MAP65-1/PRC1 reinforces microtubules through nucleation

15 Microtubules are dynamic cytoskeletal filaments that help shape cells and guide their 16 response to external cues, including mechanical stress. How cells reorganize and 17 reinforce microtubule arrays under stress remains poorly understood . Here, we show 18 that MAP65 -1/PRC1 promotes microtubule nucleation on pre-existing lattices in vitro , 19 particularly on microtubules with lattice defects and damage sites associated with 20 mechanical stress. MAP65-1 preferentially binds bent microtubules both in vitro and in 21 cells, likely because bending induces lattice damage, enhancing nucleation on and 22 bundling of these microtubules in vitro. This creates a potential feedback loop where 23 mechanical stress promotes MAP65-1 binding, which in turn stabilizes microtubules by 24 nucleation and bundling and reinforces the alignment of the array with mechanical 25 stress. Thus, we reveal a previously unknown role for MAP65/PRC1 proteins in lattice-26 based nucleation and suggest a mechanism by which cells record and respond to 27 mechanical stress through microtubule reorganization. 28 29

30 Microtubules form highly organized arrays that play critical roles in intracellular 31 transport, cell division, and morphogenesis. Their array architecture is dynamically 32 sculpted through growth and disassembly 1,2, severing 3, cross -linking4, and local 33 nucleation5. These mechanisms allow cells to build microtubule arrays adapted to their 34 shape and function, and in response to external cues. 35 Microtubule arrays can reorganize in response to mechanical stress 6, enabling cells to 36 adapt to external forces. In animal cells, mechanical cues influence microtubule 37 orientation during processes such as epithelial remodeling, cell migration, cell division, 38 or tissue folding7–10. In plants, microtubules align with the main tensile stress direction at 39 the cell cortex 6,11,12, guiding the deposition of cellulose fibers and influencing tissue 40 morphogenesis13. Such observations suggest that microtubules act as mechanosensors, 41 relaying external cues into cytoskeletal organization. 42 While centrosomes and the g-tubulin ring complex were long considered the primary 43 sites of microtubule nucleation, many di^erentiated cells – including plant cells 14, 44 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 15, 2025. ; https://doi.org/10.1101/2025.09.14.676082doi: bioRxiv preprint neurons15, and epithelial cells 16 – organize microtubules without centrosomes. 45 Alternative nucleation pathways are therefore critical for building functional microtubule 46 arrays in these cells. Several microtubule-associated proteins (MAPs) have been shown 47 to reduce the critical tubulin concentration for nucleation in solution in vitro, suggesting 48 a potential role in the generation of new microtubules 17. However, whether such 49 nucleation occurs under physiological conditions, and how it is spatially regulated within 50 cells, remains unclear. Existing microtubule lattices may serve as platforms for 51 microtubule nucleation, as observed in structures such as the spindle midzone 18 or 52 cortical microtubule arrays in plants19. 53 Critical to the formation of both the spindle midzone and plant cortical microtubule array 54 is the MAP65/PRC1 protein family4,20,21, which is best known for cross-linking antiparallel 55 microtubules. Beyond cross -linking, the MAP65 family has been shown to increase 56 microtubule flexibility 22. However, the role of MAP65 proteins in mechanosensitive 57 responses has not been explored. 58 Here, we show that MAP65 -1/PRC1 promote microtubule nucleation on existing 59 microtubule lattices. This activity is enhanced at sites of structural irregularity in the 60 lattice, such as defects or damage sites. These findings suggest a previously 61 unrecognized role of MAP65 -1/PRC1 in reinforcing microtubule arrays through lattice -62 based nucleation. We propose a model in which microtubule alignment with mechanical 63 stress in plant cells generates lattice defects and damage that promote the recruitment 64 of MAP65-1. Once recruited, MAP65-1 nucleates, bundles, and softens microtubules, 65 thereby supporting the alignment of the microtubule array with the principal direction of 66 tensile stress. 67 68

69 The MAP65 family promotes microtubule nucleation on existing microtubule lattices 70 In vitro microtubule reconstitution assays o^er a controlled and simplified environment 71 to study microtubule dynamics and protein interactions. For that reason, we examined 72 the e^ect of MAP65 -1, a well -known microtubule bundler (Fig. S1a-c), on microtubule 73 dynamics in vitro . In control conditions without MAP65 -1, polymerized microtubules 74 incubated with free tubulin at concentrations above the polymerization threshold – but 75 well below the threshold for nucleation in solution – exhibit growth only from their free 76 ends (Fig. 1a -c and Fig. S2a). Because we used a low tubulin concentration of 4 µM, 77 microtubule tips grow slowly (Fig. 1c, control). However, in presence of 100 nM MAP65-1, 78 new microtubules appeared along the existing microtubule lattices (Fig. 1c-e). These new 79 microtubules exhibited dynamic instability, growing completely along the “template” 80 microtubules (likely bundled by MAP65 -1), and their fluorescence intensity was 81 comparable to the growing microtubule tips (Fig. 1d and e). Increasing the MAP65 -1 82 concentration from 100 to 200 nM led to a significant decrease in the distance between 83 microtubule appearance events, which represents the inverse of spatial frequency (Fig. 84 S2b and c). Hence, we concluded that MAP65-1 can induce microtubule appearance in a 85 concentration-dependent manner. 86 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 15, 2025. ; https://doi.org/10.1101/2025.09.14.676082doi: bioRxiv preprint Given that the function of MAP65-1 is conserved across the eukaryotic domain, we tested 87 if the human homolog of MAP65-1, PRC1, also promotes microtubule appearance. In the 88 presence of 300 pM PRC1, microtubules appeared and grew on the existing lattices 89 similarly to MAP65 -1 (Fig. 1c). Importantly, we worked with MAP concentrations well 90 below those that induce microtubule nucleation in solution. For MAP65 -1, we only 91 observed microtubule nucleation in solution at concentrations as high as 500 nM MAP65-92 1 (Fig. S3); for PRC1, nucleation in solution was detected only above 3 nM PRC1 (Fig. S3). 93 Although MAP65 -1 shares only 24% sequence identity with PRC1, it has significantly 94 higher identity with the other Arabidopsis MAP65 family members (Table S1). This 95 suggests that these other MAP65 proteins likely share MAP65-1’s microtubule-generating 96 function. 97 Previous studies have shown that the MAP SSNA1 promotes microtubule lattice 98 extensions in the form of branches that directly emerge from one or more protofilaments 99 of the mother microtubule23. To determine whether the microtubules that appeared in the 100 presence of MAP65 -1 and PRC1 were branches of the underlying lattice or appeared 101 following genuine nucleation events, we introduced human kinesin -5 (KIF11) into our 102 assays after a 30-minute incubation of MAP65-1 with microtubules and free tubulin (Fig. 103 1f-i). Since KIF11 cross-links and slides antiparallel microtubules, the observed mobility 104 of most new microtubules upon KIF11 addition (75%, Fig. 1i) indicates that they were 105 antiparallel and structurally independent of the “template” microtubule. Thus, we 106 confirmed that these microtubules derived from bona fide nucleation events. Since both 107 MAP65-1 and PRC1 preferentially bind to antiparallel microtubules 24, it is not surprising 108 that most nucleated microtubules are oriented in an antiparallel manner (Fig. 1d). 109 With a higher PRC1 concentration (1 nM), we observed ubiquitous PRC1 -mediated 110 microtubule nucleation and growth that quickly produced thick microtubule bundles 111 (Fig. 1j and k and S4a and b). In the control (no PRC1), after the same timeframe, only 112 short polymerization stretches were observed from the microtubule ends (Fig. 1k). 113 Hence, MAP65-1/PRC1 promote microtubule nucleation on the microtubule lattice. 114 115 MAP65-1 reversibly binds free tubulin when associated with the microtubule lattice 116 To investigate the mechanism by which MAP65 -1 and PRC1 promote microtubule 117 nucleation, we examined whether MAP65 -1 can recruit free tubulin to the microtubule 118 lattice. Since microtubule nucleation and growth require GTP , we performed assays in the 119 absence of GTP to prevent polymerization and analyze free tubulin binding (Fig. 2a and 120 S5a and b). Under these conditions, MAP65-1 localized along the microtubule lattice and 121 recruited free tubulin, as hypothesized (Fig. 2b-f). 122 Quantitative analysis revealed that, at 200 nM MAP65 -1, the average enrichment of free 123 tubulin compared to the solution was approximately 43% (Fig. 2f). However, in our 124 system, a ~3.5 -fold enrichment in tubulin concentration is typically required to initiate 125 nucleation in solution in the absence of MAPs. Although we occasionally observed 126 localized patches where tubulin enrichment approached this threshold, such events 127 were considerably less frequent than the nucleation events observed, as reflected by the 128 measured distance between nucleation events under the same MAP65-1 concentration 129 (Fig. S2b and c). One possibility is that small nucleation intermediates fall well below our 130 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 15, 2025. ; https://doi.org/10.1101/2025.09.14.676082doi: bioRxiv preprint resolution limit, leading us to underestimate local tubulin concentration. Yet, we cannot 131 exclude that mechanisms other than tubulin concentration may play a role in the 132 observed microtubule nucleation. 133 To determine whether MAP65 -1 and free tubulin bound to the microtubule lattice are 134 dynamically exchanged with molecules in solution, we performed fluorescence recovery 135 after photobleaching (FRAP) experiments in the absence of GTP . Following 136 photobleaching, both GFP -MAP65-1 and free tubulin signals partially recovered on the 137 microtubule lattice ( within 300 s; Fig. 2g and h), indicating that both proteins undergo 138 dynamic exchange with the soluble pool. 139 140 MAP65-1 recognizes bent microtubules in cells and in vitro 141 Our in vitro data demonstrate that MAP65-1 promotes microtubule nucleation on existing 142 lattices, at least partly by recruiting free tubulin that undergoes exchange with the soluble 143 protein pool. Yet, the relevance of this activity and its spatial regulation in cells remained 144 unclear. 