Redox signaling by hydrogen peroxide modulates axonal microtubule organization and induces a specific phosphorylation signature of microtubule proteins distinct from distress

15 Many life processes are regulated by physiological redox signal s, referred to as oxidative 16 eustress. However, excessive oxidative stress can damage biomolecules and contribute to 17 disease. The neuronal microtubule system is critically involved in axon homeostasis, regulation 18 of axonal transport, and neurodegenerative processes. However, whether and how physiological 19 redox signals affect axonal microtubules is largely unknown. Using live cell imaging and super-20 resolution microscopy, we show that subtoxic concentrations of the central redox metabolite 21 hydrogen peroxide increase axonal microtubule dynamics, alter the structure of the axonal 22 microtubule array , and affect the efficiency of axonal transport. We report that t he 23 mitochondria-targeting antioxidant SkQ1 and the microtubule stabilizer EpoD abolish the 24 increase in microtubule dynamics. We found that oxidative eustress and distress specifically 25 modulate the phosphorylation state of the microtubule system and induce a largely non-26 overlapping phosphorylation pattern of MAP1B as the main target. Cell-wide phosphoproteome 27 analysis revealed that different signaling pathways are inversely activated by oxidative eustress 28 and distress. Signaling via casein kinase (CK2) and pyruvate dehydrogenase kinases (PDK) is 29 activated during eustress and signaling via mammalian target of rapamycin (mTOR) and 30 serum/glucocorticoid-regulated protein kinase (SGK) is activated during distress. The results 31 suggest that the redox metabolite and second messenger hydrogen peroxide induces rapid and 32 local reorganization of the microtubule array in response to mitochondrial activity or as a 33 messenger from neighboring cells by activating specific signaling cascades. 34 35

microtubules; tau; axon; redox signalling; hydrogen peroxide. 36 37 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted July 4, 2024. ; https://doi.org/10.1101/2024.07.01.601594doi: bioRxiv preprint 3

38 Microtubules are a crucial filament system involved in virtually all aspects of a neuron's life. In 39 the axon, they have a characteristic organization consisting of an array of relatively short 40 microtubules, the plus-end of which is oriented towards the distal tip (Penazzi et al., 2016a) . 41 Such organization is thought t o be cr itical for efficient axonal transport, in which cargo is 42 carried to the distal axon over long distances. Although axonal microtubules are relatively 43 stable, they still show considerable dynamics, with the plus ends of the microtubules exhibiting 44 stochastic changes between growth and shrinkage , referred to as dynamic instability. 45 Microtubule stability and dynamics are regulated by a variety of tubulin - and microtubule -46 interacting proteins , including microtubule nucleators, microtubule -binding proteins, end -47 binding proteins, tubulin -sequestering proteins and microtubule -severing proteins. Further 48 complexity arises from the presence of a variety of different tubulin isoforms, where humans 49 have eight − and nine −tubulin genes, most of which are expressed in neurons (Uhlen et al., 50 2015). Many of the components of the microtubule system have extensive posttranslational 51 modifications (PTMs) that modulate their activity. The best studied PTM of the microtubule 52 system is phosphorylation , and increased phosphorylation of specific serine and threonine 53 residues is known to influence the interaction of neuronal microtubule -associated proteins 54 (MAPs) such as MAP1B and tau , with microtubules. Of disease -related importance is the 55 modulation of the interaction of the MAP tau with axonal microtubules, as tau is 56 hyperphosphorylated in Alzheimer’s disease and other tauopathies , which largely reduces its 57 interaction with microtubules (Arendt et al., 2016). 58 Oxidative stress is associated with many neurodegenerative diseases , and increased levels of 59 reactive oxygen species (ROS) may play a role in triggering axonal degeneration and 60 microtubule disassembly (Pratico et al., 2002). High levels of ROS form additional free radical 61 compounds that have the potential to damage lipids, proteins, and nucleic acids over time 62 (Vincent et al., 2009). Indeed, high levels of ROS have been shown to inhibit axonal transport 63 (Fang et al., 2012) and induce axonal degeneration, as evidenced by the morphological features 64 of axon beading and fragmentation (Fukui et al., 2011). However, more recently it has become 65 clear that moderately elevated ROS levels can act as a crucial physiological mediator of many 66 cellular processes (Sinenko et al., 2021). This led to the distinction of “oxidative distress” as a 67 mechanism leading to pathology and “oxidative eustress” as an important physiological 68 regulator of intrinsic signalling pathways (Wilson et al., 2018). 69 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted July 4, 2024. ; https://doi.org/10.1101/2024.07.01.601594doi: bioRxiv preprint 4 The most common ROS members in mediating oxidative eustress are the superoxide anion, the 70 hydroxyl radical, and hydrogen peroxide. Because these ROS species act as relatively short -71 lived second messengers, they are able to regulate intrinsic signalling pathways (Wilson et al., 72 2018). Hydrogen peroxide was found to be the most important redox metabolite involved in 73 signal transduction and redox regulation. It is an electrically neutral molecule and is chemically 74 more stable than the other members of the ROS family. As a messenger molecule, hydrogen 75 peroxide diffuses through cells and tissues to trigger immediate cellular effects that link redox 76 biology with regulation through phosphorylation and dephosphorylation (Sies, 2017) . 77 Physiological hydrogen peroxide levels promote the establishment of neuronal polarity, neurite 78 growth and axon specification (Wilson et al., 2015) . While all of these processes cr itically 79 depend on the regulation of microtubule polymerization, the mechanisms of how hydrogen 80 peroxide influences axonal microtubules and microtubule-dependent functions under oxidative 81 eustress conditions are largely unknown. 82 Here, we determined the effect of subtoxic concentrations of hydrogen peroxide , reflecting 83 oxidative eustress, on axonal microtubule dynamics using quantitative live -cell imaging. We 84 analyzed changes in the structure of the axonal microtubule array using single-molecule 85 localization microscopy and algorithm -based reconstruction of microtubules. We examined 86 possible functional microtubule-dependent consequences by tracking individual APP vesicles 87 and determining the tau-microtubule interaction using fluorescence-decay after photoactivation 88 (FDAP) experiments. To identify downstream targets of increased ROS , we performed 89 proteomic and phosphoproteomic analysis of differentiated neur ons treated with subtoxic 90 hydrogen peroxide and compared them with arsenite treatment as an inducer of oxidative 91 distress-mediated toxicity. 92 93

94

95 Unless otherwise stated, c hemicals and cell culture material were purchased from Sarstedt 96 (Nümbrecht, Germany), Sigma -Aldrich (Deisenhofen, Germany), and Thermo -Fisher 97 Scientific (Waltham, USA). The microtubule-targeting agent epothilone D (EpoD) was a kind 98 gift from Amos Smith 3rd (University of Pennsylvania) and was prepared as previously 99 described (Lee et al., 2001; Rivkin et al., 2004) . The purity of the compound was > 95%, as 100 determined by LC-MS and NMR analyses. The spectroscopic properties were identical to those 101 reported in the literature. The mitochondria-targeted antioxidant SkQ1, and TPP, which lacks 102 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted July 4, 2024. ; https://doi.org/10.1101/2024.07.01.601594doi: bioRxiv preprint 5 the antioxidant quinone moiety, were provided by Maxim Skulachev (Mitotech S.A., 103 Luxembourg). 104 105 Cell culture and transfections 106 PC12 cells were cultured in serum -DMEM and transfections were performed with 107 Lipofectamine 2000 (Thermo -Fisher Scientific, USA) as previously described (Fath et al., 108 2002). Expression plasmids for PAGFP-α-Tubulin and mEGFP-α-Tubulin have been described 109 previously (Conze et al., 2022b) . The expression plasmid PAGFP-htau441wt (Gauthier-110 Kemper et al., 2011) was used to express the longest CNS-isoform of tau. pEGFP-n1-APP was 111 obtained from Zita Balklava and Thomas Wassmer (Addgene plasmid #69924; 112 http://n2t.net/addgene:69924; RRID: Addgene_69924). 113 114 Metabolic activity and cytotoxicity profiling using a combined LDH and MTT Assay 115 PC12 cells were cultured in 96 -well plates at 1 ×104 cells/well in serum-reduced medium 116 supplemented with 100 ng/m l 7S mouse NGF. Cells were incubated for 48 hours to initiate 117 neuronal differentiation. Concentration -response profiling of hydrogen peroxide (H2O2) was 118 performed with four triplicates and six concentrations ranging from 50-900 µM. The respective 119 H2O2 concentration was added to each well 3 hours before the measurement. Designated wells 120 for controls were untreated for negative control or a total cell lysis was induced by the addition 121 of 1% Triton-X for a positive control. For the LDH assay, 50 µl medium from each well of the 122 assay plate was transferred to a separate 96-well plate. For quantification of the LDH release, 123 50 µl of LDH reagent (4 mM iodonitrotetrazolium chloride (INT), 6.4 mM beta -nicotinamide 124 adenine dinucleotide sodium salt (NAD), 320 mM lithium lactate, 150 mM of 1 -125 methoxyphenazine methosulfate (MPMS) in 0.2 M Tris -HCl buffer, pH 8.2) was added. The 126 plate was shaken for 10 seconds and incubated in the dark for several minutes. Absorbance was 127 measured at 490 nm using a Thermomax Microplate Reader operated with SoftMaxPro Version 128 1.1 (Molecular Devices Corp., Sunnyvale, CA, U.S.A.). LDH release measurements were 129 normalized to the optical densities of the positive control wells. For the MTT assay, MTT 130 reagent (3,(4,5.dimethylthiazol-2-yl)2,5-diphenyltetrazolium bromide) at a final concentration 131 of 1 mg/ml MTT was added to the wells in the remaining assay plate. The cells were incubated 132 for a further 2 hours in the cell culture incubator before the reaction was stopped by addi ng 133 50 μl of lysis buffer (20% (wt/vol) sodium dodecyl sulfate in 1:1 (vol/vol) N,N -134 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted July 4, 2024. ; https://doi.org/10.1101/2024.07.01.601594doi: bioRxiv preprint 6 dimethylformamide/water, pH 4.7). After overnight incubation at 37 °C, the optical densities 135 of the formazan product were determined at 570 nm. MTT conversion measurements were 136 normalized to the optical densities of the negative control wells. 137 138 Detection of intracellular ROS levels 139 Intracellular ROS level s were assessed using the Cellular ROS Assay Kit Orange (abcam, 140 ab186028, Cambridge, UK). PC12 cells (2 ×104 cells/well) were seeded in 60 μl of serum -141 reduced DMEM supplemented with 100 ng/ml 7S mouse NGF in a 96-well plate. To measure 142 the effect of SkQ1, 30µl of NGF containing serum-reduced medium supplemented with 0.2 µM 143 SkQ1 or carrier control (0.2% EtOH) were added after 24 hours. After a further 24 hours of 144 incubation, the cells were treated with the respective H2O2 concentration (150 µM and 450 µM) 145 for 3 hours. A total of six independent experiments w ere conducted. Intracellular ROS level s 146 were determined according to the manufacturer’s instructions. Briefly, cells were incubated 147 with 100 µl of ROS Orange Working solution for 60 min . Changes in fluorescence intensity 148 were measured with the FLUOstar Optima (BMG Labtechnologies, Ortenberg, Germany) at 149 Ex/Em = 540/570 nm. The fluorescence intensities were normalized to the respective negative 150 control values after background subtraction. 151 152 Live-cell imaging 153 Fluorescence decay after photoactivation (FDAP) experiments with PAGFP -α-tubulin- or 154 PAGFP-tau-expressing cells were performed essentially as described previously (Niewidok et 155 al., 2016). Briefly, PC12 cells were plated on 35-mm poly-L-lysine and collagen-coated glass-156 bottom culture dishes (MatTek, USA). After transfection, PC12 cells were neuronally 157 differentiated by media exchange with DMEM with 1% (vol/vol) serum containing 100 ng/ml 158 7S mouse NGF (Alomone Laboratories, Germany). Cultivation was continued for 4 days with 159 medium exchange to DMEM with 1% (vol/vol) serum containing NGF and without phenol red 160 one day prior to live imaging. Live imaging of PC12 cells for photoactivation experiments was 161 performed using a Nikon Eclipse Ti2 -E laser scanning microscope (Nikon, Japan) equipped 162 with a LU-N4 laser unit with 488-nm and 405-nm lasers and a Fluor 60× ultraviolet-corrected 163

