Thioflavin T fluorescence and peptide fibrillation kinetics: lag phase, elongation, and what the plateau does not report

Many research peptides carry sequences with a measurable propensity to assemble into cross-β fibrils. Amylin-derived analogs, several incretin-class sequences, and a wide range of short amphipathic peptides have been characterized as fibril-forming under some combination of concentration, pH, ionic strength, and interface exposure. Fibrillation is a distinct problem from the reversible self-association discussed in the earlier post on soluble oligomers: it is typically slow to start, fast once started, largely irreversible, and invisible to the reversed-phase purity assay that reports the headline number on a certificate of analysis. The dye thioflavin T (ThT) is the most widely used tool for following it in real time, and ThT kinetic traces appear throughout the formulation and stability literature.

This post describes what ThT actually binds, how the characteristic sigmoidal curve is parameterized, which experimental variables dominate the result, and why the height of the plateau is a poor stand-in for the amount of fibril present.

What the dye reports

Thioflavin T is a cationic benzothiazole dye. In free solution the benzothiazole and dimethylaminobenzene rings rotate relative to one another around the connecting carbon-carbon bond, and this rotation provides a non-radiative decay path that keeps fluorescence quantum yield very low. When the molecule is held in a rigid, planar conformation, that path closes and emission increases by orders of magnitude. ThT is often described as a molecular rotor for this reason.

Fibrils with cross-β architecture present surface grooves running parallel to the fibril axis, formed by the side chains of residues in register along adjacent β-strands. ThT inserts into these grooves with its long axis aligned with the fibril axis, which locks the rings coplanar and switches on fluorescence. The bound dye excites near 440 to 450 nm and emits near 480 to 490 nm, with a red shift relative to the free dye that also contributes to the apparent signal gain when the instrument is set to the bound-state wavelengths.

The consequence is that ThT is reporting a specific structural state rather than aggregation in general. Monomeric peptide, disordered or molten oligomers, and amorphous precipitates generally produce little signal. Prefibrillar oligomers are variably ThT-active depending on how much ordered β-sheet they contain. A sample can therefore accumulate a substantial mass of non-fibrillar aggregate without any change in ThT fluorescence, and a turbidity or light-scattering measurement run in parallel will often diverge from the ThT trace for exactly this reason. Conversely, the first ThT-positive species may represent a small fraction of total peptide at the moment the signal begins to rise.

Anatomy of the sigmoidal curve

A typical ThT kinetic experiment holds peptide at a fixed concentration in buffer with dye present, in a multiwell plate at controlled temperature, and reads fluorescence at intervals. The resulting trace is sigmoidal. A lag phase of low, flat signal gives way to a growth phase in which fluorescence rises steeply, followed by a plateau.

The lag phase corresponds to nucleation: the stochastic formation of the first ordered assemblies capable of templating further growth. Its duration, usually reported as t_lag from the intercept of a tangent to the growth phase with the baseline, is the parameter most often compared across conditions. The growth phase reflects elongation of existing fibril ends by monomer addition, together with processes that multiply the number of ends, including fragmentation of existing fibrils and secondary nucleation on fibril surfaces. The half-time, t_1/2, at which fluorescence reaches half of its plateau value, is a more robust descriptor than t_lag because it is less sensitive to baseline noise and to the choice of tangent. An apparent rate constant, k_app, is sometimes extracted from the slope at t_1/2 or from a fit to an empirical sigmoid.

The dependence of t_1/2 on initial peptide concentration carries mechanistic information. Where the half-time scales with concentration according to a power law, the exponent has been related to the dominant process generating new fibril ends: a steep dependence is consistent with primary nucleation controlling the rate, while a shallower dependence has been interpreted as evidence of secondary processes.

Seeding is the most useful control available. Adding a small percentage of preformed fibril to a fresh monomer solution bypasses primary nucleation, and a genuine nucleation-dependent process responds by losing most or all of its lag phase. A sample whose lag time does not shorten on seeding is unlikely to be following a classical nucleation-elongation pathway, and the signal rise may reflect something other than fibril growth.

Why plateau height is not fibril mass

The most common misreading of a ThT dataset is to treat the final fluorescence intensity as a measure of how much fibril has formed and to compare plateaus between peptides, buffers, or excipient conditions. Several factors break that proportionality.

Fluorescence per unit fibril mass depends on binding site density and on the quantum yield of the bound dye, and both vary with fibril polymorph. The same peptide can form structurally distinct fibrils under different conditions, with different groove geometry and different numbers of accessible binding sites per unit length. A condition that produces a lower plateau may have produced more fibril of a type that binds ThT less efficiently.