145 Because microtubules are sti^ polymers, they would tend to align with the lowest 146 curvature cell axis by default. This is observed in non -pressurized wall-less plant cells25 147 or in non -growing hypocotyl cells 26. However, in turgid and growing plant cells, cortical 148 microtubules often align with the highest curvature axis instead. Because the highest 149 curvature axis for a pressurized cylindrical cell also corresponds to the maximal tensile 150 stress direction, it was proposed that tensile stress overrides geometrical cues 25. Since 151 MAP65-1 makes microtubules softer, we reasoned that microtubules would be less 152 responsive to the lowest curvature cell axis and more responsive to mechanical stress in 153 the presence of MAP65-1. Thus, we wondered whether MAP65-1 can specifically promote 154 microtubule nucleation on existing, bent microtubules, leading to a reinforcement of this 155 specific microtubule subset. To investigate this, we observed the distribution of MAP65-156 1 in cells of young Arabidopsis stems containing a microtubule (GFP-MBD) and a MAP65-157 1 (MAP65-1-mCherry) reporter. MAP65-1 fluorescence intensity was around twice as high 158 on bent microtubules in comparison to straight ones (Fig. 3a -c and Table S2), and we 159 observed that both MAP65 -1 and MBD fluorescence intensity positively correlated with 160 maximum microtubule curvature (Fig. 3d and e). Note that it remains unclear whether 161 MAP65-1 binds to pre-existing bundles that are more prominent in highly curved regions, 162 or whether it contributes to bundle formation in these regions by favoring microtubule 163 nucleation and bundling in highly bent microtubules. 164 The analyses above in plant tissues give an indication of MAP65 -1’s ability to recognize 165 bent microtubules. However, due to the complexity of the tissue, including water fluxes 166 and di^erent turgid statuses for instance, these analyses only o^er a correlation. To get 167 closer to causality, we used a much simpler system, consisting of wall -less plant cells 168 (protoplasts) confined in rectangular microwells (Fig. 3f). Upon osmotic pressurization, 169 achieved by transferring cells to a medium with lower osmolarity, microtubules have 170 been shown to reorient from their default longitudinal orientation to a transverse 171 alignment, which corresponds to the highest curvature and predicted highest tension 172 axis25. Under these conditions, MAP65 -1 preferentially accumulated on microtubules 173 aligning transversely (Fig. 3g). This was supported by a significant, albeit weak (Spearman 174 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 15, 2025. ; https://doi.org/10.1101/2025.09.14.676082doi: bioRxiv preprint r = -0.25), negative correlation between normalized mean MAP65 -1 intensity and mean 175 microtubule orientation two hours after pressurization (Fig. 3h), indicating that MAP65-1 176 is enriched on mechanically challenged microtubules. Microtubules may thus be able to 177 disentangle geometry from mechanical stress through a feedback loop where MAP65 -1 178 recruitment allows microtubule softening, which, in turn, promotes alignment with 179 tensile stress, bending along the highly curved axis of the cell, and further MAP65 -1 180 recruitment. 181 In cells, MAP65-1 accumulation on bent microtubules may result from di^erent factors, 182 such as the curvature of the microtubule itself, mechanical stress, or the recruitment of 183 other MAPs. Hence, to dissect these factors, we tested in our in vitro assays if MAP65-1 184 could recognize microtubule curvature alone. To do that, we bent microtubules by using 185 fluid flow and maintained them in a bent conformation by attaching the microtubule cap 186 to the coverslip. Next, we incubated microtubules with GFP -MAP65-1. We found that 187 MAP65-1 accumulated around 3 times more on bent microtubules in comparison to 188 straight ones (Fig. 4a and b). Accordingly, mean MAP65-1 intensity significantly correlated 189 with maximum microtubule curvature (Fig. 4c). 190 Next, we tested if MAP65-1-mediated microtubule nucleation happened more frequently 191 on bent microtubules as well. We incubated microtubules with free tubulin and 100 nM 192 MAP65-1 for 30 min followed by washing free tubulin away and microtubule stabilization 193 with taxol (Fig. 4d-g). The distance between nucleation events was approximately 5 times 194 smaller in bent microtubules in comparison to straight microtubules (Fig. 4f). Average 195 spatial nucleation frequency also increased with maximum microtubule curvature (Fig. 196 4g). Notably, microtubules nucleated also in regions other than those of highest 197 curvature along bent microtubules (Fig. S6), which suggests that additional factors likely 198 influence the site of nucleation. 199 200 MAP65-1 does not recognize microtubule curvature through lattice expansion or 201 compaction 202 MAPs that recognize microtubule curvature often exhibit preferential binding to either 203 expanded or compacted lattice conformations because a curved microtubule has an 204 expanded outer surface and a compacted inner surface27,28. 205 Interestingly, GMPCPP -polymerized microtubules, which adopt an expanded lattice 206 conformation29, were the predominant sites of nucleation in the presence of MAP65 -1, 207 despite comprising only a small fraction of the microtubule length (Fig. S7a and b). To test 208 whether MAP65 -1 preferentially binds to GMPCPP -containing microtubules, we 209 quantified GFP -MAP65-1 intensity on GMPCPP seeds versus GDP lattice regions. We 210 observed a significant enrichment of MAP65 -1 on the seeds (Fig. S7c -e), indicating a 211 preference for the GMPCPP lattice. 212 To determine whether MAP65 -1 specifically recognizes the expanded lattice state, we 213 treated microtubules with taxol, which also induces lattice expansion29. However, taxol-214 treated microtubules showed significantly reduced MAP65-1 accumulation (Fig. S7f and 215 g), suggesting that lattice expansion alone does not account for MAP65-1 binding. 216 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 15, 2025. ; https://doi.org/10.1101/2025.09.14.676082doi: bioRxiv preprint Since some MAPs that recognize specific lattice states can also induce lattice expansion 217 or compaction upon binding30, we examined microtubules incubated with GFP-MAP65-1 218 using microfluidics (Fig. S7h and i). Even at a high concentration (500 nM), MAP65 -1 did 219 not induce measurable changes in lattice length. Thus, we concluded that MAP65-1 does 220 not recognize microtubule curvature through lattice expansion or compaction. 221 222 MAP65-1 and PRC1 recognize microtubules with structural irregularities 223 In our experiments, we define lattice defects as irregularities that arise during 224 polymerization (for example, a change in protofilament number), whereas damage refers 225 to alterations that occur after polymerization, such as local tubulin loss or breaks in the 226 lattice. Bent microtubules are particularly prone to lattice damage, as reflected by 227 increased tubulin turnover along their shafts31. 228 Given MAP65 -1’s ability to recognize bent microtubules , we thus hypothesized that 229 MAP65-1/PRC1 might detect structural irregularities in the microtubule lattice, such as 230 defects and damage . To test this, we manipulated the frequency of microtubule 231 irregularities using two approaches: (1) by reducing the tubulin concentration for 232 polymerization to half of that used in fast -growth conditions, generating microtubules 233 with fewer defects 32 (referred to as slow growth conditions); and (2), by treating 234 microtubules with taxol to induce lattice damage 33, followed by thorough washing to 235 remove the drug (Fig. 5a-g). 236 In taxol pre -treated microtubules, we occasionally observed regions with noticeably 237 reduced lattice fluorescence (Fig. 5a-d, white arrowhead), which likely represent damage 238 sites. This observation is consistent with previous reports showing that treatment with 239 taxol overnight at room temperature causes considerable damage to the microtubule 240 lattice33. Notably, MAP65 -1-mediated microtubule nucleation events often originated 241 from these damage sites. The median distance between nucleation events was 242 significantly reduced in taxol -pre-treated microtubules (10.06 µm) compared to fast -243 growth conditions (16.88 µm; P = 0.015, Fig. 5c). In contrast, under slow -growth 244 conditions, the median distance between nucleation events increased to 63.00 µm (P = 245 0.0016). This corresponds to a spatial frequency nearly four times lower than that 246 observed in fast-growth conditions. 247 We validated these findings in an alternative approach, in which GMPCPP -stabilized 248 microtubules were polymerized under high - and low -defect conditions 34, followed by 249 incubation with MAP65 -1 and free tubulin (see Methods, Fig. S8a). Under low -defect 250 conditions, the median distance between nucleation events significantly increased from 251 3.44 µm to 6.74 µm, nearly a twofold di^erence, compared to the high-defect regime (Fig. 252 S8b and c). These results indicate that the spatial frequency of MAP65 -1-mediated 253 microtubule nucleation is promoted by the presence of structural irregularities in the 254 lattice. 255 To determine whether this function is conserved, we tested whether PRC1 also promotes 256 nucleation in a defect- and damage-dependent manner. We found that PRC1-mediated 257 nucleation occurred approximately three times more often on taxol -pre-treated 258 microtubules and about two times less often on microtubules grown under slow-growth 259 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 15, 2025. ; https://doi.org/10.1101/2025.09.14.676082doi: bioRxiv preprint conditions (Fig. 5e-g). We concluded that both MAP65-1 and PRC1 preferentially promote 260 microtubule nucleation on lattices with increased structural irregularity. 