lens (NA 1.4) enclosed in an incubation chamber at 37°C and 5% CO 2. Automated 164 image acquisition of PAGFP-tubulin- or PAGFP-tau-expressing cells after photoactivation was 165 performed essentially as described previously (Igaev et al., 2014). Briefly, photoactivation of a 166 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted July 4, 2024. ; https://doi.org/10.1101/2024.07.01.601594doi: bioRxiv preprint 7 6 μm long neurite segment was performed with a 405 -nm laser. A set of consecutive image 167 series (time stacks) was created at a frequency of 1 frame/s and 112 frames were collected per 168 activated cell with a resolution of 256×256 pixels. 169 Live-cell imaging to analyze APP transport in PC12 cells was performed essentially as 170 previously described (Conze et al., 2022a) using a Zeiss Cell Observer Z1 (Zeiss, Germany) 171 with high -speed confocal imaging using the CSU -X1 spinning disc technology from 172 Yokogawa, equipped with an optically pumped 488-nm semiconductor laser surrounded by a 173 sample incubation chamber controlled by a Zeiss TempModule S1. pH stability was maintained 174 by adding 30 mM HEPES buffer to the cell culture medium before image acquisition. For image 175 capture of eGFP-APP vesicle tracking, the 488-nm laser and Alpha Plan-Apochromat ×63 (NA 176 1.46) objective were used. An image series (time stack) of 60 s was acquired at a speed of five 177 images per second, captured with a Hamamatsu ORCA flash V3 and 2×2 binning at a resolution 178 of 640×640 pixels (pixel size 172 nm). The image series were deconvolved with the software 179 Huygens Remote Manager v3.5 (Scientific Volume Imaging B.V., Hilversum, Netherlands) 180 using a theoretical PSF and a classical maximum likelihood estimation as the deconvolution 181 algorithm with 20 iterations and a quality criterion of 0.05. 182 Treatment of PC12 cells with 150 µM H 2O2 was performed 3 hours before live imaging. To 183 assess the effect of the antioxidant SkQ1 in FDAP experiments, cells were treated with 0.2 µM 184 of the compound 24 hours before H2O2 addition. To determine the effect of H2O2 treatment on 185 stabilized MTs by epothilone D (EpoD) in FDAP experiments as well as APP vesicle tracking, 186 cells were treated with 1 nM EpoD or its carrier control (0.01% DMSO) one hour before H2O2 187 treatment. 188 189 FDAP data analysis 190 A reaction-diffusion model was used to determine the association rate k*on and the dissociation 191 rate k off constant of tubulin or tau binding , as described previously (Igaev et al., 2014) . 192 Processing and analysis of individual FDAP curves was performed as previously described 193 (Niewidok et al., 2016) using a custom C-based tool called cFDAP. The fitting procedure was 194 used to obtain k *on and koff from each individual FDAP curve, while the χ2 value was used as 195 an indicator of the goodness of fit of the model function. Effective diffusion constants were 196 obtained by fitting fluorescence decay data from photoactivation experiments using a one -197 dimensional diffusion model function for FDAP, as previously described (Weissmann et al., 198 2009). 199 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted July 4, 2024. ; https://doi.org/10.1101/2024.07.01.601594doi: bioRxiv preprint 8 200 Tracking and quantifying axonal transport 201 APP transport was tracked and analyzed using Imaris v 9.2 (Bitplane, Oxford Instruments). 202 Vesicles were detected by the Gaussian -filtered intensity of their signal within a diameter of 203 500 nm around the respective signal peak. Vesicle transport was tracked by an autoregressive 204 motion algorithm with a maximum distance of a future position of the signal of 1 µm and a 205 maximum gap size of 2 frames (400 ms). To determine the direction of the tracked vesicles, a 206

point at the transition from the cell body to the neurite was used, while only the tracks 207 that spanned a period of at least 3 s (15 frames) were considered for further analysis. Quantified 208 transport parameters were exported to spreadsheets (Excel, Microsoft Corporation, USA) and 209 further processed using TIBCO Spotfire Data Analysis Software (TIBCO Software Inc., USA). 210 Vesicles that exceeded a displacement of more than 0.75 µm over the observation time of 60 s 211 were designated as mobile, otherwise as stationary. 212 213 Super-resolution microscopy with DNA-PAINT 214 For super -resolution microscopy, PC12 cells were transfected with mEGFP -α-Tubulin and 215 neuronally differentiated for 4 days as described above. Fixation, immunostaining, and super-216 resolution microscopy with DNA -PAINT were performed essentially as previously described 217 (Conze et al., 2022b) . Briefly, PC12 cells were processed using a combined NP -40 218 permeabilization-fixation protocol, which removes membranes and cytosolic components but 219 preserves cytoskeletal structures and associated proteins (Brandt et al., 1995). Immunostaining 220 used an anti -GFP nanobody obtained from the Massive -Tag-Q Anti -GFP DNA -PAINT kit 221 (Massive Photonics GmbH, Gräfelfing, Germany). Just before imaging, cells were washed three 222 times with PBS and incubated with a 1:1000 dilution of 50 nm gold nanorods (Nanopartz, E12-223 50-600-25) that function as fiducial markers during image acquisition. Afterwards, cells were 224 washed three times with PBS before an imaging buffer (Massive Photonics) supplemented with 225 250 pM Cy3b-conjugated imager DNA-strand, complementary to the DNA strand attached to 226 the anti-GFP nanobody, was added. Cells were imaged by total internal reflection fluorescence 227 (TIRF) microscopy using an inverted microscope frame (Olympus IX -81) equipped with a 228 motorized quad-line TIR illumination condenser (cellTIRF-4-Line, Olympus) and a motorized 229 xy-stage (Märzhäuser Scan IM 120x80). Three -dimensional single-molecule localization was 230 achieved by astigmatic imaging using a cylindrical lens (Olympus) implemented directly in 231 front of the filter wheel. 232 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted July 4, 2024. ; https://doi.org/10.1101/2024.07.01.601594doi: bioRxiv preprint 9 233 Post-processing of DNA-PAINT datasets 234 Raw data sets were processed using the "Picasso" software package (Schnitzbauer et al., 2017), 235 https://github.com/jungmannlab/picasso). At the beginning of each imaging session, a z -stack 236 of immobilized fluorescent TetraSpeck™ microspheres with a diameter of 100 nm (Invitrogen, 237 T7279) was acquired. This was done in the respective imaging buffer with a step size of 10 nm 238 using a piezo z -stage (NanoScanZ, NZ100, Prior Scientific). This is a necessity for the 239 calibration of astigmatic PSFs, which is a prerequisite for three -dimensional single-molecule 240 localization microscopy. This z-stack was analyzed using “Picasso: Localize” with parameters, 241 that allow the identification of single beads in each frame. The photon conversion parameters 242 were set as follows: EM Gain: 1, Baseline: 400, Sensitivity 0.46, Quantum Efficiency: 0.80 and 243 pixel size: 130 nm. A calibration file was generated with the "Calibrate 3D" function of "Picasso 244 localize". This file, as well as the same photon conversion parameters, were used for image 245 processing of raw sample files. The Min. Net. Gradient was adjusted to remove non-specifically 246 bound imager strands or other background signals to filter for localizations with the highest 247 signal intensities. Single-molecule localizations were fitted with a Gaussian least-square fit. For 248 3D localization, the magnification factor was set to 1.0. Processed data sets were opened with 249 "Picasso: Render" and a drift correction with cross -correlation was performed, followed by a 250 correction using the fiducials. The localizations of these datasets were then exported for the 251 ImageJ plugin ThunderSTORM for 3D rendering to obtain image -stacks with a well -defined 252 voxel size set to 26 nm×26 nm×25 nm based on the overall axial resolution of 25 nm (Ovesny 253 et al., 2014). 254 255 SIFNE analysis of post-processed DNA-PAINT data 256 Computational analysis and quantification of the microtubule array in axon-like processes of 257 PC12 cells was performed using the open-source SIFNE (SMLM image filament extractor) 258 software package (Zhang et al., 2017) . The MATLAB -based tool involves the iterative 259 extraction of the filamentous structures from the image data set and the subsequent 260 identification and assignment of the detected filaments. Since the axial distance between 261 microtubules in PC12 neurites is about 70 nm (Jacobs and Stevens, 1986) , we chose optical 262 sections with a thickness of 150 nm for filament extraction by SIFNE. This allowed us to use 263 sections where the resolution was highest while avoiding overlapping microtubules from 264 another pla ne that would confound our statistics. The region of interest (ROI) in axon -like 265 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted July 4, 2024. ; https://doi.org/10.1101/2024.07.01.601594doi: bioRxiv preprint 10 processes of PC12 cells was set to the shaft of the neurite. For image enhancement using line 266 and orientation filter transformation algorithms (LFT and OFT), a radius of 10 pixels with 40 267 rotations of the scan line segment was used. S egmentation was performed using SIFNE's 268 automatic thresholding function based on Otsu's method. For creating a pool of minimal linear 269 filament fragments, areas of filament connections were removed by a local area around each 270 connection of 2×2 pixels. To recover unrecognized linear structures, iterative extraction of the 271 filaments was performed twice, followed by registration of the propagation direction of each 272 filament tip. The grouping and analysis of the detected filaments was carried out with a pixel 273 size of 26 nm and a maximum curvature of 1 rad/µm. The search angle and radius were set to 274 60° and 40 pixels, respectively. The allowable orientation difference between the endpoints was 275 set to 60°. The maximum allowable angle difference and endpoint gap vector were set to 60° 276 and 30°, respectively. The weights for similarity and continuity conditions during the scoring 277 calculations were set to 1. Due to the high complexity of the cytoskeletal network, fragment 278 overlap was not allowed as suggested by (Zhang et al., 2017). For sorting composite filaments, 279 the minimum filament length was set to 15 pixels corresponding to 390 nm, while ungrouped 280 filaments were left in the data set. 281 282 Proteome and phosphoproteome analysis 283 PC12 cells were neuronally differentiated by culture in serum-reduced DMEM with 100 ng/ml 284 7 S mouse NGF for 4 days and treated with 150 µM H2O2 for 3 hours, 0.5 mM arsenite for 20 285 min, or left as respective controls. Proteome and phosphoproteome analysis w ere essentially 286 performed as described previously (Pinzi et al., 2024) . Briefly, cells were incubated in lysis 287 buffer (8M urea in 50mM Tris/HCl, pH 7.8) supplemented with Phos-Stop tablets (Roche 288 Diagnostics GmbH, Germany), sonicated and cleared by centrifugation. Protein concentrations 289 were determined using Pierce™ BCA Protein Assay (Thermo Fisher Scientific, USA). 