JAM-C cis and trans binding sites are required for JAM adhesions in CGNs.

(A) Protein BLAST encompassing the V-Domain of Junctional Adhesion Molecule (JAM)-A, -B, and -C from Mus musculus and human JAM-C. Amino acids conserved across all four proteins are highlighted in gray. (B,C) AlphaFold prediction of JAM-C protein structure compared to the crystal structure of JAM-A (PDB:1F97) with side chains shown and labeled with one letter amino acid abbreviation for key charged residues that have been previously associated with salt bridges at JAM cis and trans binding sites. JAM-A is represented in black. JAM-C is represented in light gray with the (B) cis binding interface highlighted in pink and the (C) trans binding interface highlighted in green. (D-G) Co-immunoprecipitation (Co-IP) of JAM-C point mutants with JAM-pHluorin. (D) Diagram of Rabbit αGFP (Molecular Probes) Co-IP experimental design using JAM-pH as the bait protein with JAM-C prey molecules bearing point mutations expressed in Neuro-2a cell line. (E) Representative western blot images of inputs and elutions of JAM-pH immunoblotted with Chicken αGFP (Aves) or Goat αJAM-C (R&D Biosciences). Lanes were loaded in two sets of 5 for JAM-B-pH and JAM-C-pH respectively with negative IP of JAM-pH by Rabbit IgG isotype control (Biolegend) followed by four samples of Rb αGFP CoIP. Quantification of the relative amount of JAM-C bait detected in the IP eluate divided by JAM-C in the input for (F) JAM-B-pH and (G) JAM-C-pH. The relative amount was normalized to the Co-IP of JAM-C WT for the JAM-pH conditions independently for all biological replicates (N=4) and statistically compared to JAM-C WT using a Wilcoxon Rank Sum test. (H-K) Quantification of JAM-C-pH adhesion morphology in JAM-C-replaced CGNs. (H) Representative max projection JAM-C-pH images of excitation 488nm fluorescence in the CGNs imaged at 63x 1.46NA and subsequently CLAHE contrast-enhanced to display low signal features. Scale bar = 10 µm (I) Representative examples of Ilastik segmentation of cell body in gray and adhesion in cyan for the corresponding fluorescence images. (J) Diagram of the quantification for the proportion of total fluorescent signal measured by the adhesion mask divided by the total fluorescent signal measured within the cell body mask. (K) Plotted comparison of all biological replicates (N≥3) as daily aggregates of technical replicates that are statistically compared to JAM-C-pH WT using a Wilcoxon Rank Sum test. Statistical significance is represented by standard convention: ns p>0.05, * p<0.05, ** p<0.01, ***p <0.001, and **** p<0.0001.

The combination of JAM-C cis and trans interactions is required to initiate ML entry.

(A-C) JAM-C replacement rescues JAM-C knockdown in ex vivo cerebellar slice cultures. (A) Experimental design of JAM-C replacement by electroporating cDNA into the superficial layer of dissected cerebella to knockdown endogenous expression with MiR30 shRNA and express shRNA insensitive JAM-C. (B) Representative max projection fluorescent images of Ex 560 H2B-mCherry labeled CGNs at 24 and 48 hours captured with a 20x 0.8 NA objective. The dashed line represents the barrier between the EGL and ML at 50 µm from the cerebellar external edge. (C) Quantification of the proportion of all StarDist segmented nuclei further than 50 µm from the slice edge at 48h. Data shown represents single biological replicates (N≥3) that are an aggregate of corresponding technical replicates. Statistical comparison to the shLuc+LacZ positive control for ML entry was performed via Wilcoxon Rank Sum Test. (D) Representative tracking data of H2B-mCherry labeled CGNs at 24, 36, and 48 hour. The track color is encoded to represent time point in track based on the provided look up table. The dashed line marks the boundary between EGL and ML at 50 µm from the cerebellum edge. (E) Plot of the proportional of H2B-mCherry expressing CGNs >50 µm from the cerebellar edge in the ML. Data points represent independent biological replicates (N) collected at each 5-minute time interval from 24-48h. Data trends over time are plotted as a smoothed Generalized Additive Model (gam). Statistical comparison of proportion of ML entry to JAM-C WT rescue was performed at (F) 40 and (G) 48 hours of culture using Wilcoxon Rank Sum tests. (H) Plot of instantaneous migration angle in relation to the cerebellar edge as a function of time for a subset of the data including JAM-C WT, Δcis, and Δtrans. The trend in time was generated using gam smoothing for the aggregate of all technical and biological replicates (n) of cell motility. Gray shading on the plots represents the 95% confidence intervals of the estimated means. Scale bars = 50 µm. Statistical significance is represented by standard convention: ns p>0.05, * p<0.05, ** p<0.01, *** p<0.001, and **** p<0.0001.

