• 88Identify axonal terminals of positively transfected cells using the appropriate wavelength laser for Label 1 (7d, e).

  CRITICAL STEP: Some fluorescent labels are more prone to photobleaching than others. To minimize the effects of photobleaching, keep the amount of time a particular field is exposed to the laser to a minimum. Also, consider searching for fields of view while keeping the laser intensity low.

  TROUBLESHOOTING

• 89Select multiple fields of view for image collection and store their coordinates within the acquisition software.
• 90Acquire z-stack images from each field of view using lasers with wavelengths appropriate for detection of both Label 1 and Label 2, respectively. Ensure that the size and number of the optical sections is appropriate for the axonal sizes involved in the application. For our work with DRG neurons, we commonly acquire 20 × 0.2 μM sections to ensure the entire 3-dimensional morphology of the distal axonal terminal is captured.
• 91Take time lapse z-stack images in the desired interval. We have successfully performed these assays with time-lapse intervals of 20 minutes-1 hour, for up to 6 hours. Longer acquisition times are possible, though adjustments for axonal growth may be necessary.

  CRITICAL STEP: Acquiring images at higher frame rate provides better temporal resolution, but the increased exposure to photoenergy can cause more rapid photobleaching of less-stable fluorescent probes. It is important to compare all results to control measurements that allow for the quantification of fluorescence changes that are due to photobleaching or baseline protein regulation and not experimental manipulation.

  CRITICAL STEP: for long time-lapse imaging of live cells, we recommend using clear, serum free neuronal media minus phenol red instead of NIS.

Analysis (Timing: Steps 92–100: 2–20 min per image, depending on number of time points and manual vs. software assisted analysis)

CRITICAL: Depending on the microscope system and software used, the file type of acquired images may vary. We convert all images to TIFF before analysis in Fiji. Converting, creating directories for, and generating max intensity images of raw microscope files can be tedious. Consider using the converter software “Max Intensity Image Generation” located at (https://github.com/ycnrr/NatureProtocols_2023) to speed up analysis.

• 92Open a time-lapse, z-stacked image in .tif format in Fiji.
• 93Convert z-stacks into maximum intensity projections (Image>Stacks>Z Project…>Projection Type: “Max Intensity”).

  CRITICAL STEP: A z-stack will be generated for each time point in the time lapse.

• 94Adjust window settings to visualize max intensity projections at an appropriate brightness (keyboard shortcut: Ctrl + Shift + C, then Space bar).
• 95There will be two projections, one from each label used (Channel 1: from Label 1, Channel 2: from Label 2). For each time point in each channel, calculate, and record the modal gray value of the image. To determine the modal gray value of the image, make sure the box ‘Modal gray value’ is checked in Analyze > Set Measurements… Select Analyze > Measure to display a table containing the modal gray value for the image.
• 96In channel 1, use a line segment to approximate the distal-most 30 μm of the axonal terminal (Fig 5c). Then, using the segmented line tool, trace a line along the center of this axonal segment. Set the line width to match the width of the axons in your experiment (in our cultures this is typically ~1 μm).

  CRITICAL STEP: The length defined as “the distal axonal terminal” is arbitrary. Keeping this definition consistent across all conditions and analyzed images is more important than the exact length used. We have used lengths of between 30–60 μm to define the distal axonal ending.

• 97Calculate and record the mean gray value of this ROI (Analyze>Measure) in both Channel 1 and 2.
• 98Repeat this measurement for each time point in the series and in each channel. The ROI will need to be redefined at each time point to account for movement of the distal axonal terminal.
• 99Subtract the background measurements acquired in step 4 from the mean fluorescence measurements to generate fluorescent intensity measurements for each channel and each time point.
• 100Plot fluorescent values against time for each fluorescent channel.

  CRITICAL STEP: The fluorescent intensity of channel 1 will decrease over time as proteins labeled with Label 1 are internalized and transported away from the axon terminal (Fig 9a, c). Fluorescence measurements in channel 1 should be normalized to the intensity measurement taken at the first time point.

