The node of Ranvier influences the in vivo axonal transport of mitochondria and signaling endosomes

Summary

Efficient long-range axonal transport is essential for maintaining neuronal function, and perturbations in this process underlie severe neurological diseases. Nodes of Ranvier (NoR) are short, specialized unmyelinated axonal domains with a unique molecular and structural composition. Currently, it remains unresolved how the distinct molecular structures of the NoR impact axonal transport dynamics. Using intravital time-lapse microscopy of sciatic nerves in live, anesthetized mice, we reveal (1) similar morphologies of the NoR in fast and slow motor axons, (2) signaling endosomes and mitochondria accumulate specifically at the distal node, and (3) unique axonal transport profiles of signaling endosomes and mitochondria transiting through the NoR. Collectively, these findings provide important insights into the fundamental physiology of peripheral nerve axons, motor neuron subtypes, and diverse organelle dynamics at the NoR. Furthermore, this work has relevance for several pathologies affecting peripheral nerves and the NoR.

Graphical abstract

Highlights

• •Fast and slow motor neurons exhibit similar NoR morphology in vivo
• •Signaling endosomes and mitochondria accumulate specifically at the distal NoR
• •Transport dynamics of both organelles are altered at the NoR
• •Fastest transport speeds were observed at the internode proximal to the NoR

Molecular neuroscience, Cellular neuroscience

Introduction

Axonal transport is a fundamental biological process that maintains neuronal homeostasis with constant bidirectional shuttling of essential cargoes and structural components between the neuronal cell body and axon termini. Along the microtubule network, cytoskeletal elements (e.g., neurofilaments), as well as membranous (e.g., mitochondria, endosomes) and membrane-less (e.g., RNA granules) organelles undergo plus-end directed anterograde transport driven by members of the kinesin motor protein family, and minus-end directed retrograde transport via cytoplasmic dynein.^1,2 Effective long-range axonal transport is essential for maintaining neuronal function, and trafficking perturbations underlie several neurodevelopmental and neurological conditions.³

α-motor neurons (MNs) can be subclassified into fast MNs (FMNs) and slow MNs (SMNs) that selectively innervate specific skeletal muscle fiber subgroups. Both types of MN and the muscle fibers they innervate have distinct metabolic, functional, and transcriptional properties.^4,5 Compared to SMNs that innervate slow oxidative type I muscle fibers, FMNs have larger motor unit sizes, faster nerve conduction, and different firing patterns, and they innervate type IIa fast oxidative-glycolytic, as well as type IIx and IIb fast glycolytic muscle fibers.^4,6

We have previously determined that signaling endosome transport speeds are similar between motor axons innervating prototypical fast and slow muscles.⁷ Likewise, axonal transport speeds are comparable between hindlimb and forelimb peripheral nerves.⁸ In contrast, signaling endosome transport in MNs is faster than in sensory neurons innervating the same muscle.⁹ Several factors directly influence axonal transport dynamics,^1,10,11,12 and can be perturbed in disease.^3,13 Collectively, this suggests that the axonal transport machinery is differentially modulated in physiological and pathological contexts.

As most in vivo axonal transport studies have prioritized larger and/or consistently sized axonal segments, our understanding of transport dynamics at highly specialized axonal regions, such as the node of Ranvier (NoR), remains incomplete. NoRs are short uncovered axonal domains that facilitate action potential propagation, and each of the four regions (i.e., node, paranode, juxtaparanode, and internode) is comprised of distinct structural and functional proteins.^14,15 Electron microscopy reveals reductions in axonal diameters at the NoR, which maximize electrical conduction velocities¹⁶; these constrictions were attributed to reduced neurofilament, but not microtubule, content.^17,18,19 Most internodal neurofilaments are not continuous through the NoR and cease near the juxtaparanode, whereas microtubules extend through the NoR to connect adjacent internodes,²⁰ providing structural continuity for the continuous processive movement of cargoes.

Despite high organelle content,^18,21 unique axonal Ca²⁺ dynamics²² and high metabolic needs,²³ which are all features with the potential to regulate axonal transport, few studies have investigated transport dynamics at the NoR. Neurofilaments undergoing slow axonal transport²⁴ increase their speed through the NoR.^25,26 In contrast, muscle-administered horseradish peroxidase (HRP),²⁷ radiolabeled glycoproteins,²⁸ and lysosome-linked enzymatic activity²⁹ are enriched at peripheral nerve NoRs, but little is known about their transport.

The aim of this study was to assess in vivo axonal transport of two different organelles at the NoR in FMNs and SMNs to further our understanding of fast axonal transport dynamics through this specialized structure. Knowledge of how the node affects cargo delivery may reveal key mechanisms relevant to pathologies affecting peripheral nerves.

Results

Identifying the node of Ranvier in vivo

To confirm the identity of the NoR, we performed immunohistochemistry on fixed, teased sciatic nerve fibers from ChAT.eGFP mice probing for established markers of the node and paranode.³⁰ As expected, we observed a ring-shaped cluster of Na_V1.6 channels approximately in the middle of the ChAT.eGFP nodal constriction that is flanked by CASPR immunolabelling of the paranode (Video S1). Therefore, we concluded that the identical constrictions observed in ChAT.eGFP sciatic nerves contain the nodal and paranodal regions and can be distinguished from the larger internodal regions using intravital imaging of the endogenous eGFP fluorescence in ChAT.eGFP mice.

Video S1. Na_V1.6 clusters are located in the middle of the nodal constriction and are flanked by CASPR-positive paranodal regions in a peripheral nerve motor axon – linked to Figure 1

Representative videos of (A) a rotating max projection z-stack, and (B) slice-by-slice through the z-axis, highlighting the (i) cholinergic (i.e., ChAT-positive) motor axon, the immunolabeled (ii) nodal (Na_V1.6) and (iii) paranodal (CASPR) regions, along with the (iv) overlay, from teased axons of a ChAT.eGFP mouse sciatic nerve. Scale bar = 5 μm.