Establishing Coculture Models to Molecularly Dissect JAM Mediated Adhesion.

(A) Diagram of the conditional detection of JAM specific contacts using two pH sensitive fusion proteins expressed in apposed CGNs and glia: JAM-C-pH and LAMPshade Violet (LSV)-SNAPtag Ligand (-STL). (B) Representative time series of contact formation between JAM-C-pH in a CGN and JAM-B-SNAP on a glial cell acquired using AiryScan Microscopy with a 63x 1.4NA objective. Images shown are 0.5 gamma adjusted sum projections of pixel reassigned z-stacks captured every 7 minutes. Channels are shown split into single channel grayscale images and merged in color. Scale bar = 10 µm (C) Enlarged view of JAM-C-pH to JAM-B-SNAP contact site. Scale bar = 10 µm (D) Representative FRAP images corresponding to: i.) gamma adjusted initial max projections of GPI-2xBFP, JAM-C-pH, and LSV-STL labeled JAM-SNAP multichannel images ii.) JAM-C-pH immediately after photobleach, and iii.) JAM-C-pH at max recovery at 32 seconds. An inset is provided for JAM-C-pH to JAM-C-SNAP to show signal overlap. The orange box in ii.) is the photobleached area, and the orange triangles in iii.) points to local recovery. (E) Trace of normalized JAM-C-pH fluorescence for the data shown in (D). Statistical significance is represented by standard convention: ns p>0.05, * p<0.05, ** p<0.01, *** p<0.001, and **** p<0.0001. (F) Comparison of Mobile Fraction calculated for adhesion to JAM-C on CGNs, JAM-B on glia, JAM-B on glia treated with Latrunculin A, JAM-B on glia after washing away Latrunculin A. Technical replicates from at least 2 biological replicates are plotted as individual points within violin plots. Statistical comparison is performed using Wilcoxon Rank Sum Tests compared to JAM-B on glia. (G) Diagram of experimental setup to detect adhesion events for secondarily dissociated JAM-C-pH replaced CGNs plated on monolayers of JAM-C-SNAP expressing CGNs or JAM-B-SNAP replaced glia. (H) Representative orthoslice views of CGN JAM-C-pH adhesion to LSV-STL labeled glial JAM-B-SNAP fluorescence acquired by lattice lightsheet microscopy detection through a 50x 1.0NA objective. Orthoslice views of deskewed and deconvolved light sheet timelapses were selected to visualize the cross-sections of two CGNs as they contact JAM-B-SNAP expressing glia at a 30 second intervals. The axial position of the images was manually registered over time, and JAM-B-SNAP fluorescence was photobleach corrected using histogram equalization between time points. Images are gamma adjusted by 0.75 to highlight low fluorescence signal for both JAM-C-pH and JAM-B SNAP. Gray arrowheads are used to point out the observed first instance of JAM-C-pH to JAM-B-SNAP contact formation.

The cis and trans binding of JAM-C in CGNs initiates recognition of JAM-B on glia.

(C-F) Evaluation of the prevalence of JAM-C-pH contact formation with LSV-STL labeled JAM-SNAP at 2 hours. (C) Representative images of 0.6 gamma adjusted JAM-C-pH fluorescence with 1.2 gamma adjusted JAM-B-SNAP fluorescence for JAM-C-pH WT, Δcis, Δtrans, and Δcis/Δtrans CGNs. Instances of single glia were selected from large fields of view captured by a 20x 0.8NA objective. The gray dashed line is used to manually indicate the edges of a single glia. Scale bar = 50 µm. (D) Examples of Ilastik segmentation of JAM-C-pH adhesion and JAM-B-SNAP for the associated fluorescence images. The gray dashed line is used to manually indicate the edges of a single glia as a landmark to compare the fluorescent image. The segmentation is shown as traces of the boundaries for segmented objects. Quantification of adhesion prevalence for (E) JAM-B-SNAP on glia and (F) JAM-C-SNAP on CGNs. Adhesion prevalence is defined as the area of segmented JAM-C-pH overlapping the segmented JAM-B-SNAP divided by total JAM-B-SNAP segmented area. Data points correspond to the aggregate values of independent biological replicates (N≥3) after excluding outliers. Broken y-axis is used on the CGN graph to show spread of data at low proportion values. Pairwise statistical comparisons are performed using Wilcoxon Rank Sum Tests against JAM-C-WT replaced CGNs. Statistical significance is represented by standard convention: ns p>0.05, * p<0.05, ** p<0.01, *** p<0.001, and **** p<0.0001.

