MTs in pancreatic beta cells undergo extensive sliding driven by kinesin KIF5B.

(A) A subset of RNA-sequencing data from primary mouse beta cells showing highly expressed kinesins as indicated by mRNA counts. KIF5B (most-right bar, red data points) is the most abundant kinesin motor in this cell type. N=3. Note that this is a subset of the RNA sequencing sets published in (Sanavia et al. 2021). (B) Efficient depletion of KIF5B in MIN6 cells using two alternative shRNA sequences, as compared to a scrambled shRNA control. Based on immunofluorescent staining of KIF5B as in Supplemental Figure 1 (A-C). Fold decrease of fluorescence signal per cell normalized to cells w/o shRNA expression in the same field of view. N= 25-32 cells from 4 repeats. (C) Quantification of MT sliding FRAP assay in cells treated with scrambled control or one of the two KIF5B-specific shRNAs (see representative data in D-F). MT displacement is shown as area of MTs displaced into the bleached area after 5 minutes of recovery. One-way ANOVA test was performed for statistical significance (p-value <0.0001). N=9-20 cells per set. (D-F) Frames from representative FRAP live-cell imaging sequences. mEmerald-tubulin-expressing MIN6 cells. Inverted grayscale images of maximum intensity projections of spinning disk confocal microscopy stacks over a 1 µm-thick ventral cell layer. Scale bars 5µm. (D1-F1) Overview of the whole cell prior to photobleaching. (D2-F3) Enlarged areas from (D1-F1) immediately after photobleaching (D2-F2) and 5 minutes (300 seconds) after photobleaching (D3-F3). Light-blue dotted lines indicate the edges of the photobleached areas. Red arrows indicate MTs displaced into the bleached area. Scale bars, 5 µm (D-F, Figure 1-Video 1 “MT Sliding FRAP”). (G-I) MIN6 cells featuring fiducial marks at MTs due to co-expression of SunTag-KIF5B-560Rigor construct and Halo-SunTag ligand. Representative examples for scrambled control shRNA-treated cell (G), KIF5B shRNA #1-treated cell (H) and KIF5B shRNA #2-treated cell (I) are shown. Single slices by spinning disk confocal microscopy. Halo-tag signal is shown as inverted gray-scale image. Top panels show cell overviews (scale bars 5µm). Below, boxed insets (scale bars 2 µm) are enlarged to show dynamics of fiducial marks (color arrows) at 1-second intervals (1-5 seconds). 0- to 5-second tracks of fiducial mark movement are shown in the bottom panel, each track color-coded corresponding to the arrows in the image sequences. (J) Summarized quantification of stationary fraction of fiducial marks along MT lattice (5-second displacements below 0.15µm). Scrambled shRNA control N=1,421 tracks across 6 cells, shRNA#1 N=852 tracks across 5 cells, shRNA#2 N=2,182 tracks across 7 cells. P One-way ANOVA, p<0.001 (K) Summarized quantification of motile fraction of fiducial marks along the MT lattice (5-second displacements above 0.3µm).. Scrambled shRNA control N=2,066 tracks across 6 cells, shRNA#1 N=390 tracks across 5 cells, shRNA#2 N=412 tracks across 7 cells. P One-way ANOVA, p<0.001(G-I Figure 1-Video 2 “MT Sliding SunTag”).

(A-C) shRNA-based depletion of KIF5B in MIN6 cells. Cells expressing shRNA are detected by GFP expression (cyan in A1-C1, red arrows in A2-C2). Immunostained KIF5B signal (magenta in A1-C1 and grayscale in A2-C2) is significantly reduced by KIF5B shRNAs (B, C) and compared to scrambled control (A). See quantification in main Figure 1B. (D) Histogram of displacements over 5-second intervals for all fiducial marks in scrambled shRNA control vs KIF5B shRNA #1 and KIF5B shRNA #2.

Microtubule abundance and alignment at the cell periphery depend on KIF5B.