0.2 μg/μl 290 α-casein was added to a protein amount of 1.2 mg, and reduction and alkylation were carried 291 out in lysis buffer containing 15 mM iodoacetamide and 5 mM DL-dithiothreitol. The proteins 292 were digested with trypsin/Lys-C Mix (Promega Corporation, USA), and 10 μg of each sample 293 was used for proteomic analysis. For phosphoenrichment, the remaining samples were desalted 294 using Sep -Pak® Classic C18 cartridges (Waters, Ireland) . The eluate was lyophilized and 295 enriched with a High-Select™ TiO2 Phosphopeptide Enrichment Kit (Thermo Fisher Scientific, 296 USA). For proteome and phosphoproteome analysis, samples were collected from a PepMap 297 C18 easy spray column (Thermo Fisher Scientific, USA). MS analysis was performed as 298 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted July 4, 2024. ; https://doi.org/10.1101/2024.07.01.601594doi: bioRxiv preprint 11 previously described (Schoppe et al., 2020). The *.raw data files were analyzed using PEAKS 299 Online software (Bioinformatic Solutions Inc, Canada). PEAKS Q (de -novo-assisted 300 quantification) analysis was used for data refinement with mass correction, de novo sequencing 301 and de novo-assisted database search, and subsequent label -free quantification. The search 302 engine was applied to Rattus norvegicus *.fasta databases. MS/MS searches were performed 303 using a mass tolerance of 10 ppm parent ions and a mass tolerance of 0.2 Da fragments. Trypsin 304 with up to two missing cleavage sites was selected as the cleavage enzyme. The 305 carbamidomethylation modification was chosen as the fixed modification and the oxidation of 306 methionine, the acetylation of lysine and the phosphorylation of serine, threonine and tyrosine 307 as variable modifications. A maximum of three variable modifications were allowed per 308 peptide. Normalization to the total ion current level was performed for each sample. The 309 ANOVA test was used to calculate the significance level for each protein, outliers were 310 removed, and the top three peptides were used to quantify the protein signal where possible. 311 The peptide identification was considered valid at a false detection rate of 1% (q-value < 0.001) 312 (maximum delta Cn of the percolator was 0.05). The minimum length of acceptable identified 313 peptides was set to six amino acids. Each condition was analyzed in triplicate. All proteins were 314 assigned their gene symbol via the Uniprot knowledge database (http://www.uniprot.org/). 315 316 Bioinformatic analysis 317 The components of the microtubule system highlighted on the volcano plots were based on the 318 classification of the different functional groups of the microtubule system in (Trushina et al., 319 2019b): structure proteins (α - and β -tubulins), nucleators (γ -tubulins and γ -tubulin complex 320 proteins (GCPs)), MT-binding proteins (MAP1A, MAP1B, MAP1S, MAP2, tau (encoded by 321 the MAPT gene), MAP4, MAP6 (STOP) and MAP7 (ensconsin)), tubulin-sequestering proteins 322 (stathmins, encoded by STMN1, STMN2 (SCG10), STMN3 (SCLIP), STMN4 (RB3)), end -323 binding proteins (EB1, EB2, EB3 (encoded by MAPRE1, MAPRE2, MAPRE3, respectively), 324 CLASP1 and CLASP2), MT -severing proteins (P60 -katanin (encoded by KATNA1 and 325 KATNB1), fidgetin (FIGN) and spastin (SPAST)). Gene Ontology (GO) term enrichment 326 analysis was performed to identify overrepresented Molecular Function (MF) and Cellular 327 Component (CC) categories using EnrichR (Xie et al., 2021) . All differentially expressed 328 proteins and all proteins with upregulated phosphosites were used to create gene sets for 329 enrichment analysis. The 10 highest -scoring GO -terms for molecular function and cellular 330 component were displayed. Gene Ontology (GO) regarding molecular function and cellular 331 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted July 4, 2024. ; https://doi.org/10.1101/2024.07.01.601594doi: bioRxiv preprint 12 component was used to construct of protein classifications. Kinase enrichment analysis (KEA) 332 was performed u sing the curated kinase -substrate database KEA2 (Lachmann and Ma'ayan, 333 2009). The gene symbols with phosphorylated sites of differentially phosphorylated proteins 334 were used to infer upstream kinases whose putative substrates are overrepresented. To identify 335 the most important functional pathways upon treatment with hydrogen peroxide or arsenite, 336 significantly enriched kinases were compared. 337 338 Statistical analysis 339 Statistical analysis was performed using GraphPad Prism v8.0.1 (GraphPad Software, USA). 340 All datasets were tested for normality using the D'Agostino-Pearson and Shapiro -Wilk tests. 341 When necessary, datasets were log transformed to allow further statistical testing. Statistical 342 outliers were identified using the ROUT method. Homogeneity was assessed using the Levene 343 test. An unpaired two -tailed t -test was used to compare two datasets. In cases of unequal 344 variances, Welch’s correction was applied. To compare more than two data sets, one-way 345 ANOVA was performed followed by Dunnet post-hoc test. All statistical values are expressed 346 as mean ± SEM. 347 348

349 Subtoxic concentrations of hydrogen peroxide decrease microtubule polymer in axon-like 350 processes by decreasing kon and increasing koff rates of microtubule polymerization. 351 The redox metabolite hydrogen peroxide is known to act as a messenger molecule and diffuses 352 through cells and tissues (Sies, 2017) . While high concentrations of hydrogen peroxide are 353 considered toxic (“oxidative distress”), low levels of hydrogen peroxide have been shown to 354 promote neuronal development and axon specification (Wilson et al., 2015) , thus positively 355 modulating cellular function s through a process called “oxidative eustress” (Wilson et al., 356 2018). Therefore, we first determined the level of subtoxic concentrations of hydrogen peroxide 357 in model neurons, which are likely to induce eustress mechanisms in the cells. 358 We used a combined MTT and LDH assay to determine the metabolic activity and cytotoxicity 359 profile of exogenously added hydrogen peroxide on differentiated neuronal cells. We observed 360 that a short (3 hour) treatment with 150 µM hydrogen peroxide affected neither MTT 361 conversion, as a measure of metabolic activity, nor LDH release, as a measure of toxicity (Fig. 362 1A). In contrast, an increase to 225 µM hydrogen peroxide induced both a decrease in MTT 363 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted July 4, 2024. ; https://doi.org/10.1101/2024.07.01.601594doi: bioRxiv preprint 13 conversion and an increase in LDH release, both of which became highly significant at 450 µM. 364 Treatment with subtoxic 150 µM hydrogen peroxide increased intracellular ROS level by 365 ~30%, while toxic 450 µM hydrogen peroxide resulted in an increase of ~70% (Fig. 1B). 366 In further experiments , we therefore decided to focus on treat ing neuronal cells with the 367 subtoxic concentration of 150 µM hydrogen peroxide, which results in only a moderate increase 368 in intracellular ROS levels , likely reflecting “oxidative eustress” . Because neuronal 369 development and function cr itically depend on the regulation of microtubule polymerization, 370 we determined the effect of hydrogen peroxide on axonal microtubule dynamics using a 371 previously established live cell imaging approach. The method is based on fluorescence decay 372 after photoactivation (FDAP) measurements of model neurons transfected to express PAGFP -373 tagged α-tubulin (PAGFP-α-tubulin). Changes in microtubule dynamics were then monitored 374 by FDAP measurements on cells in which the fluorescence of PAGFP was focally activated in 375 the middle of an axon -like process (Fig. 1C). FDAP curves showed that hydrogen peroxide 376 treatment resulted in increased fluorescence decay, indicating a lower proportion of 377 polymerized tubulin (Fig. 1D). Notably, application of a reaction -diffusion model resulted in 378 greatly reduced k*on, indicating that hydrogen peroxide induces microtubule depolymerization 379 (Fig. 1E). The significant increase in koff indicates higher microtubule dynamics as it reflects a 380 shortened residence time of the tubulin dimers in the polymer. T he amount of polymerized 381 tubulin in the axon-like processes was reduced by ~10% from more than 8 0% to ~70% (Fig. 382 1F). 383 Taken together, the data indicate that subtoxic concentrations of hydrogen peroxide modulate 384 the properties of the axonal microtubule array by reducing the proportion of microtubule 385 polymer and increasing their dynamics. 386 387 The mitochondria-targeted antioxidant SkQ1 and the microtubule stabilizer epothilone D 388 prevent hydrogen peroxide-induced microtubule depolymerization. 389 A major source of ROS under physiological conditions is mitochondria. Through their NAD+ 390 pool, t hey play a predominant role in defining cellular responses to stress. While the 391 mitochondrial NAD +/NADH redox ratio is maintained separately, it appears to be strongly 392 connected to the cytosolic NAD+ pool and responsive to changes in its perturbations (Hu et al., 393 2021). Therefore, we asked whether targeting the redox status of mitochondria would influence 394 hydrogen peroxide-mediated modulation of microtubule dynamics. 395 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted July 4, 2024. ; https://doi.org/10.1101/2024.07.01.601594doi: bioRxiv preprint 14 To modulate the mitochondrial redox state, we used the mitochondria -targeted antioxidant 396 SkQ1. SkQ1 is a derivative of plastoquinone, a potent antioxidant (Antonenko et al., 2008) and 397 has already been used in preclinical studies to treat cardiovascular and renal diseases (Fedorov 398 et al., 2022). Indeed, treatment with SkQ1 completely abolished the hydrogen-peroxide induced 399 increase in intracellular ROS levels (Fig. 1G). The same was true for the hydrogen peroxide -400 induced modulation in microtubule polymer, where pretreatment with SkQ1 completely 401 abolished the effect of hydrogen peroxide on microtubule polymer reduction in axon -like 402 processes (Fig. 1H, I). In contrast, TPP lacking the antioxidant quinone moiety had no effect 403 (Fig. 1I) confirming the antioxidant activity of SkQ1 in preserving the state of axonal 404 microtubules. 405 If the cytosolic redox state indeed modulates axonal microtubule dynamics, thereby leading to 406 reduced microtubule polymer, a microtubule stabilizer may prevent this effect. To test this 407 hypothesis, we used the well-characterized microtubule-targeting agent (MTA) epothilone D 408 (EpoD). This small molecule microtubule stabilizer binds to the β -tubulin subunit on the 409 luminal surface of microtubules, induces tubulin polymerization similar to paclitaxel (Buey et 410 al., 2004) , and reduces microtubule-dependent spine loss in an Alzheimer’s disease model 411 already at subnanomolar concentration s (Penazzi et al., 2016b) . In fact, treatment with 412 nanomolar EpoD abolished the decrease in microtubule polymer caused by hydrogen peroxide 413 (Fig. 1J, K). 414 Thus, the data indicate that the mitochondrial redox state modulates microtubule dynamic s in 415 the axonal compartment. Furthermore, they show that microtubule -stabilizing drugs ove rride 416 the modulation of microtubule dynamics by the cytosolic redox state. 417 418 Subtoxic hydrogen peroxide modulates the structure of the axonal microtubule array by 419 reducing microtubule mass and increasing microtubule length. 