JAM-C recognition of glial JAM-B contacts in trans recruits Pard3 in the CGNs

(A) Diagram of experimental set up. drebrin-SNAP and Halo-Pard3 are expressed in JAM-C-pH replaced CGNs and Glia express JAM-B-pH at t0. We then track the recruitment of these cytoplasmic proteins to the point of contact at t1. (B) Representative max projection images of JAM-B-pH glial forming a contact with JAM-C WT replaced CGNs that have JFX650-Halo tag ligand (-HTL) labeled Halo-Pard3. Individual channel images of JAM-pH and Halo-Pard3 are presented alongside the multicolor merged image to show the degree of overlapping signal. (C) representative multichannel, max projection images of Halo-Pard3 recruitment to JAM contacts for mutant JAM-C replacements that include: Δcis, Δtrans, Δcis/Δtrans, and ΔPDZ which is missing the 6 amino acid C-terminal motif that binds Pard3 (D) Loess smoothed plot of Halo-Pard3 recruitment to glial JAM contacts measured by Pearson correlation for individual cells as a function of the time after first adhesion is identified including the Extracellular Domain (ECD) that lacks the entire 48 amino acid cytoplasmic tail in addition to the constructs with associated representative images. The trends are generated from all technical replicates (n) from at least 3 independent biological replicates (N≥3). The dashed line indicates the point in time used for statistical comparison. (E) Statistical comparison of Pearson correlation between JAM-pH and Halo-Pard3 at 5 minutes after first sign of adhesion. The Pearson correlation of each technical replicate is plotted within the violin plot. Pairwise statistical significance is calculated in relation to JAM-C WT replaced CGNs using Wilcoxon Rank Sum Tests. (F) Loess smoothed plot of Halo-Pard3 recruitment to glial JAM contacts measured by Pearson correlation for individual cells as a function of the segmented adhesion size. The correlation plots and statistical comparisons were generated using all technical replicates (n) from at least 3 independent biological replicates (N≥3) uniformly subsampled at 30 second intervals. Gray shading on the plots represents the 95% confidence intervals of the estimated mean values. Scale bar = 5 µm. Statistical significance is represented by standard convention: ns p>0.05, * p<0.05, ** p<0.01, *** p<0.001, and **** p<0.0001.

JAM-C trans binding is required to locally recruit Pard3 and drebrin to glial contacts.

(A) Diagram of scheme for analyzing and representing the local correlation of JAM contacts by calculating Pearson correlation in a kernel of defined dimension in x, y, and t that is subsequently scanned across the time lapse image. (B) Representative fluorescent data (first row), local recruitment of Halo-Pard3 at JAM contacts (second row), and local recruitment of drebrin-SNAP at JAM contacts (third row) for JAM-C WT replaced CGNs. Pixelwise correlation within segmented JAM-contacts is encoded from blue (correlation = -1) to red (correlation = +1) for paired fluorescence correlation. (C) Representative fluorescence images and associated local correlation for the JAM-C mutant constructs: Δcis, Δtrans, Δcis/Δtrans, and ECD that lacks the entire 48 amino acid cytoplasmic tail. Loess smoothed plots of the sum of the (D) positive (Pearson > 0) and (E) negative (Pearson<0) JAM:Pard3 local correlation as a function of the time after first adhesion. Loess smoothed plots of the sum of the (F) positive (Pearson > 0) and (G) negative (Pearson<0) JAM:drebrin local correlation as a function of the time after first adhesion. The plots of sum correlation were generated from the combination of technical replicates (n) from at least 3 independent biological replicates (N≥3) uniformly subsampled at 30 second intervals. Gray shading on the plots represents the 95% confidence intervals of the estimated mean values. Scale bar = 2×2 µm.

CGN JAM-C recognition of JAM-B on glia is a spatiotemporally defined developmental barrier to the recruitment of Pard3 and drebrin.

