Microtubules
A dividing cell must sort 46 chromosomes into two identical sets. This must happen in under an hour, with essentially zero errors. It does this using a scaffold that self-destructs and rebuilds itself every few minutes. That scaffold is the microtubule, one of the three great polymers of the The Cytoskeleton: An Overview. Unlike a girder or a rope, a microtubule behaves like a structure in constant negotiation with its own existence. It grows for a while, then abruptly collapses, then grows again. This property is called dynamic instability. It turns out to be the key to how cells build and rebuild complex internal architecture on demand. Microtubules are hollow tubes of the protein tubulin. They radiate through the cytoplasm, form the mitotic spindle, and build the core of cilia and flagella. They also provide the tracks on which motor proteins haul cargo. These are distances a plain diffusion-driven vesicle could never cover in time. Their significance is hard to overstate. Without microtubules there is no chromosome segregation, no long-range organelle positioning, no neuronal axon transport. There is also no beating cilium, and no left-right body asymmetry in vertebrate embryos. Drugs that jam microtubule dynamics — taxol, vincristine, colchicine — are among the most important anticancer and anti-gout agents in medicine. They matter precisely because they exploit this one property. The deep biological puzzle is how a simple protein polymer achieves such versatility. It is made of just two subunit types. The answer lies in the geometry of the tubulin dimer. It also lies in the chemistry of GTP hydrolysis buried within it. Finally, it lies in the existence of dedicated organizing centers that decide where and when tubes are born. This article works through the polymer's structure and the molecular logic of dynamic instability. It covers the classic experiments — Mitchison and Kirschner's in vitro reconstitution chief among them — that proved this behavior is intrinsic to tubulin itself. It also covers the cellular machines — centrosomes, basal bodies, γ-tubulin ring complexes — that nucleate and orient microtubules throughout the cell. Understanding microtubules also sets up the next major concept. This is the motor proteins kinesin and dynein that walk along these tracks. That is where the story of intracellular transport and spindle mechanics truly comes alive.
- Microtubules are hollow, 25 nm polymers of α/β-tubulin heterodimers arranged in (usually 13) parallel protofilaments.
- They are structurally polar: a fast-growing plus end and a slower minus end, usually anchored at an organizing center.
- GTP hydrolysis within β-tubulin drives dynamic instability — stochastic switching between growth, shrinkage ("catastrophe"), and rescue.
- Centrosomes (with γ-tubulin ring complexes) nucleate most cytoplasmic microtubules; basal bodies template cilia and flagella.
- Mitchison and Kirschner's 1984 in vitro experiments proved instability is intrinsic to purified tubulin, not cell-imposed.
- Drugs like taxol and colchicine treat cancer and gout by freezing or destabilizing microtubules, killing dividing cells.
Microtubules are polar, GTP-powered tubulin polymers that constantly grow and catastrophically shrink — a built-in instability that lets cells rapidly remodel their internal scaffolding for division, transport, and shape.
The core idea
A microtubule is a hollow cylinder, about 25 nm in outer diameter, built from repeating heterodimers of α-tubulin and β-tubulin, two related ~50 kDa GTPase proteins that bind tightly and never separate under normal conditions. These αβ-dimers stack head-to-tail into linear protofilaments, and typically 13 protofilaments associate side-by-side to close into a tube — though the number can vary (11–16) in specialized contexts. Because every dimer points the same direction, the whole polymer is structurally polar: one end (the plus end) exposes β-tubulin and grows quickly; the other (the minus end) exposes α-tubulin and grows slowly or is often capped and stabilized in cells. This polarity is not a minor detail — it is the reason motor proteins can be sorted into two categories (plus-end-directed and minus-end-directed) and the reason the mitotic spindle can pull chromosomes in a coordinated direction rather than randomly. The chemical engine underneath this behavior is GTP hydrolysis. Both α- and β-tubulin bind GTP, but only the GTP on β-tubulin is hydrolyzed to GDP, and only after the dimer has added onto a growing polymer. This creates a self-limiting feedback loop known as dynamic instability: as long as new GTP-tubulin dimers add to the tip faster than the existing ones hydrolyze, a stabilizing "GTP cap" persists, and the tube keeps growing. If the cap is lost — because growth outruns supply, or by chance — the underlying GDP-tubulin lattice, which prefers a curved, low-energy conformation, springs open and peels apart in a rapid, often complete disassembly called a catastrophe. A shrinking microtubule can sometimes recapture a GTP cap and resume growth, an event called rescue. The polymer therefore has no single steady length; it toggles between phases in a stochastic, not deterministic, manner — an essential trick for a structure that must rebuild itself completely within a cell cycle. Cells do not let microtubules nucleate randomly in the cytoplasm; nucleation is energetically difficult to start from scratch (forming the initial curved sheet is unfavorable), so cells use dedicated microtubule-organizing centers (MTOCs). In most animal cells the primary MTOC is the centrosome, sited near the nucleus, containing a pair of barrel-shaped centrioles surrounded by pericentriolar material rich in γ-tubulin ring complexes (γ-TuRCs) that template new protofilaments and anchor minus ends. Related structures called basal bodies template the microtubule cores (axonemes) of cilia and flagella. This organized nucleation gives the microtubule array in most cells a a clear geometric identity — a radial, aster-like network with minus ends held at the center and plus ends probing outward into the cell periphery, ready to be stabilized wherever they touch something useful, such as a chromosome kinetochore or the cell cortex.