(A-C) MT organization in MIN6 cells expressing scrambled control shRNA (A), KIF5B-targeting shRNA #1 (B), or KIF5B-targeting shRNA #2 (C). Top, immunofluorescence staining for tubulin (grayscale, inverted). Bottom, immunofluorescence staining for KIF5B (cyan). Maximum intensity projection of 1µm at the ventral side of the cell. N=12. Scale bars: 5µm. (D) Quantification of mean tubulin intensity within the outer 2µm peripheral area of a cell, in data represented in (A-C). Mean values, black bars. One-way ANOVA, p<0.0001. N=7-9 cells. (E) Histograms of MT directionality within 1µm of cell boundary using perfected thresholds (see Figure 2 - Supplemental figure 3-4 for the analysis workflow and variants) in cells treated with scrambled control versus KIF5B-targeting shRNA. Data are shown for the summarized detectable tubulin-positive pixels in the analyzed single confocal slices of shRNA-treated cell population immunostained for tubulin, as represented in (F-H). Unpaired t-test were performed across each bin for all cells, and a K-S test was performed on the overall distribution. The share of MTs parallel to the edge (bin 0-10) is significantly higher in control as compared to KIF5B depletions. Pixel numbers in the analysis: SCR N=106,780 pixels across 9 cells, shRNA #1 N=71,243 across 7 cells, shRNA #2 N= 60,087 across 7 cells. (F-H) Representative examples of MT directionality analysis in single confocal slices of shRNA-treated cells immunostained for tubulin, as quantified in (E). Single laser scanning confocal microscopy slices. (F) Scrambled control shRNA-treated cell. (G) KIF5B shRNA#1-treated cell. (H) KIF5B shRNA#1-treated cell. Overviews of cellular MT networks are shown as threshold to detect individual peripheral MTs (see Figure 2- Supplemental Figure 3, panel A5). (F1-H2) Directionality analysis outputs of regions from yellow boxes in (F-H) are shown color-coded for the angles between MTs and the nearest cell border. (I) Color code for (F1-H2): MTs parallel to the cell edge, blue; MTs perpendicular to the cell edge, red.

Microtubule abundance and alignment at the cell periphery depend on KIF5B in primary β cells in mouse islets.

(A-B) MT organization in β cells within intact islets isolated from a wt (A) or KIF5B KO mice. Immunofluorescence staining for tubulin (grayscale, inverted). Maximum intensity projections of the laser scanning confocal microscopy stacks through the cell. N=10-11. Scale bars: 5µm. (C) Quantification of mean tubulin intensity within the outer 2µm peripheral area of a cell, in data represented in (A-B). Mean values, black bars. One-way ANOVA, p<0.0001. N=10-11 cells. (D-E) Representative examples of MT directionality analysis in single confocal slices of β cells from mouse islets immunostained for tubulin, as quantified in (F). Single laser scanning confocal microscopy slices. (D) Cells from wt mouse islets. (E) Cells from KIF5B KO mouse islets. Overviews of cellular MT networks are shown as threshold to detect individual peripheral MTs (see Figure 2- Supplemental Figure 3, panel A5). (D1-E2) Directionality analysis outputs of regions from yellow boxes in (D-E) are shown color-coded for the angles between MTs and the nearest cell border. (F) Color code for (D1-E2): MTs parallel to the cell edge, blue; MTs perpendicular to the cell edge, red. (G) Histograms of MT directionality within 1µm of cell boundary using perfected thresholds (see Figure 2 - Supplemental figure 3-4 for the analysis workflow and variants) in β cells from wt versus KIF5B KO mice. Data are shown for the summarized detectable tubulin-positive pixels in the analyzed single confocal slices of shRNA-treated cell population immunostained for tubulin, as represented in (D-E). Unpaired t-test were performed across each bin for all cells, and a K-S test was performed on the overall distribution. The share of MTs parallel to the edge (bin 0-10) is significantly higher in control as compared to KIF5B depletions. Pixel numbers in the analysis: wt N=180,709 pixels across 10 cells, KIF5B KO N=103,342 across 12 cells.