Representative imaging data of fluorescence and local correlation for drebrin-SNAP, Halo-Pard3, and JAM contacts are shown for (A) JAM-B on glia or (B) JAM-C on CGNs. Similar adhesion morphology examples were selected for qualitative comparison. (C-F) The sum correlation of JAM:Pard3 and JAM:drebrin for JAM-C expressing CGNs is compared for adhesion to either JAM-B on glia or JAM-C on CGNs. Loess smoothed plots of JAM:Pard3 are shown for (C) sum positive correlation and (D) sum negative correlation as a function of time after first adhesion. Loess smoothed plots of JAM:drebrin are shown for (E) sum positive correlation and (F) sum negative correlation as a function of time after first adhesion. Loess plots were generated from the combination of all technical replicates (n) indicated in the plot legends. Gray shading on the plots represents the 95% confidence intervals of the estimated mean values. Scale bar = 2×2 µm.

AlphaFold Multimer prediction of JAM dimers.

Representative view of the predicted structure of 4 JAM subunits generated for either (A) 4 JAM-C extracellular domains or (B) 2 JAM-C with 2 JAM-B extracellular domains. The prediction yielded two independent cis interacting dimers dimers. Isolation of the predicted dimers from the 4 subunit predictions for (C) JAM-C to JAM-C and (D) JAM-C to JAM-B. The structural prediction shows that these dimers are predicted to form at the BLAST aligned charged residues for (E) JAM-C to JAM-C and (F) JAM-C to JAM-B cis adhesion sites.

Validation of JAM-pH fluorescent protein construct stability and trafficking in Neuro-2a cell line.

(A,B) Cycloheximide pulse was used to block protein translation and then protein stability was tracked over time via Western Blot. (A) Representative western blot images of cycloheximide time course for JAM-C-pH expressing Neuro-2a cells immunoblotted with Chicken αGFP (Aves). (B) Independent biological replicates of half-life calculated in GraphPad Prism by one-phase decay are plotted and statistically compared to JAM-C-pH WT via Wilcoxon Rank Sum test. (C,D) Trafficking of JAM-pH to the cell surface was measured with surface biotinylation. (C) Representative western blot images of Neuro-2a lysate and biotin pull down by streptavidin functionalized agarose beads (Thermo Scientific) immunoblotted with Chicken αGFP (Aves). (D) Surface biotinylation was quantified as the ratio of the JAM-pH detected in the pull down divided by the JAM-pH detected in the input. The ratios for biological replicates (N≥3) are plotted and statistically compared to JAM-C-pH WT using a Wilcoxon Rank Sum test. Statistical significance is represented by standard convention: ns p>0.05, * p<0.05, ** p<0.01, *** p<0.001, and **** p<0.0001.

JAM-C-pH adhesions exhibit a wide range of dynamics in live JAM-C replaced CGNs

Representative Ex 488 nm fluorescence images of the migration of JAM-C-pH WT replaced CGNs. The images shown are max projections of 3D data acquired at 63x 1.46NA that have been gamma adjusted at 0.75 to enhance low signal areas. The migrating neuron is outlined in a dashed grey line. (A) Some adhesions are observed to last longer than 12 hours. A couple of instances of this behavior are noted with the presence of white arrowheads that remain relatively stationary for the entire time course that do not exhibit strict biases in subcellular location. (B,C) Within the time course, new adhesions are observed to form and undergo rearrangement. (B) A soma localized adhesion was observed to form within a 20-minute temporal interval marked with a pink arrowhead. (C) As an example of adhesions not maintaining a constant size, the cell migrating past the adhesion site marked with the pink arrowhead appears to stretch the fluorescent signal until the soma passes the adhesion location allowing the signal to contract laterally.

Total JAM-C-pH adhesion in JAM-C replaced CGNs is at a relative steady state 24 hours after nucleofection.

(A-D) JAM-C-pH adhesions were live imaged overnight every 15 minutes to track adhesion over time. The data displayed is a Loess smoothed trend for the data points from N=3 biological replicates. In general, the (A) proportion of fluorescence at adhesions, (B) Proportion of cell area with adhesion, (C) median adhesion area measured in µm2, and (D) total cell area measured in µm2 are relatively constant over the 10-hour time course. Gray shading on the plots represents the 95% confidence intervals of the estimated means.

The quantification of JAM-C-pH adhesion is impacted by proportion of cell area and adhesion size but not total cell area.

Comparisons of (A) proportion of cell area with segmented adhesions, (B) median adhesion area measured in µm2, (C) and the total analyzed cell area measured in µm2 at 24 hours of culture for independent biological replicates each compared to JAM-C-pH WT replaced CGNs using Wilcoxon Rank Sum tests. Statistical significance is represented by standard convention: ns p>0.05, * p<0.05, ** p<0.01, *** p<0.001, and **** p<0.0001.

Trackmate and Stardist can track nuclear labled CGNs in cerebellar slice cultures.