To locate the NoR in FMN and SMN axons in vivo, we performed intramuscular injections of H_CT separately into TA or soleus muscles in ChAT.eGFP mice. Owing to the divergent muscle composition, the TA was selected to assess transport in FMN axons, whereas the soleus was chosen to assess transport in SMN axons. Indeed, TA and soleus are prototypical fast and slow muscles, respectively, with the TA being comprised of ∼45–60% type IIb, ∼30–45% type IIx, ∼10–20% type IIa and ∼0–3% type I muscle fibers, and the soleus consisting of ∼40% type I, ∼40% type IIa and ∼20% type IIx muscle fibers.^31,32 H_CT is internalized into axon termini of motor neurons through binding to polysialogangliosides, nidogens,³³ and the tyrosine phosphatases, LAR and PTPRδ.³⁴ Following sorting into Rab5-positive endosomal compartments, H_CT undergoes fast retrograde axonal transport toward the cell bodies of spinal cord motor neurons in Rab7-positive organelles^35,36 using a cytoplasmic dynein- and microtubule-dependent process.³⁷ Therefore, H_CT injections simultaneously enabled in vivo assessments of axonal signaling endosomes and the labeling of motor neurons innervating distinct muscles.

Intravital imaging of ChAT.eGFP sciatic nerve cholinergic (i.e., motor) axons enabled the identification of nodal constrictions bordered on both sides by wider internodal regions (Figure 1A). Assessing individual TA-innervated FMN or soleus-innervated SMN axons revealed similar diameters of the internodal regions (Figure 1B) and nodal constrictions (Figure 1C), as well as lengths of the nodal constrictions (Figure 1D). This analysis also revealed that the differences in axonal diameters at the internode and nodal constriction were comparable between motor axons innervating TA (Figure 1E) and soleus (Figure 1F) muscles, which was confirmed when the ratios of internodal to nodal diameters were calculated (Figure 1G). From these analyses, we concluded that morphological features of the NoR are equivalent in wild-type motor axons innervating prototypical fast and slow muscles. Furthermore, nodal morphology in TA-innervating motor axons is similar between ChAT.eGFP and Mito.CFP mice (Figure S1), which are the main two transgenic fluorescent strains used in this study.

Figure 1.

Nodal morphology is similar between fast and slow motor axons of the sciatic nerve

(A) Representative single frame image of H_CT-555 containing signaling endosomes (magenta) in a single in vivo motor axon (green) from a ChAT.eGFP mouse sciatic nerve motor axon. Scale bar = 5 μm. Yellow lines represent the axonal region assessed to quantify the internodal diameters in (B), the orange line represents the axonal region assessed to quantify the nodal constriction in (C), and the cyan line represents the axonal region assessed to quantify the length of the nodal constrictions in (D). Motor axons innervating tibialis anterior (TA) or soleus muscles display similar (B) internodal diameters (p = 0.71, unpaired two-tailed t-test), (C) nodal constriction diameters (p = 0.40, unpaired two-tailed t-test), and (D) nodal constriction lengths (p = 0.90, unpaired two-tailed t-test). Furthermore, the differences in axonal diameters at the internode and nodal constriction were comparable between motor axons innervating (E) TA and (F) soleus.

(G) The ratios of internodal to nodal diameters were also comparable.

(B, C, D, and G) were assessed by an unpaired, two-tailed t test.

(E and F), were assessed by a paired two-tailed t-test (∗∗∗p < 0.001). For all graphs, n = 6, data are represented as mean ± SEM, and the color-coding remains consistent between animals.

Linked with Video S4.

Signaling endosomes and mitochondria cluster at the distal Nodes of Ranvier

We next profiled the spatial location of signaling endosomes and mitochondria in relation to the nodal constriction in fast and slow motor axons. To label the different motor axon subtypes, we injected H_CT into either TA or soleus muscles of Mito.CFP mice, and performed simultaneous time-lapse microscopy of H_CT-containing signaling endosomes and mitochondria,³⁸ focusing specifically on the NoR (Figure 2Ai; Video S2). The individual fluorescence intensity profiles of both organelles were first created using z-projections of individual frames from time-lapse videos (Figure 2Aii-iii), and then the fluorescence intensities of signaling endosomes and mitochondria were assessed across the whole region (80 μm). We detected an accumulation of both organelles specifically at the distal side of the nodal region (Figure 2B). To quantify this, we directly compared the average axonal fluorescence intensities of signaling endosomes and mitochondria from individual videos of TA- and soleus-innervating axons from proximal and distal internodal regions. These analyses revealed that there were a greater number of immobile H_CT-containing signaling endosomes (Figure 2C) and mitochondria (Figure 2D) at the distal side of the NoR, and this distribution was similar in TA- and soleus-innervating motor axons (Figures 2C and 2D). To rule out that this clustering phenotype was caused by the administration of H_CT and generation of H_CT-containing signaling endosomes in situ, we evaluated the mitochondrial fluorescence profiles in naive Mito.CFP axons (i.e., without intramuscular H_CT injections). This analysis revealed the same mitochondrial distribution as shown in Figure 2D, demonstrating that the accumulation of mitochondria seen distally at the NoR is independent of H_CT (Figure S2).

Figure 2.

Signaling endosomes and mitochondria selectively accumulate at sites distal to the node of Ranvier

(A) (i) Representative image demonstrating the spatial distribution of H_CT-containing signaling endosomes (magenta) and mitochondria (cyan) from a single tibialis anterior-innervating motor axon in a Mito.CFP mouse. Representative z stack projection (sum of the slices) from a time-lapse video (see Video S2) indicates that the fluorescence profiles of (ii) H_CT-containing signaling endosomes and (iii) Mito.CFP-labeled mitochondria are enriched on the distal side of the NoR. Scale bar = 10 μm.

(B) In motor axons innervating the (i) tibialis anterior and (ii) soleus muscles, we observed increased average relative fluorescence intensities of both H_CT-containing signaling endosomes (magenta) and Mito.CFP-labeled mitochondria (cyan), exclusively on the distal side of the NoR. Enhanced fluorescence profiles of (C) H_CT-containing signaling endosomes between 1 and 11 μm from the center of the NoR (i) TA - Axonal Location: p < 0.001; Mean Relative Fluorescence: p < 0.001; Interaction: p < 0.001. (ii) Soleus: Axonal Location: p < 0.001; Mean Relative Fluorescence: p < 0.001; Interaction: p < 0.001), and (D) mitochondria 2–8.5 μm from the center of the NoR (i) TA - Axonal Location: p < 0.001; Mean Relative Fluorescence: p < 0.001; Interaction: p = 0.0002. (ii) Soleus - Axonal Location: p < 0.001; Mean Relative Fluorescence: p < 0.001; Interaction: p < 0.001). For all graphs, n = 5 animals, 16 axons, data are represented as mean (solid line) ± SEM (dashed lines).

