Amyloid fibrils do not have a fixed structure: the same protein assembles into different conformations depending on pH, salt, concentration, and cofactors. This environment-dependence, together with frustrated folding surfaces in disease-related proteins, undermines conformer-specific drug design,…
Amyloid fibrils do not have a single structure for each protein. The same sequence assembles into different fibril conformations depending on pH, salt, monomer concentration, cofactors, and the path the sample took to reach its final state. That environment-dependence, combined with the frustrated folding surfaces that most disease-related amyloid proteins occupy, creates three linked problems. Drug designers cannot assume a stable target. Material scientists cannot assume a stable product. Biophysicists cannot assume that the fibril in the cuvette during a denaturation experiment is the fibril that was formed in the tube. Researchers in all three settings should treat fibril structure as an experimental variable, something to control, measure, and report rather than a fixed property of the sequence.
The default model of amyloid formation is nucleated polymerization : monomers fluctuate into a critical nucleus, and fibrils elongate by sequential monomer addition. Siegel argues that for pathological amyloid proteins this model is too simple. Aggregation reactions populate species other than monomer and fibril, and which intermediates accumulate depends on aggregation conditions and monomer concentration.
Supersaturation does much of the work in deciding what forms. A review of supersaturation-dependent fibril formation concludes that breakdown of supersaturation is central to fibrillization and describes the HANABI system, a high-throughput ultrasonication-based method that triggers fibril formation on demand PMID 35889461 . The fact that a mechanical perturbation can initiate fibrillization shows how far fibril formation is from a simple sequence-driven assembly: the physical state of the solution is a determining factor. Polyphosphate acts in the same register, promoting fibril formation of β2-microglobulin and α-synuclein through counter-anion binding and preferential hydration, with bimodal concentration-dependent effects that also occur with salt and heparin PMID 37448594 . The same additive can push a sample toward fibrils at one concentration and away from them at another. The product of an aggregation reaction is decided by the solution as much as by the sequence.
β2-microglobulin β2m is a useful worked example. It is the main fibril component in dialysis-related amyloidosis, a condition in which uremia may create carbonyl overload and lead to modified β2m PMID 10372842 . Fibril-derived β2m carries specific chemical modifications: N-terminal truncation, partial deamidation at Asn17 and Asn42, and oxidation of Met99, with all other modifications together accounting for less than 1 to 2% of the molecule PMID 16154394 . The molecules incorporated into the fibril are not the pristine recombinant protein of the textbook; they are processed, oxidized, and truncated, and that processing is itself a product of the uremic environment.
The in vivo evidence points the same way. Because structures depend on their formation environment, Siegel argues, amyloidogenic proteins likely form multiple distinct structures in vivo depending on the cellular compartment or extracellular location involved. Deposit morphology bears this out. A six-case series reported globular amyloid deposits 3 to 40 μm in diameter in the gastrointestinal lamina propria, Congo red-positive and AA-immunoreactive, that did not react with antibodies to β2-microglobulin, transthyretin, or immunoglobulin light chains PMID 12000197 . Globular deposits are a different material form from classic fibrillar deposits, arising in a tissue-specific context. Cofactors multiply the structural possibilities: light-chain amyloid fibrils are invariably associated with glycosaminoglycans, predominantly heparan sulfate, which may enhance fibril stability and protease resistance, and the extensive sequence diversity of light chains affects amyloidogenicity and glycosaminoglycan interactions PMID 10823245 . The amyloid in a patient is a composite of protein, post-translational modification, and bound cofactor, assembled under local conditions. A single structure solved in one buffer cannot represent it.
One main goal of determining amyloid fibril structures is drug design against the toxic structures. If a protein populates different conformers in different compartments, the toxic species may be one conformer among many, and the target is a moving one. Structures solved from fibrils made in one set of conditions may not match the conformer that drives disease in a particular organ.
