Chemotherapy drug taxol reshapes microtubules to switch on a key cell signal

Paclitaxel and its chemical cousins, the taxanes, are among the most widely prescribed chemotherapy drugs in the world, yet decades after their introduction, scientists are still discovering unexpected ways in which they act on cells. A new study published in Nature Cell Biology by Joyce Meiring, Varsha Mahapatra and colleagues, working in the laboratories of Anna Akhmanova and Lukas Kapitein at Utrecht University together with an international team of collaborators, reveals a surprising signalling function for these drugs. The researchers show that taxanes do far more than freeze the mitotic spindle: they physically reshape the wall of microtubules, the hollow protein filaments that give cells their shape and internal highways, and in doing so they fling a dormant signalling protein off the microtubule surface, triggering a cascade that remodels the actin cytoskeleton and rounds cells up within minutes.

The protein at the heart of this story is GEF-H1, a guanine nucleotide exchange factor that activates the small GTPase RhoA, a master regulator of cell contractility, migration and shape. In healthy interphase cells, most GEF-H1 sits parked on microtubules in an inactive state, like a loaded spring held in check. When microtubules fall apart, for example under the influence of the destabilizing drug nocodazole, GEF-H1 is released, rushes to the plasma membrane and switches on RhoA. Because taxanes stabilize microtubules rather than destroy them, the textbook expectation was that they should leave GEF-H1 untouched. The new work overturns that assumption and shows that stabilization itself, when it comes with a specific structural change, can be just as effective at liberating the signalling protein.

To follow RhoA activity in living cells, the team used HT1080 fibrosarcoma cells equipped with a Förster resonance energy transfer biosensor called RhoA2G, which reports RhoA activation as a changing fluorescence ratio. Serum-starved cells had low baseline activity, which surged when serum was added. Adding nocodazole, which dismantles the microtubule network completely, produced a slower but persistent RhoA rise, and paclitaxel did the same, albeit more weakly. Crucially, when the researchers deleted the GEF-H1 gene using CRISPR-Cas9, both drug responses largely vanished, while the response to serum survived intact. This genetic control established that taxanes, like microtubule poisons, drive RhoA activation through GEF-H1 rather than through any of the alternative pathways that growth factors use.

The most telling comparison came from discodermolide, a marine sponge-derived compound that also binds the taxane site on tubulin and stabilizes microtubules just as effectively as paclitaxel. Despite blocking microtubule growth in cells just as efficiently, discodermolide failed to activate RhoA, failed to displace GEF-H1 and left the actin cytoskeleton looking essentially normal. The difference between the two drugs lies not in what they do to microtubule dynamics but in the geometry they impose on the microtubule wall. Taxanes expand the lattice, increasing the longitudinal spacing between tubulin dimers, whereas discodermolide locks the lattice into a compacted state that closely resembles the conformation found in normal, dynamic microtubules.

Downstream of RhoA, the consequences were dramatic and visible within an hour. Serum-starved fibrosarcoma, Swiss 3T3 and U2OS cells, normally elongated with sparse stress fibres, responded to paclitaxel by forming dense actin stress fibres, membrane blebs and rounded morphologies, mirroring the response to nocodazole. In GEF-H1 knockout cells, none of these changes appeared. The team also detected increased levels of phosphorylated myosin light chain, a canonical readout of RhoA-driven actomyosin contractility, after paclitaxel treatment. Live imaging of breast cancer, melanoma and fibrosarcoma cells captured the sequence in real time: rapid rounding and blebbing, followed by partial re-spreading decorated with prominent stress fibres. Taxanes, in other words, hijack the same contractility programme that cells normally deploy during migration and invasion.

Direct visualization of GEF-H1 itself sealed the argument. The researchers expressed GFP-tagged GEF-H1 at near-endogenous levels in U2OS cells and watched it decorate microtubules along their length while avoiding the acetylated, long-lived subset. Within one minute of paclitaxel addition, the fluorescent protein peeled off the microtubules, an effect captured in striking live-cell movies. Discodermolide, again, did nothing. Because one minute is far too short for meaningful changes in tubulin acetylation, and because artificially boosting acetylation with the deacetylase inhibitor tubacin or the acetyltransferase αTAT1, or inducing detyrosination with VASH2-SVBP or MATCAP, did not remove GEF-H1 from microtubules, the team concluded that post-translational modifications could not explain the drug effect. The culprit had to be the lattice itself.