(C and D) Data were compared by two-way ANOVA and Šídák’s multiple comparisons tests (∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001).

See also Figures S1 and S2.

Video S2. In vivo axonal transport of mitochondria and signaling endosomes is observed through the nodal constrictions of a single sciatic nerve axon – linked to Figures 2-3

Representative time-lapse microscopy video of dual-organelle in vivo axonal transport through the nodal constriction of a Mito.CFP mouse. Frame interval = 1.44 s, number of frames: 201, playback rate: 15 fps, scale bar = 10 μm.

Previous studies suggest that axon-Schwann cell interactions regulate several critical functions. For example, CXCL12α/SDF-1 released from perisynaptic Schwann cells promotes motor axon regeneration,³⁹ and ATP released from neurons activates Schwann cells.⁴⁰ We, therefore, aimed to evaluate whether H_CT-positive signaling endosomes are released by exocytosis from this site of communication between motor axons and Schwann cells. To do so, we injected H_CT into TA muscles in the PLP-GFP mouse, in which GFP is exclusively expressed in Schwann cells.⁴¹ 24 h after H_CT injections, teased individual axons were isolated from the PLP-GFP sciatic nerves, and imaged. H_CT clusters were again found in the distal portion of the NoR; however, no signaling endosomes were detected in the cytosol of Schwann cells (Video S3), ruling out that H_CT is transferred from motor axons to Schwann cells. Altogether, using three separate transgenic reporter mouse models, we have shown that H_CT-positive signaling endosomes, along with mitochondria, cluster specifically at the distal side of the NoR.

Video S3. Signaling endosomes that traverse the nodal constrictions are not present in the Schwann cell cytosol

Representative video of a (A) rotating max projection z-stack, and (B) slice-by-slice through the z-axis, highlighting the (i) H_CT-containing signaling endosomes (magenta), within the (ii) Schwann cell cytosol (eGFP), and (iii) overlay, from the same teased axon of a PLP-eGFP mice sciatic nerve. Scale bar = 10 μm.

Axonal transport of signaling endosomes and mitochondria through the node of Ranvier

We next sought to determine the axonal transport dynamics of both signaling endosomes and mitochondria through the NoR. First, we established that in ChAT.eGFP motor axons, H_CT-containing signaling endosomes narrow their trajectories as they traverse the nodal constriction (Figure 3A; Video S4). We found that bidirectionally moving mitochondria also funnel their paths through the NoR when transitioning from the larger internodal regions to the nodal constriction, and then widen their trajectories when entering the subsequent internodal region (Figure 3B; Videos S2 and S4). We also observed non-linear courses that individual signaling endosomes would take as they approached the nodal constriction (e.g., moving in an “S” shape (Figure 4A; Video S5) or traveling in a circular motion (Figures 4B and 4C; Video S6)). These unusual trajectories might be caused by signaling endosomes traversing areas where the normal microtubule distribution present in internodal regions is altered due to proximity to nodal constrictions, e.g., radial axonal expansions,⁴² or circling areas with a high density of immobile organelles (e.g., mitochondria, Figure 4C).

Figure 3.

Axonal cargoes are bi-directionally funnelled through the node of Ranvier

(A) In vivo axonal transport analysis at the node of Ranvier (NoR) reveals that H_CT-containing signaling endosomes (circles) narrow their trajectories (lines) from the wider intranodal regions to the cytoskeletal confines at the nodal constriction. Representative image of a sciatic nerve motor axon from a ChAT.eGFP mouse.

(B) Time-lapse images taken every 3.3 s of in vivo axonal transport of Mito.CFP-labeled mitochondria (cyan) and H_CT-containing signaling endosomes (magenta) specifically focused at the NoR. Green circles identify retrogradely moving signaling endosomes, orange triangles identify an anterogradely moving mitochondrion, and yellow triangles/circles identify paused/immobile organelles. Anterograde movement is from left to right, and retrograde movement is in the opposite direction. Scale bars = 5 μm.

Figure 4.

Non-linear trajectories of signaling endosomes during their transit through a node of Ranvier

(A) Time lapse image series of an individual H_CT-containing signaling endosome (magenta, with orange circle) traveling on an “S-shaped” trajectory (orange dotted lines) during in vivo axonal transport through the NoR from a ChAT.eGFP sciatic nerve motor axon. Frame interval = 4.5 s; scale bar = 10 μm.

Linked with Video S5.

(B) Time lapse image series demonstrating an individual H_CT-containing signaling endosome (magenta, with yellow arrowheads) traversing in a “circular” pathway (contained inside the white dashed lines) around an individual mitochondrion (cyan) on the distal side of the NoR in a Mito.CFP sciatic nerve axon. Frame interval = 1.6 s; scale bar = 10 μm.

Linked with Video S6.

(C) z stack maximum projection of (B), highlighting the circular pathway exhibited by an individual signaling endosome (magenta) as it traverses around a mitochondrion. C″ are enlargements of the boxed areas in C’. Scale bar: C = 10 μm, Scale bar: C’’ = 2 μm.

Video S4. Representative video of in vivo axonal transport of H_CT-containing signaling endosomes traversing the node of Ranvier in a single sciatic motor axon – Linked to Figures 3-5

Top panel = ChAT.eGFP signal; Middle panel = retrogradely transporting H_CT-containing signaling endosomes; Bottom panel = overlay. Frame interval: 0.53 s, number of frames: 400, frames per second: 20, scale bar = 10 μm.

Video S5. Retrograde axonal transport of H_CT (magenta) can display an unusual trajectory through the node of Ranvier in peripheral nerve motor axons (green) – Linked to Figure 4A

Frame interval: 0.45 s, playback rate = 25 fps, acquisition time = 41 s, scale bar: 10 μm.

Video S6. Retrograde axonal transport of H_CT-containing signaling endosomes (magenta) traveling in a circular path (green arrow) around mitochondria (cyan) on the distal side of the node of Ranvier – Linked to Figure 4B

Frame interval: 1.64 s, playback rate = 8 fps, acquisition time = 3 min 11 s, scale bar = 10 μm.