  CRITICAL STEP: The fluorescent intensity of channel 2 will increase over time as newly inserted proteins are labelled by free fluorogenic ligands in the bath solution (Fig 9a, b).

CRITICAL STEP: The methods described here can be extended to investigating protein insertion and removal from the somatic membrane as well. To do this, simply apply labels to neuronal somas instead of the axons as described here. In the analysis step, define the ROI along the circumference of the cell body.

Timing

Steps 1–10: Construct generation: −10 weeks

Steps 11–16, Preparation of MFCs: 2.5 h

Step 17, DRG neuron harvest and transfection ⁶⁷: 3h

Steps 18–21, DRG neuron culturing: 24h

Step 22, Incubation of DRGs: 5 – 7 d

Steps 23–26: Washing, 30 min – 1 h, depending on the assay

Step 27, Imaging: 10 min – 24 h, depending on the assay (see assay specific timings below)

Step 28, Analysis: 2 – 30 min per image, depending on the assay and use of software assistance

Steps 29–33, Surface Expression Assay Labeling, 25 – 35 min, depending on the ligand used

Steps 34–37, Surface Expression Assay Imaging, 10 min – 1 h, depending on experiment

Steps 38–45, Surface Expression Assay Analysis, 1 – 10 min per image analyzed, depending on manual vs. software assisted analysis

Steps 46–49, Trafficking Assay Labeling, 25 – 35 min, depending on the assay and ligand used

Step 50, Trafficking Assay Imaging, 1 – 2 h, depending on experiment

Step 51–60, Trafficking Assay Analysis, 5 – 30 min per image, depending on manual vs. software assisted analysis

Steps 61–65, Co-trafficking Assay Labeling, 25 – 35 min, depending on the assay and ligands used

Steps 66–69, Co-trafficking Assay Imaging, 1 – 2 h, depending on experiment

Steps 70–80, Co-trafficking Assay Analysis, 5 – 30 min per image, depending on manual vs. software assisted analysis

Steps 81–87, Exo/endocytosis Assay Labeling, 30 – 40 min, depending on the assay and ligands used

Steps 88–91, Exo/endocytosis Assay Imaging, 1 – 24 h, depending on experiment

Steps 92–100, Exo/endocytosis Assay Analysis, 2 – 20 min per image, depending on number of time points and manual vs. software assisted analysis.

Troubleshooting

Step	Problem	Possible Reason	Solution
17	Neurons do not express transfected proteins	Failed transfection	Occasionally transfections are unsuccessful, try again. Consider increasing the amount of plasmid DNA (taking care to use the minimal amount necessary).
			Insufficient DNA dose. Larger plasmids (such as those encoding NaVs) have relatively low transfection and translation efficiency, requiring more transfection DNA.
19–20	Construct fails validation steps	Tag insertion is causing an alteration in protein behavior	Design a new construct with the tag in a different location
22	Axons are not contained within microfluidic channels	MFC detached from glass	Ensure MFCs are well-attached to glass.
23	Axons wash away/lift off during washing	Wash step done too vigorously	Wash MFCs slower or more gently
24	High background fluorescence along axons in unlabeled chamber	Label is leaking across the microgroove barrier of the MFC	Ensure MFCs are well-attached to glass.
			Maintain lower fluid level in labeled chamber to bias fluid flow away from the unlabeled chamber
50A	Vesicles cannot be seen moving from the proximal to distal chamber	Axons are constricted within the microfluidic channels	Seed fewer axons. Perform the assay earlier when fewer axons have grown through.
		The protein of interest is not transported to axons	Run a positive control experiment by co-transfecting with a protein that is known to be trafficked to axons in the cell type being studied.
		Failed labelling	Ensure ligand has been stored properly and is appropriate for the tagging enzyme used.
	Not enough axons in the distal chamber	Wash step done too vigorously	Wash MFCs slower or more gently
58	KymoButler labels vesicle tracks erroneously	Thresholds for KymoButler inputs are not optimized	Manipulate thresholds for inputs until vesicle tracks are marked appropriately
66	Vesicles cannot be seen moving from the proximal to distal chamber	Axons are constricted within the microfluidic channels	Seed fewer axons. Perform the assay earlier when fewer axons have grown through.
		The protein of interest is not transported to axons	Run a positive control experiment by co-transfecting with a protein that is known to be trafficked to axons in the cell type being studied.
		Failed labelling	Ensure ligand has been stored properly and is appropriate for the tagging enzyme used.
	Not enough axons in the distal chamber	Wash step done too vigorously	Wash MFCs slower or more gently
88	Not enough axons in the distal chamber	Wash step done too vigorously	Wash MFCs slower or more gently

