Nerve biopsy material was obtained from a male patient who was diagnosed with anti-pan-NF-associated autoimmune nodopathy. The diagnosis was based on a typical clinical picture with tetraplegia, areflexia and cranial nerve involvement, requiring mechanical ventilation for several months, and detection of autoantibodies against NF155 and NF186 by enzyme-linked immunosorbent assay (ELISA) and cell-based assays with NF-transfected human embryonic kidney 293 cells and confirmation by nodal and paranodal binding to murine teased fibers as previously described (4). CSF protein was highly elelvated (>2000 mg/l), CSF cell count was normal. Nerve conduction studies of the tibial, peroneal, sural, ulnar and median nerve did not detect any recordable sensory or compound motor nerve action potentials. The patient exhibited a classic monophasic disease progression, with complete elimination of anti-pan-neurofascin autoantibodies following treatment with plasma exchange, rituximab, and bortezomib. This was accompanied by a gradual and steady improvement of symptoms. The nerve biopsy of the sural nerve was obtained for diagnostic work-up six months after the onset of symptoms according to standard surgical procedures. The study was approved by the Ethics Committee of the University of Würzburg (number 158/21).

Murine sciatic nerves were dissected from adult C57/Bl6 mice. Human biopsy samples and murine nerves were cut into 5-mm slices, fixed in 4% paraformaldehyde for 10 minutes, then washed three times for 5 minutes each in phosphate-buffered saline (PBS) (1 M, pH 7.4), and processed for further analysis as previously described (13):

The epineurium and perineurium were removed with fine forceps under a stereo microscope. The remaining nerve fascicles were teased into single fibers using fine forceps in phosphate-buffered saline (1 M, pH 7.4). The main purpose of this procedure was to expose single nerve fibers and/or fiber bundles for better staining and visualization.

The 24-mm glass coverslips (Carl Roth, Marienfeld - Germany) on which the fibers were teased for dSTORM and fluorescence imaging received special treatment beforehand: they were washed in isopropanol for 15 minutes, plasma cleaned for 10 minutes, and coated with poly-D-lysin (1 mg/ml) for 1 hour at 37°C. After the procedure, the samples were stored at -20°C.

The teased fibers were immunostained using primary antibodies against Caspr-1 (polyclonal rabbit IgG, 0.2µg/ml, AbCam, ab34151), PanNF (polyclonal chicken, IgY, 0.4µg/ml, R and D Systems, AF3235), beta-IV spectrin (polyclonal rabbit IgG, 5 µg/ml, Invitrogen, PA5-62972), beta-II spectrin (polyclonal rabbit IgG, 5 µg/ml, BD Biosciences, ab34151), and Pan-sodium (polyclonal mouse IgG, 0.2µg/ml, Sigma, S8809). Prior to staining, the teased fibers were pre-permeabilized and fixed with ice-cold methanol at -20°C for 10 minutes, then washed with PBS three times for 5 minutes each. Afterward, the samples were permeabilized with 0.3% Triton-X in PBS for 30 minutes. Subsequently, the samples were blocked in a solution of 1% BSA + 0.3% Triton-X in PBS for 1 hour. The samples were then incubated with primary antibodies overnight at 4°C.

Secondary antibodies, including anti-rabbit Alexa Fluor 647 (A647, polyclonal goat IgG, 10 µg/ml, Jackson, #711-605-152), anti-chicken Cy3 (polyclonal goat IgG, 5 µg/ml, Jackson, 15914), and anti-mouse Alexa Fluor 488 (A488, polyclonal goat IgG, 5 µg/ml, Invitrogen, #A-110019), were used. The incubation with secondary antibodies was performed overnight at 4°C. For conventional fluorescence imaging, the teased fibers were embedded in a Vectashield antifade mounting medium (Vector Laboratories, VEC-H-1000). For dSTORM, the samples were immersed in PBS until imaging in the dSTORM-specific imaging buffer, inducing fluorophore blinking.