Drug binding complicates matters further. Siegel argues that a drug interacting with an amyloidogenic protein can alter the protein's folding surface and stabilize a different amyloid conformation than the one that forms without the drug. The outcome could be harmful or beneficial, and it is hard to predict in advance. A compound selected to bind one fibril conformer with high affinity may, in the cell, shift the population toward another conformer with different toxicity. The drug is, in effect, a new environmental condition, and the structure it stabilizes may be neither the toxic structure nor the structure used for the screening campaign. Measured affinity against a single structure does not predict the effect on the ensemble.
These difficulties point toward a different therapeutic strategy. Siegel proposes aiming not at a particular fibril structure but at complete inhibition of any aggregation by disease-related amyloid proteins. This may be more beneficial than conformer-specific targeting, but it is very challenging, because it requires intervening at the level of the monomer-to-fibril transformation rather than at the level of one assembled state. Understanding that transformation is the route to treatment for amyloid diseases, but the transformation cannot be summarized by a single rate constant when multiple structures compete.
Registered trials illustrate how far the clinic is from conformer-specific design. Three completed studies bracket the current approach:
| Trial ID | Phase | Status | Enrollment | Condition |
| --- | --- | --- | --- | --- |
| NCT03237494 | N/A | Completed | 5,028 | Transthyretin amyloidosis ATTR : familial polyneuropathy and cardiomyopathy |
| NCT03542656 | Phase 3 | Completed | 160 | Cerebral amyloid angiopathy, intracranial hemorrhage, Alzheimer disease |
| NCT01855360 | Phase 1/2 | Completed | 40 | Transthyretin amyloid cardiomyopathy |
The first screens an at-risk population for hereditary ATTR and monitors TTR-positive subjects NCT03237494 . The second applies amyloid PET in cerebral amyloid angiopathy, intracranial hemorrhage, and Alzheimer disease NCT03542656 . The third tests doxycycline and TUDCA in patients with transthyretin amyloid cardiomyopathy NCT01855360 . None of these designs depends on knowing which fibril conformer is present. Screening and imaging are structure-agnostic, and the doxycycline/TUDCA combination is a repurposed, broadly acting intervention rather than a conformer-selected inhibitor. Completion status in the registry says nothing about efficacy; the point is that the clinical pipeline operates without resolving the conformer question that structural biology is working to answer.
Siegel's analysis lists proposed bionanomaterial applications for amyloid fibrils: nanowires, gels, liquid crystals, and structural scaffolds. Amyloid fibrils are attractive for these uses because they are ordered self-assemblies. Real applications, Siegel argues, require a better understanding of amyloid energetics, because the same environment-dependence that complicates drug design threatens material stability. Changing environmental conditions after fibril formation can profoundly affect fibril structure and stability over time. The clearest example is β2-microglobulin: fibrils formed at low pH dissolve when returned to neutral conditions. A fibril-based material with that behavior would fail at the first buffer exchange.
The choice of sequence is the main lever available to the material scientist. Siegel argues that native amyloids may be better suited as bionanomaterials because their sequences evolved to form amyloid. Evolution has produced funneled folding surfaces for functional amyloid proteins: the energy surface slopes toward one functional structure, so assembly is reliable and relatively insensitive to small environmental changes. Disease-related amyloid proteins, by contrast, occupy frustrated surfaces with many minima of similar energy, and the structure that forms depends on the path and the concentration. Frustration here is a technical term from protein folding theory: competing interactions that cannot all be satisfied leave the protein with many nearly isoenergetic states rather than one dominant folded basin. This distinction between funneled and frustrated surfaces is the organizing idea of the argument.
The practical rule for sequence selection follows. Siegel advises choosing sequences for amyloid-like bionanomaterials carefully, avoiding de novo designed sequences and disease-related amyloid sequences when those exhibit rough, concentration-dependent folding surfaces. Native amyloid sequences, having been shaped by selection to form amyloid reliably, are the safer starting point. Whether that rule holds in practice is not yet demonstrated by the literature cited here; it is a reasoned hypothesis from evolutionary design principles. A researcher choosing a scaffold should treat the native amyloid recommendation as a screening heuristic and then test empirically whether the candidate sequence forms the same structure across the buffer changes, drying steps, and temperature shifts of the intended process, and whether that structure persists over time.