Overexpression of truncated KIF5B motor lacking cargo binding does not affect MT density at β cell periphery.

(A-D) MT organization in MIN6 cells expressing scrambled control shRNA (A-B) or KIF5B-targeting shRNA #2 (C-D). BFP-tagged kinesin-1 motor construct is ectopically expressed in (B, D). Top, immunofluorescence staining for tubulin (grayscale, inverted). Bottom, BFP immunofluorescence for motor expression (cyan). Maximum intensity projections of the laser scanning confocal microscopy stacks through the whole cell. N=8. Scale bars: 5µm.

Workflow of MT directionality analysis.

(A) Representation of the analysis workflow using a control DMSO-treated cell immunostained for tubulin as an example. (A1) An image of the original inverted grayscale confocal slice. (A2) A deconvolved image. (A3) Mask of the cell boundary. (A4) An image within the mask after application of standard % threshold. (A5) An image within the mask after application of a threshold optimizing detection of peripheral MTs for a particular cell. (A3-A5) are derivatives of (A2). (A6) A schematic illustrating map of distances from the nearest cell (mask) border per pixel (not to scale). (A7) A schematic illustrating map of angles per pixel (not to scale). A6 and A7 can be produced from A4 or A5. (A8) Color-coded output map of MT directionalities. A8 is a derivative of A7.

Illustration of thresholding variations and their influence on the output analysis.

(A) KIF5B-depleted cell as an example of contrast/threshold optimization for different MT densities and corresponding outputs. Upper row, contrast is adjusted for distinction of single MTs at the cell periphery. Lower row, contrast is adjusted for distinction of single MTs at the cell center. (A1) A deconvolved inverted grayscale confocal slice image contrasted to highlight peripheral MTs. (A2) The same deconvolved inverted grayscale confocal slice image contrasted to highlight central MTs. (A3) The image from A1 after application of a threshold optimized to highlight peripheral MTs. (A4) The image from A1 after application of a standard % threshold. (A5) Analysis outcome from (A3). (A6) Analysis outcome from (A4). (B) Histograms of MT directionality within 1um of cell boundary using standard thresholds (see Supplemental figure 3/1 for the analysis workflow) in cells treated with scrambled control versus KIF5B-targeting shRNA. Data are shown for the summarized detectable tubulin-positive pixels in the analyzed shRNA-treated cell population, as represented in (F-H). Unpaired t-test were performed across each bin for all cells, and a K-S test was performed on the overall distribution. The share of MTs parallel to the edge (bin 0-10) is significantly higher in control as compared to KIF5B depletions. Pixel numbers in the analysis: SCR N=61,437 pixels across 9 cells, shRNA#1 N=15,215 pixels across 7 cells, shRNA#2 N= 21,125 pixels across 7 cells. (C) Histograms of MT directionality presented in this Figure panel B with depicted outliers. (D) Histograms of MT directionality presented in Figure 2 panel E with depicted outliers. (E) Histograms of MT directionality presented in Figure 2 – Supplemental Figure 1 panel G with depicted outliers.

MT sliding is facilitated through the ATP-independent MT-binding domain of kinesin-1.