Following this observation, we then used semi-automated tracking of H_CT-positive signaling endosomes in TA- and soleus-innervating motor axons of the Mito.CFP mouse using the TrackMate plugin⁴³ to assess transport dynamics. Strikingly, the mean velocities of retrogradely moving H_CT-containing signaling endosomes are reduced when approaching the nodal constriction, with a concomitant increase in the relative frequency of pausing. This was followed by an increase in retrograde velocities on the proximal side of the NoR in both TA- (Figure 5A) and soleus-innervating motor axons (Figure S3). Quantitative analyses in TA-innervating axons indicated that the mean velocities of signaling endosomes in the nodal constriction were ∼48% and ∼40% slower compared to the proximal and distal internodal regions, respectively (Figure 5B). Furthermore, the retrograde transport speeds through the proximal internode were ∼14% faster compared to distal internode regions (Figure 5B). In addition, pausing in TA-innervating motor axons was reduced by ∼86% and ∼67% in the proximal and distal internodal regions, respectively, compared to the nodal constriction (Figure 5C). Similarly in soleus-innervating motor axons, the mean velocities of H_CT-containing signaling endosomes in the nodal constriction were ∼46% and ∼35% slower compared to the proximal and distal internodal regions, respectively (Figures S3B and S3C). As the transport dynamics of signaling endosomes at the NoR largely overlap in FMN and SMN motor axons (Figures 5D, 5E, S3 D, and S3E), we conclude that motor neuron subtypes do not determine overt changes in axonal transport dynamics at the NoR, which is consistent with our previous observations across the larger internodal axonal region⁷

Figure 5.

Signaling endosomes decelerate and pause more at nodes of Ranvier regardless of motor neuron subtype

(A) Retrograde axonal transport dynamics (mean moving velocity [purple] and relative frequency of mean pausing [gray bars]) of H_CT-containing signaling endosomes plotted across the nodal constriction and beyond (80 μm distance) in motor axons innervating the tibialis anterior muscle. The x axis represents the distance from the center of the node of Ranvier (μm) and is split into three segments: (1) Proximal = 38 μm of the proximal internode (dark green); (2) Center = 4 μm representing the mean nodal length, (as determined in Figure 1D; bright green); and (3) Distal = 38 μm of the distal internode (light green). Comparisons across the proximal internode, nodal constriction and distal internode of the (B) mean moving velocity (p < 0.001), and (C) relative frequency of mean time paused (p < 0.001).

(B and C) Data were compared by one-way ANOVA, followed by Holm-Šídák’s multiple comparisons test. Histograms comparing the (D) mean moving velocity and (E) relative frequency of mean time paused from motor axons innervating the tibialis anterior (salmon) and soleus (teal). For all graphs, data are represented as mean (solid line) ± SEM (dashed lines/error bars), n = 4 (soleus) and n = 5 (tibialis anterior) from Mito.CFP mice, ∗∗p < 0.01, ∗∗∗p < 0.001.

See also Figures S3 and S5.

To determine whether these phenotypes are specific for the organelle and/or direction of travel, we assessed the axonal transport of mitochondria in TA-innervating motor axons, using the same videos from which signaling endosomes were analyzed (e.g., Figure 5). We separately assessed the anterograde and retrograde axonal transport of mitochondria through the nodal constrictions using the manual tracking feature of TrackMate⁴³ (Video S7). Similar to H_CT-containing signaling endosomes, the mean velocity of mitochondria is also reduced at the nodal constriction, which was then followed by a return to faster velocities for both anterogradely and retrogradely moving mitochondria (Figures 6A and S4). Anterograde mitochondrial velocities in the nodal constriction were ∼28% and ∼21% slower compared to the proximal and distal internodes, respectively (Figure 6B). Similarly, the mean velocities of retrogradely moving mitochondria were ∼24% and ∼23% slower in the nodal constriction compared to the proximal and distal internodal regions, respectively (Figure 6B). We also observed faster mean velocity for anterogradely moving mitochondria in the proximal internode compared to the distal internode (Figure 6B), replicating the increased speeds of signaling endosomes proximal to the node (Figure 5B). However, an increase in pausing frequency is observed only in mitochondria moving anterogradely, but not retrogradely (Figures S4 and 6C). Despite such regional differences in mitochondrial transport, no significant correlation emerged between transport speeds or pausing events and the axonal diameter across each of the individual axonal sub-domains (Figure S5). A higher percentage of mitochondria move anterogradely than retrogradely (Figure 6D), with mitochondrial flux being more than three times greater in the anterograde than retrograde direction (Figure 6E). Consistent with the accumulation distal to the node, anterograde mitochondrial flux is lower in the distal internode than the preceding axonal subdomains, whereas no regional differences in retrograde flux were found (Figure 6F). This suggests that anterogradely moving mitochondria are the main source for nodal accumulation. Finally, we determined that independently from their location, stationary mitochondria are longer than motile mitochondria, and that, only in the proximal internode, retrogradely moving mitochondria are shorter than those moving anterogradely (Figures 6G and S6).

Figure 6.

Mitochondrial transport dynamics are bi-directionally altered at the node of Ranvier

(A) Combined anterograde and retrograde mitochondrial dynamics (mean moving velocity [blue diamonds] and relative frequency of mean pausing [pink bars]) across the nodal constriction and beyond (80 μm distance) in sciatic nerve axons innervating the tibialis anterior muscle. The x axis represents the distance from the center of the node of Ranvier (μm) and is split into three segments: (1) Proximal = 38 μm of the proximal internode (navy); (2) Center = 4 μm representing the mean nodal length (as determined in Figure 1D; cyan); and (3) Distal = 38 μm of the distal internode (light blue). Comparisons between the nodal constriction, and proximal and distal internodes of in vivo axonal transport of anterogradely and retrogradely moving mitochondria of (B) mean moving velocity (∗∗∗Axonal Location: p < 0.001; Directionality: p = 0.83; Interaction: p = 0.19) and (C) relative frequency of mean time paused (∗Axonal Location: p = 0.0156; Directionality: p = 0.81; Interaction: p = 0.39).

(D) A higher percentage of mitochondria move in the anterograde than retrograde direction, (E) resulting in greater flux of anterogradely moving mitochondria across all axonal sub-domains (p < 0.001). In addition, (F) there is a selective decline in anterograde flux in the distal internode, whereas there is no difference in retrograde flux across axonal sub-domains (Axonal Location: p = 0.0021; Directionality: p < 0.001; Interaction: p = 0.0124).

