DEER of Singly Labelled Proteins to Evaluate Supramolecular Packing of Amyloid Fibrils

Reviewer #1 (Public review):

Summary:

Proteins' misfolding into amyloid fibrils is the hallmark of neurodegenerative disorders. Tau fibrils, in particular, exhibit subtle structural variations that distinguish different pathologies. Understanding the mechanism of amyloid formation requires structural characterization, usually done by NMR or cryo-EM, and insights into fibril packing order and homogeneity remain limited.

Here, the authors exploit DEER echo decays of singly spin-labeled proteins to quantify packing order. While DEER is most used to measure intramolecular distances between two spin labels within a single protein, it also provides access to intermolecular distance distributions through the so-called background decay. This background decay has been theoretically described and can be used to characterize the spatial distribution of spins in terms of local spin concentration and the dimensionality of their arrangement. In the case of singly labeled proteins, the DEER signal contains only this intermolecular information. The authors propose using the extracted dimensionality as a reporter of packing disorder along the fibril axis and demonstrate this approach on the tau protein.

The background decay follows an exponential form with a time constant proportional to alphaD, where D is the dimensionality of the spin distribution and ranges from 1 to 3. For a homogeneous frozen solution of singly spin-labeled proteins, D = 3, and alpha is proportional to pbCL, where pb is the probability of changing the orientation of the spins excited by the DEER pump pulse, and CL is the local spin concentration. In a homogeneous system, CL equals the spin bulk concentration. The parameter pb is instrument-dependent and can be experimentally determined. When 𝐷<3, alpha takes a more complex form (given by Eq. 3), but remains linear C with a pre-factor that depends on 𝑝𝑏 and a defined function of D. For known C and pb, a plot of alpha vs C yields a linear curve, the slope of which can be used to determine D.

This approach was applied to the tau fragment tau187, labeled with a nitroxide spin label at positions 272C, 313C, 322C, and 404C. DEER measurements were performed on mixtures of labeled and unlabeled proteins at different ratios, and D was determined. DEER measurements were performed on mixtures of labeled and unlabeled protein at varying ratios to determine D. Fibril formation was induced by heparin, and the resulting decrease in D was monitored over time, reaching a final value of ~1.5. The authors find that the final dimensionality (D) is reached within 12 minutes and is independent of concentration. Consistent values of D ≈ 1.5 are observed for residues 272C, 313C, and 322C located in the protein core, whereas residue 404C, positioned in the C-terminal "fuzzy" region, yields a higher value of D ≈ 2.

Comparisons across tau variants show that heparin-induced fibrils of longer constructs are mispacked, whereas shorter tau fragments form well-ordered, seeding-competent fibrils with lower conformational variability. Seeded aggregation further improves templating and packing, as indicated by reduced dimensionality. Finally, the authors demonstrate that the local spin density derived from the α parameter can be used to estimate the number of protofilaments.

With the method now established, its application to other amyloid systems may reveal correlations between fibril packing order and disease-related properties.

Strengths:

This study presents an original, conceptually clear method for quantifying fibril packing using a single parameter (dimensionality). The approach is experimentally accessible and straightforward to analyze, making it broadly applicable with standard pulse EPR instrumentation.

Weaknesses:

A discussion about the meaning of D<1 is missing. In addition, the treatment of multi-protofilament fibrils is limited. In particular, it remains unclear how increases in dimensionality arising from multiple protofilaments start to affect D and how it can be distinguished from packing disorder.

Reviewer #2 (Public review):

This manuscript by Tsay et al. reports an EPR (electron paramagnetic resonance) approach based on double electron electron resonance spectroscopy (DEER) to characterize the supramolecular packing of amyloid fibrils. The authors claim that this approach can "deliver an apparent dimensionality of the supramolecular organization of tau fibrils", "assess the amyloid core location and packing order, and track time-resolved formation of aggregation intermediates".