(A) Representative examples of manually tracked H2B-mCherry-labeled CGNs from 3 independent scientists (B) Representative example of CGN tracking using StarDist and Trackmate. (C-F) Quantification of the reproducibility of tracking performed on 4 separate data sets. Dashed lines are placed on bar charts at the ideal value for reproducibility for the associated metrics. Data points represent quantified values of independent biological replicates (C) Reproducibility of individual migration events was quantified as the ratio of migration speeds for the comparisons listed on the x-axis. Overall track quality was quantified by categorically annotating successful tracks and plotting the output (D) precision, (E) recall, and (F) F1-Score.

Tracking CGN migration for JAM-C rescue with StarDist and Trackmate reveals subtle changes in migration behavior at 24-48 hours.

(A) Representative tracking data for validating JAM-C rescue at 24, 36, and 48 hours. Track position in time is color encoded based on the provided lookup table. Quantification and plotting of the proportion of CGNs in the (B) oEGL, (C) iEGL, and (D) ML over time. Aggregated proportion for each biological replicates is plotted directly on the graph at each time point, and the smoothed trend over time was plotted using gam. (E-H) Quantification of the migration tracks for each condition, aggregating data for tracked migration events across biological and technical replicates (n). Data is plotted as a gam smoothed trend as a descriptive representation of migration behavior. Gam plots of (E) migration speed and (F) migration angle as functions of time. Gam plots of (G) migration speed and (H) migration angle data as functions of distance from tissue edge. Gray shading on the plots represents the 95% confidence intervals of the estimated mean values.

Migration track data for JAM replacement in slice cultures.

(A-D) Quantification of migration tracks for each JAM replacement is shown, aggregating data for tracked migration events across biological and technical replicates (n). Data is plotted using gam smoothing as descriptive representations of migration behavior overall. Gam plots of (A) migration speed and (B) migration angle as functions of time. Gam plots of (C) migration speed and (D) migration angle data as functions of distance from tissue edge. Gray shading on the plots represents the 95% confidence intervals of the estimated mean values.

Stable de novo JAM contacts are more likely to form between CGN JAM-C and Glia JAM-B than between CGNs with JAM-C.

(A) Representative 0.25 gamma adjusted, max projected, fluorescent images for JAM-C-pH WT replaced CGNs adhering to either JAM-C-SNAP expressing CGNs or JAM-B-SNAP expressing glia at 2 hours for entire fields of view acquired by a 20x 0.8 NA objective. Scale bar = 100 µm (B) Ilastik segmentation for JAM-C-pH and JAM-SNAP adhesions corresponding to the fluorescent imagers in (A). Segmentation is shown as the traces of the boundaries for segmented objects. (C-F) Quantification and statistical comparison of SNAP-pH adhesion parameters for JAM-SNAP expressing monolayers. (C) Fraction of JAM-SNAP with JAM-C-pH is defined as the overlapped area of Ilastik segmented JAM-C-pH and JAM-SNAP divided by total JAM-SNAP segmented area. (D) Fraction of monolayer with JAM-SNAP is defined as the proportion of SNAP segmented area within the cell body mask divided by the total area of the cell mask. (E) Fraction of monolayer with JAM-C-pH is defined as the proportion of pHluorin segmented area within the cell body mask divided by the total area of the cell mask. (F) Median JAM:JAM adhesion size is determined as the median area of JAM-SNAP segmented objects that overlap with JAM-C-pH segmented signal. Data points are reported as individual biological replicates statistically compared using Wilcoxon Rank Sum Tests. Statistical significance is represented by standard convention: ns p>0.05, * p<0.05, ** p<0.01, *** p<0.001, and **** p<0.0001.

JAM-C binding in cis or trans did not significantly alter other parameters of SNAP-pH adhesion.

(A-H) Quantification and statistical comparison of SNAP-pH adhesion parameters for (A-D) JAM-B-SNAP expressing glia and (E-H) JAM-C-SNAP expressing CGNs. (A,E) Total cell monolayer area is defined as the summed area of the cell body mask for biological replicates. (B,F) Fraction of monolayer with JAM-SNAP is defined as the proportion of SNAP segmented area within the cell body mask divided by the total area of the cell mask. (C,G) Fraction of monolayer with JAM-C-pH is defined as the proportion of pHluorin segmented area within the cell body mask divided by the total area of the cell mask. (D,H) Median JAM:JAM adhesion size is determined as the median area of JAM-SNAP segmented objects that overlap with JAM-C-pH segmented signal. Data points are reported as individual biological replicates statistically compared using Wilcoxon Rank Sum Tests. Statistical significance is represented by standard convention: ns p>0.05, * p<0.05, ** p<0.01, *** p<0.001, and **** p<0.0001.