261 262 MAP65-1 mediated microtubule nucleation often co-localizes with annealing sites 263 A remaining question was whether MAP65 -1/PRC1 promote microtubule nucleation 264 directly at defect or damage sites. To test this, we generated microtubules with defined 265 annealing sites, which often form structural defects due to the imperfect alignment of 266 frayed microtubule ends. We polymerized two populations of GMPCPP -stabilized 267 microtubules, mixed them in equal amounts and allowed them to anneal overnight (Fig. 268 5h). This approach generated a mixed population of annealed and non -annealed 269 microtubules. We then incubated the mixture with MAP65-1 and free tubulin, followed by 270 washing and stabilization with taxol. 271 Strikingly, approximately 45% of the nucleation events co-localized with annealing sites 272 (25 out of 55 events; Fig. 5i and j). This was significantly higher than expected from 273 randomly distributed nucleation sites: among 55 randomly selected microtubule 274 patches, only 6 contained nucleation events (Fisher’s exact test, P < 0.0001). 275 Furthermore, the median distance between nucleation events significantly decreased in 276 annealed microtubules (5.11 µm) compared to non -annealed ones (14.86 µm; P = 277 0.0099, Fig. 5k), almost a three -fold di^erence. Thus, we concluded that nucleation 278 occurs at annealing sites more frequently than expected by chance, supporting the 279 hypothesis that MAP65 -1/PRC1 can recognize structural defects, where they promote 280 microtubule nucleation. 281 282 Bundled microtubules are more resistant to breakage after free tubulin removal 283 Considering that MAP65 -1 preferentially promotes nucleation – and thus bundling – on 284 microtubules with structural defects and damage, we wondered whether microtubules 285 bundled by MAP65-1 are more resistant to breakage. To test this, we removed free tubulin 286 from our in vitro assays, a condition that causes microtubules to gradually lose tubulin 287 dimers from their lattices, soften, and eventually break. After polymerizing and capping 288 microtubules, we simultaneously removed free tubulin and added 100 nM MAP65-1 and 289 imaged microtubules over time (Fig. 6a and b). Remarkably, bundled microtubules 290 survived almost 3 times longer compared to single ones (Fig. 6b), showing that MAP65-1 291 protects microtubule bundles from breakage. 292 Mechanical stress in the form of tension has been shown to accelerate microtubule 293 polymerization in vitro 35. Based on our findings that MAP65 -1 promotes nucleation on 294 microtubules with more lattice defects and damage, we propose a model in which 295 mechanical stress enhances microtubule polymerization speed in plant cells. Under 296 higher polymerization speeds, microtubules accumulate structural irregularities, for 297 instance in the form of lattice defects, which are recognized by MAP65 -1. Microtubules 298 aligning with mechanical stress in plant cells also likely accumulate lattice damage 299 caused by bending. MAP65-1 thus preferentially stabilizes, softens, nucleates on, and 300 bundles these microtubules. As a consequence, MAP65-1 activity promotes microtubule 301 array alignment along mechanical stress patterns (Fig. 6c). 302 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 15, 2025. ; https://doi.org/10.1101/2025.09.14.676082doi: bioRxiv preprint

303 In summary, we showed that MAP65 -1 preferentially binds to and promotes nucleation 304 on bent microtubules, supporting a model in which MAP65-1 acts as a mechanosensitive 305 regulator of cytoskeletal organization. This is particularly relevant in plant cells, where 306 cortical microtubules typically align with tensile stress patterns rather than local 307 geometric cues. For instance, in many plant tissues, adjacent cells with di^erent 308 geometries exhibit consistent supracellular microtubule alignment 11,36. Our finding 309 provides a scenario for microtubules to be di^erentially sensitive to geometrical and 310 mechanical cues. In turgid cells, tensile stress would promote fast microtubule 311 polymerization and the formation of defects in the lattice; the subsequent recruitment of 312 MAP65-1 would secure the viability of defective microtubules through microtubule 313 addition (nucleation and bundling) . As a consequence of MAP65 recruitment , MAP65 -314 decorated microtubules would also become softer. They would then be less inclined to 315 align with the flattest part of the cell cortex (compared to when they were sti^) and 316 instead align more in the direction of tensile stress at the cell cortex, i.e., under high 317 curvature, in a feedback loop . Importantly, we do not expect single microtubules to 318 reorient and align with the direction of highest tensile stress or curvature; instead, a few 319 microtubules stochastically growing in t hat direction are expected to be su^icient to 320 support the proposed feedback loop. 321 The co-localization of nucleation events with microtubule annealing sites supports the 322 idea that MAP65 -1 recognizes structurally vulnerable regions and reinforces them 323 through nucleation and bundling. Importantly, we demonstrate that microtubules 324 bundled by MAP65 -1 are more resistant to breakage following tubulin depletion, 325 suggesting a stabilizing role for MAP65-1-mediated bundling under stress conditions (Fig. 326 6c). Interestingly, an alternative scenario where tensile stress would promote the 327 expanded lattice conformation of microtubules and further recruitment of MAP65 was 328 not supported by our experiments. MAP65-1-dependent stress perception would thus be 329 mediated through the accumulation of defects and lattice damage. In this scenario, 330 MAP65-1 recruitment to damaged microtubules would serve a dual purpose: it would 331 acutely reinforce microtubules in response to mechanical stress, and, because the 332 resulting bundles are particularly stable, it would also leave a lasting record of past stress 333 events within the cell. This would, in turn, make the microtubule array align better with 334 stress. 335 The exact molecular mechanism by which MAP65 -1/PRC1 promote microtubule 336 nucleation remains to be elucidated. Since the tubulin concentration on the lattice was 337 not high enough to explain the spatial frequency of nucleation activity, it is likely that 338 MAP65-1/PRC1 stabilize nucleation intermediates by enhancing lateral or longitudinal 339 a^inity between tubulin dimers 17. Accordingly, the yeast homolog Ase1 has been 340 proposed to reduce the detachment of terminal tubulin subunits at depolymerizing 341 microtubule tips 37, which shows that MAP65 -1/PRC1 could influence the tubulin 342 dissociation rate from the microtubule lattice. 343 Note that further synergies may be envisioned here. In particular, higher osmolarity does 344 not only reduce cortical tension, but also leads to reduced microtubule growth 38. Thus, 345 in non -pressurized cells, slow -growing microtubules might experience less defects in 346 their lattice, leading to reduced recruitment of MAP65 -1. The resulting sti^er 347 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 15, 2025. ; https://doi.org/10.1101/2025.09.14.676082doi: bioRxiv preprint microtubules would be more sensitive to cell geometry and align with the flat part of the 348 cell cortex, by default. 349 The mechanism by which MAP65 -1 recognizes bent microtubules seems to di^er from 350 other MAPs that have been shown to bind bent microtubules through a preference for the 351 expanded or compacted lattice states 27,28,39. Although we did not observe preferential 352 binding to the taxol -expanded lattice, MAP65-1 exhibited a preference for the GMPCPP 353 lattice. This suggests that MAP65-1 may recognize bent microtubules through alternative 354 features of the microtubule lattice that are not explained solely by expansion or 355 compaction.28,37,38 Furthermore, one interesting possibility is that MAP65 -1 detects 356 microtubules under tension, a hypothesis that remains to be tested. 357 Because MAP65-1/PRC1 have not been described to preferentially bind GTP-tubulin, the 358 recognition of lattice defects also seems di^erent from other MAPs that recognize defect 359 sites through the recognition of GTP-tubulin incorporation, like CLASP , CLIP-170, and the 360 EBs40–43. Moreover, MAP65-1 may recognize bent microtubules through their increased 361 lattice damage, as bent microtubules show much higher levels of repair (with an 362 increased incorporation of tubulin from solution) along their lattice than straight ones31. 363 Finally, the interplay of MAP65 -1/PRC1 with other MAPs that are responsive to 364 mechanical stress is likely complex in cells. For example, PRC1 and Ase1 interact with 365 and recruit CLASP44,45, and microtubule bundles formed by MAP65-1 are protected from 366 severing by KATANIN 46. Recent simulations of plant cortical microtubule arrays also 367 support a key role for bundling in helping microtubule self -organization together with 368 KATANIN’s severing function 47. Therefore, MAP65 -1 might further contribute to the 369 microtubule response to mechanical stress in cells by recruiting CLASP and counter -370 acting the severing by KATANIN. Overall, it will be interesting to test how depletion of 371 MAP65 in plant cells can a^ect their response to changes in mechanical stress, although 372 this is a challenging endeavor because the MAP65 family has nine members in A. 373 thaliana48. Future work could address whether other MAP65 members also localize to 374 bent microtubules or promote nucleation, clarifying whether they have a conserved or 375 specialized role. 376 Together, our findings position MAP65 -1 and PRC1 as key players in the dynamic 377 regulation of the microtubule cytoskeleton, contributing to a form of cytoskeletal 378 “memory” , where mechanical stress-induced microtubule patterns are reinforced and 379 likely maintained over time. Since both MAP65-1 and PRC1 have conserved functions in 380 cell division, it also remains open whether the nucleation mechanism described here 381 contributes to spindle or phragmoplast formation and has further implications in cell 382 proliferation that were previously unknown. Thus, both proteins have the potential to 383 actively contribute to tissue morphogenesis across the eukaryotic kingdom – particularly 384 in plants, where growth anisotropy mainly relies on the cortical microtubule -cellulose 385 deposition nexus. 386 387 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 15, 2025. ; https://doi.org/10.1101/2025.09.14.676082doi: bioRxiv preprint