420 Axonal microtubules are not continuous but are present in an array of relatively short, uniformly 421 oriented fragments . In developing mammalian axons, average microtubule s are only a few 422 micrometers long (Bray and Bunge, 1981; Yu and Baas, 1994) . The question therefore arises 423 as to whether hydrogen peroxide also changes the microtubule arra ngement in axon s, for 424 example their length distribution or their physical properties. 425 To determine microtubule organization in axon -like processes, we used single molecule 426 localization microscopy (SMLM) using DNA -PAINT (Point Accumulation in Nanoscale 427 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted July 4, 2024. ; https://doi.org/10.1101/2024.07.01.601594doi: bioRxiv preprint 15 Topography) (Jungmann et al., 2010), followed by algorithm-based filament extraction (Zhang 428 et al., 2017) (Fig. 2A). We have previously validated the approach in axon-like processes of 429 PC12 cells and dendrites of primary neurons (Conze et al., 2022b). As expected from the FDAP 430 data, hydrogen peroxide resulted in reduced microtubule polymer , as evidenced by lower 431 microtubule mass and density (Fig. 2B, left). Interestingly, hydrogen peroxide led to an increase 432 in mean microtubule length (Fig. 2B, middle), suggesting a trend toward fewer but longer 433 microtubules in the processes. The physical properties of the microtubules, indicated by their 434 straightness, did not change (Fig. 2B, right). To determine how hydrogen peroxide affect s 435 microtubule length, we plotted the length distribution in a relative frequency histogram (Fig. 436 2C). The data show that hydrogen peroxide increases mean microtubule length primarily by 437 reducing the proportion of short microtubules (0.5-2.5 µm length) and increasing the number 438 of long microtubules (>8 µm). 439 Thus, the data indicate that hydrogen peroxide alters the organization of microtubule arrays in 440 axon-like processes by reducing their mass and increasing their length, but does not affect their 441 physical properties. 442 443 Hydrogen peroxide-induced changes in the microtubule array reduce the proportion of 444 mobile vesicles but have no effect on the speed and velocity of axonal transport. 445 A change in axonal microtubule arra ngement may have a direct impact on the properties of 446 axonal transport, as C. elegans motor neurons have been shown to pause axonal transport at the 447 ends of microtubules before switching to a new polymer (Yogev et al., 2016) . This suggests 448 that altering microtubule tracks can affect the efficiency of axonal transport. 449 To determine whether the hydrogen peroxide-induced change in the microtubule array affects 450 axonal transport parameters, we used single-vesicle tracking of eGFP-tagged amyloid precursor 451 protein (APP), a key axonal transport cargo (Morotz et al., 2019) (Fig. 3A). Hydrogen peroxide 452 treatment reduced the proportion of mobile vesicles by a pproximately 20% compared to a 453 control (Fig. 3B, left ), indicating a reduction in the total amount of cargo transported. To 454 quantify the movement of mobile vesicles, we determined the effect of hydrogen peroxide on 455 velocity (displacement per time) and speed (run length per time) from the trajectories of each 456 cell analyzed (Fig 3B, middle and right). Both the velocity and speed of APP-vesicles were the 457 same under both conditions, indicating that the change in microtubule length distribution had 458 no effect on the transport of the vesicles on the microtubule tracks once they were moving. 459 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted July 4, 2024. ; https://doi.org/10.1101/2024.07.01.601594doi: bioRxiv preprint 16 To confirm that the observed reduction in the proportion of mobile vesicles is due to the change 460 in the organization of the microtubule array, we determined the effect of the microtubule 461 stabilizer EpoD. We previously observed that nanomolar concentrations of EpoD shift ed the 462 length distribution of microtubules in axon -like processes toward shorter and denser 463 microtubules (Conze et al., 2022b). Indeed, EpoD abolished the effect of hydrogen peroxide on 464 reducing the fraction of mobile vesicles (Fig. 3C), suggesting that the effect was caused by the 465 altered organization of the microtubule array. 466 Thus, the data indicate that hydrogen peroxide has the potential to influence axonal transport 467 by reducing the total amount of cargo transported, likely due to its effect on modulating the 468 microtubule array. 469 470 Hydrogen peroxide-induced changes in the microtubule array do not affect the dynamic 471 interaction of the axonal tau protein with microtubules. 472 Tau protein is an axonal ly enriched microtubule -associated protein (MAP) th at is th ought to 473 regulate axonal microtubule polymerization. Pathological changes in tau are involved in a class 474 of neurodegenerative diseases collectively referred to as tauopathies (Arendt et al., 2016) . 475 Under physiological conditions , tau dynamically interacts with microtubules and exhibits a 476 “kiss and hop” behavior that is likely required to regulate microtubule polymerization without 477 affecting the efficiency of axonal transport (Janning et al., 2014) . Indeed, post -translational 478 modifications of tau that make tau microtubule interaction less dynamic can lead to axonal 479 transport defects that cause dendritic atrophy in tauopathies (Bakota and Brandt, 2024; Conze 480 et al., 2022a). 481 Therefore, hydrogen peroxide could induce changes in t he interaction of tau with axonal 482 microtubules, which may affect microtubule dynamics , cause the observed reduction in 483 microtubule polymer , and impair microtubule-dependent cargo transport. We used FDAP 484 experiments to determine a po ssible change in the dynamics of tau interaction with 485 microtubules in axon-like processes of the model neurons. Tau was N -terminally tagged with 486 photoactivatable GFP (PAGFP) and expressed exogenously in PC12 cells , which were 487 differentiated to a neuronal phenotype. Following focal activation of tau in hydrogen peroxide-488 treated or control cells, tau showed similar dissipation from the activation area (Fig. 3D, left). 489 Application of a refined reaction -diffusion model of the tau-microtubule interaction allowed 490 the determination of the binding constants (k*on and koff rates) of the tau microtubule-interaction 491 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted July 4, 2024. ; https://doi.org/10.1101/2024.07.01.601594doi: bioRxiv preprint 17 (Igaev et al., 2014) . We did not detect any difference in either kinetic constant as a result of 492 hydrogen peroxide treatment (Fig. 3D, right). 493 Thus, the data indicate that hydrogen peroxide does not affect the dynamics of tau-microtubule 494 interaction in axon -like processes. Consequently, the data also indicates that the increase in 495 microtubule dynamics as a result of hydrogen peroxide treatment is not caused by a change in 496 the interaction of tau with microtubules. 497 498 Subtoxic hydrogen peroxide modulates the phosphorylation state of components of the 499 microtubule system. 500 Hydrogen peroxide , a s a second messenger molecule, is thought to link redox biology to 501 intrinsic signalling pathways (Sies, 2017). Thus, it is likely that subtoxic hydrogen peroxide 502 shapes the axonal microtubule array by affecting the expression or phosphorylation of proteins 503 of the microtubule system. To identify the respective target molecules and their modification, 504 we performed proteomics and phosphoproteomics analysis of differentiated model neurons 505 treated with hydrogen peroxide compared to control conditions (Fig. 4A). GO-term analysis for 506 differentially expressed proteins revealed mainly NADPH-binding and RNA polymerase -507 binding proteins (Fig. 4B). The only microtubule -related protein with differential expression 508 was MAPRE3 (Microtubule-associated protein RP/EB family member 3, EB3), which showed 509 slightly increased expression in hydrogen peroxide -exposed cells (Fig. 4C). EB3 is known to 510 be a microtubule plus -end binding protein (Akhmanova and Steinmetz, 2015) and promotes 511 microtubule growth by su ppressing catastrophe (Komarova et al., 2009) . However, it was 512 recently shown to be also relevant to microtubule minus-end organization (Yang et al., 2017). 513 Therefore, the increase in mean microtubule length which we observed in the microtubule array 514 could be due to the increased EB3 expression. 515 In contrast to the moderate effect on the expression of microtubule system proteins, hydrogen 516 peroxide had a strong effect on the phosphorylation of various microtubule proteins , as 517 demonstrated by phosphoproteomic analysis (Fig. 4D, E). GO-term analysis of proteins with 518 upregulated phosphosites revealed the presence of many cytoskeletal proteins, particularly 519 components of the microtubule skeleton (Fig. 4D). Further examination of the proteins with 520 upregulated phosphosites revealed that various functional subgroups of the microtubule system, 521 including structur al proteins, tubulin -sequestering proteins and microtubule -associated 522 proteins, exhibited increased phosphorylation (Fig. 4E). 523 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted July 4, 2024. ; https://doi.org/10.1101/2024.07.01.601594doi: bioRxiv preprint 18 Thus, the data indicate that hydrogen peroxide shapes the axonal microtubule array primarily 524 by modulating the phosphorylation status of different functional groups of the microtubule 525 system. 526 527 Hydrogen peroxide and ROS-generation by arsenite induce differential phosphorylation 528 of MAP1B as the major target of microtubule proteins. 529 If hydrogen peroxide acts as a second messenger molecule, it would be expected to selectively 530 regulate signalling pathways under eustress conditions that differ from the action of other redox 531 modulators, particularly when compared to conditions that induce oxidative distress. Therefore, 532 we decided to compare the effect of subtoxic hydrogen peroxide with ROS-generation by 533 arsenite, a naturally occurring toxicant (Huber et al., 2022). 534 Phosphoproteomic analysis showed that arsenite also affected the phosphorylation of several 535 functional groups of the microtubule system, including structural proteins, microtubule-binding 536 proteins, microtubule nucleators , and tubulin -binding proteins (Fig. 5A). In fact, arsenite 537 induced a higher number of upregulated phosphosites than hydrogen peroxide, which was more 538 than twice as high for the microtubule system proteins (Fig. 5B). Notably, the overlap of 539 upregulated phosphosites induced by hydrogen peroxide and arsenite was small, and less than 540 5% (4 out of 82) of the phosphosites increased by arsenite were also increased in the presence 541 of hydrogen peroxide. 542 A comparison of the target proteins that showed increased phosphorylation revealed that twice 543 as many proteins of the microtubule system were altered in arsenite-exposed cells (Fig. 5C, D). 544 In both cases, the microtubule -binding protein MAP1B was the main target. MAP1B is 545 predominantly expressed in the nervous system and has been implicated in the regulation of 546 axonal elongation and guidance (Bouquet et al., 2004; Gonzalez-Billault et al., 2001). MAP1B 547 showed a complex pattern of increased phosphorylation sites with both hydrogen peroxide (22 548 sites) and arsenite (27 sites) (Fig. 5E). Notably, only one of the sites (S1772) showed increased 549 phosphorylation under both conditions. This specificity of increased phosphorylation at 550 selected sites was also present in other MAPs such as tau (total of 6 phosphosites, no overlap) 551 and MAP2 (total of 16 phosphosites, no overlap). 