Optical effects compound this. At high fibril loads, light scattering and inner-filter absorption attenuate both excitation and emission. Fibrils that clump or sediment during the experiment move out of the read volume in a bottom- or top-reading plate reader, which can cause the plateau to drift downward even as total fibril mass is stable or increasing. Dye depletion is a further consideration: if ThT is present in large excess the plateau is set by the fibril, but at high peptide-to-dye ratios the signal saturates at the dye rather than at the fibril.

Where the quantity of fibril matters, orthogonal mass measurements are generally used. Residual soluble peptide is quantified by reversed-phase HPLC of the supernatant after centrifugation or after filtration through a membrane that retains fibrils, and the difference from the starting concentration is taken as the aggregated fraction. These measurements are slower and are usually taken at endpoints rather than continuously, which is why ThT remains the kinetic readout while mass balance anchors the interpretation.

Experimental variables that dominate the outcome

Fibrillation kinetics are among the least reproducible measurements in peptide science when experimental details are not controlled, and inter-laboratory comparisons of lag times for the same peptide have historically spanned an order of magnitude. Most of the variance traces to a handful of variables.

Pre-existing seeds in the starting material shorten or abolish the lag phase. Lyophilized peptide that has been stored in solution at any point, or that was exposed to interfaces during purification, may contain trace fibrillar material. Monomerization protocols include dissolution in hexafluoroisopropanol followed by evaporation, filtration through very fine pore membranes, or isolation of the monomer peak by size-exclusion chromatography immediately before the experiment. The method chosen affects both the lag time and its reproducibility, and should be stated.

Interfaces drive nucleation. Agitation, whether by orbital shaking or by beads added to the well, dramatically shortens lag times by renewing the air-water interface and by fragmenting fibrils. Quiescent experiments are slower and more variable. The plate material matters for the same reason: non-binding surface treatments reduce peptide adsorption and interfacial nucleation, and two plate types from different suppliers can give different kinetics with identical solutions.

Solution conditions set the thermodynamic drive. pH relative to the isoelectric point, ionic strength, buffer species, and the presence of organic cosolvent carried over from a stock solution all change the rate. Dimethyl sulfoxide at even a few percent is known to alter aggregation kinetics of many sequences. ThT concentration itself is a variable; the dye is usually held in the range of 10 to 25 micromolar, and at higher concentrations ThT has been reported both to accelerate and to inhibit fibrillation of particular peptides.

Instrumental choices complete the list. Read interval and excitation intensity determine photobleaching of bound dye over a long experiment. Temperature control across the plate affects nucleation rates exponentially. Top versus bottom reading interacts with fibril sedimentation as noted above. A method description that does not specify plate type, agitation mode, read interval, temperature, dye concentration, and monomerization procedure is not sufficient to reproduce a lag time.

Interferences and orthogonal confirmation

ThT fluorescence is not perfectly specific for cross-β structure. Certain proteins with hydrophobic pockets, nucleic acids, cationic surfactant micelles, and some polymers bind the dye and restrict its rotation, producing fibril-like signal without fibrils. In screening experiments where test compounds are added to a peptide solution, intrinsically fluorescent or absorbing compounds produce false increases or false decreases through inner-filter effects and spectral overlap, and aromatic compounds can quench bound ThT or compete for binding sites. An apparent inhibitor of fibrillation identified by ThT alone requires confirmation by a dye-independent method before the result is interpreted structurally.

The dye is also sensitive to its environment. ThT signal is attenuated under strongly acidic conditions, so peptides formulated at low pH are often diluted into neutral buffer for the measurement. That dilution changes the aggregation state being measured, and the kinetics in the read buffer are not the kinetics in the formulation. Reading in situ at the formulation pH, where feasible, avoids the perturbation at the cost of sensitivity.

For these reasons ThT data are normally paired with at least one structural method. Circular dichroism tracks the transition from the starting secondary structure to the β-sheet signature with a minimum near 218 nm. Infrared spectroscopy of the amide I band distinguishes cross-β fibrils, which show a characteristic low-frequency component near 1620 to 1630 cm⁻¹, from native β-sheet. Transmission electron microscopy and atomic force microscopy confirm fibril morphology and dimensions directly. Dynamic light scattering and turbidity report total particle load including the amorphous material ThT misses.

ThT remains the reference kinetic readout because nothing else offers continuous, inexpensive, multiwell monitoring of fibril formation with a signal that is nearly silent before fibrils appear. Its value lies in the shape of the curve: the duration of the lag phase, the response to seeding, the concentration dependence of the half-time, and the ranking of conditions by their effect on those parameters. Its limits lie in the plateau, which encodes polymorph, optics, and dye stoichiometry as much as fibril quantity, and in the long list of experimental variables that each shift the curve independently of the peptide’s intrinsic propensity. Studies that report ThT kinetics alongside a mass-balance endpoint, a structural spectroscopy, and a full description of the plate, agitation, and monomerization conditions are the ones whose lag times can be compared, and that combination is what has made the method durable despite its well-documented quirks.