JAM-C recognition of glial JAM-B recruits Pard3 in seconds and drebrin in minutes.

(A-D) Representative max projection images of JAM-pH, drebrin-SNAP, and Halo-Pard3 during JAM-C WT contact formation to JAM-B in glia acquired at 3 second intervals. Data is presented with individual channels represented in grayscale and merged data presented in color. (A) First sign of contact between the CGN and glia as a proportionally bright area of JAM-pH marked with orange arrowhead. (B) A second JAM contact forms after 12 seconds that directly overlaps with Halo-Pard3 marked by a gray arrowhead. (C) The two initial contacts fuse at 57 seconds with Halo-Pard3 coincidence. (D) At 5 minutes, more Halo-Pard3 has been recruited to the contact site and drebrin-SNAP is excluded to the contact periphery.

Δtrans JAM-C delays Pard3 recruitment and adhesion formation.

(A) Statistical comparison of Pearson correlation between JAM-pH and Halo-Pard3 at 10 minutes after first sign of adhesion. The Pearson correlation of each technical replicate is plotted within the violin plot. Pairwise statistical significance is calculated in relation to JAM-C WT replaced CGNs using Wilcoxon Rank Sum Tests. (B) Loess smoothed plot of adhesion size as a function of time after first sign of adhesion is shown. The dashed line at 5 minutes indicates the time point for subsequent statistical comparison. (C) Statistical comparison of adhesion size at 5 minutes after first sign of adhesion. The adhesion size of each technical replicate is plotted within the violin plot. The plots and statistical comparisons were generated for the combined technical replicates (n) from at least 3 independent biological replicates (N≥3) uniformly subsampled at 30 second intervals. Pairwise statistical significance is calculated in relation to JAM-C WT replaced CGNs using Wilcoxon Rank Sum Tests. Gray shading on the plots represents the 95% confidence intervals of the estimated mean values. Statistical significance is represented by standard convention: ns p>0.05, * p<0.05, ** p<0.01, *** p<0.001, and **** p<0.0001.

JAM contact formation on glia does not directly recruit drebrin within the contact site.

(A) Loess smoothed plot of drebrin-SNAP recruitment to glial JAM contacts measured by Pearson correlation for individual cells as a function of the time after first adhesion. The dashed line indicates the point in time used for statistical comparison. (B) Statistical comparison of Pearson correlation between JAM-pH and drebrin-SNAP at 5 minutes after first sign of adhesion. The Pearson correlation of each technical replicate is plotted within the violin plot. Pairwise statistical significance is calculated in relation to JAM-C WT replaced CGNs using Wilcoxon Rank Sum Tests. (C) Loess smoothed plot of drebrin-SNAP recruitment to glial JAM contacts measured by Pearson correlation for individual cells as a function of the segmented adhesion size. (D) Diagrammatic explanation of low Pearson correlation of the JAM-pH contact and drebrin. Since drebrin-SNAP borders the contact, regions of correlation and anticorrelation yield a correlation of 0. When zoomed in at the contact edge, the Pearson Correlation will trend towards 1. Plots and statistical comparisons were generated for the combined technical replicates (n) from at least 3 independent biological replicates (N≥3) uniformly subsampled at 30 second intervals. Gray shading on the plots represents the 95% confidence intervals of the estimated mean values. Statistical significance is represented by standard convention: ns p>0.05, * p<0.05, ** p<0.01, *** p<0.001, and **** p<0.0001.

Sensitivity analysis of correlation kernel indicates 9 pixels strikes a balance between sensitivity and noise.

(A-D) Representative time points in the timelapse data of the local correlation of JAM:drebrin used for empirical selection of kernel size. The first row in in each image shows the fluorescence of individual channels in grayscale and merge in color. The second row shows the raw data output of local correlation analysis at varied pixel window sizes indicated in the top left corner ranging from 3-15 pixels. The third row applies a manually determined segmentation mask to the raw correlation to restrict our analyses to the JAM contact as our feature of interest. The time frames included represent (A) prior to adhesion, (B) first sign of adhesion, (C) adhesion formation, and (D) stable adhesion.

Local correlation of JAM contacts reveals JAM:Pard3 correlation is mostly positive whereas JAM:drebrin has regions of positive and negative correaltion.