552 Thus, the data indicate that hydrogen peroxide leads to a specific phosphorylation pattern of 553 proteins of the microtubule system that is largely distinct from the effects of arsenite as a toxic 554 redox modulator, reflecting conditions of oxidative distress. The data also point to MAP1B as 555 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted July 4, 2024. ; https://doi.org/10.1101/2024.07.01.601594doi: bioRxiv preprint 19 the main target for both redox modulators , suggesting that MAP1B is the master regulator of 556 redox-mediated axonal microtubule dynamics. 557 558 Cell-wide phosphoproteome analysis of predicted upstream kinases in neuronally 559 diffentiated cells reveals a pattern of inversely regulated kinases by hydrogen peroxide 560 and arsenite. 561 The fact that the two redox modulators hydrogen peroxide and arsenite each have a di fferent 562 effect on the phosphorylation of microtubule -regulating proteins suggests that oxidative 563 eustress and distress influence specific signalling cascades differently. Therefore, we performed 564 a kinase enrichment analysis using all changed phosphosites as input to predict the upstream 565 kinases responsible for the different phosphorylation events observed (Kuleshov et al., 2021). 566 Remarkably, w e observed no overlap between the predicted upstream kinases that were 567 activated by the two redox modulators (Fig. 6A). Prominent upstream kinases activated by 568 hydrogen peroxide included kinases known to su ppress apoptosis such as DNA -dependent 569 protein kinase PRKDC (Yue et al., 2020) or the casein kinase 2 subunit CSNK2A2 (Ahmad et 570 al., 2008). Other kinases are known to be involved in regulating brain metabolism such as the 571 pyruvate dehydrogenase kinases PDK2 and 3 (Wang et al., 2023). On the other hand, prominent 572 predicted upstream kinases activated by arsenite were several members of the MAP kinase 573 signal transduction pathway (MAPK3, MAP3K4, MAP2K1, 3, 4, 6, and 7), mammalian target 574 of rapamycin (mTOR) , and serum/glucocorticoid regulated protein kinases SGK1, 2 and 3. 575 Deregulation of the MAPK signaling and mTOR pathway has been implicated in the 576 development of several neurodegenerative diseases (Ahmed et al., 2020; Perluigi et al., 2015) 577 and upregulation of SKG is associated with various neurodegenerative disorders (Kwon et al., 578 2021). 579 Of particular importance for the regulation of differential phosphorylation by certain redox 580 modulators could be those signalling pathways that lead to a reverse change in phosphorylation, 581 i.e. to activation under one condition, but to inhibition under the other. The respective activated 582 upstream kinases for hydrogen peroxide versus arsenite include pyruvate dehydrogenase 583 kinases (PDKs), which play a crucial role in aerobic metabolism , linking glycolysis to the 584 tricarboxylic acid cycle and ATP generation (Wang et al., 2021), and a subunit of casein kinase 585 1, a major contributor to the generation of the human phospho-proteome (Borgo et al., 2021) 586 (Fig. 6B). In turn, the upstream kinases for arsenite versus hydrogen peroxide include mTOR 587 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted July 4, 2024. ; https://doi.org/10.1101/2024.07.01.601594doi: bioRxiv preprint 20 and glucocorticoid-regulated kinase 3 (SGK3), kinases whose dysregulation has been linked to 588 metabolic dysfunction and disease (Liao et al., 2022; Maiese, 2020). 589 Taken together, cell-wide phosphoproteome analysis shows that different signalling pathways 590 are mutually exclusively activated by oxidative eustress and distress . While treatment with 591 subtoxic hydrogen peroxide (reflecting oxidative eustress) is mainly associated with kinases 592 suppressing apoptosis and regulating brain metabolism (PRKDC and CK2 ) and cellular 593 metabolism (PDKs), treatment with the toxic redox modulator arsenite (reflecting oxidative 594 distress) resulted in activati on of signal ling pathways associated with neurodegeneration 595 (mTOR, SGKs). The different phosphorylation patterns of microtubule-regulating proteins, in 596 particular MAP1B, is reflected by a mutually exclusive activation of signaling pathways at 597 eustress versus distress conditions. 598 599

600 Reactive oxygen species are involved in a variety of physiological cellular functions such as 601 cell proliferation, differentiation and maturation (Beckhauser et al., 2016). However, when ROS 602 accumulation exceeds antioxidant defense mechanisms , it leads to oxidative distress and 603 promotes pathological conditions in the brain (Pratico et al., 2002) . The development and 604 maturation of neurons critically depends on the microtubule cytoskeleton. Pathological changes 605 in the microtubule system and microtubule-dependent functions such as axonal transport occur 606 early in the disease process and disruption of microtubule dynamics may be a key mechanism 607 contributing to neurodegeneration (Penazzi et al., 2016a). However, it is largely unknown how 608 physiological, nontoxic ROS levels (oxidative eustress) influence axonal microtubule dynamics 609 and microtubule-dependent functions and how they differ from conditions that trigger oxidative 610 distress. 611 Here, we show that (1) a subtoxic concentration of the redox messenger hydrogen peroxide 612 increases microtubule dynamics and shape s microtubule organization in axon -like processes, 613 that (2) hydrogen peroxide modulates the phosphorylation state of different functional groups 614 of the microtubule system, and that (3) subtoxic hydrogen peroxide induces a complex and 615 ROS-specific phosphorylation pattern of MAP1B as the main target protein of the microtubular 616 system. By cell-wide phosphoproteome analysis, we further provide evidence that (4) different 617 signalling pathways are mutually exclusive activated by oxidative eustress and distress, 618 reflecting the distinct phosphorylation patterns of microtubule-related proteins. While oxidative 619 eustress is associated with the activation of pathways related to kinases that suppress apoptosis 620 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted July 4, 2024. ; https://doi.org/10.1101/2024.07.01.601594doi: bioRxiv preprint 21 and regulat e brain metabolism (PRKDC , CK2, PDKs), treatment with the toxic redox 621 modulator arsenite led to the activation of signalling pathways related to neurodegeneration 622 (mTOR, SGKs). 623 Axonal microtubules exhibit a unique organization in that they exist in relatively short 624 fragments with a uniform orientation and their dynamic plus -ends pointing towards the axon 625 tip. Although most axonal microtubules are more stable than the microtubules of dendrites, 626 axons also possess dynamic microtubules that undergo phases of polymerization and 627 depolymerization, a process known as dynamic instability (Baas et al., 2016) . The dynamic 628 microtubules in the axon can act as sensors of the cellular microenvironment and enable the 629 rapid reorganization of the cytoskeleton in response to changes in the environment (Coles and 630 Bradke, 2015) . The r egulation of microtubule nucleation and the dynamics of their 631 polymerization could therefore play an important role in local axon homeostasis for axon 632 maintenance, function and pathology (Hahn et al., 2019). Regulation by ROS can thus couple 633 neuronal activity to the local organization of the microtubule array. High neuronal activity 634 would lead to an increase in ROS species due to more active mitochondria. A physiologically 635 increased hydrogen peroxide content would then lead to a rearrangement of the microtubule 636 array by decreasing microtubule density and increasing mean microtubule length, accompanied 637 by a reduced amount of transported cargo, as we observed in our experiments (Fig. 7). This 638 would create a negative feedback loop to dampen over-activation of the neuron. 639 Hydrogen peroxide may also play a role in communication between glial cells and neurons. As 640 a messenger molecule, hydrogen peroxide diffuses through cells and tissues, which would allow 641 myelinating oligodendrocytes, which are known to provide metabolic and functional support to 642 the underlying axon (Duncan et al., 2021; Simons and Nave, 2015) , to modulate the structure 643 and dynamics of axonal microtubule s. Oligodendrocytes, in turn, could also act as a sink for 644 hydrogen peroxide produced locally in the axon. Therefore, it would be interesting to determine 645 the possible crosstalk of oligodendrocytes and neurons with respect to redox signalling. 646 Our data show that several functional groups of the microtubule system are differentially 647 phosphorylated due to hydrogen peroxide signalling. By far the largest group is microtubule-648 associated proteins (MAPs), followed by the tubulin sequestering protein stathmin 2. This 649 suggests that the other groups of microtubule-regulating proteins such as microtubule-severing 650 factors, end -binding proteins or microtubule nucleators are not involved in the immediate 651 response to hydrogen peroxide with re gard to axonal microtubule homeostasis. Interestingly, 652 microtubule-binding proteins and tubulin-sequestering proteins were also identified as the most 653 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted July 4, 2024. ; https://doi.org/10.1101/2024.07.01.601594doi: bioRxiv preprint 22 important drivers for the development of increased neuronal complexity during vertebrate 654 evolution, as they showed the largest increase in the number of orthologs and predicted protein-655 coding splice variants (Trushina et al., 2019b) . Thus, MAPs and tubulin -sequestering factors 656 appear to be the most important microtubule -regulating proteins in adapting the microtubule 657 skeleton to changes in the environment, both on an evolutionary scale and in terms of regulating 658 axonal microtubule homeostasis. Under oxidative distress conditions (arsenite treatment), end-659 binding proteins and a microtubule nucleator also joined the functional groups modified by 660 phosphorylation, which may indicate that increased phosphorylation of end-binding proteins in 661 particular could be responsible for the pathological changes of the axonal microtubule array. 662 Our data show that MAP1B is the main target of phosphorylation in both eustress and distress 663 conditions and approximately two third of the phosphorylation sites of microtubule-related 664 proteins that are upregulated in eustress conditions belong to MAP1B. MAP1B appears to 665 preferentially associate with tyrosinated (dynamic) microtubules, rather than detyrosinated 666 (stable) microtubules (Tymanskyj et al., 2012) , which may increase the pool of dynamic 667 microtubules (Tortosa et al., 2013; Utreras et al., 2008). Such behavior could be important for 668 the local regulation of axonal microtubule dynamics and polymerization. Phosphorylation of 669 MAP1B at different sites can regulate the local fine -tuning of neurite branching and 670 microtubule dynamics (Barnat et al., 2016; Scales et al., 2009; Ulloa et al., 1993). Historically, 671 two types of MAP1B phosphorylation have been distinguished. Mode I phosphorylation 672 induces a shift in electropho retic mobility and decreases with development , while mode II 673 phosphorylation does not affect electrophoretic mobility and remains unchanged (Kawauchi et 674 al., 2005). Phosphorylation at mode I sites results in a loss of microtubule-stabilizing ability 675 (Goold et al., 1999) and mode I phosphorylated MAP1B is also observed in neurofibrillary 676 tangles (NFTs) and dystrophic neurites in Alzheimer’s disease brains (Ulloa et al., 1994). This 677 suggests that mode I phosphorylation may be associated with nervous system pathogenesis and 678 reflects the MAP1B phosphorylation at oxidative distress conditions . In contrast, m ode II 679 phosphorylation is mediated for example by casein kinase 2 (Ulloa et al., 1993) , a predicted 680 upstream kinase with inverse regulation with hydrogen peroxide and arsenite. Regions of 681 phosphorylated epitopes recognized by the monoclonal antibody SMI -31 implicated in 682 detecting mode I phosphorylation sites (Johnstone et al., 1997) are present both after arsenite 683 treatment (1836 -2076) and hydrogen peroxide treatment ( 1244-1264) (Fig. 6E) . It will be 684 interesting to determine the functional consequences of the hydrogen peroxide-induced change 685 in phosphorylation compared to the arsenite-induced change, in order to determine sites that 686 may be cr ucial for the physiological versus pathological regulation of microtubule 687 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted July 4, 2024. ; https://doi.org/10.1101/2024.07.01.601594doi: bioRxiv preprint 23 polymerization. However, the remarkable complexity of MAP1B phosphorylation makes this a 688 difficult undertaking. 