388 MAP65-1, PRC1 and KIF11 purification 389 Plasmids encoding His -MAP65-1-His (referred to as MAP65 -1) and GFP -MAP65-1-His 390 (referred to as GFP -MAP65-1) with MAP65-1 sequences from Arabidopsis thaliana were 391 previously generated24. The recombinant proteins were purified from Rosetta 2(DE3) E. 392 coli cells. Bacteria were grown at 37 °C to an OD of 0.5 followed by transfer to 20 °C for 393 one hour. Protein expression continued at 20 °C overnight with 0.5 mM IPTG. The next day, 394 cells were collected by centrifugation and frozen. Cell pellets were resuspended in lysis 395 bu^er containing 50 mM Sodium Phosphate Bu^er, pH 7.9, 200 mM NaCl, 20 mM 396 imidazole, 0.5% Triton X -100, 0.5 mM DTT and a protease inhibitor cocktail (cOmplete, 397 EDTA-free). Samples were then sonicated in a beaker on ice (Bandelin Sonopuls, 9X 20 398 seconds on/o^, 40% duty cycle). The lysate was then centrifuged for 30 min at 16,000 399 rpm at 4 °C and applied to a column containing Ni Sepharose High Performance beads 400 (Cytiva Life Sciences). The column was washed with a bu^er containing 50 mM Sodium 401 Phosphate Bu^er, pH 7.9, 100 mM NaCl, 30 mM imidazole and 0.5 mM DTT. The proteins 402 were eluted with 500 mM imidazole and dialyzed overnight in a bu^er containing 50 mM 403 Sodium Phosphate Bu^er, pH 7.9, 100 mM NaCl and 0.5 mM DTT. Further purification 404 followed by loading the protein on a gel filtration column (HiLoad 16/600 Superdex 200 405 pg, Cytiva Life Sciences) connected to the NGC Chromatography system (Bio-Rad) in 50 406 mM Sodium Phosphate Bu^er, pH 7.9, 100 mM NaCl and 0.5 mM DTT. Proteins were 407 concentrated using Amicon Ultra-15 Centrifugal Filters (Merck). 408 Human PRC1 was expressed for 96 h in SF9 cells from a pOCC7 plasmid encoding PRC1 409 labeled with a His6 tag and a 3C precision cleavage site. For purification, cell pellets were 410 thawed on ice and resuspended in purification bu^er (50 mM NaH2PO4, 500 mM NaCl, 411 2 mM MgCl2, 1 mM DTT, 0.1% Tween20, pH 7.8) with protease inhibitor. The lysate was 412 cleared with an ultracentrifuge spin with 40,000 rpm for 1 h at 4 °C. The supernatant was 413 filtered through a 0.45 μm filter and loaded on a 1 mL HiTrap column with a superloop. 414 The column was washed with IMAC wash bu^er (purification bu^er with 20 mM 415 imidazole) and the protein was eluted with IMAC elution bu^er (purification bu^er with 416 300 mM imidazole) with an elution gradient. Protein -containing fractions were pooled 417 and concentrated with Amicon filters (cuto^ 100 kDa). 3C protease was added (1:150, 418 v/v) and the His6 tag was cleaved overnight at 4°C. The protein solution was diluted 6-fold 419 to reduce the imidazole concentration and passed over the HiTrap column again. The 420 protease remained bound to the column with its His6 tag. The flow through was 421 concentrated to 0.5 mL, cleared at 17,000 g for 10 min and gel-filtered over a Superose6 422 column with purification bu^er. 10% glycerol was added, and the protein was flash frozen 423 in liquid nitrogen and stored at -80°C. 424 Human KIF11 was expressed and purified as previously described49. 425 426 Tubulin purification and labeling 427 Fresh bovine brains were used as the source of brain tubulin, which was purified by three 428 cycles of temperature-dependent polymerization and depolymerization in Brinkley Bu^er 429 80 (BRB80 bu^er; composed of 80 mM PIPES, pH 6.8, 1 mM EGTA and 1 mM MgCl 2 430 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 15, 2025. ; https://doi.org/10.1101/2025.09.14.676082doi: bioRxiv preprint supplemented with 1 mM GTP). We obtained MAP -free tubulin by using low (32.6% 431 glycerol, 1.5 mM ATP , 0.5 mM GTP , 3 mM MgCl2) and high salt bu^ers (High Molarity PIPES 432 bu^er; 1 M PIPES, pH 6.9, adjusted with KOH, 10 mM MgCl 2, 20 mM EGTA) and cation -433 exchange chromatography (EMD SO, 650 M, Merck) in 50 mM PIPES, pH 6.8, 1 mM MgCl2 434 and 1 mM EGTA. 435 For the labeling (with ATTO-488, ATTO-565, Alexa-Fluor-647 or biotin), microtubules were 436 polymerized with purified brain tubulin at 37 °C for 30 min and layered onto cushions of 437 0.1 M NaHEPES, pH 8.6, 1 mM MgCl 2, 1 mM EGTA, 60% glycerol, followed by 438 sedimentation by ultracentrifugation at 37 °C. Microtubules were then resuspended in 439 0.1 M NaHEPES, pH 8.6, 1 mM MgCl 2, 1 mM EGTA, 40% glycerol and labeled by adding 440 1:10 volume 100 mM NHS-ATTO (ATTO Tec) or NHS-LC-LC-Biotin (EZ-link, Thermo) for 10 441 min at 37 °C. Two volumes of 2X BRB80 with 100 mM potassium glutamate and 40% 442 glycerol were used to stop the labeling reaction, followed by microtubule sedimentation 443 onto cushions of BRB80 supplemented with 60% glycerol. Finally, microtubules were 444 resuspended in BRB80, and a last cycle of polymerization and depolymerization was 445 performed before storage. 446 447 Coverslip treatment 448 Coverslips were cleaned by successive treatment with the following solutions: 30 min 449 acetone and 15 min 96% ethanol followed by two washes with ultrapure water, then 2 450 hours in Hellmanex III (2% in water) followed by two washes with ultrapure water. 451 Coverslips were then airdried and treated with UV for 25 min. Next, coverslips were 452 incubated for 3 days in a solution containing a 1:9 mix of triethoxysilane -PEG-biotin and 453 triethoxysilane-PEG (30 kDa, Creative PEGWorks) at 1 mg/ml in 96% ethanol and 0.1% 454 HCl with gentle agitation at room temperature. Coverslips were then rinsed once in 455 absolute ethanol and twice in ultrapure water, airdried and stored at 4 °C. 456 457 Microtubule growth, capping and nucleation dynamics 458 Microtubules seeds were polymerized in a total volume of 100 μl with 6 μM tubulin 459 (labeled with 30% ATTO-565 or Alexa Fluor 647 and 70% biotinylated tubulin) in 1X BRB80 460 supplemented with 0.5 mM GMPCPP at 37 °C for 1 hour. 2 μl of 50 μM taxol (Merck) were 461 then added followed by incubation at room temperature for 30 min and 462 ultracentrifugation at 156,000 xg at 25 °C for 10 min. Seeds were then resuspended in 1X 463 BRB80 supplemented with 0.5 mM GMPCPP and 1 μM taxol. 464 A flow cell chamber with a volume of approximately 20 μl was built using double -sided 465 adhesive tape and a glass coverslip functionalized and passivated as mentioned above. 466 The top and bottom pieces were cut into the desired sizes using a diamond engraving 467 pen. Flow chambers were first perfused with 50 μg/ml Neutravidin (Fisher Scientific) in 468 1X BRB80 for 1 min, followed by passivation with 0.1 mg/ml PLL -g-PEG (PII 20 K -G35-469 PEG2K, Jenkam Technology) in 10 mM Na-HEPES, pH 7.4, for 1 min and washed with 1X 470 BRB80. Microtubules seeds were then flushed into the chamber. Non -attached seeds 471 were washed out by using 1X BRB80 supplemented with 1 mg/ml casein (BRB80/casein). 472 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 15, 2025. ; https://doi.org/10.1101/2025.09.14.676082doi: bioRxiv preprint Microtubule polymerization with seeds as template was achieved with a mix containing 473 11 μM tubulin (10 to 20% labeled with ATTO -565 or Alexa Fluor 647) in 0.7X BRB80 and 474 0.38X MAP bu^er (500 mM Phosphate bu^er, 1 mM KCl, 10 mM DTT, pH 7.9) 475 supplemented with 1 mM GTP , an oxygen scavenger cocktail (22 mM DTT, 1.2 mg/ml 476 glucose, 8 μg/ml catalase and 40 μg/ml glucose oxidase), 1 mg/ml casein and 0.033% 477 methyl cellulose (1,500 cP , Sigma) at 37 °C. Microtubules were capped by substituting 478 GTP with 0.5 mM GMPCPP (Jena Bioscience) and using 3 µM 100% labeled tubulin (with 479 ATTO-565 or Alexa Fluor 647) at 37 °C. Capping microtubules extends their lifetime for 480 long-term observation (for instance up to 30 min) and overcomes dynamic instability in 481 vitro. To observe microtubule nucleation in the presence of PRC1 or MAP65 -1, the same 482 bu^er as in microtubule polymerization was used, supplemented with 4 µM tubulin 100% 483 labeled with ATTO-488 and the corresponding protein at the desired concentration (the 484 protein stock was diluted in 1X BRB80). Microtubule nucleation dynamics in the presence 485 of PRC1/MAP65-1 was observed for 30 minutes, unless stated otherwise. 486 For the experiments varying the amount of microtubule defects: for fast growth 487 conditions, 11 µM tubulin was used for polymerization; for slow growth conditions, 5.5 488 µM tubulin was used. 489 To observe stabilized microtubules nucleated by varying MAP65 -1 concentrations, 490 microtubule incubation with 100% ATTO-488-labeled tubulin and MAP65-1 (in the same 491 bu^er as for microtubule polymerization) proceeded for 20 min, followed by a wash 492 bu^er (with the same composition as the microtubule polymerization bu^er, but without 493 free tubulin and with 20 µM taxol). 494 495 Nucleated microtubule transport by KIF11 496 Microtubules were polymerized in microcentrifuge tubes as described below with 10% 497 labeled ATTO-565 and with biotin on their caps and seeds. After washing away taxol, 498 microtubule nucleation proceeded for 30 min in the presence of 100 nM MAP65 -1 and 4 499 µM 100% ATTO-488-labeled tubulin followed by flushing of a kinesin bu^er containing 10 500 nM KIF11, 2 µM of tubulin 100% labeled with Alexa Fluor 647 and 0.7X BRB80 501 supplemented with 1 mM GTP , 1 mM ATP , an oxygen scavanger cocktail (22 mM DTT, 1.2 502 mg/ml glucose, 8 μg/ml catalase and 40 μg/ml glucose oxidase), 1 mg/ml casein and 503 0.033% methyl cellulose (1,500 cP , Sigma). Microtubules were observed for 20 min. 504 505 Microtubule polymerization for bending and taxol treatment 506 Microtubules were first polymerized in microcentrifuge tubes by using 11 µM tubulin 507 (labeled with 10% ATTO -565 or Alexa Fluor 647) in 200 µl of a bu^er containing 1.2X 508 BRB80, 0.6X MAP bu^er, 0.5 mM GTP and previously polymerized seeds for 40 min at 37 509 °C. 5 µl of 30 µM taxol were then added, following by centrifugation for 30 min at 15,000 510 rpm at room temperature. Microtubules were then resuspended in capping mix 511 containing 0.5 µM tubulin (labeled with 60% biotin and 40% ATTO-565 or Alexa Fluor 647) 512 in 1.2X BRB80, 0.6X MAP bu^er, 0.5 mM GMPCPP and 10 µM taxol. Stepwise capping of 513 microtubules was achieved by adding 0.5 µM tubulin at a time followed by incubation for 514 15 min at 37 °C for a total of ten times. Microtubules were diluted 1:200 in BRB80/taxol 515 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 15, 2025. ; https://doi.org/10.1101/2025.09.14.676082doi: bioRxiv preprint (1X BRB80, 10 µM taxol) until usage. The same conditions were used for microtubules 516 referred to as taxol pre-treated, which were incubated overnight at room temperature in 517 BRB80/taxol. 