Thermodynamic parameters for amyloid fibrils are usually extracted from denaturation curves in which a chaotrope such as urea or guanidinium chloride is titrated into the sample. The analysis relies on two-state approximations . Two-state analysis assumes no intermediates are populated during the transition, a single folded structure exists, and equilibrium holds at the time of measurement. For fibrils, the "single folded structure" means a single fibril conformer. For a globular protein, a two-state fit describes a single molecule unfolding in a fixed solvent; for a fibril, the folded state is a multimolecular lattice, and the titration changes the solvent that the lattice sits in. Each assumption can therefore fail.
The chaotrope itself is an environmental change. Adding chaotrope may change which fibril structure is most stable, so the titration can drive a conformational switch before it drives unfolding. A two-state fit of such a curve yields parameters that do not correspond to the free energy of unfolding of any single structure. The β2m behavior is a warning: fibrils that form at low pH dissolve at neutral pH, so an experiment that crosses that boundary is not measuring the stability of the original fibril. Bound cofactors add another complication. If glycosaminoglycans stabilize fibrils and slow their degradation PMID 10823245 , then a chaotrope titration of a tissue-derived fibril measures the stability of the protein-cofactor complex, not the bare fibril.
Siegel's caution is specific: before deriving thermodynamic constants from denaturation curves, verify that the system is at equilibrium and that the same fibril structure is populated during chaotrope addition. Practical checks include: incubating each titration point to steady state before measuring; comparing forward titrations with reverse dilutions to test for reversibility; and assaying fibril structure across the entire chaotrope range using electron microscopy, X-ray fiber diffraction, solid-state NMR, or a conformation-sensitive dye such as thioflavin T. If structure changes mid-titration, or if the curve depends on the order of addition, a two-state fit is not justified, and the numbers should not be reported as thermodynamic constants.
The reporting burden follows from the same logic. Parameters that emerge from a justified fit are conditional, not intrinsic. They should be reported with the full protocol: pH, ionic strength, chaotrope identity, incubation time, protein concentration, the fibril preparation method, and the supersaturation state of the sample, since fibril formation is driven by breakdown of supersaturation PMID 35889461 . A stability constant that cannot be reproduced when the buffer is reproduced is not a property of the amyloid; it is a property of one preparation.
The arguments and evidence above reduce to a short list of operating rules, each tied to a specific failure mode.
The open questions follow directly from the limits of the current evidence. No registered trial has tested a therapeutic designed around complete inhibition of disease-related amyloid aggregation, and the trial record shows the clinic working around the conformer problem rather than through it. Predicting how drug binding will reshape the folding surface of an amyloidogenic protein is not yet possible from structure or sequence alone, and the biological outcome of such reshaping remains difficult to forecast.
Which native amyloid sequences are suitable as bionanomaterials, and how stable they are under environmental change, has not been established experimentally; the evolutionary argument is plausible but untested. Accurate thermodynamic measurement of amyloid stability in the face of environment-driven structural change still lacks a general protocol, and the nucleated polymerization mechanism, though useful as a first approximation, is too simple to describe pathological amyloid aggregation, where intermediate species and condition-dependent conformations dominate. Each of these gaps is a research project in its own right, and each sits on the same unresolved fact: amyloid structure is decided by the environment, and the environment is always changing.
Related reading: Common Cosmetic Peptides: Copper, Carnosine, Glutathione, and More, Animal, Plant and Synthetic Peptides: Sources, Uses and Differences, Five Enzyme Families That Build Short Oligopeptides and Peptide Drugs, Peptide Targeting Agents for In Vivo Tumor Imaging: A Practical Primer.