689 Increased phosphorylation of the microtubule -associated protein tau is associated with the 690 development of tauopathies such as Alzheimer’s disease (Arendt et al., 2016; Trushina et al., 691 2019a). Remarkably, our data show that tau is not a major target in either oxidative eustress or 692 distress. This is consistent with our observation that treatment with hydrogen peroxide does not 693 affect the tau-microtubule interaction, as many of the disease-related tau phosphorylation events 694 negatively affect tau binding to microtubules. Thus, the data indicate that increased 695 phosphorylation of tau is not involved in the regulation of microtubule polymerization under 696 oxidative eustress conditions and that increased phosphorylation of tau and the concomitant 697 dysregulation of axonal microtubule polymerization is a later event during neurodegeneration. 698 699 ACKNOWLEDGMENTS: 700 We thank Maxim Skulachev (Mitotech S.A., Luxembourg) for providing SkQ1 and TPP, and 701 Amos Smith 3rd (University of Pennsylvania) for providing EpoD. This work was supported by 702 the Deutsche Forschungsgemeinschaft (DFG BR1192/14-1 to RB). 703 704 AUTHOR CONTRIBUTIONS: 705 Christian Conze: Investigation, Writing - Original Draft; Nataliya I. Trushina : Investigation, 706 Writing - Original Draft; Nanci Monteiro-Abreu: Investigation, Writing - Original Draft; Daniel 707 Villar Romero : Investigation, Writing - Original Draft ; Eike Wienbeuker : Investigation, 708 Writing - Original Draft ; Anna-Sophie Schwarze : Investigation; Michael Holtmannspötter : 709 Investigation, Writing - Original Draft; Lidia Bakota: Writing - Original Draft, Supervision; 710 Roland Brandt: Conceptualization, Writing - Original Draft, Supervision, Funding acquisition. 711 The authors declare that no generative AI or AI -assisted technologies were used in the 712 preparation of this manuscript. 713 714 COMPETING INTERESTS STATEMENT: 715 The authors declare no competing interests. 716 717 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted July 4, 2024. ; https://doi.org/10.1101/2024.07.01.601594doi: bioRxiv preprint 24

718 Ahmad, K.A., G. Wang, G. Unger, J. Slaton, and K. Ahmed. 2008. Protein kinase CK2--a key 719 suppressor of apoptosis. Adv Enzyme Regul. 48:179-187. 720 Ahmed, T., A. Zulfiqar, S. Arguelles, M. Rasekhian, S.F. Nabavi, A.S. Silva, and S.M. 721 Nabavi. 2020. Map kinase signaling as therapeutic target for neurodegeneration. 722 Pharmacol Res. 160:105090. 723 Akhmanova, A., and M.O. Steinmetz. 2015. Control of microtubule organization and 724 dynamics: two ends in the limelight. Nat Rev Mol Cell Biol. 16:711-726. 725 Antonenko, Y.N., V.A. Roginsky, A.A. Pashkovskaya, T.I. Rokitskaya, E.A. Kotova, A.A. 726 Zaspa, B.V. Chernyak, and V.P. Skulachev. 2008. Protective effects of mitochondria-727 targeted antioxidant SkQ in aqueous and lipid membrane environments. J Membr Biol. 728 222:141-149. 729 Arendt, T., J.T. Stieler, and M. Holzer. 2016. Tau and tauopathies. Brain Research Bulletin. 730 126:238-292. 731 Baas, P.W., A.N. Rao, A.J. Matamoros, and L. Leo. 2016. Stability properties of neuronal 732 microtubules. Cytoskeleton (Hoboken). 73:442-460. 733 Bakota, L., and R. Brandt. 2024. Why kiss-and-hop explains that tau does not stabilize 734 microtubules and does not interfere with axonal transport (at physiological 735 conditions). Cytoskeleton (Hoboken). 81:47-52. 736 Barnat, M., M.N. Benassy, L. Vincensini, S. Soares, C. Fassier, F. Propst, A. Andrieux, Y. 737 von Boxberg, and F. Nothias. 2016. The GSK3-MAP1B pathway controls neurite 738 branching and microtubule dynamics. Molecular and cellular neurosciences. 72:9-21. 739 Beckhauser, T.F., J. Francis-Oliveira, and R. De Pasquale. 2016. Reactive Oxygen Species: 740 Physiological and Physiopathological Effects on Synaptic Plasticity. J Exp Neurosci. 741 10:23-48. 742 Borgo, C., C. D'Amore, S. Sarno, M. Salvi, and M. Ruzzene. 2021. Protein kinase CK2: a 743 potential therapeutic target for diverse human diseases. Signal Transduct Target Ther. 744 6:183. 745 Bouquet, C., S. Soares, Y. von Boxberg, M. Ravaille-Veron, F. Propst, and F. Nothias. 2004. 746 Microtubule-associated protein 1B controls directionality of growth cone migration 747 and axonal branching in regeneration of adult dorsal root ganglia neurons. The Journal 748 of neuroscience : the official journal of the Society for Neuroscience. 24:7204-7213. 749 Brandt, R., J. Leger, and G. Lee. 1995. Interaction of tau with the neural plasma membrane 750 mediated by tau's amino-terminal projection domain. The Journal of cell biology. 751 131:1327-1340. 752 Bray, D., and M.B. Bunge. 1981. Serial analysis of microtubules in cultured rat sensory 753 axons. J Neurocytol. 10:589-605. 754 Buey, R.M., J.F. Diaz, J.M. Andreu, A. O'Brate, P. Giannakakou, K.C. Nicolaou, P.K. 755 Sasmal, A. Ritzen, and K. Namoto. 2004. Interaction of epothilone analogs with the 756 paclitaxel binding site: relationship between binding affinity, microtubule 757 stabilization, and cytotoxicity. Chem Biol. 11:225-236. 758 Coles, C.H., and F. Bradke. 2015. Coordinating neuronal actin-microtubule dynamics. Curr 759 Biol. 25:R677-691. 760 Conze, C., M. Rierola, N.I. Trushina, M. Peters, D. Janning, M. Holzer, J.J. Heinisch, T. 761 Arendt, L. Bakota, and R. Brandt. 2022a. Caspase-cleaved tau is senescence-762 associated and induces a toxic gain of function by putting a brake on axonal transport. 763 Mol Psychiatry. 27:3010-3023. 764 Conze, C., N.I. Trushina, M. Holtmannspotter, M. Rierola, S. Attanasio, L. Bakota, J. Piehler, 765 and R. Brandt. 2022b. Super-resolution imaging and quantitative analysis of 766 microtubule arrays in model neurons show that epothilone D increases the density but 767 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted July 4, 2024. ; https://doi.org/10.1101/2024.07.01.601594doi: bioRxiv preprint 25 decreases the length and straightness of microtubules in axon-like processes. Brain 768 Res Bull. 190:234-243. 769 Duncan, G.J., T.J. Simkins, and B. Emery. 2021. Neuron-Oligodendrocyte Interactions in the 770 Structure and Integrity of Axons. Front Cell Dev Biol. 9:653101. 771 Fang, C., D. Bourdette, and G. Banker. 2012. Oxidative stress inhibits axonal transport: 772 implications for neurodegenerative diseases. Mol Neurodegener. 7:29. 773 Fath, T., J. Eidenmüller, and R. Brandt. 2002. Tau-Mediated Cytotoxicity in a 774 Pseudohyperphosphorylation Model of Alzheimer's Disease. The Journal of 775 Neuroscience. 22:9733. 776 Fedorov, A.V., M.A. Chelombitko, D.A. Chernyavskij, Galkin, II, O.Y. Pletjushkina, T.V. 777 Vasilieva, R.A. Zinovkin, and B.V. Chernyak. 2022. Mitochondria-Targeted 778 Antioxidant SkQ1 Prevents the Development of Experimental Colitis in Mice and 779 Impairment of the Barrier Function of the Intestinal Epithelium. Cells. 11. 780 Fukui, K., H. Takatsu, T. Koike, and S. Urano. 2011. Hydrogen peroxide induces neurite 781 degeneration: Prevention by tocotrienols. Free Radic Res. 45:681-691. 782 Gauthier-Kemper, A., C. Weissmann, N. Golovyashkina, Z. Sebo-Lemke, G. Drewes, V. 783 Gerke, J.J. Heinisch, and R. Brandt. 2011. The frontotemporal dementia mutation 784 R406W blocks tau's interaction with the membrane in an annexin A2-dependent 785 manner. The Journal of cell biology. 192:647-661. 786 Gonzalez-Billault, C., J. Avila, and A. Caceres. 2001. Evidence for the role of MAP1B in 787 axon formation. Molecular biology of the cell. 12:2087-2098. 788 Goold, R.G., R. Owen, and P.R. Gordon-Weeks. 1999. Glycogen synthase kinase 3beta 789 phosphorylation of microtubule-associated protein 1B regulates the stability of 790 microtubules in growth cones. Journal of cell science. 112 ( Pt 19):3373-3384. 791 Hahn, I., A. Voelzmann, Y.T. Liew, B. Costa-Gomes, and A. Prokop. 2019. The model of 792 local axon homeostasis - explaining the role and regulation of microtubule bundles in 793 axon maintenance and pathology. Neural Dev. 14:11. 794 Hu, Q., D. Wu, M. Walker, P. Wang, R. Tian, and W. Wang. 2021. Genetically encoded 795 biosensors for evaluating NAD(+)/NADH ratio in cytosolic and mitochondrial 796 compartments. Cell Rep Methods. 1. 797 Huber, S.M., U. Begley, A. Sarkar, W. Gasperi, E.T. Davis, V. Surampudi, M. Lee, J.A. 798 Melendez, P.C. Dedon, and T.J. Begley. 2022. Arsenite toxicity is regulated by 799 queuine availability and oxidation-induced reprogramming of the human tRNA 800 epitranscriptome. Proc Natl Acad Sci U S A. 119:e2123529119. 801 Igaev, M., D. Janning, F. Sundermann, B. Niewidok, R. Brandt, and W. Junge. 2014. A 802 refined reaction-diffusion model of tau-microtubule dynamics and its application in 803 FDAP analysis. Biophysical journal. 107:2567-2578. 804 Jacobs, J.R., and J.K. Stevens. 1986. Changes in the organization of the neuritic cytoskeleton 805 during nerve growth factor-activated differentiation of PC12 cells: a serial electron 806 microscopic study of the development and control of neurite shape. The Journal of cell 807 biology. 103:895-906. 808 Janning, D., M. Igaev, F. Sundermann, J. Bruhmann, O. Beutel, J.J. Heinisch, L. Bakota, J. 809 Piehler, W. Junge, and R. Brandt. 2014. Single-molecule tracking of tau reveals fast 810 kiss-and-hop interaction with microtubules in living neurons. Molecular biology of the 811 cell. 25:3541-3551. 812 Johnstone, M., R.G. Goold, D. Bei, I. Fischer, and P.R. Gordon-Weeks. 1997. Localisation of 813 microtubule-associated protein 1B phosphorylation sites recognised by monoclonal 814 antibody SMI-31. Journal of neurochemistry. 69:1417-1424. 815 Jungmann, R., C. Steinhauer, M. Scheible, A. Kuzyk, P. Tinnefeld, and F.C. Simmel. 2010. 816 Single-molecule kinetics and super-resolution microscopy by fluorescence imaging of 817 transient binding on DNA origami. Nano Lett. 10:4756-4761. 818 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted July 4, 2024. ; https://doi.org/10.1101/2024.07.01.601594doi: bioRxiv preprint 26 Kawauchi, T., K. Chihama, Y.V. Nishimura, Y. Nabeshima, and M. Hoshino. 2005. MAP1B 819 phosphorylation is differentially regulated by Cdk5/p35, Cdk5/p25, and JNK. 820 Biochemical and biophysical research communications. 331:50-55. 821 Komarova, Y., C.O. De Groot, I. Grigoriev, S.M. Gouveia, E.L. Munteanu, J.M. Schober, S. 822 Honnappa, R.M. Buey, C.C. Hoogenraad, M. Dogterom, G.G. Borisy, M.O. 823 Steinmetz, and A. Akhmanova. 2009. Mammalian end binding proteins control 824 persistent microtubule growth. The Journal of cell biology. 184:691-706. 825 Kuleshov, M.V., Z. Xie, A.B.K. London, J. Yang, J.E. Evangelista, A. Lachmann, I. Shu, D. 826 Torre, and A. Ma'ayan. 2021. KEA3: improved kinase enrichment analysis via data 827 integration. Nucleic acids research. 49:W304-W316. 828 Kwon, O.C., J.J. Song, Y. Yang, S.H. Kim, J.Y. Kim, M.J. Seok, I. Hwang, J.W. Yu, J. 829 Karmacharya, H.J. Maeng, J. Kim, E.H. Jho, S.Y. Ko, H. Son, M.Y. Chang, and S.H. 830 Lee. 2021. SGK1 inhibition in glia ameliorates pathologies and symptoms in 831 Parkinson disease animal models. EMBO Mol Med. 13:e13076. 