{"id":"49d073eb-1902-42df-af69-408be60d8cce","arxiv_id":"2505.09151","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Using DFT and nudged elastic band calculations on an Aβ fibril-end dimer, the authors find twisted conformations are local minima separated by activation barriers controlled by sidechain steric hindrance.","lead":"The paper uses quantum chemistry calculations to map how the end of an amyloid-beta fibril can twist into bent shapes and how hard it is to twist back. This matters for Alzheimer's research because fibril growth and toxicity depend on what happens at the fibril tip, a region experiments struggle to resolve.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Starting structure PDB 6OIZ contains isoaspartate at residue 23, so the D23 torsional barriers used to support the sidechain-size claim may not describe the native Aβ fibril end.","rationale":"The reader's weakest assumption was the neglected solvent effect, which the paper explicitly discusses and attempts to justify via analogy. I find a more fundamental and unacknowledged problem: the structural model is built from PDB 6OIZ, a crystal structure of Aβ(20–34) containing isoaspartate at position 23, as stated in the reference the authors themselves cite. The manuscript treats residue 23 as standard Asp and computes ψ/φ torsions for it, but isoaspartate has a different backbone connectivity, so those dihedral angles and the resulting steric contacts do not represent the native Aβ fibril end. This directly undermines the D23 barriers in Table 1 and Figure 3, which are among the strongest evidence for the sidechain-size rule. The concern is concrete, verifiable from the PDB entry, and does not rely on speculation about solvent effects. However, the paper's qualitative conclusion still has support from the G25/S26 and E22/V24 comparisons, so conditional acceptance remains appropriate, provided the authors either correct the model or clearly reframe the D23 results as pertaining to the isoAsp23 variant. The verdict therefore stays conditional, but with an additional explicit condition that the reader had not identified. I mark agreement as 'disagree' because the reader's stated weakest assumption (solvent) is not the most load-bearing issue; the structural anomaly is more specific and falsifiable. The proposed test—rebuilding the dimer from a native fibril structure and recomputing the D23 barriers—would settle whether the D23 data are valid for the native sequence or are artifacts of isoaspartate.","tokens_in":160,"tokens_out":4456,"duration_ms":58100,"concrete_test":"Inspect the sequence and backbone connectivity of residue 23 in PDB 6OIZ (it is isoaspartate, not Asp). Then rebuild the dimer from a native Aβ fibril-end structure containing standard L-Asp at position 23, using coordinates from an Aβ40 cryo-EM fibril structure (e.g., PDB 2M4J or 5OQV) trimmed to residues 20–34, and repeat the same PBEsol+D3 structural optimizations and NEB calculations for the ψ and φ torsions of residue 23. If the D23(ψ) activation barrier (0.61 eV) and the close O–O contact at the barrier top (2.65 Å in Table 1) change substantially, the sidechain-size rule loses a key supporting example and the manuscript must explicitly state that the D23 results are specific to the isoaspartyl variant, not the native sequence.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim—that a twisted conformation's reversion propensity is controlled by sidechain size near the torsional axis—relies on comparing small-barrier G25 with larger-barrier E22, D23, V24, and S26. The strongest supporting data are the D23 torsions, which give a 0.61 eV barrier for D23(ψ) and a 0.43 eV barrier for D23(φ), with the D23(ψ) barrier attributed to steric contact between the D23 carboxyl group and the lower monomer. However, the starting structure, PDB 6OIZ, is not a native Aβ(20–34) fibril fragment. The title of the cited experimental paper [29] explicitly states that this crystal structure contains 'Alzheimer's-associated isomerization at Asp23,' i.e., an L-isoaspartate residue at position 23. In isoaspartate, the aspartate sidechain carboxyl is incorporated into the polypeptide backbone and an extra methylene is inserted into the mainchain, changing the local backbone connectivity, the sidechain position, and the definition of ψ/φ dihedral angles. The manuscript treats residue 23 as standard Asp and defines ψ/φ torsions for D23 on this non-standard backbone without any acknowledgment. The computed D23(ψ) barrier and the O–O distances in Table 1 may therefore be artifacts of the isoaspartyl linkage rather than properties of the native Aβ fibril end. If the D23 data are excluded, the remaining comparison (E22, V24, G25, S26) still shows a glycine outlier, but the sidechain-size argument loses one of its two charged-residue examples and a key quantified steric contact. The solvent assumption, while acknowledged in the Discussion, is a modeling limitation; the isoAsp issue is an unacknowledged error in the structural model itself, making it the more load-bearing concern.","agreement_with_reader":"disagree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper uses PBEsol+D3 density functional theory with gamma-point PWscf sampling and nudged elastic band calculations to study torsional conformational transitions at a model of the amyloid-beta fibril end. The model is