518 Microtubules were flushed into passivated and functionalized chambers as described 519 above, followed by successive addition of 100 µl BRB80/taxol from both ends of the flow 520 chamber in an alternate fashion to cause microtubule bending due to fluid flow, since 521 microtubules could attach via the seed and the cap to the coverslips. Taxol was washed 522 away by using BRB80/casein. To observe GFP -MAP65-1 binding to bent microtubules, 523 MAP65-1 was added with 15% GFP -MAP65-1 and 85% MAP65 -1 in the presence of 2 µM 524 non-fluorescent tubulin. To observe microtubule nucleation on bent microtubules, 525 MAP65-1-mediated nucleation proceeded for 20 min followed by stabilization with a 526 wash bu^er with no tubulin and 10 µM taxol. 527 528 Microtubule nucleation using annealed microtubule population 529 Microtubules were polymerized in a bu^er containing 1X BRB80, 1.25 mM GMPCPP , 1.25 530 mM MgCl2 and 2.5 µM tubulin (labeled with 20% ATTO -565 or Alexa Fluor 647 and 20% 531 biotin) for 5 hours at 28 °C, followed by addition of 120 µl 1X BRB80 and centrifugation at 532 13,000 rpm for 15 min. Microtubules were then resuspended in 150 µl of 1X BRB80 and 533 mixed in equal amounts, followed by incubation at 30 °C overnight to allow for annealing 534 to happen. Microtubules were then flushed into flow chambers and MAP65 -1-mediated 535 nucleation proceeded by 20 min followed by stabilization with a wash bu^er with no 536 tubulin and 10 µM taxol. 537 538 Microtubule nucleation using a population of microtubules with high and low defect 539 regime 540 For the high defect regime, microtubules were polymerized in 10 µl of a bu^er containing 541 1X BRB80, 2 mM GMPCPP , 0.1 mM MgCl and 20 µM tubulin (labeled with 20% Alexa Fluor 542 647 and 20% biotin) for 30 min at 37 °C, followed by addition of 190 µl of 1X BRB80 and 543 centrifugation at 156,000 xg for 15 min. Microtubules were then resuspended in 150 µl of 544 1X BRB80. For the low defect regime, microtubules were polymerized in 80 µl of a bu^er 545 containing 1X BRB80, 1.25 mM GMPCPP , 1.25 mM MgCl2 and 2.5 µM tubulin (labeled with 546 20% ATTO-565 and 20% biotin) for 5 hours at 28 °C, followed by addition of 120 µl of 1X 547 BRB80 and centrifugation at 13,000 rpm for 15 min. Microtubules were then resuspended 548 in 150 µl of 1X BRB80. Microtubules were then flushed in sequentially and MAP65 -1-549 mediated nucleation proceeded by 20 min followed by stabilization with a wash bu^er 550 with no tubulin and 10 µM taxol. 551 552 Plant growth conditions 553 Arabidopsis thaliana seeds were surface -sterilized by treatment with a solution 554 containing 2% bleach and 0.05% Triton X -100 for 5 min followed by three washes with 555 sterile distilled water and resuspension in 0.05% agarose. Seeds were sown on ½ MS 556 medium (basal salt mixture, Duchefa Biochemie) supplemented with 0.5% sucrose and 557 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 15, 2025. ; https://doi.org/10.1101/2025.09.14.676082doi: bioRxiv preprint 0.8% plant agar (Duchefa Biochemie). Plates containing seeds were stratified at 4 °C for 558 2 to 3 days in the dark. Plants were grown in a 16-hour/21 °C light and 8-hour/18 °C dark 559 regime with 60% humidity. 560 561 Protoplast isolation and observation 562 Roots from seedlings grown for eleven days were dissected and inserted in an enzyme 563 solution composed of solution A (2 mM CaCl2, 2 mM MgCl2,10 mM MES, pH 5.5 adjusted 564 with KOH, and 600 mOsmol/l Mannitol) and the following enzymes: 17 mg/ml Cellulysine 565 (Calbiochem), 17 mg/ml Cellulase RS (Duchefa Biochemie) and 0.4 mg/ml Pectyolase 566 from Aspergillus japonicus (Sigma-Aldrich). Roots were digested for two to four hours at 567 room temperature on a rotating stage at 15 rpm. Next, the solution was filtered through a 568 70 µm filter and the filter was washed with 1 ml of solution A. The filtered protoplasts 569 were then centrifuged for 4 min at 1,000 rpm. The supernatant was removed and 1 ml of 570 solution A was added, followed by gentle flicking. The protoplasts were centrifuged again 571 for 4 min at 1,000 rpm, followed by supernatant removal and resuspension in 200 µl of 572 solution A. The protoplasts were then applied on microwells and allowed to sediment for 573 10 min. 3 ml of solution B (the same as solution A, but with 280 mOsmol/l Mannitol) were 574 then added to promote protoplast pressurization. Microscopic observation started 2 575 hours after treatment with solution B and continued for another 2 hours. The 12 X 40 µm 576 microwells were produced as previously described 25. We selected protoplasts that had 577 an aspect ratio of at least 1.1 (major axis divided by minor axis) to make sure that only 578 enclosed protoplasts were taken into account for our analysis. We included protoplasts 579 with an aspect ratio range of 1.28 to 1.12. 580 581 Imaging conditions and image analysis 582 To observe plant cells, a point scanning confocal microscope (Zeiss LSM 900) with 583 Airyscan 2 and Axiocam 705 camera was used. In vitro microtubules were either 584 visualized with the Zeiss LSM 900 with a stage that was kept at 37 °C through a cage 585 incubator (PECON), or an objective -based orbital TIRF microscope (Nikon Eclipse Ti2, 586 modified by Visitron Systems) and EMCCD camera (Andor iXon Life) at minimal laser 587 intensity with a stage kept at 37 °C through a warm stage controller (OkoLabs). For the 588 Zeiss LSM 900 microscope, the ZenBlue software version 3.2 was used. For the TIRF 589 microscope, the VisiView software version 6.0 was used. 590 To observe microtubule nucleation dynamics in the presence of PRC1/MAP65-1, images 591 were taken every 4 s for a total of 30 minutes. Nucleation dynamics were observed 592 through kymographs. The distance between nucleation events was measured manually 593 using the segmented line on Fiji. 594 To distinguish between incorporation and nucleation events in the case of stabilized 595 microtubules, typically 50 images were taken per field of view (FOV) with an interval of 596 0.5 to 1 s. Images were processed for background subtraction and smoothing. Line scans 597 of the green fluorescence intensity were drawn along maximum intensity projections of 598 individual microtubules. The elongation of microtubule ends was used as a reference for 599 a full microtubule for every FOV – an event that surpassed the intensity found at the 600 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 15, 2025. ; https://doi.org/10.1101/2025.09.14.676082doi: bioRxiv preprint polymerized microtubule ends was scored as a nucleation event if it co -localized with 601 the microtubule lattice. 602 To quantify the recruitment of free tubulin on the microtubule lattice by MAP65-1, a 1-µm 603 section in length spanning the whole microtubule width was drawn using the polygon tool 604 in Fiji. This section was moved along the microtubule from the beginning to the end. For 605 each individual microtubule, the integrated density measured from each microtubule 606 section was divided by the integrated density in a background region with the same area 607 right next to the microtubule. 608 To assess whether microtubule nucleation sites overlapped with microtubule annealing 609 sites, we measured the distance between the peak maximum of the nucleation 610 fluorescence signal and the nearest annealing site. If this distance was less than 500 nm, 611 the nucleation site was classified as overlapping with the annealing site. As a control, we 612 selected 55 random 500 nm -long patches along annealed microtubules. A patch was 613 scored as containing a nucleation event if the peak maximum of a nucleation 614 fluorescence signal fell within the 500 nm interval. 615 616 FRAP assay 617 The GFP-MAP65-1 and free tubulin signals were bleached by using 5 cycles of the 405 618 laser line at 100% at a speed of 50 ms per pixel. Images were acquired every 0.9 s for 5 619 min. Bleaching events that resulted in less than 30% of the initial fluorescence intensity 620 were used for the analysis. The fluorescence intensity was normalized to the initial 621 maximum value and plotted over time by using the Stowers Plugins Collection 622 (https://research.stowers.org/imagejplugins) after background subtraction (rolling ball 623 50 pixels). 624 625 Microtubule curvature and orientation analysis 626 Microtubules were tracked with the JFilament Fiji plugin 627 (https://imagej.net/plugins/jfilament). Using custom-written python code, the curves of 628 the tracked filaments were first smoothed by parametric Spline interpolation. Then, the 629 menger curvature was computed and the orientation was calculated via finite di^erence. 630 GFP-MBD and MAP65 -1-RFP fluorescence values were normalized for every protoplast 631 against the mean fluorescence intensity (for each corresponding channel) in a circle with 632 a diameter of 2 µm drawn in a region of the cytoplasm with no microtubules present. 633 634 Statistical methods 635 All statistical analyses were performed with GraphPad Prism. For the spatial frequency 636 quantifications, microtubules for each experimental condition were concatenated in a 637 random order and the distance between two adjacent nucleation events was measured 638 (similarly to what was done previously for the quantification of spatial incorporation 639 frequency32). 640 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 15, 2025. ; https://doi.org/10.1101/2025.09.14.676082doi: bioRxiv preprint Data sets were tested for their normality. All of the tested datasets had at least one group 641 that was non -normally distributed, thus non -parametric tests were employed to test if 642 di^erences were statistically significant (Mann-Whitney and Kruskal-Wallis). 643 To test if the microtubule nucleation frequency di^ered in regions containing microtubule 644 annealing sites compared to randomly selected patches of the annealed microtubule 645 population, a Fisher’s exact test was used. 646 647 Data availability 648 The datasets generated and analyzed in this study are available from the corresponding 649 authors upon request. 650 651

