Polarized microtubule remodeling transforms the morphology of reactive microglia and drives cytokine release

Abstract

Microglial reactivity is a pathological hallmark in many neurodegenerative diseases. During stimulation, microglia undergo complex morphological changes, including loss of their characteristic ramified morphology, which is routinely used to detect and quantify inflammation in the brain. However, the underlying molecular mechanisms and the relation between microglial morphology and their pathophysiological function are unknown. Here, proteomic profiling of lipopolysaccharide (LPS)-reactive microglia identifies microtubule remodeling pathways as an early factor that drives the morphological change and subsequently controls cytokine responses. We find that LPS-reactive microglia reorganize their microtubules to form a stable and centrosomally-anchored array to facilitate efficient cytokine trafficking and release. We identify cyclin-dependent kinase 1 (Cdk-1) as a critical upstream regulator of microtubule remodeling and morphological change in-vitro and in-situ. Cdk-1 inhibition also rescues tau and amyloid fibril-induced morphology changes. These results demonstrate a critical role for microtubule dynamics and reorganization in microglial reactivity and modulating cytokine-mediated inflammatory responses.

Fig. 1. Reactive microglia show similar morphological changes in late-stage pathology and acute systemic stimulation models.

a Immunostaining of Iba1 in murine hemibrain of wild-type mouse. Cortex and hippocampal regions of interest that were subsequently quantified are highlighted in red boxes. b–d Single-cell segmentation workflow of microglia in mouse brain sections as shown in a. All steps are shown for regions highlighted in yellow boxes in a. Immunostaining and segmented cell bodies are shown in (b). Watershed-based segmentation individual ramified microglia cells and overlay with original staining is shown in c and d in higher magnification for the areas highlighted in red. Arrowheads in d show cell branches stemming from cells located outside the section stained that are purposefully omitted in the segmentation result. e Montage of representative cell morphologies illustrating ramified and ameboid cell morphologies along with a selection of their morphological descriptors. f Immunostaining for Iba1 and single-cell segmentation results in hippocampi of 6- and 12-month-old Tau^WT and Tau^P301S mice. g Quantification of ramification index of microglia measured in (f) per age and averaged per animal, n = 14, 16, 5, 5 animals per group. h Immunostaining for Iba1 and single-cell segmentation results in cortex (Ctx) and hippocampus (HC) of 6-month-old Trem2^WT and Trem2^KO animals crossed into PS2APP amyloidosis model. i Quantification of ramification index of microglia measured in h per brain region and averaged per animal, n = 15 animals per group. j Immunostaining for Iba1 and single-cell segmentation results of WT mice injected with LPS i.p. at timepoints indicated. k Quantification of ramification index of microglia measured in j averaged per animal, n = 5 animals per group. Scale bars are 500 µm in a, 100 µm in (b, c, f, h, j) and 30 µm in (d). Boxplots show all datapoints, median, 25th and 75th percentile, whiskers are 1.5*IQR. Statistical significance was calculated with unpaired, two-sided t-tests in (g, i, k). Significance intervals p: ****<1e−04 <***<0.001 <**<0.01 <*<0.05 <ns. Source data are provided as a Source Data file.

Fig. 2. An in vitro model of microglial LPS-reactivity recapitulates morphological changes seen in situ and demonstrates dependence on dynamic microtubules.

a Representative images of primary microglia stained with CellMask dye and single-cell segmentation results after indicated stimulations for 24 h. b Quantification of cell size and ramification index of segmented results shown in (a) averaged per well relative to untreated cells, n = 32, 32, 8, 24, 16, 24 wells from 4 independent cultures. c Representative images of primary microglia stained with CellMask dye and single-cell segmentation results in a LPS stimulation time course. d Quantification of cell size and ramification index of segmented results shown in (c) averaged per well relative to untreated cells, n = 23, 23, 23, 15, 22 wells from 3 independent cultures. e Representative images of microglia stained with CellMask dye and single-cell segmentation results after treatment with LPS and microtubule poisons Taxol and Nocodazole for 24 h. f Quantification of cell size and cell perimeter of segmented cells shown in (e) averaged per well relative to untreated cells, n = 22, 22, 6, 22 wells from 4 independent cultures. g Quantification of cytokines secreted into the supernatant of microglial cultures over 24 h after LPS and Taxol or Nocodazole treatment. Measurements out of detection range are indicated by n.d. n = 3 replicates. Scale bars are 100 µm. Bars indicate mean ± SD in (g), scatters show average values per well in (b, d, f) and replicates in (g). Boxplots show all datapoints, median, 25th and 75th percentile, whiskers are 1.5*IQR. Statistical significance was calculated with one-way ANOVA and Tukey HSD in (b, d, f, g). Significance intervals p: **** <1e−04 <*** <0.001 <** <0.01 <* <0.05 <ns. Source data are provided as a Source Data file.