(A) Schematic of kinesin-1 (KIF5) and the Dominant Negative (KIFDN) and heterodimerization strategy. Top schematic shows full length KIF5s, consisting of the motor domain, stalk coil-coil domain and the tail. Three constructs utilized here include (1) The KIF5C motor domain tagged with a blue fluorescent protein (BFP) and the FRB for heterodimerization; (2) KIFDNwt construct with KIF5B Tail domain tagged with the mCherry fluorescent protein and the FKBP for heterodimerization. (3) KIFDNmut construct is the same as (2) but features a set of point mutations (magenta) making the ATP-independent MT-binding domain unable to bind MT lattice. (B) Quantification of MT sliding in FRAP assay in cells subjected to DN construct expression and heterodimerization. MT displacement is shown as area of MTs displaced into the bleached area after 5 minutes of recovery. See representative data (C-G). N= 5-25 per condition. One-way ANOVA test was performed for statistical significance (p-value <0.0001; ns, non-significant). (C-G”) Frames from representative FRAP live-cell imaging sequences of MIN6 cells expressing mEmerald-tubulin. Inverted grayscale images of maximum intensity projections over 1 µm-thick stacks by spinning disk confocal microscopy. (C1-G1) The first frame after photobleaching. (C2-G2) A frame 5 minutes (300 seconds) after photobleaching. Light-blue dotted lines indicate the edges of the photobleached areas. Red arrows indicate MTs displaced into the bleached area. Scale bars, 5 µm. (C3-G3) Schematics of experimental manipulation: green represents MTs, blue represents endogenous KIF5B, magenta represents KIFDNwt, purple represents KIFDNmut, gray represents KIF5C motor, orange bracket represents heterodimerizing agent (rap, rapalog). Conditions: (C1-C3) Untreated control. Only endogenous KIF5B is present. (D1-D3) KIFDNwt overexpression. Endogenous KIF5B is unable to bind MTs. (E1-E3) KIFDNmut overexpression. It does not bind MTs and does not interfere with endogenous KIF5B. (C-E, Figure 3-Video 1 “KFDN FRAP”) (F1-F3) KIFDNwt and KIF5C motor overexpression plus rapalog treatment. Heterodimerization creates a large pool of motors capable of MT sliding. (G-G”) KIFDNmut and KIF5C motor overexpression plus rapalog treatment. Heterodimerization creates a large pool of the motor non-functional in MT sliding (F-G, Figure 3-Video 2 “KFDN + Motor FRAP”).

Effects of ATP-independent MT-binding domain of KIF5B on microtubule abundance and alignment at the β-cell cell periphery.

(A-C) MT organization in MIN6 cells expressing (B) KIFDNwt and KIF5C motor heterodimerized via rapalog treatment, (C) KIFDNmut and KIF5C motor heterodimerized via rapalog treatment and compared to a control cell with no ectopic expressions (A). Top, immunofluorescence staining for tubulin (grayscale, inverted). Blue dotted line indicates the borders of a cell expressing constructs of interest. Bottom in A, f-actin (phalloidin, red) and DAPI (blue). Bottom in B and C, ectopically expressed mCherry-labeled KIFDN constructs (magenta) and BFP-labeled KIF5C motor (green). Laser scanning confocal microscopy maximum intensity projection of 1µm at the ventral side of the cell. Scale bars: 5um. (D) Quantification of mean tubulin intensity within the outer 2µm peripheral area of a cell, in data represented in (A-C). Mean values, black bars. One-way ANOVA, p<0.0001. N=7-15 cells. (E) Histograms of MT directionality within 1um of cell boundary (see Supplemental figure 2-1 for the analysis workflow) in control cells compared to cells expressing heterodimerized KIFDN variants. Data are shown for the summarized detectable tubulin-positive pixels in the analyzed cell population, as represented in (F-H). Unpaired t-test were performed across each bin for all cells, and a K-S test was performed on the overall distribution. The share of MTs parallel to the edge (bin 0-10) is significantly higher in control as compared to the over-expressions. NT control N=138,810 pixels across 9 cells, KIFDNwt + motor N= 48,285 pixels across 9 cells, KIFDNmut + motor N= 40,832 pixels across 10 cells. (F-H) Representative examples of MT directionality analysis quantified in (E). (F) Control cell, no ectopic expressions. (G) Cell expressing KIFDNwt+ Motor. (H) Cell expressing KIFDNmut+ Motor. Overviews of cellular MT networks are shown as threshold to detect individual peripheral MTs (see Supplemental figure 2-1 panel A5). (F1-H2) Directionality analysis outputs of regions from yellow boxes in (F-H) are shown color-coded for the angles between MTs and the nearest cell border (see Supplemental figure 2-1 panel A8). (I) Color code for (F1-H2): MTs parallel to the cell edge, blue; MTs perpendicular to the cell edge, red.