Dual-color dSTORM microscopy was performed using an inverted light microscope (Zeiss Observer Z.1, Carl Zeiss AG) customized for single-molecule localization microscopy (SMLM) ( Supplementary Table 1 ). The dSTORM imaging buffer contained 125 mM cysteamine hydrochloride (Sigma, M6500), 20 mM D-glucose (Sigma, G7528), 0.55 mg/ml glucose oxidase (Roth, 60,281), and 0.011 mg/ml catalase (Sigma, C1345) in PBS at pH 7.7. The desired pH (7.7) was achieved using a 5M KOH solution. The dual-color dSTORM images were acquired sequentially as summarized in Supplementary Table 2 . Reconstruction of dSTORM images was performed by ImageJ plugin ThunderSTORM (14, 15) and resulted in super-resolved images with a pixel size of 102 nm. For more details about the reconstruction process see Supplementary Table 2 .

Autocorrelation analysis was performed on selected regions of interest (ROIs) to determine periodic protein arrangements as described previously (11) using the respective MATLAB script. Several regions of interest (ROIs) within the paranodal and internodal regions were manually selected in the super-resolved reconstruction of the respective node. The lengths along the axon were determined by selecting a start and end point of the centerline. Super-resolved images of Caspr-1, PanNF, beta-II and -IV spectrin, and Voltage-Dependent Sodium Channels (PanNaV+) were analyzed. Due to severely damaged ultrastructures, the periodicities from PanNF and Caspr-1 could not be retrieved and therefore not be quantified. All images with clear periodic structures were included in the peak-to-peak autocorrelation analysis, while those with low resolution or undefined structures were excluded.

Colocalization analysis was performed on the super-resolved images stained for Caspr-1 (to define the paranodal region) and stained for PanNF to visualize two isoforms of NF (-155 and -186) as previously described (11). For featuring the colocalization, we assessed the Manders A and B correlation coefficients (16, 17).

The dSTORM images of structural proteins from the para- and nodal regions of the nodes of Ranvier (specifically Caspr-1 and PanNF), from the patient under study and the mouse control, were compared by the number of localization events acquired over 20,000 frames (20 ms per frame) within the selected ROI. From a single node, a total of 3 ROIs each for Caspr-1 and PanNF were analyzed for patient and mouse control. The average localization events/µm² were calculated to provide a relative measure of protein density. It is important to note that all parameters for dSTORM acquisition and image reconstruction, as well as the concentration of antibodies for IHC and the methodology, were kept constant to allow for quantification (18).

SMLM images of nodes of Ranvier were analyzed using Shannon entropy as a metric for the nodes by their structural order. Nodes were automatically extracted based on their typical size using Gaussian filtering and Otsu’s thresholding. Shannon entropy was calculated for each node to quantify its complexity and information content This automated analysis enabled a comparison of nodal patterns with those of mouse controls, offering valuable insights into structural differences and organization. It also provided the analytical foundation for identifying and explaining the complete loss of periodicity in some samples ( Supplementary Figure 1 )

High-content images (large field-of-view) (Leica DMi8 microscope, magnification of the objective lens: 20 x) were used to obtain a general overview of the nodes of Ranvier. ( Supplementary Figure 2 ). Nodes were categorized into severely damaged nodes if the nodal architecture was apparently disrupted, with dispersion of Caspr-1 beyond the paranodes, severe elongation of the nodal gap and/or disruption of the staining of nodal/paranodal proteins, and normal-looking nodes if there was no obvious paranodal and nodal disruption and a clear block-like staining of anti-Caspr-1 and anti-pan-NF could be identified, together with a small anti-Nav-positive nodal gap ( Figure 1 ). The percentage was calculated based on the number of normal-looking and severely damaged nodes over the total number of nodes in the image. A total of nine images were analyzed. For morphometric analysis, the length of the nodal gap, i.e. the distance between the paranodes as visualized by immunofluorescence staining with anti-Caspr-1, was measured. This was done manually using previously established methods (9).

Statistical analysis was performed using Origin Pro software (Origin2021b, OriginLab Corp.). Normality was tested with the Shapiro-Wilk test and the comparison of nodal length between the groups (murine nodes, normal-looking human nodes, severely damaged human nodes) was performed using Welch’s test. Results with p-values < 0.05 were considered significant.