(G) Stationary mitochondria are longer compared to both the anterogradely and retrogradely moving mitochondria, independent of the axonal sub-domain (Axonal Location: p = 0.79; Movement: p < 0.001; Interaction: p = 0.36). For all graphs, data are represented as mean (solid line) ± SEM (dashed lines/error bars). Data were compared by two-way ANOVA and Šídák’s multiple comparisons tests (B, C, F, and G) or an unpaired, two-tailed t-test (E). n = 5 Mito.CFP mice. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.

See also Figures S4–S6.

Video S7. Representative videos of tracked mitochondria separated by direction of travel – Linked to Figure 6 and Figure S4

(A) Representative unprocessed time-lapse video. Using the TrackMate plugin in FIJI/ImageJ, mitochondria (green dots and lines) were manually tracked in the (B) anterograde (magenta) and (C) retrograde (gold) directions. Frame interval: 1.77 s, playback rate = 20 fps, acquisition time = 10 min 21 s, scale bar: 10 μm.

In summary, we have characterized an in vivo axonal transport profile of signaling endosomes and mitochondria across the NoR in multiple motor neuron sub-types within the sciatic nerve.

Discussion

Axons are reliant upon efficient transport to maintain neuronal homeostasis. Using our intravital imaging approach,^38,44 we aimed to resolve the in vivo axonal transport dynamics of multiple organelles at the NoR in intact and synaptically connected motor axons in the sciatic nerves of healthy mice. Firstly, we observed equivalent morphological features of the NoR in motor axons innervating prototypical fast and slow muscles. Next, we identified the co-clustering of signaling endosomes and mitochondria specifically at the distal side of the NoR. In vivo axonal transport analyses revealed a slowing of mitochondria and signaling endosomes as they approach the NoR, with a concomitant rise in pausing events. This was followed by an increase in velocity in the adjacent intranodal region, irrespective of directionality and motor neuron subtype. Collectively, these findings further our understanding of the morphology and physiology of NoR in peripheral nerve axons.

Node of Ranvier morphology in peripheral nerve

We have previously reported that peripheral motor axons transporting H_CT upon uptake at neuromuscular junctions are larger in diameter than H_CT-positive peripheral sensory axons innervating the same muscle,⁹ consistent with the observation that ventral roots have a larger average caliber than dorsal roots in healthy mice.⁴⁵ Such differences have also been shown in central nervous system (CNS) axons⁴⁶ and in ventral funiculi axons in the thoracic spinal cord.⁴⁷ Conversely, the mean intranodal diameters of motor axons innervating the TA, lateral gastrocnemius, and soleus muscles do not differ.⁷

Here, we reveal comparable morphological features of the NoR between H_CT-labeled FMN and SMN axons, as well as between ChAT.eGFP and Mito.CFP FMN axons, which are in turn, also comparable to the previously reported sciatic nerve profiles⁴⁷ and morphometric results obtained from tibial nerve explants.²⁵ In contrast, morphological features of the NoR in CNS axons, such as those in the rat optic nerve and cerebral cortex, are heterogeneous, with differences observed between axons rather than within axons, even when comparing axons with similar intranodal diameters.³⁰ However, altering neuronal activity can modify NoR properties,⁴⁸ suggesting the axonal morphology is dynamically regulated.

Organelle accumulations at the NoR

Using intravital imaging, we show dual-organelle accumulation at the NoR in vivo. As the peak fluorescence signal from both signaling endosomes and mitochondria was approximately 3–4.5 μm distal to the center of the NoR, it is likely that these clusters were located in the distal juxtaparanode region. Indeed, the peak signal from the distal region was approximately 50% greater than from the proximal internode for both organelles. Our data are in line with previous reports of expansions in the distal juxtaparanode of myelinated axons, which have been attributed to accumulations of diverse organelles, including retrogradely transported glycoproteins,²⁸ retrogradely transported horseradish peroxidase (HRP),²⁷ multivesicular bodies,¹⁸ lysosomes²⁹ and mitochondria.²¹ However, there may be regional and/or neuronal subtype differences in organelle clustering at the NoR, since mitochondria do not accumulate at the NoR in small-diameter myelinated CNS axons,⁴⁹ nor in the optic nerve.⁵⁰

Mechanistically, it is unclear whether organelle accumulation at the NoR is an active process, or is simply a consequence of a localized cytoskeletal bottleneck. Although neurofilaments are reduced up to 10-fold at the NoR,^25,26,51 the microtubule network must be preserved to enable axonal transport over the long distances covered by peripheral nerve axons.⁵² Hence, the accumulation of organelles at the NoR might be structurally and functionally linked, with several potential mechanisms that could cause such focal accumulations, including differences in: 1) axonal protein translation⁵³; 2) metabolic demands²³; 3) axo-glial communication⁵⁴; 4) activity dependent alterations to microtubule structure⁵⁵; 5) differential distributions of motor proteins and their regulators¹²; 6) hotspots for cytoskeletal anchoring (e.g., syntaphilin-mediated)⁵⁶; or 7) microtubule post-translational modifications.⁵⁷ Furthermore, myelination and axonal electrical activity also play a regulatory role in mitochondrial localization at the CNS NoR.⁵⁸

On the other hand, arrested mitochondrial motility is linked to activity-dependent axonal Ca²⁺ elevations.^22,59 Furthermore, radial contractility alters axonal size to accommodate the movements of larger organelles.⁶⁰ Thus, focal organelle accumulations might be linked to activity and can be transient in nature; however considerable follow up assessments are required to determine their mechanisms and dynamics. Moreover, direct interactions between Rab7-positive signaling endosomes and mitochondria, as well as co-transport with other organelles, could also influence nodal accumulations.^61,62,63

Organelle axonal transport dynamics at the nodes of Ranvier

While we have previously characterized signaling endosome and mitochondrial transport in physiological and pathological conditions,^7,9,64,65 this study quantitatively assesses in vivo axonal transport dynamics of these organelles specifically at the NoR. We observed similar patterns of transport dynamics of signaling endosomes and mitochondria across the node, with both organelles displaying faster velocities at the internodes, and more pausing at the NoR, which contrasts with the report of transient accelerations of neurofilaments through nodal constrictions.²⁵ Such transport differences between signaling endosomes and mitochondria compared to neurofilaments might be directly attributed to the transport machinery that regulates fast and slow axonal transport or their unique biological functions.^66,67 We also identified faster speeds in the proximal internode compared to the distal internode of both retrogradely transported signaling endosomes and anterogradely, but not retrogradely, moving mitochondria. This, along with the observations of organelle accumulations at the distal NoR, is suggestive of asymmetrical structural or functional differences in the nodal compartments of peripheral motor axons, which warrants further investigation. Given the similar composition and function of the NoR and the axon initial segment (e.g., distribution of Na_V1.6 channels,⁶⁸ Ank-G,⁶⁹ and Neurofascin-186⁷⁰), it would be interesting to compare transport dynamics between these two key axonal specializations.