DN effect of KIF5B tail domain on MT abundance at the β- cell cell periphery requires ATP-independent MT-binding domain.

(A-B) MT organization in MIN6 cells expressing (A) KIFDNwt, (B) KIFDNmut, Top, immunofluorescence staining for tubulin (grayscale, inverted). Blue dotted line indicates the borders of a cell expressing constructs of interest. Bottom, ectopically expressed mCherry-labeled KIFDN constructs (magenta). Laser scanning confocal microscopy maximum intensity projection of 1µm at the ventral side of the cell. Scale bars: 5um. (C) Quantification of mean tubulin intensity within the outer 2µm peripheral area of a cell, in data represented in (A-B) as compared to untreated controls (see Figure 4A). Mean values, black bars. One-way ANOVA, p<0.0001. N=7-11 cells.

Influence of thresholding variations on the output analysis of MT directionality in Figure 4 F-H.

(A) Histograms of MT directionality within 1um of cell boundary using standard thresholds (see Supplemental figure 2-1 for the analysis workflow) in cells expressing ectopic heterodimerized KIF5C motor. Data are shown for the summarized detectable tubulin-positive pixels in the analyzed shRNA-treated cell population, as represented in (Figure 4 F-H). Unpaired t-test were performed across each bin for all cells, and a K-S test was performed on the overall distribution. The share of MTs parallel to the edge (bin 0-10) is significantly higher in control as compared to heterodimerized KIF5B over-expression. Pixel numbers in the analysis: NT control N=120,959 pixels across 9 cells, KIFDNwt + motor N= 26,110 pixels across 7 cells, KIFDNmut + motor N= 41,091 pixels across 10 cells. (B) Histograms of MT directionality presented in this Figure panel A with depicted outliers. (C) Histograms of MT directionality presented in main Figure 4 panel E with depicted outliers.

Enhanced MT sliding results in loss of peripheral MT alignment at the border.

(A-B) MT organization in MIN6 cells pretreated with DMSO and Kinesore respectively. Immunofluorescence staining for tubulin (grayscale, inverted). Laser scanning confocal microscopy maximum intensity projection of 1µm at the ventral side of the cell. Scale bars: 5um. (C) Quantification of mean tubulin intensity within the outer 2µm peripheral area of a cell, in data represented in (A-B). Mean values, black bars. One-way ANOVA, p<0.0001. N=10 cells per condition. Histograms of MT directionality within 1um of cell boundary (see Supplemental figure 2-1 for the analysis workflow) in DMSO treated control cells compared to kinesore treated cells. Data are shown for the summarized detectable tubulin-positive pixels in the analyzed cell population, as represented in (D-E). Unpaired t-test were performed across each bin for all cells, and a K-S test was performed on the overall distribution. The share of MTs parallel to the edge (bin 0-10) is significantly higher in control as compared to the over-expressions. DMSO control N=136,840 pixels across 10 cells, kinesore treated N= 87,361 pixels across 9 cells. (D-E) Representative examples of MT directionality analysis quantified in (F). Directionality analysis outputs of regions from yellow boxes in (D-E) are shown color-coded for the angles between MTs and the nearest cell border (see Supplemental figure 2-1 panel A8). (G) Color code for (D1-E2): MTs parallel to the cell edge, blue; MTs perpendicular to the cell edge, red.

Influence of thresholding variations on the output analysis of MT directionality in Figure 5 D-E.

(A) Histograms of MT directionality within 1um of cell boundary using standard thresholds (see Supplemental figure 2-1 for the analysis workflow) in cells treated with DMSO or kinesore. Data are shown for the summarized detectable tubulin- positive pixels in the analyzed respective cell population, as represented in (Figure 5 D-E). Unpaired t-test were performed across each bin for all cells, and a K-S test was performed on the overall distribution. The share of MTs parallel to the edge (bin 0-10) is significantly higher in control as compared to kinesore treated. Pixel numbers in the analysis: DMSO control N= 94,841 pixels across 9 cells, kinesore N= 29,796 pixels across 9 cells. (B) Histograms of MT directionality presented in this Figure panel A with depicted outliers. (C) Histograms of MT directionality presented in main Figure 5 panel F with depicted outliers.