To prove the point with purified components, the group turned to in vitro reconstitution. Using total internal reflection fluorescence microscopy, they grew dynamic microtubules from stabilized seeds in the presence of nanomolar concentrations of purified GFP-GEF-H1. The protein bound avidly to the dynamic GDP lattice but conspicuously avoided the GMPCPP-stabilized seeds, which have an expanded conformation. This exclusion could not be attributed to nucleotide state, because microtubules polymerized with the non-hydrolysable analogue GTPγS, which keeps a compacted lattice, bound GEF-H1 just as well as GDP lattices. Interference reflection microscopy confirmed that the seeds and growing lattices contained the expected protofilament numbers, indicating that GEF-H1 is not fussy about protofilament count but is exquisitely sensitive to longitudinal lattice spacing.

The wash-in experiments provided the most elegant demonstration. When paclitaxel was flowed onto microtubules sparsely labelled with fluorescent tubulin, the speckles visibly shifted apart, a direct optical readout of lattice expansion, and simultaneously the bound GFP-GEF-H1 dissociated within seconds. Discodermolide wash-in produced neither expansion nor release. Cryo-electron microscopy and single-particle analysis then quantified the geometry: discodermolide-stabilized microtubules with 13 or 14 protofilaments had dimer rise values of 81.7 and 81.6 angstroms, essentially identical to compacted GTPγS lattices, confirming their compacted character. Remarkably, at high concentration GEF-H1 could flip the relationship and act as a code writer rather than a code reader: adding a 200-fold excess of the protein rapidly compacted microtubules that paclitaxel had expanded, imposing its preferred geometry on the drug-stabilized lattice. This concentration-dependent bidirectionality likely explains why earlier studies using strongly overexpressed GEF-H1 missed the taxane sensitivity altogether.

The structural logic is intriguing. Previous cryo-EM work showed that the C1 zinc-finger domain of GEF-H1, spanning amino acids 28 to 100, binds in a groove between two tubulin dimers in adjacent protofilaments, a site far from the exchangeable nucleotide pocket where taxane-induced expansion is thought to originate. The new study shows that a dimerized fragment containing the first 136 amino acids is sufficient to confer taxane sensitivity in cells, and that the GEF-H1-bound microtubule lattice adopts a partially compacted twist resembling that of GTPγS and discodermolide structures. How binding at the interdimer interface allosterically favours compaction near the β-tubulin E-site remains an open question, but the authors suggest that the drug and the protein compete for control of the same conformational degree of freedom.

The clinical implications are potentially far-reaching. Docetaxel, a taxane with higher microtubule affinity that is a mainstay of cancer chemotherapy, activated RhoA at concentrations as low as 100 nanomolar, well within the range measured in the plasma of treated patients, and the response grew stronger over hours as the drug accumulated in cells, a process the team followed directly with a fluorescent taxane probe. Because RhoA signalling can promote anti-tumour immune responses and drive apoptosis in cells that lose adhesion, taxane-induced RhoA activation may contribute to therapeutic efficacy, not just to toxicity. Conversely, misregulated RhoA contractility in neurons offers a plausible route to the dose-limiting peripheral neuropathy that plagues taxane therapy. Beyond the clinic, the study elevates lattice conformation to a recognized layer of the tubulin code, alongside isotypes and post-translational modifications, and provides the community with discodermolide as a powerful tool compound for separating conformational effects from effects on microtubule dynamics. If other microtubule-binding proteins turn out to read the lattice as keenly as GEF-H1 does, taxanes may be quietly rewiring many more signalling pathways than anyone suspected.

Subject of Research: How taxane-induced microtubule lattice expansion releases GEF-H1 to activate RhoA signalling

Article Title: Taxane-induced conformational changes in the microtubule lattice activate GEF-H1-dependent RhoA signalling

Article References: Meiring, J. C. M., Mahapatra, V., Liu, R., Gravett, M. S. C., Morren, D., Saunders, H. A. J., Choi, S. R., Pelster José, A., Karhanova, A., Metallidou, I., Moore, A., Spoelstra, M. F. M., Grigoriev, I. S., Giono, M., Iyer, S. S., Stehbens, S. J., Díaz, J. F., Lansky, Z., Stecker, K. E., … Akhmanova, A. (2026). Taxane-induced conformational changes in the microtubule lattice activate GEF-H1-dependent RhoA signalling. Nature Cell Biology. https://doi.org/10.1038/s41556-026-02079-4

Image Credits: AI Generated

DOI: 10.1038/s41556-026-02079-4

Keywords: microtubules, taxanes, paclitaxel, GEF-H1, RhoA, tubulin code, lattice conformation, discodermolide, actin cytoskeleton, chemotherapy, cryo-EM, cell signalling