Mitochondrial axonal transport ensures that specific neuronal regions, such as synapses and NoRs, can respond to localized high energy and Ca²⁺ handling demands.^71,72,73 We speculate that the observed mitochondrial axonal transport phenotypes are linked to the Ca²⁺ dynamics at the NoR. Indeed, an increase in intracellular Ca²⁺ decreases mitochondrial motility, independent of directionality, with Miro directly interacting with the motor domain of kinesin-1 to prevent its binding to microtubules.^22,74 Electrical activity also regulates mitochondrial dynamics, as Na⁺ channel activation in the nodal and paranodal regions slows mitochondrial transport (coupled to Na/K-ATPase),²² and can even modulate the size of stationary mitochondria by influencing fission/fusion events.⁵⁸ However, we did not observe spatial differences in the lengths of stationary mitochondria, but rather we detected major differences in mitochondrial length between stationary and motile mitochondria. On the other hand, kinases, such as GSK3, JNK3, and p38 MAPK, regulate kinesin-1 and/or cytoplasmic dynein motor protein activity and are considered as key modifiers in neurodegeneration.^2,10 Indeed, we and others have previously reported that inhibiting p38 MAPK rescues axonal transport deficits in a variety of models of motor neuron disease (MND).^75,76,77

The physiological role of signaling endosome trafficking at the NoR is inherently different from that of mitochondria. In motor neurons, signaling endosomes are Rab7-positive organelles that deliver distally activated signaling complexes from axon termini to the cell body to impact transcription and translation across the neuron.⁷⁸ In addition to the factors mentioned above (e.g., Ca²⁺, electrical activity, and kinases), the trafficking journey of signaling endosomes can be modulated by additional factors, including neurotrophins and their receptors^{7,31,35,79,80,81,82,83} and components of the extracellular matrix,³⁴ with glycolytic enzymes (e.g., GAPDH) sufficient to provide the in-transit ATP necessary for processive movement.^84,85 The reduction in speed and increased pausing that we observed at the NoR might be caused by the cytoskeletal constraints associated with the reduced axonal diameter. On the other hand, Rab7-positive organelles are involved in local mRNA translation⁶¹ and the majority of axonal mRNAs are undergoing active translation,⁸⁶ which may link the signaling endosome transport and accumulation phenotypes with local translation at the NoR.⁸⁷ Additionally, the accumulation of synaptic vesicle proteins, including SV2 and synaptophysin, at the NoR has been reported to function in axonal membrane processing and/or turnover.⁸⁸ Determining the precise molecular mechanisms regulating axonal transport dynamics at the NoR and their physiological roles will be important targets of future experiments.

In conclusion, using intravital imaging, we have characterized NoR morphologies in FMNs and SMNs, identified accumulations of signaling endosomes and mitochondria at the distal NoR, and determined the axonal transport dynamics of both organelles through the NoR. Finally, this work has clear implications for the peripheral nervous system and its numerous disorders (e.g., MND, multiple sclerosis),^89,90 and suggests that both the internodal and nodal transport modalities should be monitored across neurodegenerative disease models.

Limitations of the study

Our axonal transport method enables the in vivo visualization of both signaling endosomes and mitochondria from the same axon. Due to several factors, we performed manual tracking of the organelles, which might lead to subjective differences between the researchers performing the tracking. To minimize this risk, the same individual tracked the entire dataset (i.e., FA for signaling endosomes; APT for mitochondria). Furthermore, to avoid potential bias, quantitative analyses of the tracking outputs were only analyzed upon the completion of organelle tracking from all videos. Tracking was performed by randomly selecting a minimum of 20 endosomes/mitochondria per axon. We assume that the parameters gleaned from this analysis (e.g., mean speed, pausing events, and so forth) carried out in three or more axons per animal, accurately represent the physiological state of the organelles, the axons, and the individual mice. Another limitation of our study is that we assume that organelle dynamics at the NoR are similar between peripheral nerves. To address this point experimentally, validation of our conclusions in forelimb⁸/cranial nerves should be carried out in future studies.

Resource availability

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Giampietro Schiavo (giampietro.schiavo@ucl.ac.uk).

Materials availability

Requests for resources used in this study are available from the lead contact.

Data and code availability

• •Data: All data reported in this article will be shared by the lead contact upon request.
• •Code: This article does not report any original code.
• •All other items: Any additional information required to reanalyze the data reported in this article is available from the lead contact upon request.

Acknowledgments

We thank the personnel of the Denny Brown Laboratories (Queen Square Institute of Neurology, University College London) for assistance in maintaining the mouse colonies, and Elena R. Rhymes and David Villarroel-Campos (Queen Square Institute of Neurology, University College London) for critical reading of the article. This work was supported by a Junior Non-Clinical Fellowship from the Motor Neurone Disease Association (Tosolini/Oct20/973-799) (APT); a Col Bambrick MND Research Grant from Motor Neuron Disease Research Australia (IG 2450) (APT); a FightMND Drug Development Grant awarded to Giovanni Nardo (Istituto di Ricerche Farmacologiche Mario Negri - IRCCS) (DDG-73; for APT); EMBO short-term fellowship (SN); Medical Research Council fellowships (MR/S006990/1 and MR/Y010949/1) (JNS); Wellcome Trust Senior Investigator Awards (107116/Z/15/Z and 223022/Z/21/Z) (GS), and a UK Dementia Research Institute award (UKDRI-1005) (GS).

Author contributions

Conceptualization: APT and GS; methodology: APT, FA, SN, and JNS; Investigation: APT, FA, and SN; writing – original draft: APT; writing – review and editing: APT, SN, JNS and GS; funding acquisition: APT and GS; resources: APT, SN, JNS, and GS; supervision: APT and GS.