Kinesore has no effect in MT network in KIF5B-depleted cells.

(A-F) Representative examples of MT organization in MIN6 cells expressing scrambled control shRNA (A, D), KIF5B-targeting shRNA #1 (B, E), or KIF5B-targeting shRNA #2 (C, F). A DMSO-treated control (A-C) and kinesore-treated cells (D-F) are shown. Immunofluorescence staining for tubulin (grayscale, inverted). Maximum intensity projections of the laser scanning confocal microscopy stacks through the whole cell. N=8 cells/condition. Scale bars: 5µm.

MT sliding in β cells is stimulated by glucose.

(A-B) Frames from representative FRAP live-cell imaging sequences of MT sliding response to glucose stimulation. mEmerald-tubulin-expressing MIN6 cells. Inverted grayscale images of maximum intensity projections over 1 µm-thick stacks by spinning disk confocal microscopy. (A) A cell pretreated with 2.8mM glucose before the assay. (B) A cell pretreated with 2.8mM glucose and stimulated with 20 mM glucose before the assay. (A1-B1) The first frame after photobleaching. (A2-B2) A frame 5 minutes (300 seconds) after photobleaching. Light-blue dotted lines indicate the edges of the photobleached areas. Red arrows indicate MTs displaced into the bleached area. Scale bars, 5 µm. (C) Quantification of MT sliding FRAP assay in cells in 2.8mM versus 20mM glucose (see representative data in A-B). MT displacement is shown as area of MTs displaced into the bleached area after 5 minutes of recovery. One-way ANOVA test was performed for statistical significance (p-value <0.0001). N=16-24 cells per set (A-B Figure 6-Video 1 “ FRAP Low and High Glucose”). (D-E) MIN6 cells featuring fiducial marks at MTs due to co-expression of SunTag-KIF5B-560Rigor construct and Halo-SunTag ligand. Representative examples for cells in 2.8mM glucose (D) and a cell stimulated by 20mM glucose (E) are shown. Single-slice spinning disk confocal microscopy. Halo-tag signal is shown as inverted gray-scale image. Top panels show cell overviews (scale bars 5µm). Below, boxed insets are enlarged to show dynamics of fiducial marks (color arrows) at 1 second intervals (1-5 seconds). 0- to 5-second tracks of fiducial mark movement are shown in the bottom panel, each track color-coded corresponding to the arrows in the image sequences. N=6-11 cells (A-B Figure 6-Video 2 “ SunTag Low and High Glucose”). (F) Histogram of all 5-second displacement of fiducial marks in Low vs High glucose (G) Summarized quantification of stationary fiducial marks along MT lattice (5-second displacements below 0.15µm). Low glucose N=5,615 tracks across 11 cells, High glucose 5min N=2,259 tracks across 6 cells, High Glucose 20min N=3,059 tracks across 6 cells. P One-way ANOVA, p<0.001 (H) Summarized quantification of moving fiducial marks along the MT lattice (5-second displacements over 0.3µm). Low glucose N=2,595 tracks across 11 cells, High glucose 5min N=2,642 tracks across 6 cells, High Glucose 20min N=4,049 tracks across 6 cells. One-way ANOVA, p<0.001.

Schematic of the main results and predictions.

(A) Role of KIF5B-dependent MT sliding in MT density at the cell periphery. (B) Role of KIF5B-dependent MT sliding in the alignment of peripheral MTs. (C) Potential role of glucose-facilitated KIF5B-dependent MT sliding in forming temporal openings in the peripheral MT array by moving fragmented MTs aside or MT looping. (D) Potential role of glucose-facilitated KIF5B-dependent MT sliding in repair of the peripheral MT array, partially destabilized/fragmented downstream of glucose. Track MTs, green. Cargo MTs, blue. Kinesin-1, black. Direction of kinesin movement shown as solid arrow. Subsequent steps of the process shown as dashed double arrows.