832 Lachmann, A., and A. Ma'ayan. 2009. KEA: kinase enrichment analysis. Bioinformatics 833 (Oxford, England). 25:684-686. 834 Lee, C.B., Z. Wu, F. Zhang, M.D. Chappell, S.J. Stachel, T.C. Chou, Y. Guan, and S.J. 835 Danishefsky. 2001. Insights into long-range structural effects on the stereochemistry 836 of aldol condensations: a practical total synthesis of desoxyepothilone F. J Am Chem 837 Soc. 123:5249-5259. 838 Liao, Q.Q., Q.Q. Dong, H. Zhang, H.P. Shu, Y.C. Tu, and L.J. Yao. 2022. Contributions of 839 SGK3 to transporter-related diseases. Front Cell Dev Biol. 10:1007924. 840 Maiese, K. 2020. Dysregulation of metabolic flexibility: The impact of mTOR on autophagy 841 in neurodegenerative disease. Int Rev Neurobiol. 155:1-35. 842 Morotz, G.M., E.B. Glennon, J. Greig, D.H.W. Lau, N. Bhembre, F. Mattedi, N. Muschalik, 843 W. Noble, A. Vagnoni, and C.C.J. Miller. 2019. Kinesin light chain-1 serine-460 844 phosphorylation is altered in Alzheimer's disease and regulates axonal transport and 845 processing of the amyloid precursor protein. Acta Neuropathol Commun. 7:200. 846 Niewidok, B., M. Igaev, F. Sundermann, D. Janning, L. Bakota, and R. Brandt. 2016. 847 Presence of a carboxy-terminal pseudorepeat and disease-like 848 pseudohyperphosphorylation critically influence tau's interaction with microtubules in 849 axon-like processes. Molecular biology of the cell. 27:3537-3549. 850 Noble, M., S.A. Lewis, and N.J. Cowan. 1989. The microtubule binding domain of 851 microtubule-associated protein MAP1B contains a repeated sequence motif unrelated 852 to that of MAP2 and tau. The Journal of cell biology. 109:3367-3376. 853 Ovesny, M., P. Krizek, J. Borkovec, Z. Svindrych, and G.M. Hagen. 2014. ThunderSTORM: 854 a comprehensive ImageJ plug-in for PALM and STORM data analysis and super-855 resolution imaging. Bioinformatics (Oxford, England). 30:2389-2390. 856 Penazzi, L., L. Bakota, and R. Brandt. 2016a. Microtubule Dynamics in Neuronal 857 Development, Plasticity, and Neurodegeneration. Int Rev Cell Mol Biol. 321:89-169. 858 Penazzi, L., C. Tackenberg, A. Ghori, N. Golovyashkina, B. Niewidok, K. Selle, C. Ballatore, 859 A.B. Smith, 3rd, L. Bakota, and R. Brandt. 2016b. Abeta-mediated spine changes in 860 the hippocampus are microtubule-dependent and can be reversed by a subnanomolar 861 concentration of the microtubule-stabilizing agent epothilone D. Neuropharmacology. 862 105:84-95. 863 Perluigi, M., F. Di Domenico, and D.A. Butterfield. 2015. mTOR signaling in aging and 864 neurodegeneration: At the crossroad between metabolism dysfunction and impairment 865 of autophagy. Neurobiology of disease. 84:39-49. 866 Pinzi, L., C. Conze, N. Bisi, G.D. Torre, A. Soliman, N. Monteiro-Abreu, N.I. Trushina, A. 867 Krusenbaum, M.K. Dolouei, A. Hellwig, M.S. Christodoulou, D. Passarella, L. 868 Bakota, G. Rastelli, and R. Brandt. 2024. Quantitative live cell imaging of a tauopathy 869 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted July 4, 2024. ; https://doi.org/10.1101/2024.07.01.601594doi: bioRxiv preprint 27 model enables the identification of a polypharmacological drug candidate that restores 870 physiological microtubule interaction. Nat Commun. 15:1679. 871 Pratico, D., C.M. Clark, F. Liun, J. Rokach, V.Y. Lee, and J.Q. Trojanowski. 2002. Increase 872 of brain oxidative stress in mild cognitive impairment: a possible predictor of 873 Alzheimer disease. Arch Neurol. 59:972-976. 874 Rivkin, A., F. Yoshimura, A.E. Gabarda, Y.S. Cho, T.C. Chou, H. Dong, and S.J. 875 Danishefsky. 2004. Discovery of (E)-9,10-dehydroepothilones through chemical 876 synthesis: on the emergence of 26-trifluoro-(E)-9,10-dehydro-12,13-desoxyepothilone 877 B as a promising anticancer drug candidate. J Am Chem Soc. 126:10913-10922. 878 Scales, T.M., S. Lin, M. Kraus, R.G. Goold, and P.R. Gordon-Weeks. 2009. Nonprimed and 879 DYRK1A-primed GSK3 beta-phosphorylation sites on MAP1B regulate microtubule 880 dynamics in growing axons. Journal of cell science. 122:2424-2435. 881 Schnitzbauer, J., M.T. Strauss, T. Schlichthaerle, F. Schueder, and R. Jungmann. 2017. Super-882 resolution microscopy with DNA-PAINT. Nat Protoc. 12:1198-1228. 883 Schoppe, J., M. Mari, E. Yavavli, K. Auffarth, M. Cabrera, S. Walter, F. Frohlich, and C. 884 Ungermann. 2020. AP-3 vesicle uncoating occurs after HOPS-dependent vacuole 885 tethering. EMBO J. 39:e105117. 886 Sies, H. 2017. Hydrogen peroxide as a central redox signaling molecule in physiological 887 oxidative stress: Oxidative eustress. Redox Biol. 11:613-619. 888 Simons, M., and K.A. Nave. 2015. Oligodendrocytes: Myelination and Axonal Support. Cold 889 Spring Harb Perspect Biol. 8:a020479. 890 Sinenko, S.A., T.Y. Starkova, A.A. Kuzmin, and A.N. Tomilin. 2021. Physiological Signaling 891 Functions of Reactive Oxygen Species in Stem Cells: From Flies to Man. Front Cell 892 Dev Biol. 9:714370. 893 Tortosa, E., N. Galjart, J. Avila, and C.L. Sayas. 2013. MAP1B regulates microtubule 894 dynamics by sequestering EB1/3 in the cytosol of developing neuronal cells. EMBO J. 895 32:1293-1306. 896 Trushina, N.I., L. Bakota, A.Y. Mulkidjanian, and R. Brandt. 2019a. The Evolution of Tau 897 Phosphorylation and Interactions. Frontiers in aging neuroscience. 11:256. 898 Trushina, N.I., A.Y. Mulkidjanian, and R. Brandt. 2019b. The microtubule skeleton and the 899 evolution of neuronal complexity in vertebrates. Biological chemistry. 900 Tymanskyj, S.R., T.M. Scales, and P.R. Gordon-Weeks. 2012. MAP1B enhances microtubule 901 assembly rates and axon extension rates in developing neurons. Molecular and 902 cellular neurosciences. 49:110-119. 903 Uhlen, M., L. Fagerberg, B.M. Hallstrom, C. Lindskog, P. Oksvold, A. Mardinoglu, A. 904 Sivertsson, C. Kampf, E. Sjostedt, A. Asplund, I. Olsson, K. Edlund, E. Lundberg, S. 905 Navani, C.A. Szigyarto, J. Odeberg, D. Djureinovic, J.O. Takanen, S. Hober, T. Alm, 906 P.H. Edqvist, H. Berling, H. Tegel, J. Mulder, J. Rockberg, P. Nilsson, J.M. Schwenk, 907 M. Hamsten, K. von Feilitzen, M. Forsberg, L. Persson, F. Johansson, M. Zwahlen, G. 908 von Heijne, J. Nielsen, and F. Ponten. 2015. Proteomics. Tissue-based map of the 909 human proteome. Science. 347:1260419. 910 Ulloa, L., J. Avila, and J. Diaz-Nido. 1993. Heterogeneity in the phosphorylation of 911 microtubule-associated protein MAP1B during rat brain development. Journal of 912 neurochemistry. 61:961-972. 913 Ulloa, L., E. Montejo de Garcini, P. Gomez-Ramos, M.A. Moran, and J. Avila. 1994. 914 Microtubule-associated protein MAP1B showing a fetal phosphorylation pattern is 915 present in sites of neurofibrillary degeneration in brains of Alzheimer's disease 916 patients. Brain Res Mol Brain Res. 26:113-122. 917 Utreras, E., E.M. Jimenez-Mateos, E. Contreras-Vallejos, E. Tortosa, M. Perez, S. Rojas, L. 918 Saragoni, R.B. Maccioni, J. Avila, and C. Gonzalez-Billault. 2008. Microtubule-919 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted July 4, 2024. ; https://doi.org/10.1101/2024.07.01.601594doi: bioRxiv preprint 28 associated protein 1B interaction with tubulin tyrosine ligase contributes to the control 920 of microtubule tyrosination. Dev Neurosci. 30:200-210. 921 Vincent, A.M., K. Kato, L.L. McLean, M.E. Soules, and E.L. Feldman. 2009. Sensory 922 neurons and schwann cells respond to oxidative stress by increasing antioxidant 923 defense mechanisms. Antioxid Redox Signal. 11:425-438. 924 Wang, C., C. Cui, P. Xu, L. Zhu, H. Xue, B. Chen, and P. Jiang. 2023. Targeting PDK2 925 rescues stress-induced impaired brain energy metabolism. Mol Psychiatry. 28:4138-926 4150. 927 Wang, X., X. Shen, Y. Yan, and H. Li. 2021. Pyruvate dehydrogenase kinases (PDKs): an 928 overview toward clinical applications. Biosci Rep. 41. 929 Weissmann, C., H.J. Reyher, A. Gauthier, H.J. Steinhoff, W. Junge, and R. Brandt. 2009. 930 Microtubule binding and trapping at the tip of neurites regulate tau motion in living 931 neurons. Traffic (Copenhagen, Denmark). 10:1655-1668. 932 Wilson, C., E. Munoz-Palma, and C. Gonzalez-Billault. 2018. From birth to death: A role for 933 reactive oxygen species in neuronal development. Seminars in cell & developmental 934 biology. 80:43-49. 935 Wilson, C., M.T. Nunez, and C. Gonzalez-Billault. 2015. Contribution of NADPH oxidase to 936 the establishment of hippocampal neuronal polarity in culture. Journal of cell science. 937 128:2989-2995. 938 Xie, Z., A. Bailey, M.V. Kuleshov, D.J.B. Clarke, J.E. Evangelista, S.L. Jenkins, A. 939 Lachmann, M.L. Wojciechowicz, E. Kropiwnicki, K.M. Jagodnik, M. Jeon, and A. 940 Ma'ayan. 2021. Gene Set Knowledge Discovery with Enrichr. Curr Protoc. 1:e90. 941 Yang, C., J. Wu, C. de Heus, I. Grigoriev, N. Liv, Y. Yao, I. Smal, E. Meijering, J. 942 Klumperman, R.Z. Qi, and A. Akhmanova. 2017. EB1 and EB3 regulate microtubule 943 minus end organization and Golgi morphology. The Journal of cell biology. 216:3179-944 3198. 945 Yogev, S., R. Cooper, R. Fetter, M. Horowitz, and K. Shen. 2016. Microtubule Organization 946 Determines Axonal Transport Dynamics. Neuron. 92:449-460. 947 Yu, W., and P.W. Baas. 1994. Changes in microtubule number and length during axon 948 differentiation. The Journal of neuroscience : the official journal of the Society for 949 Neuroscience. 14:2818-2829. 950 Yue, X., C. Bai, D. Xie, T. Ma, and P.K. Zhou. 2020. DNA-PKcs: A Multi-Faceted Player in 951 DNA Damage Response. Front Genet. 11:607428. 952 Zhang, Z., Y. Nishimura, and P. Kanchanawong. 2017. Extracting microtubule networks from 953 superresolution single-molecule localization microscopy data. Molecular biology of 954 the cell. 28:333-345. 955 956 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted July 4, 2024. ; https://doi.org/10.1101/2024.07.01.601594doi: bioRxiv preprint 29 FIGURES 957 958 Fig. 1. Subtoxic concentrations of hydrogen peroxide decrease microtubule polymer in 959 axon-like processes by decreasing k on and increasing k off rates of microtubule 960 polymerization. 961 A. Combined MTT (blue) and LDH (red) assay showing the effect of 3 h H2O2 exposure on 962 neuronally differentiated PC12 cells. Mean ± SEM of 4 experiments , each carried out in 963 triplicates, are shown. Statistically significant differences as determined by o ne-way ANOVA 964 followed by a Dunnett post hoc test showed significant differences from a control at 965 concentrations ≥ 225 µM. B. Bar graphs showing the increase in intracellular ROS level in 966 response to H2O2. Each data point represents an independent experiment normalized to a 967 control. Mean ± SEM is shown. Statistically significant differences between treated and control 968 cells as determined by o ne-sample t-tests are indicated (**p<0.01) . C. Representative time-969 lapse images of a fluorescence decay after photoactivation (FDAP) experiment in an axon-like 970 process. A 6 µm long segment (white box) in the middle of a process was photoactivated and 971 the fluorescence decay in this area was monitored over time. A schematic representation of the 972 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted July 4, 2024. ; https://doi.org/10.1101/2024.07.01.601594doi: bioRxiv preprint 30 FDAP approach and the expressed construct is shown on the left. D. FDAP diagrams after 973 photoactivation of PAGFP--Tubulin expressing cells show an increased fluorescence decay 974 after treatment with H2O2. Mean values ± SEM of 29 (control) and 13 (150 µM H2O2) cells are 975 shown. E. Scatterplots of the association (k*on) and dissociation rate constants (koff), determined 976 by modelling the FDAP plots from (D) , show that H 2O2 decreases k*on and increases koff. 977 Statistically significant differences between treated and control cells, determined by unpaired 978 two-tailed Student’s t-tests, are indicated. *p<0.05, **p<0.01. F. The scatterplot of the amount 979 of polymerized tubulin determined from the association and dissociation constants in (E) shows 980 that H 2O2 reduces the amount of polymerized tubulin in axon -like processes. Statistically 981 significant differences determined by unpaired two -tailed Student’s t -tests are indicated. 