(A-D) Data of the proportional makeup of local correlation at WT JAM-C adhesions to glial JAM-B for all technical replicates (n=44) uniformly subsampled at 30 second intervals. Loess smoothed plots of the proportions of positive correlation (Pearson > 0.25), no correlation (-0.25<Pearson<0.25), or negative correlation of JAM:Pard3 as a function of (A) time after first adhesion and (B) adhesion area. Loess smoothed plots of the proportions of positive correlation, no correlation, or negative correlation of JAM:drebrin as a function of (C) time after first adhesion and (D) adhesion area. Gray shading on the plots represents the 95% confidence intervals of the estimated mean values.

JAM:Pard3 and JAM:drebrin sum correlation is associated with increasing JAM adhesion size.

(A-D) Data of the local recruitment of drebrin and Pard3 to JAM contacts based on JAM adhesion size. Loess smoothed plots of the sum of the (A) positive (Pearson > 0) and (B) negative (Pearson<0) JAM:Pard3 local correlation as a function of JAM adhesion area. Loess smoothed plots of the sum of the (C) positive (Pearson > 0) and (D) negative (Pearson<0) JAM:drebrin local correlation as a function of JAM adhesion area. The plots of sum correlation were generated from the combination of technical replicates (n) from at least 3 independent biological replicates (N≥3) uniformly subsampled at 30 second intervals. Gray shading on the plots represents the 95% confidence intervals of the estimated mean values.

Paired comparisons of sum local correlation highlight the role of intracellular interactions and JAM adhesion.

(A-C) Loess smoothed curves of the sum of positive JAM:Pard3 correlation as a function of time compared to JAM-C WT for (A) JAM-C Δcis/Δtrans, (B) ΔCyto, and (C) ΔPDZ. (D-F) Loess smoothed curves of the sum of positive JAM:drebrin correlation as a function of time compared to JAM-C WT for (D) JAM-C Δcis/Δtrans, (E) ΔCyto, and (F) ΔPDZ. (G-I) Loess smoothed curves of the sum of negative JAM:drebrin correlation as a function of time compared to JAM-C WT for (G) JAM-C Δcis/Δtrans, (H) ΔCyto, and (I) ΔPDZ. Loess smoothed plots were generated from the data of all technical replicates (n) as shown in plot legends. (J) Diagrammatic summary of correlation data. Gray circles indicate no change from WT JAM-C, red inverted triangles indicate decrease relative to WT JAM-C, and green triangles indicate increased value relative to WT JAM-C. Diagrams in the bottom row propose the molecular arrangement associated with each JAM construct. Gray shading on the plots represents the 95% confidence intervals of the estimated mean values.

Local correlation of JAM:drebrin but not JAM:Pard3 is spatially distribution to the outer 0.5 µm of JAM contacts.

(A-C) Example images used to calculate the spatial distribution of local correlation are shown. (A) Representative grayscale fluorescence images gamma adjusted by 0.7 at the first sign of adhesion, 1 minute after first adhesion, and 15 minutes after first adhesion. (B) Data associated with the fluorescent images in (A) showing the segmentation mask of JAM adhesion with the associated correlation maps for JAM:Pard3 and JAM:drebrin denoted by the subscript ‘A’. (C) Data associated with only the 0.5 µm periphery of the JAM adhesion in (B) denoted by the subscript ‘P’. (D-G) The deviation of proportional correlation from adhesion area reveals spatial distribution of Pard3 and drebrin for WT JAM-C adhering to glial JAM-B (n=44). Loess smoothed plots of the proportion of positive correlation or adhesion area within the 0.5 µm periphery of the JAM contact divided by the entire adhesion are shown as a function of (D) time after first adhesion or (E) adhesion size. Loess smoothed plots of the proportion of negative correlation or adhesion area within the 0.5 µm periphery for the JAM contact divided by the entire adhesion are shown as a function of (D) time after first adhesion or (E) adhesion size. (F-I) JAM-C mutants exhibit distribution of JAM:Pard3 and JAM:drebrin local correlation from the combined data for all technical replicates (n) indicated in associated plot legends. (F) Loess smoothed plot of the ratio of the sum adhesion area within 0.5 µm of adhesion edge divided by total adhesion area for JAM-C constructs to glial JAM-B is shown. (G-I) Plots show the associated deviation from peripheral adhesion area by subtracting the ratio from the proportional sum correlation. (G) Loess smoothed plot of the spatial distribution of JAM:Pard3 sum positive correlation as a function of time after first adhesion is shown. Loess smoothed plots of the spatial distribution are shown for JAM:drebrin (H) sum positive correlation and (I) sum negative correlation as a function of time after first adhesion. Gray shading on the plots represents the 95% confidence intervals of the estimated mean values.