Mechanism
1. Tubulin dimer structure and lattice assembly. α-tubulin and β-tubulin are each single globular domains of similar fold, related by ancient gene duplication, both structurally homologous to the bacterial cell-division protein FtsZ. They form an obligate heterodimer roughly 8 nm long. GTP is bound at the interface between α- and β-tubulin within a dimer (the "N-site," non-exchangeable, never hydrolyzed) and at the exposed top of β-tubulin where the next dimer will stack (the "E-site," exchangeable). Dimers polymerize head-to-tail; lateral bonds between neighboring protofilaments zip the sheet into a closed tube. Straight, GTP-bound dimers pack well into a tube; GDP-bound dimers prefer to curve outward, storing mechanical strain in the lattice as long as they remain constrained inside it.
2. Nucleation. Spontaneous nucleation of a new microtubule from free dimers is kinetically slow because small oligomers are unstable. Cells solve this with γ-TuRC, a lock-washer-shaped complex of γ-tubulin and accessory GCP proteins that mimics the helical geometry of a microtubule's minus end, providing a template that dramatically lowers the energy barrier to start a tube. γ-TuRCs are concentrated at the centrosome's pericentriolar material and at other MTOCs (spindle pole bodies in yeast, Golgi-associated sites, chromatin in acentrosomal spindles), which is why new microtubules overwhelmingly appear at these locations rather than diffusely throughout the cytoplasm.
3. Growth and the GTP cap. At the plus end, incoming GTP-tubulin dimers add rapidly (typically several to over 10 dimers/second at physiological concentrations, translating to growth rates on the order of 5–15 µm/min in cells, faster in vitro with more free tubulin). Each newly added dimer briefly holds the tube in a straight, capped, stable conformation. Hydrolysis of the E-site GTP to GDP occurs shortly after incorporation, converting the dimer to a strained, GDP-bound state — but as long as addition outpaces hydrolysis, a small cap of GTP-tubulin persists at the very tip, holding the whole polymer together like a cork in a bottle under pressure.
4. Catastrophe and rescue. If growth slows — tubulin concentration drops locally, or by simple stochastic fluctuation — hydrolysis catches up to the tip and the GTP cap is lost. The exposed GDP-lattice, no longer held straight, springs apart at the protofilaments, which peel outward like a banana skin, and the microtubule depolymerizes rapidly (shrinkage rates often several-fold faster than growth rates, up to tens of µm/min). This catastrophic switch can, however, be reversed — a rescue — if the shrinking tip re-acquires a GTP cap, for instance by encountering a stabilizing protein or a burst of local tubulin addition. The net result is that any individual microtubule's length over time looks less like smooth growth and more like a jagged, unpredictable trace: growing, catastrophizing, shrinking, sometimes rescuing, over and over.