To evaluate the extent of damage to the nodes of Ranvier, we conducted a comprehensive assessment of a substantial number of nodes (n=100) using high-content imaging with anti-pan-NF and anti-Caspr-1 staining, following previously established protocols (11). Our analysis revealed that 14% of the nodes exhibited a significantly increased nodal length (mean: 3.16 ± 0.67 µm) compared to the normal values reported in a prior study (mean: 0.9 ± 0.3 µm) (11), with this difference being statistically significant (p = 2.10 x 10^-10) ( Table 1 ). These elongated nodes also demonstrated considerable disruption of the nodal protein architecture ( Figures 1B, E ). In contrast, the majority of the nodes (86%) maintained a normal architecture and length ( Figures 1A, D ) (mean: 0.943 ± 0.14 µm). We further refer to ‘normal-looking’ nodes as those with preserved nodal architecture, characterized by block-like staining of the paranodes with anti-Caspr1, a clearly identifiable nodal gap positively stained for Nav. Conversely, ‘abnormal-looking’ nodes exhibit disrupted nodal architecture, with paranodal protein staining dispersed along the fiber and/or a nodal gap that is significantly enlarged or absent (see Figure 2 for an example).These two populations were compared qualitatively as well to a mouse control as shown in Figures 1C , F . Normal- and abnormal-looking nodes occurred within the same nerve fiber.

Furthermore, sodium channels (NaV+) were assessed in both populations: normal-looking and severely damaged nodes. Qualitatively, the normal-looking nodes exhibited intact sodium channel integrity ( Figure 1J ) and normal architecture of PanNF proteins ( Figure 1G ). However, in the severely damaged nodes, this integrity was lost for both the sodium channels and the PanNF proteins ( Figures 1L , H , respectively). These qualitative observations were compared with the mouse control, stained for sodium channels and PanNF protein, as shown in Figures 1M , I .

By dSTORM we investigated subtle changes in the normal-looking nodes and evaluated nodal architecture at the nanoscale. We assessed 20 nodes using our custom dSTORM setup and compared the results with data from a reference mouse control cohort (11). Our analysis revealed a qualitative decrease in the density of the paranodal proteins Caspr-1 and NF (NF-155) and nodal protein (NF-186) ( Figures 2A–D ) in the normal-looking nodes. This decrease was quantified by analyzing the density of localization events over the dSTORM image acquisition from the respective super-resolved images of the normal-looking nodes and the dSTORM images of the same protein structures in a mouse control ( Figures 2E–H ). This decrease ranged from 0.106 (mouse control) to 0.052 (patient) for Caspr-1 and 0.32 (mouse control) to 0.024 (patient), for PanNF, with both units expressed as average localization events per frame/µm² ( Table 2 ).

Despite the decrease, the colocalization of these proteins was preserved, indicating no axoglial disjunction ( Figure 2I ).

The periodicity of anti-Caspr-1 and anti-NF staining, however, was severely impaired and could not be quantified ( Figures 2B, D ). Further entropy analysis (Shannon entropy analysis) was done to confirm the total loss of the periodicity compared to a mouse control ( Supplementary Figure 1 ).

In contrast, the nodal sodium channels exhibited their characteristic periodicity (mean value: 198 ± 42 nm) ( Figures 3C, D, E ), despite the complete loss of periodicity in nodal (NF-186) and paranodal (NF-155) Neurofascin ( Figures 3A, B ).

To evaluate the potential destruction of cytoskeletal architecture at the nodes of Ranvier and internodes as a marker for axonal damage, the ultrastructure of beta-IV and beta-II spectrin (nodal and internodal, respectively) was analyzed by dSTORM imaging ( Figures 3F–H ). We did not detect any abnormalities of the periodicities of beta-IV and beta-II spectrin (187 ± 29 nm, beta-IV and 188 ± 18, beta-II) (graph 3i) indicating preservation of the nodal and axonal cytoskeleton. The periodicity measurement of beta-II spectrin was compared with the mouse control cohort from (9) and the beta-IV spectrin was compared with a mouse control cohort analyzed for this project ( Supplementary Figure 3 ). We did not find any significant differences between the patient and the mouse control cohorts ( Table 3 ). However, in severely damaged nodes, decreased or even absent beta-IV spectrin staining was observed. ( Supplementary Figure 4 ).