Anticipated results

The construction of neuronal membranes is a dynamic process involving the biogenesis, packaging, transport, insertion, and recycling of membrane proteins. We describe a system that enables the direct visualization of several aspects of axonal membrane protein trafficking in real time (Fig 1).

The system relies on two critical components: 1) the generation of proteins of interest fused to enzymes that bind fluorescently conjugated ligands (e.g., HaloTag, SNAPTag) and 2) the spatial segregation of neuronal axons from cell bodies using microfluidic chambers (Fig 3). After neurons are transfected with these enzymatically tagged protein constructs and plated in MFCs, a host of imaging assays become possible using the tagging strategies and microscopic parameters described in our protocol (Fig 4).

The surface expression assay enables the experimenter to evaluate the amount of protein expressed on the membrane at the axonal terminal using cell-impermeable ligands (Fig 5a). One can use this technique to quantitatively evaluate how protein surface expression changes in response to pharmacologic treatment (Fig 5b, d), or potentially between neuronal subtypes. These techniques are also extendable to the study of somatic or dendritic protein expression (Fig 5e).

The OPAL imaging technique allows for real-time visualization of vesicular trafficking and subsequent deep learning-based quantitation (Fig 6; Supplementary Video 1). A successful OPAL experiment generates kymographs that show clear vesicle tracks that can be segmented by the machine learning algorithm (Fig 6b–c). Occasionally, too many or too few vesicle tracks will be detected (i.e., noise may be incorrectly classified as a moving vesicle). In these cases, the detection settings of the algorithm can be modified to detect vesicle tracks in a more or less stringent manner (Fig 6d). This assay enables the investigation of vesicular behavior in neurons in different conditions (e.g., in the presence of inflammatory mediators vs. control; Fig 6e–f). If a perfusion system is installed on the imaging apparatus, before-and-after treatment recordings of vesicular transport in the same neuron is also possible^24,34.

Halo and SNAPTag systems can be used for super-resolution imaging applications. These methods can be used to interrogate co-localization of tagged proteins in subcellular compartments such as the lysosome (Fig 7).

The OPAL imaging system can be extended to evaluate the co-transport of two or more proteins in the same vesicles (Fig 8; Supplementary Video 2). By imaging multiple spectrally distinct fluorescent tags, vesicles expressing two or more proteins of interest can be followed in real-time (Fig 8a). Proteins that are trafficked together co-localize in time and space (Fig 8b), while those that are carried in different vesicles exhibit distinct movement patterns (Fig 8c). These techniques can be used to study how treatment with compounds affects vesicular co-trafficking (Fig 8d). Dual-color imaging with rapid filter switching results in a small discrepancy in the apparent location of the respective fluorescent signals due to continued motion of the vesicle during the delay between acquisition of each color. This could be eliminated by simultaneously imaging both colors with two cameras and a dichroic color filter.

The insertion and removal of proteins from the axonal membrane is yet another step capable of regulation by cellular processes. Taking advantage of fluorogenic labels and the high binding affinities of enzymatic labeling systems like Halo and SNAPTag allows for the quantification of protein insertion and internalization in our system (Fig 9). Like the other assays, protein insertion and removal experiments can be conducted on neurons exposed to different conditions (Fig 9b–c) or by imaging before and after treatment of the same neurons. The quantification of protein removal can be normalized to the baseline value of labeled proteins present on the membrane surface at time 0. Then, as proteins are endocytosed, the amount of fluorescence will decrease over time (Fig 9a, c). Unlabeled proteins present in the cytoplasm at time 0 on the other hand, will pick up free fluorogenic fluorescent probes in the bath solution as they are inserted into the membrane and the fluorescence intensity will rise over time (Fig 9a–b).