5. Regulation by associated proteins. Cells tune this intrinsic instability with a large cast of microtubule-associated proteins (MAPs). Stabilizing MAPs (e.g., tau, MAP2) bind along the lattice and suppress catastrophe, important in neurons for maintaining long, stable axonal microtubules. Plus-end tracking proteins (+TIPs, e.g., EB1) ride the growing tip and recruit other regulators there. Catastrophe-promoting factors like kinesin-13 family members (e.g., MCAK) actively strip tubulin from ends using ATP hydrolysis rather than simple diffusion. Severing enzymes (katanin, spastin) cut microtubules internally, generating new ends and new dynamic behavior. This regulatory layer means the "raw" dynamic instability seen with pure tubulin in a test tube is heavily edited inside a living cell to produce the specific, purposeful arrays needed for mitosis, transport, or ciliary beating.
6. Cellular consequences — search-and-capture. Because plus ends are intrinsically unstable, they behave like probing fingers: growing outward, and if they fail to find a stabilizing partner (a kinetochore, the cell cortex, an organelle), they catastrophize and retract, then try again in a new direction. This "search-and-capture" behavior, formalized by Marc Kirschner and Tim Mitchison, explains how a centrosome-nucleated aster can efficiently locate and attach to all 92 kinetochores (46 chromosomes × 2 sister chromatids) during mitosis despite starting with no prior information about where they are.
The evidence
The founding observation — instability seen but not yet explained. Electron microscopy in the 1970s and 1980s repeatedly showed microtubules with strikingly different appearances at their ends: some tips were blunt and sheet-like, others frayed into curled, splayed protofilament "ram's horns" characteristic of rapid disassembly. Live-cell light microscopy of spindle and interphase microtubules (using differential interference contrast and later fluorescence) showed individual microtubules abruptly shrinking while their neighbors kept growing nearby, at the same time, in the same cytoplasm. This was the central puzzle: if all microtubules experience the same average tubulin concentration, why do some grow while others catastrophically shrink at the very same moment?
Mitchison and Kirschner (1984). Tim Mitchison and Marc Kirschner resolved this with a landmark set of in vitro experiments using pure tubulin purified from bovine brain. They nucleated microtubules from centrosomes or seed fragments in a defined buffer, then used video-enhanced light microscopy to track the length of individual microtubule plus ends over time, frame by frame. If microtubule length changed simply according to a size-versus-free-tubulin-concentration equilibrium (as classical polymer theory predicted for a single population at steady state), all filaments should slowly converge toward a similar average length, with small thermal fluctuations. Instead, they observed something the equilibrium model could not explain: at a single, fixed tubulin concentration, individual microtubules coexisted in two distinct, sharply different behavioral populations — some persistently growing, others persistently and rapidly shrinking — and filaments could switch abruptly between these states. This showed that microtubules do not behave as a single population hovering near one equilibrium length; instead, each filament occupies one of two phases, and switches between them stochastically and discontinuously. Because this behavior appeared with nothing but purified tubulin, GTP, and buffer — no cellular signaling, no other proteins — it proved that dynamic instability is an intrinsic biochemical property of the tubulin polymer itself, driven by nucleotide state, not something imposed by the cell's regulatory machinery. Later structural and biochemical work (including cryo-EM of GTP- versus GDP-lattice conformations, and mutant tubulins that hydrolyze GTP more slowly) confirmed the mechanistic basis: GTP-bound dimers favor a straight conformation compatible with the closed tube, while GDP-bound dimers favor a curved conformation that, once the stabilizing GTP cap is lost, drives rapid protofilament peeling and depolymerization.
Complementary evidence — the GTP cap directly visualized. Later work using antibodies and, more decisively, fluorescently tagged EB1-family plus-end tracking proteins (which specifically recognize the GTP/GDP-Pi transition zone at growing tips) allowed direct visualization of a small, transient "comet" of stabilized lattice at growing plus ends in living cells — providing direct, real-time confirmation of the GTP cap model predicted from the original in vitro kinetics.
Nucleation evidence — γ-tubulin and the centrosome. The identity of centrosomes as dominant nucleating sites was established by classic cell biology: microinjecting fluorescent tubulin into cells and watching new microtubules appear radiating from the centrosome first; laser-ablating or physically removing centrosomes and observing that microtubule nucleation is delayed and disorganized (though not entirely abolished, since acentrosomal nucleation pathways also exist); and biochemical purification identifying γ-tubulin as a conserved, essential component whose depletion (by antibody injection or gene knockout in yeast and Drosophila) blocks microtubule nucleation and causes mitotic spindle defects.