Declaration of interests

The authors declare no competing interests.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
NaV1.6 [SCN8a]	Alomone Labs	Cat # ASC-009 RRID AB_2040202
Anti-CASPR/Neurexin IV	Antibodies Incorporated	Cat# 75-001 RRID:AB_2083496
Goat anti-Mouse IgG (H + L) Cross-Adsorbed Secondary Antibody, Alexa Fluor™ 555	Thermo Fisher Scientific	Cat# A-21422 RRID:AB_2535844
Donkey anti-Rabbit IgG (H + L) Highly Cross-Adsorbed Secondary Antibody, Alexa Fluor™ 647	Thermo Fisher Scientific	Cat# A-31573 RRID:AB_2536183
Alexa Fluor 555 C2 maleimide	Thermo Fisher Scientific	Cat# A-20346
Chemicals, peptides, and recombinant proteins
HCT 441	Restani et al.84	N/A
Experimental models: organisms/strains
Mouse: Tg(Chat-EGFP)GH293Gsat/Mmucd (ChAT-eGFP)	Mutant Mouse Resource and Research Center	RRID: MMRRC: 000296-UCD
Mouse: B6.Cg-Tg(Thy1-CFP/COX8A)S2Lich/J (Mito.CFP)	Jackson Laboratory	RRID: IMSR_JAX: 007967
Mouse: B6; CBA-Tg(Plp1-EGFP)10Wmac/J (PLP-GFP)	Jackson Laboratory	RRID: IMSR_JAX:033357
Software and algorithms
TrackMate plugin	FIJI/ImageJ	Ershov et al.41
Prism 10 (Version 10.2.3)	GraphPad	https://www.graphpad.com/features
Other
Fluoromount G Mounting Medium	Thermo Fisher Scientific	Cat# 00-4958-02 RRID: SCR_015961
Dako fluorescence mounting medium	Agilent Technologies	Cat# S3023

Experimental model and study participant details

Animals

Animal experiments performed in the United Kingdom were conducted in accordance with the European Community Council Directive of 24 November 1986 (86/609/EEC), under license from the UK Home Office in accordance with the Animals (Scientific Procedures) Act 1986, and were approved by the UCL Institute of Neurology Ethical Review Committee. Procedures carried out in Italy were approved by the ethical committee and by the animal welfare coordinator of the OPBA from the University of Padua. All procedures specified in the projects are approved by the Italian Ministry of Health, Ufficio VI (authorisation numbers: 359/2015PR; 81/2017 PR; 521/2018 PR; 439/2019 PR) and were conducted in accordance with National laws and policies (D.L. n. 26, March 14, 2014), following the guidelines established by the European Community Council Directive (2010/63/EU) for the care and use of animals for scientific purposes.

3-6 month old heterozygous male and female mice of the following transgenic strains were used: 1) Tg(Chat-EGFP)GH293Gsat/Mmucd mice (RRID: MMRRC_000296-UCD,⁹¹; referred to as ‘ChAT.eGFP’ mice; 2) B6.Cg-Tg(Thy1-CFP/COX8A)S2Lich/J (RRID: IMSR_JAX: 007967),⁹² referred to as ‘Mito.CFP’ mice; and 3) B6; CBA-Tg(Plp1-EGFP)10Wmac/J (RRID: IMSR_JAX:033357),⁹³ referred to as ‘PLP-GFP’ mice. As we have previously determined that sex does not influence basal transport dynamics in ChAT.eGFP mice,⁹ all results are from both males and females. Mice were housed in individually ventilated cages in a controlled temperature/humidity environment and maintained on a 12 h light/dark cycle with access to food and water ad libitum.

Method details

Intramuscular injections of HcT

Fluorescently labeled atoxic fragment of tetanus neurotoxin (H_CT-555) was prepared as previously described.⁹⁴ Briefly, H_CT (residues 875–1315) fused to an improved cysteine-rich region was expressed in bacteria as a glutathione-S-transferase fusion protein, cleaved and subsequently labeled with Alexa Fluor 555 C2 maleimide (Thermo Fisher Scientific, A-20346). Mice were anesthetized using isoflurane, and after the fur on the ventrolateral lower leg was shaved, mice were placed on a heat-pad ready for intramuscular injections. A small incision was made on the ventral surface below the patella for the tibialis anterior muscle (TA), whereas for the soleus muscle a vertical incision was made on the skin covering the lateral surface of lower hindlimb between the patella and tarsus to expose the underlying musculature. Guided by previously established motor endplate maps,⁹⁵ intramuscular injections were performed, with 7.5–10 μg of H_CT in PBS in a volume of ∼3.5 μL using a 701 N Hamilton syringe (Merck, 20,779) for TA, or 1 μL in PBS using pulled graduated, glass micropipettes (Drummond Scientific,5-000-1001-X10) for soleus, as previously described.⁹⁶ Upon H_CT administration, the skin was sutured, and mice were monitored for up to 1 h before returning to the home cage.

In vivo imaging

Signaling endosomes were visualised in vivo after administration of H_CT, as previously described.^38,44,97 At least 4 h after HcT intramuscular injections, mice were re-anaesthetised with isoflurane, and the sciatic nerve was exposed by first removing the skin and then the overlying biceps femoris muscle. To aid the imaging process a small piece of parafilm was inserted between the underlying connective tissue and the sciatic nerve. The anesthetized mouse and nosepiece were then transferred to an inverted LSM780 confocal microscope (Zeiss) enclosed within an environmental chamber maintained at 37°C. Superficially located axons containing H_CT-loaded signaling endosomes were selected at random for imaging, as previously described.^38,44,97Time-lapse microscopy was performed using a 40×, 1.3 NA DIC Plan-Apochromat oil-immersion objective (Zeiss) focusing on axons around the NoR using an 80× digital zoom (1024 × 1024, <1% laser power). Frame intervals of ∼0.4–0.5 s were used when acquiring transport videos of motile H_CT-positive signaling endosomes, whereas frame intervals of ∼1.5–2.0 s were used when acquiring videos of motile Mito.CFP-positive mitochondria alone or in combination with H_CT-positive signaling endosomes. All imaging was concluded within a maximum of 1 h from initiating re-anaesthesia.