652 1. Mitchison, T. & Kirschner, M. Dynamic instability of microtubule growth. Nature 653 312, 237–242 (1984). 654 2. Burbank, K. S. & Mitchison, T. J. Microtubule dynamic instability. Current Biology 655 16, R516–R517 (2006). 656 3. Kuo, Y.-W. & Howard, J. Cutting, Amplifying, and Aligning Microtubules with 657 Severing Enzymes. Trends Cell Biol 31, 50–61 (2021). 658 4. Walczak, C. E. & Shaw, S. L. A MAP for Bundling Microtubules. Cell 142, 364–367 659 (2010). 660 5. Tovey, C. A. & Conduit, P . T. Microtubule nucleation by γ-tubulin complexes and 661 beyond. Essays Biochem 62, 765–780 (2018). 662 6. Hamant, O., Inoue, D., Bouchez, D., Dumais, J. & Mjolsness, E. Are microtubules 663 tension sensors? Nat Commun 10, 2360 (2019). 664 7. Matis, M. The Mechanical Role of Microtubules in Tissue Remodeling. BioEssays 665 42, 1900244 (2020). 666 8. Ju, R. J. et al. Compression-dependent microtubule reinforcement enables cells 667 to navigate confined environments. Nat Cell Biol 26, 1520–1534 (2024). 668 9. Lechler, T. & Mapelli, M. Spindle positioning and its impact on vertebrate tissue 669 architecture and cell fate. Nat Rev Mol Cell Biol 22, 691–708 (2021). 670 10. Seetharaman, S. et al. Microtubules tune mechanosensitive cell responses. Nat 671 Mater 21, 366–377 (2022). 672 11. Hamant, O. et al. Developmental Patterning by Mechanical Signals in 673 Arabidopsis. Science (1979) 322, 1650–1655 (2008). 674 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 15, 2025. ; https://doi.org/10.1101/2025.09.14.676082doi: bioRxiv preprint 12. Yan, Y ., Sun, Z., Yan, P ., Wang, T. & Zhang, Y . Mechanical regulation of cortical 675 microtubules in plant cells. New Phytologist 239, 1609–1621 (2023). 676 13. Bringmann, M. et al. Cracking the elusive alignment hypothesis: the microtubule–677 cellulose synthase nexus unraveled. Trends Plant Sci 17, 666–674 (2012). 678 14. Lee, Y .-R. J. & Liu, B. Microtubule nucleation for the assembly of acentrosomal 679 microtubule arrays in plant cells. New Phytologist 222, 1705–1718 (2019). 680 15. Vinopal, S. & Bradke, F . Centrosomal and acentrosomal microtubule nucleation 681 during neuronal development. Curr Opin Neurobiol 92, 103016 (2025). 682 16. Sallee, M. D. & Feldman, J. L. Microtubule organization across cell types and 683 states. Current Biology 31, R506–R511 (2021). 684 17. Roostalu, J. & Surrey, T. Microtubule nucleation: beyond the template. Nat Rev 685 Mol Cell Biol 18, 702–710 (2017). 686 18. Wadsworth, P . The multifunctional spindle midzone in vertebrate cells at a glance. 687 J Cell Sci 134, jcs250001 (2021). 688 19. Ehrhardt, D. W. Straighten up and fly right—microtubule dynamics and 689 organization of non-centrosomal arrays in higher plants. Curr Opin Cell Biol 20, 690 107–116 (2008). 691 20. Chan, J., Jensen, C. G., Jensen, L. C. W., Bush, M. & Lloyd, C. W. The 65-kDa carrot 692 microtubule-associated protein forms regularly arranged filamentous cross-693 bridges between microtubules. Proceedings of the National Academy of Sciences 694 96, 14931–14936 (1999). 695 21. Mollinari, C. et al. PRC1 is a microtubule binding and bundling protein essential to 696 maintain the mitotic spindle midzone. Journal of Cell Biology 157, 1175–1186 697 (2002). 698 22. Portran, D. et al. MAP65/Ase1 promote microtubule flexibility. Mol Biol Cell 24, 699 1964–1973 (2013). 700 23. Basnet, N. et al. Direct induction of microtubule branching by microtubule 701 nucleation factor SSNA1. Nat Cell Biol 20, 1172–1180 (2018). 702 24. Gaillard, J. et al. Two Microtubule-associated Proteins of Arabidopsis MAP65s 703 Promote Antiparallel Microtubule Bundling. Mol Biol Cell 19, 4534–4544 (2008). 704 25. Colin, L. et al. Cortical tension overrides geometrical cues to orient microtubules 705 in confined protoplasts. Proceedings of the National Academy of Sciences 117, 706 32731–32738 (2020). 707 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 15, 2025. ; https://doi.org/10.1101/2025.09.14.676082doi: bioRxiv preprint 26. Lindeboom, J. J. et al. A Mechanism for Reorientation of Cortical Microtubule 708 Arrays Driven by Microtubule Severing. Science (1979) 342, 1245533 (2013). 709 27. Paquette, A. L. et al. Competition for microtubule lattice spacing between a 710 microtubule expander and compactor. bioRxiv 2025.02.28.640185 (2025) 711 doi:10.1101/2025.02.28.640185. 712 28. Bechstedt, S., Lu, K. & Brouhard, G. J. Doublecortin Recognizes the Longitudinal 713 Curvature of the Microtubule End and Lattice. Current Biology 24, 2366–2375 714 (2014). 715 29. Alushin, G. M. et al. High-Resolution Microtubule Structures Reveal the Structural 716 Transitions in αβ-Tubulin upon GTP Hydrolysis. Cell 157, 1117–1129 (2014). 717 30. Liu, H. & Shima, T. Preference of CAMSAP3 for expanded microtubule lattice 718 contributes to stabilization of the minus end. Life Sci Alliance 6, e202201714 719 (2023). 720 31. Nandakumar, S. et al. Kinesin-induced buckling reveals the limits of microtubule 721 self-repair. bioRxiv 2025.09.08.672697 (2025) doi:10.1101/2025.09.08.672697. 722 32. Schaedel, L. et al. Lattice defects induce microtubule self-renewal. Nat Phys 15, 723 830–838 (2019). 724 33. Reid, T. A., Coombes, C. & Gardner, M. K. Manipulation and quantification of 725 microtubule lattice integrity. Biol Open 6, 1245–1256 (2017). 726 34. Reuther, C., Santos-Otte, P ., Grover, R., Korten, T. & Diez, S. Microtubule lattice 727 defects facilitate spastin-mediated severing. bioRxiv 2025.08.29.673059 (2025) 728 doi:10.1101/2025.08.29.673059. 729 35. Trushko, A., Schä^er, E. & Howard, J. The growth speed of microtubules with 730 XMAP215-coated beads coupled to their ends is increased by tensile force. 731 Proceedings of the National Academy of Sciences 110, 14670–14675 (2013). 732 36. Hervieux, N. et al. A Mechanical Feedback Restricts Sepal Growth and Shape in 733 Arabidopsis. Current Biology 26, 1019–1028 (2016). 734 37. Krattenmacher, J. et al. Ase1 selectively increases the lifetime of antiparallel 735 microtubule overlaps. Current Biology 34, 4071-4080.e6 (2024). 736 38. Molines, A. T. et al. Physical properties of the cytoplasm modulate the rates of 737 microtubule polymerization and depolymerization. Dev Cell 57, 466-479.e6 738 (2022). 739 39. Peet, D. R., Burroughs, N. J. & Cross, R. A. Kinesin expands and stabilizes the 740 GDP-microtubule lattice. Nat Nanotechnol 13, 386–391 (2018). 741 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 15, 2025. ; https://doi.org/10.1101/2025.09.14.676082doi: bioRxiv preprint 40. Aher, A. et al. CLASP Mediates Microtubule Repair by Restricting Lattice Damage 742 and Regulating Tubulin Incorporation. Current Biology 30, 2175-2183.e6 (2020). 743 41. Li, Y . et al. Compressive forces stabilize microtubules in living cells. Nat Mater 22, 744 913–924 (2023). 745 42. de Forges, H. et al. Localized Mechanical Stress Promotes Microtubule Rescue. 746 Current Biology 26, 3399–3406 (2016). 747 43. Aumeier, C. et al. Self-repair promotes microtubule rescue. Nat Cell Biol 18, 748 1054–1064 (2016). 749 44. Liu, J. et al. PRC1 Cooperates with CLASP1 to Organize Central Spindle Plasticity 750 in Mitosis. Journal of Biological Chemistry 284, 23059–23071 (2009). 751 45. Bratman, S. V & Chang, F . Stabilization of Overlapping Microtubules by Fission 752 Yeast CLASP. Dev Cell 13, 812–827 (2007). 753 46. Burkart, G. M. & Dixit, R. Microtubule bundling by MAP65-1 protects against 754 severing by inhibiting the binding of katanin. Mol Biol Cell 30, 1587–1597 (2019). 755 47. Deinum, E. E., Tindemans, S. H., Lindeboom, J. J. & Mulder, B. M. How selective 756 severing by katanin promotes order in the plant cortical microtubule array. 757 Proceedings of the National Academy of Sciences 114, 6942–6947 (2017). 758 48. Smertenko, A. P . et al. The C-Terminal Variable Region Specifies the Dynamic 759 Properties of Arabidopsis Microtubule-Associated Protein MAP65 Isotypes. Plant 760 Cell 20, 3346–3358 (2008). 761 49. Meißner, L., Niese, L., Schüring, I., Mitra, A. & Diez, S. Human kinesin-5 KIF11 762 drives the helical motion of anti-parallel and parallel microtubules around each 763 other. EMBO J 43, 1244-1256–1256 (2024). 764 765 Acknowledgments 766 MRM and OH were supported by a grant from the European Research Council (ERC, grant 767 agreement number 101019515, “Musix” , awarded to OH). MRM was also supported by a 768 grant from the German Research Foundation (DFG, project number 545084341, awarded 769 to MRM). LS was supported by the DFG grant SFB 1027 and the ERC grant StG 101115795 770 “CROSSTALK”. 771 772 Author contributions 773 MRM, LS, OH and SD conceived and guided the project. MRM, LS, OH and SD designed 774 the experiments. MRM and CG performed the experiments. MG and BK purified the 775 tubulin. LM, LN and SD provided KIF11 and PRC1 and related expertise. MRM and CG 776 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 15, 2025. ; https://doi.org/10.1101/2025.09.14.676082doi: bioRxiv preprint analyzed the experiments. MRM, LS and OH wrote the manuscript with input from all 777 authors. 778 779 Competing interest declaration 780 The authors declare no competing interests. 781 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 15, 2025. ; https://doi.org/10.1101/2025.09.14.676082doi: bioRxiv preprint .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 15, 2025. ; https://doi.org/10.1101/2025.09.14.676082doi: bioRxiv preprint Fig. 1: MAP65-1/PRC1 promote microtubule nucleation on existing lattices a, Schematic representation of the experimental setup used to study the function of MAP65-1 and PRC1 in lattice-templated microtubule appearance. Microtubules were grown with ATTO-565-labeled tubulin at a concentration of 11 μM (step I) before they were capped with GMPCPP (step II) and exposed to ATTO-488-labeled free tubulin at a concentration of 4 μM (step III). Step III was observed live for a total of 30 minutes. b, Schematic representation of the flow chambers used as an experimental setup throughout the manuscript. 