Worked examples
Example 1 — how fast can a microtubule reach a kinetochore? Suppose a centrosome sits 8 µm from a kinetochore at the start of prometaphase, and microtubule plus ends grow at 12 µm/min but undergo catastrophe on average every 25 seconds if they fail to find a target, followed by shrinkage at 18 µm/min back toward the centrosome, then renucleation and another attempt. How long, roughly, might "search and capture" take? Convert 25 seconds to minutes: 25/60 ≈ 0.42 min. Distance covered per growth attempt before catastrophe = growth rate × time = 12 µm/min × 0.42 min ≈ 5 µm. Since 5 µm < 8 µm, a single attempt on average falls short of the kinetochore; the microtubule catastrophizes before arrival, shrinks back, and must try again in a new, randomly chosen direction. Because attempt direction is stochastic (the aster probes many angles simultaneously via multiple filaments, not just one at a time), the effective search time depends on how many filaments are probing in parallel — a real spindle nucleates hundreds of microtubules from each pole simultaneously, so even if any single filament's odds of reaching the target on one attempt are low, the aggregate probability of some filament capturing the kinetochore within a minute or two is high. This is precisely why real spindles assemble bipolar attachments to all chromosomes within roughly 10–20 minutes in animal cells, despite each individual microtubule's search being inefficient and failure-prone — a probabilistic, population-level solution rather than a deterministic, single-filament one.
Example 2 — interpreting a taxol dose-response in a cancer cell line. A researcher treats a cultured tumor cell line with increasing concentrations of taxol (paclitaxel), which binds the microtubule lattice and suppresses depolymerization (stabilizing microtubules rather than destroying them, unlike colchicine). She measures the fraction of cells arrested in mitosis (identified by condensed chromosomes and spindle morphology) at each dose:
| Taxol (nM) | % cells arrested in mitosis |
|---|---|
| 0 | 3% |
| 2 | 15% |
| 10 | 48% |
| 50 | 81% |
| 250 | 83% |
Reasoning: the plateau near 80–85% (rather than 100%) even at high dose suggests a subpopulation of cells is either not cycling (quiescent, so never enters mitosis to be arrested) or has an efflux/resistance mechanism reducing effective intracellular drug. The steep rise between 2 and 50 nM indicates the drug's target (microtubule dynamics) is highly sensitive within this narrow range — consistent with the fact that taxol needs to occupy only a modest fraction of tubulin subunits to suppress catastrophe network-wide, because a few "frozen" GTP-like-conformation dimers per lattice segment can propagate stability locally. Mechanistically, why does stabilizing microtubules (rather than destroying them) block mitosis? Because chromosome segregation requires the dynamic capture-and-release, tension-sensing, and eventual coordinated shortening of kinetochore-fiber microtubules to pull sisters apart; a spindle frozen rigid by taxol can attach chromosomes but cannot complete the choreographed disassembly needed for anaphase, so cells stall at the metaphase-to-anaphase checkpoint and, if arrest is prolonged, often die by apoptosis — the basis for taxol's use in ovarian, breast, and lung cancer chemotherapy.