Morphological analysis of the NoR

Axonal morphologies at the NoR were assessed using the H_CT transport videos obtained in TA- or soleus-innervating motor axons from ChAT.eGFP and Mito.CFP mice, as previously described.^7,9 Orthogonal measurements were made between the upper and lower or proximal and distal ends of axonal regions containing motile H_CT signaling endosomes (Figure 1A). A minimum of ten measurements (i.e., combined proximal and distal internodal axonal diameter, axon diameter in the nodal constriction, and axon length in the nodal constriction) from at least three different axons were used to calculate the average diameter/length per animal.

Spatial analysis of organelles at the NoR

The distributions of signaling endosomes and mitochondria at the NoR were assessed from the axonal H_CT transport videos in TA- or soleus-innervating motor axons in Mito.CFP mice. Using FIJI/ImageJ, the relative fluorescence profiles of H_CT-containing signaling endosomes and mitochondria across the NoR and internodal regions were measured after drawing an 80 μm region centered around the nodal constriction from the z stack projection. After the background was subtracted, the relative fluorescence intensity profiles were then plotted, and averaged from a minimum of 3 axons per animal.

Immunohistochemistry of the sciatic nerve

NoR staining

Sciatic nerves were dissected from euthanised ChAT.eGFP mice and fixed in 4% paraformaldehyde in PBS for approximately 1 h at room temperature. Sciatic nerves were washed in PBS three times for 5 min, teased into individual fibers/bundles and were then permeabilised and blocked in a blocking solution containing 1% Triton X-100 and 10% bovine serum albumin in PBS for 1 h at room temperature. Sciatic nerve fibers were then incubated in a blocking solution containing primary antibodies against Na_V1.6 [SCN8a] (1:400; Alomone, ASC-009) and Caspr [clone K65/35] (1:500; Antibodies Incorporated 75-001) for ∼3 days at 4°C with mild agitation. Following three washes in PBS at room temperature, fibers were then immersed in a solution containing anti-mouse-555 (1:500; Thermo Fisher Scientific; A-21422) and anti-rabbit-647 (1:500; Thermo Fisher Scientific; A31573) secondary antibodies in PBS for ∼2 h at room temperature. Fibers were then washed three times in PBS, mounted in Fluoromount-G (Thermo Fisher Scientific, 00-4958-02) and covered with 22 × 50 mm cover glass (VWR, 631-0137). Slides were dried and imaged with a LSM780 confocal microscope using a 63× Plan-Apochromat oil immersion objective (Zeiss).

H_CT and Schwann cell cytosol

Anesthetized PLP-GFP mice were injected into the TA muscle with H_CT³⁸; after 24 h, mice were euthanised and the sciatic nerves dissected and fixed in 4% paraformaldehyde in PBS for 2 h at room temperature. After 3 × 5 min washes in PBS, sciatic nerves were then de-sheathed, teased into individual fibers and mounted using Dako fluorescence mounting medium (Agilent Technologies, cat S3023). z stack images were obtained using a with a confocal microscope (Zeiss LSM900 Airyscan2) equipped with an EC Plan-Neofluar 40×/1.30 oil objective.

Tracking analysis

Confocal “.czi” images were opened in FIJI/ImageJ (http://rsb.info.nih.gov/ij/), converted to “.tiff”, and using the TrackMate plugin⁴³ were tracked in a semi-automatic way for H_CT-positive signaling endosomes⁷ or manually for mitochondria.⁹⁸ The following criteria were used for the tracking analysis: 1) only endosomes and mitochondria that were moving for ≥10 consecutive frames were included, and terminal pauses, as defined by the absence of movement in ≥10 consecutive frames, were excluded; 2) each axon required a minimum of 20 trackable organelles per axon, except for retrogradely moving mitochondria, which only required a minimum of 10 per axon; 3) at least three separate axons were assessed per mouse. A pause was defined by a previously motile organelle with a velocity of ≤0.1 μm/s between consecutive frames (to account for a potential breathing/arterial pulsing artifact), and the time paused (%) was determined by dividing the number of pauses by the total number of frame-to-frame movements assessed from an individual axon.

Transport dynamics of the retrogradely moving H_CT signaling endosomes, as well as anterogradely and retrogradely moving mitochondria were separately assessed with TrackMate,⁴³ using the above criteria. Each frame-to-frame recording (e.g., velocity, pausing) was matched with its x and y co-ordinates, binned every 2 μm across an 80 μm axonal length centered at the NoR, and subsequently separated into either the proximal internode, nodal constriction, or distal internode locations. The mean moving velocity represents the average of all frame-to-frame movements in a particular axonal area (excluding pausing events), whereas the relative frequency of mean pausing was determined by assessing the relative number of pauses in relation to the axonal location. The correlation figures were generated by plotting the mean velocity of signaling endosome and mitochondrion transport, as well as the relative frequency of signaling endosome and mitochondrion pausing, against the mean value of axonal diameters for each individual axon.

Mitochondrial directionality, flux and length

Mitochondrial directionality, flux and length were quantified using the same videos used to analyze in vivo axonal mitochondrial transport at the NoR. For directionality, every movement was separately counted for individual anterogradely and retrogradely moving mitochondria across the proximal, nodal constriction and distal regions using Cell Counter plugin (FIJI). Outputs were averaged across the subdomains, and then presented as a relative fraction directly comparing anterogradely with retrogradely moving mitochondria from individual axons, and then averaged across the three axons per animal. For mitochondrial flux, the outputs from the directionality assessments were then normalised to the time of individual videos and presented as the mean number of moving mitochondria per minute, either combining the axonal subdomains (e.g., Figure 6E) or presenting them separately for both anterogradely and retrogradely moving mitochondria (e.g., Figure 6F). For mitochondrial length, a minimum of 10 randomly selected stationary, anterogradely or retrogradely moving mitochondria from the proximal and distal regions were measured from every axonal transport video, and presented for individual mice as the mean value from a minimum of 3 axons (Figure 6F) or individually (Figure S6).

Quantification and statistical analysis

GraphPad Prism 10 Software (Version 10.2.3) was used for all statistical analyses and figure production. Data were assumed to be normally distributed, and parametric data were assessed using paired or unpaired two-tailed t-tests, as well as one-way or two-way analyses of variance (ANOVA) with Holm-Sidaks multiple comparison tests. Pearson correlation coefficients (two-tailed) were also computed. Specific statistical details of each experiment can be found in the figure legends. n = number of animals. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.

Published: October 11, 2024