Supplementary Material

Supplementary Information

Supplementary Methods. Molecular biology protocols for generation of enzymatically-tagged full length proteins for use in imaging assays.

Supplementary Video 1

Supplementary Video 1. OPAL imaging enables direct visualization of vesicles carrying Halo-tagged membrane proteins with high signal to noise ratio.

The use of bright, photostable synthetic fluorophores combined with low background provided an excellent signal-to-noise ratio for clear visualization of vesicles. Anterograde Halo-Na_V1.7 vesicles are visualized using OPAL imaging. This high-resolution imaging technique allows visualization of individual Halo-Na_V1.7 vesicles with single-molecule resolution.

Movie taken from Akin et al. 2019 Science Advances²⁴ with permission of the authors.

Supplementary Video 2

Supplementary Video 2. OPAL imaging can be extended to investigate co-trafficking of 2 proteins in the same vesicles.

Two-color time-lapse OPAL imaging was performed using cell-permeable Halo-tag ligand-JF646 (magenta) and SNAP-tag ligand-JF549 (green). White arrowheads indicate vesicles doubly positive for Halo-Na_V1.8 and SNAP-Na_V1.7 as they move anterogradely along the axon outlined in white. Magenta and green arrowheads indicate vesicles that are singly positive for Halo-Na_V1.8 or SNAP-Na_V1.7, respectively. As described previously, the ~1 μm separation of the SNAP and Halo signals is because of the continued motion of the vesicle during the time between acquisition of images in separate channels. When the vesicle is moving, the SNAP signal is always “behind” the Halo signal because the SNAP (561 nm) channel is acquired first. When the vesicle stops, however, the two signals overlap completely.

Movie taken from Higerd-Rusli et al. 2022 J Neurosci²⁵ with permission of the authors.

Extended Data Figure 1

Extended Data Figure 1: HaloTag and SNAPTag ligands do not label DRG neurons non-specifically

(a) Confocal Z-stacks of a DRG neuron transfected with eGFP only, imaged in different spectral channels. The neuron was incubated with JF549-cpSNAPTag ligand and JF646-HaloTag ligand, washed, then imaged. Panels from left to right: 1) DIC image of neuron showing normal DRG morphology. 2) 488 nm channel (pseudo colored yellow) showing robust eGFP fluorescence. 3) 561 nm channel (pseudo colored green) showing lack of JF549-cpSNAPTag fluorescence. 4) 637 nm channel (pseudo colored magenta) showing lack of JF646-HaloTag fluorescence.

DRG neurons were transfected with either Halo-Na_V1.8 only (b), or SNAP-Na_V1.7 only (c), incubated with JF549-cpSNAPTag ligand and JF646-HaloTag ligand, washed, then imaged.

(b) Kymographs of vesicles detected in axons of DRG neurons transfected with Halo-Na_V1.8 show that the SNAPTag ligand is not trafficked in the absence of SNAPTag.

(c) Kymographs of vesicles detected in axons of DRG neurons transfected with SNAP-Na_V1.7 show that the HaloTag ligand is not trafficked in the absence of HaloTag.

Panels a, b, and c are adapted from Higerd-Rusli et al. J Neurosci, 2022²⁶ with permission of the authors.

Data Availability

The authors declare that the main data discussed in this protocol are available in the supporting primary research papers (https://doi.org/10.1126/sciadv.aax4755, https://doi.org/10.1093/brain/awab113, https://doi.org/10.3389/fnmol.2023.1161028, https://doi.org/10.1073/pnas.2215417120, https://doi.org/10.3389/fnmol.2023.1130123, https://doi.org/10.1016/j.jbc.2022.102816 and https://doi.org/10.1523/JNEUROSCI.0058-22.2022). The raw datasets are too large to be publicly shared but are available for research purposes from the corresponding authors upon reasonable request.

All data in this protocol are available within the paper and its Supplementary Information.