Common misconceptions & traps
| Misconception | Why it's wrong |
|---|---|
| "Microtubules only grow, or only shrink — they're stable structures like actin stress fibers." | Every microtubule cycles stochastically between growth, catastrophe, and rescue; instability is intrinsic, not a malfunction. |
| "Both tubulin subunits hydrolyze GTP the same way." | Only the E-site GTP on β-tubulin is hydrolyzed after polymerization; the α-tubulin GTP is non-exchangeable and never hydrolyzed. |
| "Colchicine and taxol do the same thing to microtubules." | Colchicine binds free tubulin and blocks polymerization (net depolymerization); taxol binds the polymer lattice and suppresses depolymerization (net stabilization) — opposite mechanisms, same mitotic-arrest outcome. |
| "Centrosomes are required for spindle assembly in all cells." | Many cell types (plant cells, oocytes of several species) assemble functional bipolar spindles without centrosomes, nucleating microtubules from chromatin and other acentrosomal pathways instead. |
| "The plus end is simply the end that grows; the minus end never grows at all." | Minus ends can grow too, just much more slowly, and are frequently capped/anchored at an MTOC in cells, which suppresses their dynamics rather than eliminating them entirely. |
The scientist behind it
Tim Mitchison, working as a postdoctoral researcher with Marc Kirschner at the University of California, San Francisco in the early 1980s, set out to resolve a paradox that had stalled the microtubule field: why did microtubules in vitro fail to reach the single stable average length predicted by simple polymer equilibrium theory? Rather than assuming measurement noise, Mitchison and Kirschner took the anomaly seriously, developing video-enhanced differential interference contrast microscopy sensitive enough to track individual microtubule ends over minutes in real time — a technical achievement in itself, since single microtubules are far below the classical diffraction-limited resolution needed to distinguish by eye. Their 1984 paper in Nature, "Dynamic instability of microtubule growth," reframed how an entire generation of cell biologists thought about cytoskeletal polymers: not as passive structural fibers but as far-from-equilibrium, energy-consuming, self-organizing structures whose instability is the very feature that makes them biologically useful. Mitchison went on to co-found much of modern quantitative cell biology, including systematic small-molecule screens (that yielded the kinesin-inhibitor monastrol) and later work on cytoplasmic organization and cell size control.
Origins · 1 minTimothy J. Mitchison and how this was discovered
Where this matters today
Cancer chemotherapy. Taxanes (paclitaxel, docetaxel) and vinca alkaloids (vincristine, vinblastine) are frontline drugs for breast, ovarian, lung cancers, and lymphomas, respectively, because they exploit the fact that rapidly dividing cancer cells depend absolutely on functional spindle dynamics; stabilizing or destabilizing microtubules preferentially kills cells attempting mitosis while sparing many non-dividing tissues (though neurotoxicity occurs because axonal transport also depends on microtubules, explaining chemotherapy-induced peripheral neuropathy).
Gout treatment. Colchicine, one of the oldest continuously used drugs in medicine, treats acute gout flares by disrupting microtubule-dependent neutrophil migration and inflammasome activation, reducing the inflammatory response to urate crystals.
Ciliopathies and reproductive medicine. Defects in microtubule-based axonemes of cilia and flagella cause primary ciliary dyskinesia (chronic respiratory infections, infertility, and situs inversus from failed left-right patterning), directly linking basic microtubule biology to clinical genetics and IVF counseling.
Neurodegeneration. Pathological aggregation of the microtubule-stabilizing protein tau, when hyperphosphorylated and released from microtubules, forms the neurofibrillary tangles characteristic of Alzheimer's disease — a direct link between microtubule-associated protein regulation and dementia research, and a target of ongoing tau-directed therapeutics.
Agricultural and veterinary use. Benzimidazole compounds (albendazole, mebendazole) selectively bind parasite β-tubulin, disrupting microtubule function in helminths and protozoa while sparing host tubulin, forming the backbone of livestock and human deworming programs worldwide.
Biotechnology and structural biology. Cryo-electron microscopy of microtubule-motor and microtubule-drug complexes, enabled by the field's deep mechanistic understanding of the tubulin lattice, continues to drive structure-based drug design for both cancer therapeutics and next-generation antiparasitics.
Ask the tutor: How exactly do kinesin and dynein "read" microtubule polarity to walk in opposite directions along the same track?
Why do microtubules need GTP if they aren't motor proteins doing mechanical work directly?
GTP hydrolysis here doesn't power movement the way ATP powers kinesin's steps; instead it powers a conformational switch within the lattice itself. GTP-bound tubulin favors a straight shape compatible with a closed tube, while GDP-bound tubulin favors a bent shape. As long as unhydrolyzed GTP-tubulin caps the growing tip, the whole polymer stays straight and stable; once hydrolysis catches up to the tip, the underlying strained GDP-lattice is free to spring apart. So the energy from GTP hydrolysis is stored as mechanical strain in the lattice, later released explosively during catastrophe — a form of stored potential energy rather than direct mechanical work, conceptually similar to how actin's ATP hydrolysis regulates filament stability rather than driving actin's own movement.
How is a microtubule different from an actin filament, mechanistically?