Fig. 3. Phosphoproteomic analysis of LPS-reactive microglia highlights extensive regulation cytoskeletal proteins.

a Experimental outline: primary murine microglia were purified and exposed to LPS for a time course of 30 min to 24 h before lysis, TMT-multiplexing and proteomic and phosphoproteomic analysis. b Quantification of significantly up- and downregulated proteins in proteomic analysis of microglia treated with LPS. For full table see Supplementary Data 1. c, d Gene ontology analysis of significantly upregulated proteins for cellular component and biological process after 8 h LPS treatment. For full table see Supplementary Data 2. e Volcano plots of significantly down- and upregulated proteins at indicated timepoints highlighting proteins associated with “Inflammatory response” (green) and selected inflammatory markers (purple). Non-significantly regulated proteins are omitted. GO: Gene oncology. f Volcano plot as in e, but highlighting proteins associated with “Microtubules” or “Microtubule organizing centers” (green) and selected examples (orange). g Number of significantly (de-) phosphorylated proteins after LPS treatment at indicated timepoints. For full table see Supplementary Data 1. h, i Gene ontology analysis of significantly upregulated peptides for cellular component and biological process in the 8 h LPS treatment. For full table see Supplementary Data 2. j Volcano plot of de-phosphorylated and phosphorylated peptides after 0.5 h LPS treatment highlighting proteins associated with “Microtubules” (blue) or “Microtubule organizing centers” (green). Non-significantly regulated peptides are omitted. k Volcano plot of de-phosphorylated and phosphorylated peptides at indicated timepoints highlighting phosphorylated peptides of Map4 (red) and Stmn1 (teal). Non-significant peptides are omitted. Source data are provided as a Source Data file.

Fig. 4. LPS-reactive microglia establish a radial microtubule array through centrosome activation and loss of Golgi-associated nucleation.

a Super-resolution immunostaining for alpha-tubulin. Regions in yellow dashes at centrosome and periphery are shown enlarged. b Immunostaining for microtubule plus-tip marker EB1 and minus-end marker Camsap2. c Quantification of microtubule phenotypes. n = 58, 76, 76, 67, 84, 43, 58, 54, 54 cells per time point. d Top: MT + TIP tracking in microglia treated with LPS over 5 min. Bottom: a selection of tracks and their direction: inward = blue, outward = orange. Red dot marks position of MTOC. Asterisks indicate tracks turning at plasma membrane. See Supplementary Video 1. e Quantification of MT + TIP track direction, speed and number shown in (d). n = 31, 31, 27, 30 cells from 3 independent cultures. f Immunostaining for γ-tubulin and pericentrin. g Left: Quantification of γ-tubulin and pericentrin levels at the centrosome and in the cytoplasmic periphery. Right: Quantification of centrosome/cytoplasm ratio. n = 134, 407, 107, 84 cells from 3 independent cultures. h Immunostaining for α-tubulin and Golgi marker GM-130 after treatment with nocodazole for 1 h and 1 min recovery in microglia treated with LPS for 24 h. Arrowheads point to microtubules associated with Golgi structures. See Supplementary Video 2. i Quantification of α-tubulin intensity at centrosome in microglia treated with or recovered from nocodazole treatment as in (h). n = 7, 109, 82 cells from 3 independent cultures. j Quantification of number of non-centrosomal microtubules and golgi-associated microtubules after recovery from nocodazole treatment. n = 92, 77, (left) 59, 48 (right) cells from 4 independent cultures. Scale bars are 5 µm in a; 10 µm in (b, d, f, h). Bars indicate mean ± SE in (i, j); mean ± SD in (e, g). Statistical significance was calculated with one-way ANOVA and Tukey HSD in (e, g), one-way ANOVA and Dunnett’s in (i), and unpaired, two-sided t-tests in (j). Significance intervals p: **** <1e−04 <*** <0.001 <** <0.01 <* <0.05 <ns. Source data are provided as a Source Data file.