CGN monolayers are preferential to glia for drebrin and Pard3 recruitment to JAM contacts even though JAM-C affects CGN-CGN adhesion size.

(A-B) Loess smoothed plots of the median correlation within segmented adhesions are shown for (A) JAM:Pard3 and (B) JAM:Drebrin as a function of time after first adhesion for WT JAM-C expressing CGNs adhering to monolayers of glia or CGNs expressing either JAM-C or JAM-B. (C) Loess smoothed plot of adhesion area are shown for WT JAM-C expressing CGNs adhering to monolayers of glia or CGNs expressing either JAM-C or JAM-B. (D) Statistical comparison of JAM adhesion size at 2.5 minutes after first adhesion. Pairwise statistical comparisons were performed as Wilcoxon rank sum comparisons in relation to either WT JAM-C to JAM-C CGN adhesion or WT JAM-C to JAM-B glia adhesion as indicated by the brackets. Loess plots and statistical comparisons were generated from the combination of all technical replicates (n) across at least 3 biological replicates replicates (N≥3). Gray shading on the plots represents the 95% confidence intervals of the estimated mean values. Statistical significance is represented by standard convention: ns p>0.05, * p<0.05, ** p<0.01, *** p<0.001, and **** p<0.0001.

Recruitment of Pard3 and drebrin is mostly driven by cell substrate rather than JAM receptor.

(A-D) Comparison of the sum correlation of JAM:Pard3 and JAM:drebrin for JAM-C expressing CGNs adhering to JAM-B or JAM-C on glia or CGNs as a function of time after first adhesion. Loess smoothed plots of JAM:Pard3 (A) sum positive correlation and (B) sum negative correlation. Loess smoothed plots of JAM:drebrin (C) sum positive correlation and (D) sum negative correlation. (E-H) Comparison of the sum correlation of JAM:Pard3 and JAM:drebrin for JAM-C expressing CGNs adhering to JAM-B or JAM-C on glia or CGNs as a function of adhesion area. Loess smoothed plots of JAM:Pard3 are shown for (E) sum positive correlation and (F) sum negative correlation. Loess smoothed plots of JAM:drebrin are shown for (G) sum positive correlation and (H) sum negative correlation. Loess plots were generated from the combination of all technical replicates (n) indicated in the plot legends. Gray shading on the plots represents 95% confidence intervals of the estimated mean values.

JAM contacts on glia elicit a stronger peripheral spatial distribution for drebrin than CGNs but not for Pard3.

(A) Loess smoothed plot of the ratio of the sum adhesion area within 0.5 µm of adhesion edge divided by total adhesion area for WT JAM-C expressing CGNs adhering to JAM-C or JAM-B on glia or CGN monolayers. (B-D) Plots show the associated deviation from peripheral adhesion area by subtracting the ratio from the proportional sum correlation. (B) Loess smoothed plot of the spatial distribution of JAM:Pard3 sum positive correlation as a function of time after first adhesion is shown for WT JAM-C expressing CGNs added to types of monolayers specified in plot legends. Loess smoothed plots of the spatial distribution are shown for JAM:drebrin (H) sum positive correlation and (I) sum negative correlation as a function of time after first adhesion is shown for WT JAM-C expressing CGNs added to the types of monolayers specified in plot legends. Gray shading on the plots represents the 95% confidence intervals of the estimated mean values.

JAM-B replacement in ex vivo slice cultures delays CGN iEGL entry and halts ML entry.

(A) Representative images at 24 and 48 hours of CGNs tracked in ex vivo slice cultures by live imaging slices electroporated to express either shJAM-C and WT JAM-C or shJAM-C and WT JAM-B. Time position of the tracked CGNs is color encoded as shown by the provided lookup table. (B-C) Plots of the layer occupancy of CGNs are shown from the proportions calculated for individual biological replicates (N) are shown. (B) Loess smoothed plot of the proportion of CGNs in the iEGL as a function of time is shown. (C) Loess smoothed plot of the proportion of CGNs that have entered the ML as a function of time is shown. (D-G) Plots of migration data for all cell migration events (n) aggregated across all the biological replicates (N) are shown. Loess smoothed plots of (D) migration speed and (E) instantaneous migration angle relative to the cerebellar surface as a function of time are shown. Loess smoothed plots of (F) migration speed and (G) instantaneous migration angle relative to the cerebellar surface distance from the cerebellar surface are shown. Gray shading on the plots represents the 95% confidence intervals of the estimated mean values. Scale bar = 50 µm.