Both are dynamic, polar, nucleotide-hydrolyzing polymers, but they differ in scale, geometry, and dynamics. Microtubules are hollow 25 nm tubes of tubulin heterodimers with 13 protofilaments, far stiffer (larger persistence length) than the two-stranded 7 nm actin helix. Microtubules undergo dynamic instability — discontinuous catastrophe/rescue switching — while actin filaments exhibit treadmilling, a smoother net flux of subunits from one end to the other at steady state. Functionally, microtubules dominate long-range transport, spindle mechanics, and cilia, while actin dominates cell shape, cortex mechanics, and short-range force generation, as covered in Actin Filaments. Both systems, however, share the underlying logic of polarity plus nucleotide-driven dynamics.
Why does the number of protofilaments matter, and is it always 13?
Thirteen protofilaments is the dominant number in most animal cells and gives a nearly straight, cylindrically closed lattice with minimal seam mismatch between adjacent protofilaments. But the number is not fixed by chemistry alone — it's influenced by the nucleating template and accessory proteins, and some organisms and specialized structures use different numbers (for example, C. elegans touch-receptor neurons commonly build 15-protofilament microtubules). In vitro, tubulin polymerizing without a centrosome template can produce microtubules with anywhere from about 9 to 16 protofilaments, and the resulting lattice geometry subtly affects motor protein movement and mechanical properties, since it changes the helical pitch and the "seam" where the tube doesn't quite close symmetrically.
Do all cells have centrosomes, and what happens if a cell doesn't?
No. Plant cells lack centrosomes entirely and nucleate microtubules from dispersed sites including the nuclear envelope and existing microtubules, using γ-TuRC-based nucleation without centriole-containing MTOCs. Many oocytes (including those of most mammals, notably humans and mice) lack centrosomes after early development and assemble spindles via chromatin-mediated and other acentrosomal nucleation pathways, relying on motor proteins and self-organization to establish bipolarity. This shows that the centrosome, while the dominant MTOC in most animal somatic cells, is not an absolute requirement for building a functional microtubule array — it is one solution among several to the nucleation problem.
How does dynamic instability actually help a cell, rather than just being wasteful?
Constant polymer turnover looks wasteful in raw energetic terms (every GTP hydrolyzed is "spent"), but it buys enormous adaptability. Because plus ends are always sampling growth and collapse, the cell can rapidly redeploy its microtubule array to a completely new geometry within minutes — from a diffuse interphase network to a bipolar spindle at mitotic entry, for instance — without needing to disassemble a stable structure through some separate, slower mechanism. Dynamic instability also enables error correction: a microtubule that attaches incorrectly to a kinetochore can be selectively destabilized and retried, which is central to how the spindle assembly checkpoint achieves such low chromosome mis-segregation rates.
What is the difference between microtubule "nucleation" and "growth," and why separate the two conceptually?
Nucleation is the difficult, rate-limiting formation of a new, minimally stable oligomeric seed capable of subsequent tube elongation; growth is the comparatively easy, fast addition of further dimers onto an existing, already-stable end. The two are governed by different kinetics: nucleation has a large energy barrier because small aggregates of tubulin are unstable and prone to falling back apart, whereas growth onto an existing lattice is fast and favorable. Cells exploit this asymmetry by using γ-TuRC to bypass the nucleation barrier entirely (mimicking a pre-formed, stable minus-end template), which is why virtually all new cytoplasmic microtubules appear at MTOCs rather than spontaneously throughout the cytosol.
Why are microtubules essential for the next topic, molecular motors?
Kinesin and dynein, covered in Molecular Motors: Kinesin and Dynein, are mechanochemical enzymes that convert ATP hydrolysis into stepwise movement along the microtubule lattice, using the periodic, polar arrangement of tubulin dimers as a literal set of footholds. Because microtubules are structurally polar, most kinesins walk toward the plus end and dynein walks toward the minus end, giving the cell a built-in one-way/other-way logistics system: outbound cargo (secretory vesicles toward the plasma membrane, for instance) typically rides kinesins toward peripheral plus ends, while inbound cargo (endocytic vesicles, viruses hijacking transport) often rides dynein back toward the centrosome-anchored minus ends near the nucleus.