20-μl flow chambers were built from passivated coverslips cut with a diamond pen and attached by double-sided tape. Solutions were sequentially flushed in through one side of the flow chamber. After step III shown in a, flow chambers were sealed to prevent evaporation and observed for 30 min. c, Microtubule dynamics imaged by TIRFM in the presence of ATTO-488-labeled free tubulin (green) and ATTO-565 microtubules (magenta). In the control (top), only growth from the free microtubule ends (elongation) was observed. In the presence of both 100 nM MAP65-1 (middle) and 300 pM PRC1 (bottom), microtubules appeared and grew in parallel to the existing lattices (microtubule appearance events are indicated with a white star). Scale bars, 5 µm. d, Schematic representation of lattice-templated microtubule appearance. Microtubule appearance and growth could be clearly distinguished from elongation at the free ends because it occurred overlapping with the original microtubule lattice “template”. e, Kymograph of the corresponding stills in c in the presence of MAP65-1. Appearance events are indicated with a white star. Scale bars, 5 µm (horizontal) and 1 min (vertical). f, Experimental setup used to test if microtubules that appear in the presence of MAP65-1 are independent from the “template” microtubule lattice. 10 nM KIF11 was added in the presence of 2 μM free tubulin, and some of the new microtubules were effectively used as cargo. g, KIF11-mediated microtubule sliding observed by TIRFM after incubation of microtubules with 100 nM MAP65-1 and 4 μM ATTO-488-labeled tubulin. Scale bar, 5 μm. h, Kymograph of the corresponding microtubule shown in g. Scale bars, 5 μm (horizontal) and 1 min (vertical). i, Quantification of nucleated microtubule behavior (mobile or static) upon addition of KIF11 (n = 24 nucleated microtubules, N = 3). j, Schematic representation of the experimental setup used to study PRC1-mediated microtubule nucleation in k. The same steps were taken as in a, except microtubule seeds and caps were polymerized with Alexa-Fluor-647-labeled tubulin (blue) instead of ATTO 565. k, Microtubule dynamics in the absence (control) and in the presence of 1 nM PRC1 observed by TIRFM. Scale bar, 50 μm. .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 15, 2025. ; https://doi.org/10.1101/2025.09.14.676082doi: bioRxiv preprint .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 15, 2025. ; https://doi.org/10.1101/2025.09.14.676082doi: bioRxiv preprint Fig. 2: MAP65-1 concentrates tubulin on the microtubule lattice a, Schematic representation of the experimental setup used to study free tubulin recruitment by MAP65-1. Microtubules were grown with Alexa-Fluor-647-labeled tubulin at a concentration of 11 μM (step I) before they were capped with GMPCPP (step II) and exposed to ATTO-565-labeled free tubulin at a concentration of 4 μM and 100 nM or 200 nM GFP-MAP65-1 (step III). b, Recruitment of free tubulin in the absence of GTP to the microtubule lattice by 100 and 200 nM GFP-MAP65-1 (cyan) observed by CLSM. Microtubules in magenta and free tubulin in green. Scale bar, 5 μm. c, Schematic representation of the setup used to analyze free tubulin recruitment by MAP65-1 to the microtubule lattice. ROIs with a length of 1 µm were drawn to measure fluorescence intensity across the full width of the microtubule, which was then divided to the fluorescence intensity in the background to estimate tubulin concentration. d, Graphs represent line scans along the microtubule shown in b in the presence of 200 nM GFP-MAP65-1. e, Line scan of free tubulin with the intensity normalized to the

free tubulin level. The graph corresponds to the microtubule shown in b with 200 nM GFP-MAP65-1 and the corresponding graph in d. f, Quantification of tubulin recruitment to the microtubule lattice (normalized to the background level) in the presence of 100 nM (n = 150 microtubule segments of 1 µm) and 200 nM (n = 113 microtubule segments of 1 µm) MAP65-1. Bars represent the median values. Statistics: Mann-Whitney test, **** P < 0.0001. g, FRAP assay observed by TIRFM. The dashed white rectangle indicates the bleached region where fluorescence intensity was measured over time. Scale bar, 5 μm. h, Graph with icons representing the average normalized fluorescence intensity recovery over time for both GFP-MAP65-1 (cyan) and free tubulin (green; n = 17 microtubules). Shaded region above and below the icons corresponds to the standard deviation for each time point. .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 15, 2025. ; https://doi.org/10.1101/2025.09.14.676082doi: bioRxiv preprint .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 15, 2025. ; https://doi.org/10.1101/2025.09.14.676082doi: bioRxiv preprint Fig. 3: MAP65-1 preferentially localizes to bent microtubules in cells a, Average intensity Z-projection of CLSM of epidermal hypocotyl cells from 5-day-old plants co- expressing p35S::GFP-MBD and pMAP65-1::MAP65-1-mCherry. Scale bar, 5 μm. b, Arrowheads indicate examples of straight and bent microtubules extracted from the image shown in a. Scale bar, 5 μm. c, Quantification of mean MAP65-1 intensity on straight (maximum curvature 0.1 μm-1, n = 31) microtubules, N = 5 cells. Bars indicate the median values. Statistics: Mann-Whitney test, **** P < 0.0001. d, Quantification of mean MBD intensity per microtubule against microtubule maximum curvature. The pink line indicates a simple linear regression (Spearman r = 0.67, P < 0.0001, n = 58 microtubules). e, Quantification of mean MAP65-1 intensity per microtubule against microtubule maximum curvature. The pink line indicates a simple linear regression (Spearman r = 0.64, P < 0.0001, n = 58 microtubules). f, Schematic representation of the protoplast experiment. Protoplasts are first extracted in a solution with 600 mOsmol/L mannitol, then allowed to sediment into rectangular microwells followed by an exchange to a solution with 280 mOsmol/L mannitol that causes the protoplasts to inflate. g, Average intensity Z-projection of CLSM of enclosed protoplasts extracted from roots of plants co-expressing p35S::GFP-MBD and pMAP65- 1::MAP65-1-mCherry. Scale bar, 5 μm. h, Quantification of normalized mean MAP65-1 intensity against mean microtubule orientation. The pink line indicates a simple linear regression (Spearman r = -0.25, P = 0.0028, n = 135 microtubules). .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 15, 2025. ; https://doi.org/10.1101/2025.09.14.676082doi: bioRxiv preprint .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 15, 2025. ; https://doi.org/10.1101/2025.09.14.676082doi: bioRxiv preprint Fig. 4: MAP65-1 promotes microtubule nucleation on bent microtubules in vitro a, Microtubules attached to the surface at both ends, straight or forced to statically bend by fluid flow, observed by TIRFM. 100 nM MAP65-1 was added at a ratio of 15% GFP-MAP65-1 to 85% non- fluorescent MAP65-1. Scale bar, 5 μm. b, Quantification of mean MAP65-1 intensity on straight (maximum curvature 0.1 μm-1, n = 29) microtubules, N = 3. Bars indicate the median values. Statistics: Mann-Whitney test, **** P < 0.0001. c, Quantification of mean MAP65-1 intensity per microtubule against microtubule maximum curvature. The pink line indicates a simple linear regression (Spearman r = 0.52, P < 0.0001, n = 53). d, Microtubule nucleation mediated by 100 nM MAP65-1 on straight and bent microtubules observed by TIRFM. Nucleation events are indicated with a white star. Scale bar, 5 μm. e, Graphs represent line scans along the microtubules shown in d (original microtubule in magenta, nucleated/elongated microtubules in green). f, Quantification of the distance between nucleation events on straight (maximum curvature 0.1 μm-1, n = 40 events) microtubules. Bars indicate the median values. T otal microtubule lengths of 723.52 μm (straight) and 819.45 μm (bent) were analyzed, N = 3. Statistics: Mann-Whitney test, ** P = 0.0012. g, Spatial nucleation frequency (number of events per µm) according to microtubule maximum curvature (n = 14 events for 0–0.1 µm-1, 17 events for 0.1–0.3 µm-1, 10 events for 0.3–0.5 µm-1, 8 events for 0.5–0.7 µm-1 and 6 events for 0.7–0.9 µm-1). T otal microtubule lengths = 723.52 µm (0–0.1 µm-1), 461.18 µm (0.1–0.3 µm-1), 143.07 µm (0.3–0.5 µm-1), 94.91 µm (0.5–0.7 µm-1) and 72.87 µm (0.7–0.9 µm-1). .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 15, 2025. ; https://doi.org/10.1101/2025.09.14.676082doi: bioRxiv preprint .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 15, 2025. ; https://doi.org/10.1101/2025.09.14.676082doi: bioRxiv preprint Fig. 5: MAP65-mediated microtubule nucleation depends on the amount of microtubule defects a, Microtubule nucleation on taxol pre-treated microtubule mediated by 100 nM MAP65-1 imaged by TIRFM. Free tubulin (green) labeled with ATTO 488 and microtubules (magenta) with ATTO 565. White arrowhead indicates presumed microtubule damage site from where nucleation (white star) happens. Scale bar, 5 µm. b, Kymograph of the corresponding microtubule shown in a. Scale bars, 5 µm (horizontal) and 1 min (vertical). c, Quantification of the distance between nucleation events on fast grown, taxol pre-treated and slowly grown microtubules. A total of 16 (fast growth), 26 (taxol pre- treated) and 6 (slow growth) nucleation events were observed. Bars indicate the median values. T otal microtubule lengths of 430.04 µm, 398.67 µm and 453.21 µm were analyzed respectively, N  3. Statistics: Mann-Whitney test, Bonferroni-corrected, * P = 0.015 and ** P = 0.0016. d, Graph represents line scan along the microtubule shown in a, showing fluorescence intensity at presumed damage site (white arrowhead). e, Microtubule nucleation on taxol pre-treated microtubule mediated by 300 pM PRC1 imaged by TIRFM. Free tubulin (green) labeled with ATTO 488 and microtubules (magenta) with ATTO 565. Nucleation events are indicated with white stars. Scale bar, 5 µm. f, Kymograph of corresponding microtubule shown in e. Scale bars, 5 µm (horizontal) and 1 min (vertical). g, Quantification of the distance between nucleation events on fast grown, taxol pre-treated and slowly grown microtubules. A total of 26 (fast growth), 46 (taxol pre-treated) and 13 (slow growth) nucleation events were observed. Bars indicate the median values. T otal microtubule lengths of 536.38 µm, 254.24 µm and 446.95 µm were analyzed respectively, N  3. Statistics: Mann-Whitney test, Bonferroni-corrected, ** P = 0.0024 and *** P = 0.0002. h, Schematic illustration of the experimental setup to generate annealed microtubules. Microtubules were slowly polymerized with GMPCPP and in the presence of ATTO-565-labeled or Alexa-Fluor-647-labeled tubulin (step I). The two microtubule populations were then mixed and incubated overnight at 30°C (step II). The next day, the mix of non-annealed and annealed microtubules was used for the experiments (step III). i, Microtubule nucleation mediated by 100 nM MAP65-1 observed on annealed microtubules by TIRFM. Nucleation event is indicated with white star. Scale bar, 5 µm. j, Graph represents line scan along the microtubule shown in i (microtubule 1 in magenta, microtubule 2 in blue, nucleated microtubule in green). k, Quantification of the distance between nucleation events on non-annealed and annealed microtubules. A total of 9 (non-annealed) and 54 (annealed) nucleation events were observed. Bars indicate the median values. T otal microtubule lengths of 244.35 µm and 406.52 µm were analyzed respectively, N = 2. Statistics: Mann-Whitney test, ** P = 0.0099. .