982 ***p<0.001. G. A s chematic representation of the conversion between the oxidized and 983 reduced forms of SkQ1 as a mitochondria-targeted antioxidant is shown. TPP, which lacks the 984 antioxidant quinone moiety, is indicated. A bar graph is displayed on the right showing that 985 pretreatment with SkQ1 abolished the increase in intracellular ROS levels in response to H2O2. 986 H. Schematic representation of the timeline of the experiment to determine the effect of 987 pretreatment with SkQ1 or a control (TPP) on microtubule polymerization. I. Scatterplot of the 988 amount of polymerized tubulin with the control (TPP) and SkQ1 , showing that pretreatment 989 with SkQ1 prevents H2O2-induced microtubule depolymerization in axon -like processes as 990 determined by FDAP experiments. Shown are mean values ± SEM of 16, 13 (TPP) and 28, 23 991 (SkQ1) cells for control and H 2O2-treatment, respectively. Statistically significant differences 992 determined by unpaired two -tailed Student’s t-tests are indicated. **p<0.01. J. Shown is the 993 time course of the experiment to determine the effect of pretreatment with the microtubule -994 stabilizer EpoD on H2O2-induced microtubule depolymerization. K. Scatterplot of the amount 995 of polymerized tubulin showing that pretreatment with EpoD prevents H 2O2-induced 996 microtubule depolymerization. Shown are m ean values ± SEM of 21, 14 ( carrier, 0.01% 997 DMSO) and 29, 17 (1 nM EpoD) cells for control and H2O2-treatment, respectively. Statistically 998 significant differences determined by unpaired two -tailed Student’s t -tests are indicated. 999 **p<0.01. 1000 1001 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted July 4, 2024. ; https://doi.org/10.1101/2024.07.01.601594doi: bioRxiv preprint 31 1002 Fig. 2. Subtoxic hydrogen peroxide modulates the structure of the axonal microtubule 1003 array by reducing microtubule mass and increasing microtubule length. 1004 A. Shown on the left is a schematic representation of the single molecule localization 1005 microscopy (SMLM) approach to quantify MT organization . An indication of the average 1006 microtubule spacing in axon-like processes of differentiated PC12 is included. A total internal 1007 reflection fluorescence ( TIRF) image of the 3D -rendered SML data and a 150 nm optical 1008 section with fire color code are shown in the middle. Microtubule filaments as extracted from 1009 a selected ROI using SIFNE (single-molecule localization microscopy image filament network 1010 extractor) are shown in the rainbow color code on the right . Scale bar 10µm. B. Boxplots 1011 showing the mass, density, m ean length and straightness of microtubules . Each data dot 1012 represents the average of a single cell (control, 6 cells with n=902 individual microtubules ; 1013 H2O2, 7 cells with n=457 individual microtubules). H2O2 treatment reduces microtubule mass 1014 but increases mean microtubule length. Statistically significant differences determined by 1015 unpaired two -tailed Student’s t -tests are indicated. *p<0.05, **p<0.01. C. Distribution of 1016 microtubule lengths in a relative abundance histogram. Dotted lines show mean microtubule 1017 length under control conditions and with H2O2. The bin width was set to 0.5 µm and the area 1018 of the histogram bars is 1.0. H2O2 shifts the length distribution by reducing the proportion of 1019 short microtubules (0.5-2.5 µm length) and increasing the number of long microtubules (>8 1020 µm). 1021 1022 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted July 4, 2024. ; https://doi.org/10.1101/2024.07.01.601594doi: bioRxiv preprint 32 1023 Fig. 3. Hydrogen peroxide -induced changes in the microtubule array reduce the 1024 proportion of mobile vesicles but have no effect on the speed and velocity of axonal 1025 transport, or the interaction with the axonal tau protein. 1026 A. Tracking APP vesicles in a PC12 cell process using eGFP-tagged APP and an autoregressive 1027 motion algorithm. The first micrograph shows an overview of a cell expressing APP-eGFP. The 1028 images on the right show selected time s of the APP vesicle movement from the part of the 1029 process marked by the white box in the overview image. Arrows point to a moving (white) and 1030 a stationary vesicle (red). Scale bars, 10 μm (overview) and 1 μm (time lapse). B. The bar graph 1031 shows proportions of mobile vesicles in the axon -like process under control conditions and in 1032 response to H2O2. Velocity and speed of mobile vesicles are shown in the scatter plots on the 1033 right (mean ± SEM of n = 25 cells with 1595 trajectories (control) and 23 cells with 1209 1034 trajectories (H2O2); each point represents an average value for one analyzed cell). Statistically 1035 significant differences determined by unpaired two-tailed Student’s t-tests are indicated. **p < 1036 0.01. C. Proportion of mobile vesicles in cells pretreated with 1 nM EpoD according to the 1037 timeline shown in Fig 2D. Each data point represents an average value for a corresponding cell 1038 (mean ± SEM of 20 cells with 1276 trajectories ( control) and 25 cells with 1720 trajectories 1039 (H2O2)). D. Effect of H2O2 on the interaction of tau with microtubules. A schematic 1040 representation of the FDAP approach and the expressed tau construct is shown on the left. 1041 FDAP plots after photoactivation of PAGFP -Tau expressing cells (middle) show a similar 1042 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted July 4, 2024. ; https://doi.org/10.1101/2024.07.01.601594doi: bioRxiv preprint 33 fluorescence decay with and without H 2O2. Likewise, the scatterplots of the association (k *on) 1043 and dissociation rate constants (k off) (right) show no statistically significant differences. Mean 1044 values ± SEM of 25 (control) and 14 (H2O2) cells are shown. 1045 1046 1047 Fig. 4. Subtoxic hydrogen peroxide modulates the phosphorylation state of components of 1048 the microtubule system. 1049 A. Schematic representation of the approach for proteomics and phosphoproteomics analyses 1050 of differentiated model neurons treated with hydrogen peroxide compared to control. B. Bar 1051 plot with the top enriched GO-terms for molecular function of all differentially regulated 1052 proteins in response to H2O2. The length of the bar indicates the p -value, reflecting the 1053 enrichment significance of each term. Note an enrichment of proteins mostly involved in RNA 1054 processing and redox modulation. C. Volcano plot showing up- and down-regulated proteins 1055 in hydrogen peroxide -treated cultures compared to controls. Log2 fold changes are plotted 1056 against -log10 p -values. Significant upregulation upon H 2O2 treatment is shown in blue, 1057 downregulation in dark grey. Members of different groups of microtubule -regulating proteins 1058 are indicated. The axes are cut for representation purposes (x -axis from -3 to 3, y-axis from 0 1059 to 60); all the proteins above the limits are shown as points at each limit border. Members of 1060 different groups of microtubule -regulating proteins are indicated in different colors, with 1061 significantly changed ones labelled with their gene names. Only one protein of the more than 1062 30 identified proteins of the microtubule system (MAPRE3) shows a slight but significant 1063 upregulation. D. Bar plot with the top enriched GO -terms for cellular components of all 1064 proteins with upregulated phosphosites in response to H2O2. The length of the bar indicates the 1065 p-value, reflecting the enrichment significance of each term. Note enrichment of nuclear 1066 components, organelles and proteins of the cytoskeleton. E. Volcano plot showing up - and 1067 down-regulated phosphosites in hydrogen peroxide-treated cultures compared to controls. The 1068 axes are cut for representation purposes and all phosphosites above the limits are shown as 1069 points at each limit border. Coloring and labelling as in C. 1070 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted July 4, 2024. ; https://doi.org/10.1101/2024.07.01.601594doi: bioRxiv preprint 34 1071 1072 Fig. 5. Subtoxic hydrogen peroxide and ROS-generation by arsenite induce differential 1073 phosphorylation of MAP1B as the major target of microtubule proteins. 1074 A. Volcano plot showing upregulated phosphosites in arsenite -treated c ells compared to 1075 control. Members of different groups of microtubule -regulating proteins are indicated by the 1076 same color code as in Fig. 5C. Log2 fold changes are plotted against -log10 p-values. Significant 1077 upregulation upon arsenite treatment is shown in orange, downregulation in dark grey. The axes 1078 are cut for representation purposes and all phosphosites above the limits are shown as points at 1079 each limit border. B. Venn diagram showing low overlap of phosphosites of microtubule -1080 regulating proteins upregulated in response to hydrogen peroxide or arsenite. C, D. Distribution 1081 of upregulated phosphosites on different proteins of the microtubule system in response to 1082 hydrogen peroxide (C) or arsenite (D). Note that MT-binding proteins and especially MAP1B 1083 are the main target. E. Graphical representation of the different phosphosites on MAP1B that 1084 are upregulated in response to hydrogen peroxide or arsenite. The blue bar shows the 1085 microtubule-binding region according to the deletion study by (Noble et al., 1989) . 1086 Phosphorylated epitopes recognized by the monoclonal antibody SMI -31 (Johnstone et al., 1087 1997), which detects disease-associated mode I phosphorylation sites, which cause a loss of the 1088 microtubule stabilizing activity, are indicated by the dark green bar. 1089 1090 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted July 4, 2024. ; https://doi.org/10.1101/2024.07.01.601594doi: bioRxiv preprint 35 1091 Fig. 6. Cell-wide phosphoproteome analysis of predicted upstream kinases in neuronally 1092 differentiated cells reveals a pattern of inversely regulated kinases by hydrogen peroxide 1093 and arsenite. 1094 A. Kinase enrichment analysis of the phosphoproteomics data to identify the pattern of kinases 1095 responsible for increased phosphorylation of all cellular proteins in response to hydrogen 1096 peroxide (left) and arsenite (right). Note that there is no overlap between significantly 1097 upregulated upstream kinases with hydrogen peroxide and arsenite. B. Venn diagrams showing 1098 upstream kinases leading to a reverse change in phosphorylation, i.e. induction of increased 1099 phosphorylation with hydrogen peroxide and reduced phosphorylation with arsenite (top) and 1100 increased phosphorylation with arsenite and reduced phosphorylation with hydrogen peroxide 1101 (bottom). The respective upstream kinases with reverse change are indicated below the 1102 corresponding Venn diagram. 1103 1104 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted July 4, 2024. ; https://doi.org/10.1101/2024.07.01.601594doi: bioRxiv preprint 36 1105 Fig. 7. Schematic representation of the effect of hydrogen peroxide on axonal microtubule 1106 organization and microtubule-dependent transport. 1107 Hydrogen peroxide diffuses through cells and tissues and can reach neighboring axons when 1108 produced in oligodendrocytes or astrocytes. In addition, it is also produced by mitochondria in 1109 the axons . In axons, H 2O2 causes a reorganization of the microtubule cytoskeleton toward 1110 longer but less dense microtubules, which reduces the efficiency of vesicle transport because it 1111 becomes more difficult for vesicles to change the microtubule track. Figure created with 1112 BioRender.com. 1113 (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted July 4, 2024. ; https://doi.org/10.1101/2024.07.01.601594doi: bioRxiv preprint