Specifically, the authors used the electron spin echo (ESE) decay to report the arrangement of spin labels in the amyloid fibrils. When the spin labels are arranged in a straight line, a planar surface, or a 3D space, the dimensionality of the ESE decay would be 1, 2, and 3, respectively. To demonstrate their methods, the authors used a singly spin-labeled tau protein, which is involved in several amyloid diseases, including Alzheimer's and other tauopathies. For the truncated 0N4R tau (residues 244-441, named tau187), four labeling sites were studied (272, 313, 322, and 404). Residues 272, 313, and 322 gave a dimensionality of ~1.5, while residue 404 gave a dimensionality of ~2.0. The authors explained that residues 272, 313, and 322 are expected to be part of the amyloid core, while 404 is part of the so-called fuzzy coat. However, the authors then explained that all three amyloid core sites are misaligned because their dimensionality is ~1.5 instead of 1. Using a short tau fragment of 16 amino acids (residues 295-313), the authors show that this peptide formed fibrils with a dimensionality of 0.8. Using the short tau fragment fibrils as seeds, the authors obtained tau187 fibrils with a dimensionality of 1.3. Furthermore, the α parameter (a fitting parameter used to obtain the dimensionality) was used to interpret the protofilament composition.

While this approach has great potential in providing structural insights into amyloid fibrils, there are several critical flaws in experimental design, data analysis, and interpretation in the current version.

(1) The authors didn't rigorously establish the central premise of the DEER approach to characterize the supramolecular structure of amyloid fibrils. The parallel in-register β-sheet structure of amyloid fibrils is supposed to give a dimensionality of 1 in the ESE decay analysis. For tau187 fibrils, the authors obtained 1.5. For tau16 fibrils, the authors obtained 0.8. Because the theoretical lower limit of dimensionality is 1, tau16 fibrils do not serve as evidence that this approach can identify a perfectly aligned parallel in-register β-sheets. A 20% deviation from the theoretical value suggests the low accuracy of using ESE decay to report amyloid core structures. The high-resolution structures of tau fibrils have been widely reported using cryo-EM methods; it shouldn't be difficult for the authors to identify a good protein candidate to obtain a dimensionality of 1 to establish their methods. With a good protein candidate, rigorous data analysis should be presented to show how reliable a core site can be distinguished from a supposedly disordered site.

(2) Regarding the claim of probing protofilament composition using the α parameter, the authors should prepare fibrils with defined protofilament composition. A number of amyloid fibril structures have been solved to show different numbers of protofilaments.

(3) Regarding the claim of tracking "time-resolved formation of aggregation intermediates", the authors need to show more than a couple of data points, and the real-time aggregation needs to be accompanied by characterizations with complementary methods such as TEM.

(4) The authors largely ignored progress that has been made on the previous spin labeling studies of amyloid fibrils. A lot of the claims, such as identifying amyloid core, real-time aggregation, and the effects of seeding on structures, have been characterized extensively using continuous-wave EPR. It would be to the benefit of the readers to show what additional values this approach provides over existing methods.

Reviewer #3 (Public review):

In this work, Tsay et al. examine the challenge of inferring the ordering of amyloid fibrils. There is a clear need for such methodology. In their work, they computationally analyze the case of the expected decay in the DEER signal for spins randomly distributed in one, two, and three dimensions and show that (not unexpectedly) the decay is sensitive to dimensionality for a range of spin label concentrations. More intriguingly, they measure the dimensionality of tau amyloid labeled at several positions. Intriguingly, they show uniform (but unexpected) dimensionality when the label is in the fibril core. Through further simulations, they show that this anomalous dimensionality cannot arise from label attraction or repulsion (which can lead to deviations from random positions). Instead, this dimensionality is interpreted (again using compelling simulations) to arise from mis-registering due to changes in packing. Taken together, this paper convincingly shows that the DEER signal can be used to get site-specific information on amyloid dimensionality and can discriminate between regions of fibril core vs the "fuzz coat". Overall, this paper moves forward the methodology and opens up the technique to attractive applications in the areas of amyloid formation. More substantively, the field of DEER has been fixated on the dipolar modulation, and it is only once in a while now that one comes across a paper with a fresh breath of air - this paper certainly is!