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 15, 2025. ; https://doi.org/10.1101/2025.09.14.676082doi: bioRxiv preprint .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 15, 2025. ; https://doi.org/10.1101/2025.09.14.676082doi: bioRxiv preprint Fig. 6: Bundled microtubules survive longer in the absence of free tubulin a, Microtubule dynamics upon removal of free tubulin and incubation with 100 nM MAP65-1 imaged by TIRFM (microtubules labeled with ATTO-565 in magenta, seeds and caps in blue). White arrowheads indicate microtubule breakage. Scale bar, 5 µm. b, Quantification of microtubule survival over time upon removal of free tubulin in the presence of 100 nM MAP65-1 for single (N = 8, n = 33 microtubules) and bundled (N = 5, n = 69 microtubules) microtubules. Statistics: Logrank (Mantel- Cox) test, **** P < 0.0001. c, Schematic representation of the hypothetical function of MAP65-1 in reinforcing microtubule alignment with mechanical stress. Microtubules polymerize faster under tensile stress (lower right part of the panel); this induces more defects in the lattice (along with more damage sites due to the high curvature of these microtubules); these defects and damage sites recruit MAP65, which then stabilizes, nucleates on and bundles microtubules. MAP65 binding also leads to microtubule softening. The alignment of microtubule arrays is the emerging property of their stiffness: stiff microtubules tend to align along the flattest direction of the cell cortex. For soft microtubules, this emerging property is disrupted, and microtubules instead align to a greater extent along the direction of maximal tensile stress, i.e., the surface experiencing the highest curvature for a pressurized cell. These mechanisms bring forth a positive feedback loop, where then microtubules under stress again polymerize faster, have more defects (and damage due to bending) and recruit MAP65. MAP65 thus protects microtubule under stress and promotes its own recruitment. .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 15, 2025. ; https://doi.org/10.1101/2025.09.14.676082doi: bioRxiv preprint Fig. S1, related to Fig. 1: MAP65-1 is a well-known microtubule bundler a, Confocal laser scanning microscopy (CLSM) of epidermal hypocotyl cells from 5-day-old plants co- expressing p35S::GFP-MBD and pMAP65-1::MAP65-1-mCherry. White arrowhead indicates a microtubule that grows and gets bundled in the last time point (marked with a dashed white square). Scale bar, 5 µm. b, Close-up of dashed white square indicated in panel a. c, In vitro microtubule dynamics imaged by total internal reflection fluorescence microscopy (TIRFM) in the presence of 8 μM ATTO-565-labeled free tubulin (magenta) and 50 nM GFP-MAP65-1 (cyan). Stable GMPCPP seeds labeled with Alexa Fluor 647 appear in blue. White arrowhead indicates a microtubule that grows and gets bundled in the last time point. Scale bar, 5 µm. .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 15, 2025. ; https://doi.org/10.1101/2025.09.14.676082doi: bioRxiv preprint Fig. S2, related to Fig. 1: Nucleation does not happen in the control and is MAP65-1 concentration-dependent a, Kymograph of the corresponding control microtubule shown in Fig. 1c (top panel), exhibiting growth from the free ends in the absence of MAP65-1/PRC1. Free tubulin labeled with ATTO 488 (green) and microtubule labeled with ATTO 565 (magenta). Scale bars, 5 µm (horizontal) and 1 min (vertical). b, TIRFM images showing microtubule nucleation by MAP65-1 with different concentrations. White stars indicate nucleation events. Scale bar, 5 µm. c, Quantification of the distance between nucleation events using different MAP65-1 concentrations. A total of 11 (100 nM), 34 (150 nM) and 32 (200 nM) nucleation events were observed. Bars indicate the median values. Total microtubule lengths of 1009.49 µm, 705.76 µm and 587.54 µm were analyzed respectively, N = 3. Statistics: Kruskal-Wallis test, *** P = 0.0001 and **** P < 0.0001. .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 15, 2025. ; https://doi.org/10.1101/2025.09.14.676082doi: bioRxiv preprint .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 15, 2025. ; https://doi.org/10.1101/2025.09.14.676082doi: bioRxiv preprint Fig. S3, related to Fig. 1: Microtubule nucleation thresholds in solution for MAP65-1 and PRC1 4 µM ATTO-488-labeled free tubulin was incubated either in the presence of PRC1 or MAP65-1 with the indicated concentrations. For the condition of 500 nM MAP65-1, no seeds are visible. White stars indicate nucleation events in solution. Scale bars, 25 µm. .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 15, 2025. ; https://doi.org/10.1101/2025.09.14.676082doi: bioRxiv preprint Fig. S4, related to Fig. 1: 1 nM PRC1 promotes extensive nucleation a, Close-up of microtubule from Fig. 1k showing nucleation events (white stars) along the microtubule lattice in the presence of 4 µM free ATTO-488-labeled tubulin (green) and 1 nM PRC1. Scale bar, 5 µm. b, Kymograph of corresponding microtubule shown in a. White stars indicate nucleation events. Scale bars, 5 µm (horizontal) and 1 min (vertical). .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 15, 2025. ; https://doi.org/10.1101/2025.09.14.676082doi: bioRxiv preprint Fig. S5, related to Fig. 2: In the absence of GTP, no growth happens from microtubule ends a, Microtubule incubated with 100 nM GFP-MAP65-1 and 4 µM free ATTO-565-labeled tubulin in the absence of GTP imaged by TIRFM. Scale bar, 5 µm. b, Kymograph of the corresponding microtubule shown in a. Scale bars, 5 µm (horizontal) and 1 min (vertical). .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 15, 2025. ; https://doi.org/10.1101/2025.09.14.676082doi: bioRxiv preprint Fig. S6, related to Fig. 4: Nucleation events do not always happen at the highest curvature regions Examples of site of microtubule nucleation on bent microtubules, showing the stochastic nature of this event. The nucleation does not always happen at the site of highest microtubule curvature. Scale bar, 5 µm. .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 15, 2025. ; https://doi.org/10.1101/2025.09.14.676082doi: bioRxiv preprint .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 15, 2025. ; https://doi.org/10.1101/2025.09.14.676082doi: bioRxiv preprint Fig. S7, related to Fig. 4: MAP65-1 does not specifically recognize an expanded lattice a, Quantification of the frequency of the nucleation origin site in microtubules grown with 11 µM of 10% ATTO-565-labeled tubulin and incubated with 100 nM MAP65-1 and 4 µM ATTO-488-labeled tubulin as shown in Fig. 1a. n = 46 nucleation events, N = 4. b, Average composition of the microtubule lattice in length from the microtubules that were used for the quantification in a. Total microtubule length analyzed (GMPCPP seed, GMPCPP cap, and GDP lattice) = 578.94 µm. c, Microtubule incubated with 100 nM GFP-MAP65-1 and 3 µM black tubulin and observed by TIRFM. Scale bar, 5 µm. d, Graph represents line scans along microtubule shown in c, with MAP65-1 in cyan and the microtubule in magenta. The peak corresponds to a GMPCPP region of the microtubule, the seed. e, Quantification of GFP-MAP65-1 intensity on the seed (GMPCPP) and lattice (GDP) portions of the microtubule, n = 17 microtubules. Bars indicate the median values. Statistics: Mann-Whitney test, *** P = 0.0001. f, Microtubules incubated with 100 nM GFP-MAP65-1 and 3 µM black tubulin or with 20 µM taxol observed by TIRFM. Scale bar, 5 µm. g, Quantification of GFP-MAP65-1 intensity on the GDP lattice with and without 20 µM taxol, n = 15-17 microtubules. Bars indicate the median values. Statistics: Mann-Whitney test, **** P < 0.0001. h, Stills of a microtubule before and after addition of 500 nM GFP-MAP65-1 and 4 µm black tubulin imaged by TIRFM in a microfluidics setup. Scale bar, 5 µm. i, Kymograph of the corresponding microtubule shown in h, showing no obvious change in microtubule length upon binding of MAP65-1. White arrow indicates moment in which MAP65-1 is flushed in. Scale bars, 5 µm (horizontal) and 1 min (vertical). .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 15, 2025. ; https://doi.org/10.1101/2025.09.14.676082doi: bioRxiv preprint Fig. S8, related to Fig. 5: Microtubule nucleation frequency depends on the amount of microtubule defects a, Schematic representation of the experimental setup to obtain microtubules with high and low amounts of defects. Microtubules are either polymerized in a high or low defect regime (step I), labeled respectively with 20% Alexa Fluor 647 and 20% ATTO 565. Next, 1X BRB80 is added and microtubules are centrifuged and resuspended in fresh 1X BRB80 (step II). Microtubules are then sequentially flushed in the flow chamber for incubation with 100 nM MAP65-1 and free tubulin (step III). b, Microtubule nucleation mediated by 100 nM MAP65-1 on microtubules polymerized under a high or low defect regime imaged by TIRFM. Scale bar, 5 µm. c, Quantification of the distance between nucleation events on microtubules polymerized under a high or low defect regime. A total of 95 (high defect) and 67 (low defect) nucleation events were observed. Bars indicate the median values. Total microtubule lengths of 545.41 µm and 626.598 µm were analyzed respectively, N = 3. Statistics: Mann-Whitney test, * P = 0.0433. .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted September 15, 2025. ; https://doi.org/10.1101/2025.09.14.676082doi: bioRxiv preprint