Fig. 5. Microglial LPS stimulation results in microtubule polymerization and stabilization through Stmn1 and Map4 pathways.

a Representative immunoblot for α-tubulin and posttranslational modifications. Quantification is shown for n = 3 independent experiments. b Immunostaining for acetylated tubulin. c Quantification of Stmn1 protein levels after LPS treatment compared to control, n = 2, 3, 3, 2, 3 replicates. d Quantification of Stmn1 phosphorylation on S25 and S38 after LPS treatment, n = 2, 3, 3 replicates. e Representative immunoblot and quantification of tubulin spin-down assay of microglia treated as indicated or expressing Stmn1^2xSA. Quantification shows fraction of polymerized tubulin for n = 5, 6, 4, 6, 6 assays from 3 independent cultures. f MT + TIP tracking for 5 min in microglia treated with LPS expressing STMN1^2xSA color coded for their average speed. g Quantification of track characteristics in (f). n = 26, 26, 25, 25 cells from 3 independent cultures. h Quantification of Map4 S1046 and S914 phosphorylation, in MT binding domains, after LPS treatment, n = 2, 3, 3 replicates. i Immunostaining for Map4. j Representative immunoblot and quantification of tubulin spin-down assay of microglia treated with LPS and siMap4, n = 6 assays from 3 independent cultures. k Quantification MT + TIP track characteristics in microglia treated with siMap4 and LPS. n = 28, 31, 25, 30 cells from 4 independent cultures. l Immunostaining for acetylated and tyrosinated tubulin quantification for n = 57, 81, 44, 95 cells from 2 independent cultures. Scale bars are 10 µm. Bars indicate mean ± SE in (a, e, g, j, k, l) and mean ± SD in (c, d, h). Statistical significance was calculated with one-sided ANOVA and Dunnett’s in (a), and Tukey HSD in (c, d, e, g, h, j, k) and unpaired, two-sided t-tests in (l). Significance intervals p: **** <1e−04 <*** <0.001 <** <0.01 <* <0.05 <ns. Source data are provided as a Source Data file.

Fig. 6. Cdk1 activation is required for microglial reactivity and rearrangement of microtubule cytoskeleton.

a Quantification of Cdk1 T14 and Y15 phosphorylation after LPS treatment, n = 2, 3, 3 replicates. b Representative images of CellMask-stained microglia and single-cell segmentation. c Quantification of cell size, ramification index and perimeter from (b), n = 22, 22, 20 wells from 4 independent cultures. d Immunostaining and quantification of γ-tubulin at the centrosome. n = 191, 201 cells from 4 cultures. e Quantification of polymerized shown in Fig. 5e. n = 5, 6, 6 assays from 2 independent cultures. f Quantification of MT + TIP track characteristics in microglia treated with RO-3306 and LPS. n = 26, 26, 25, 25 cells from 5 independent cultures. g Immunostaining and quantification of acetylated tubulin. n = 102, 82 cells from 4 independent cultures. h Immunostaining and quantification of Map4. n = 85, 82 cells from 4 independent cultures. i Up- and downregulated proteins after 8 and 24 h of LPS and RO3-306 treatment compared to LPS alone. Proteins associated with “inflammatory response” (green) and selected examples (red) are highlighted. Non-significantly regulated proteins are omitted. j Immunostaining and quantification of TNFα. n = 52, 40 cells from 2 independent cultures. Right panel quantifies radius of a circle enclosing 80% of the TNFα signal. n = 39, 31 cells from 2 independent cultures. k Quantification of cytokines secretion by microglial cultures, n = 2 replicates. Scale bars are 50 µm in (b); 10 µm in (d, g, h, j). Bars indicate mean ± SE in (a, c, d, e, f, g, h, j); mean ± SD in (k). Boxplots show all datapoints, median, 25th and 75th percentile, whiskers are 1.5*IQR. Statistical significance was calculated with one-sided ANOVA and Tukey HSD in (a, c, e, f), and unpaired two-tailed t-tests in (d, g, h, j). Significance intervals p: **** <1e−04 <*** <0.001 <** <0.01 <* <0.05 <ns. Source data are provided as a Source Data file.

Fig. 7. Cdk1 inhibition prevents morphological remodeling of LPS-reactive microglia in ex-vivo brain tissue.

a Experimental outline: 300 µm thick acute slices of Cx3CR1^wt/GFP brains were incubated in treatments indicated in ASCF for 4 h before fixation and 2-photon microscopy. b Representative 3D segmentation result of GFP signal from 150 µm thick image stacks of indicated treatments. Cells are color coded by their volume. Grid spacing is 20 µm in xy and 40 µm in z. See also Supplementary Video 3. c Quantification of cell volumes relative to control condition in acute slices shown in (b). Graph shows mean ± SE for n = 13 acute sections recorded in 3 ROIs each from 5 mice. Symbols indicate datapoints originating from the same animals. Statistical significance was calculated with ANOVA and Tukey HSD. Source data are provided as a Source Data file. Significance intervals p: **** <1e−04 <*** <0.001 <** <0.01 <* <0.05 <ns. d Example 3D segmentation of cells from b with measured volumes indicated. Color-coding as in (b). Scale bar is 5 µm. e Working model of microglial inflammatory stimulation and resulting MT reorganization as proposed in this paper.