a two-monomer parallel beta-sheet built from PDB 6OIZ (A-beta residues 20-34), and the paper examines six twisted conformations generated by psi/phi torsions at E22, D23, V24, G25, and S26. All twisted conformations are found to be endothermic relative to the fibril conformation, with endothermicity correlated with the number of broken mainchain hydrogen bonds and, in some cases, with the loss of van der Waals contacts. NEB barrier heights for the twisted-to-fibril transition range from 0.43 to 0.61 eV, except for the psi torsion of G25, which gives 0.05 eV. The authors attribute the barrier heights to local steric hindrance between the sidechain near the torsional axis and the lower monomer, and conclude that the propensity of a twisted conformation to revert to the fibril conformation depends strongly on the sidechain size and orientation near the torsional axis.","tokens_in":10455,"tokens_out":4458,"duration_ms":44153,"significance":"If the results hold, the paper provides an ab initio, parameter-free characterization of a local conformational transition at an amyloid fibril end and yields a falsifiable prediction: twisted conformations near bulky residues are kinetically trapped, while glycine-site twists revert readily. The strengths of the work are that no quantities are fitted to the target data, the NEB protocol directly computes barrier heights rather than inferring them from heuristics, and the paired comparisons within the same residue segment control some structural confounds. The structural analysis in terms of specific interatomic distances is also a useful step. However, the quantitative reliability of the barrier ranking is limited by the absence of convergence tests and error bars, by gamma-point-only sampling, by the gas-phase dimer model, and by a significant concern about the residue-23 stereochemistry in the starting crystal structure. The central claim is qualitatively plausible but is not quantitatively established in the present form.","major_comments":[{"comment":"The dimer model is built from PDB 6OIZ, which according to reference [29] contains an L-isoaspartate at residue 23. The manuscript nevertheless treats residue 23 as standard aspartate and defines psi/phi torsions for D23, and Table 1 reports a 0.61 eV barrier for D23(psi) together with O-O distances attributed to the D23 carboxyl group. In isoaspartate, the sidechain carboxyl is incorporated into the mainchain and an extra methylene inserts into the backbone, so the computed D23 torsions and the associated steric analysis do not describe the native A-beta fibril end. This weakens the sidechain-size argument, which uses D23 as one of its charged-residue examples; the remaining E22, V24, G25, and S26 comparison still shows a glycine outlier, but the D23 data must either be re-computed on a native Asp23 model or explicitly removed and the conclusions re-derived.","section":"Section 2.1 and Table 1"},{"comment":"The calculations are performed only at the gamma point for an isolated dimer in a 45 x 45 x 40 Angstrom cell, and no convergence tests, DFT-D3 benchmarking, or error bars are reported. The G25(psi) barrier of 0.05 eV is almost certainly below the intrinsic uncertainty of PBEsol+D3 for hydrogen-bonded peptide systems, so the quantitative contrast between 'low' and 'high' barriers is not established. Please add cutoff, k-point, and cell-size convergence checks and a realistic uncertainty estimate, or recast the comparison as a qualitative ordering with explicit caveats about the accuracy of the functional.","section":"Section 2.2 and Table 1"},{"comment":"The solvent discussion assumes that water does not disrupt the local conformation around the torsional axis and that the steric barrier survives in solution, citing only the authors' own adsorption studies [39,40] as an analogy. Because the twisted minima are gas-phase minima and the endothermicity would likely change substantially in an aqueous environment, the experimental prediction (twisted conformations at bulky residues are observable at fibril ends) rests on an untested assumption. At minimum, a continuum-solvent calculation or an explicit-water test of one representative barrier is needed before this prediction is stated as a conclusion.","section":"Section 4"},{"comment":"The NEB protocol is not fully specified: the number of images, the spring constants, the convergence threshold for the band, and whether a climbing-image or improved-tangent variant was used are not reported. Without these details the barrier heights in Table 1 cannot be reproduced or independently assessed, which is a serious reproducibility issue for a quantitative barrier comparison.","section":"Section 2.2"}],"minor_comments":[{"comment":"The row labeled 'S25 (phi)' should read 'S26 (phi)' to match the text, Figure 4, and the residue numbering used throughout the manuscript.","section":"Table 1"},{"comment":"The text refers to the 'phi torsion of D24' twice when describing the barrier analysis; the intended label is presumably 'phi torsion of V24', as the torsional axis is in the 23-24 segment.","section":"Section 3.3"},{"comment":"The phrase 'Hellman-Feynman force' should be 'Hellmann-Feynman force' (double 'n').","section":"Section 2.2"},{"comment":"The statement that 'similar results are expected for torsions that lift the peptide chain from the C-terminus' is an untested extrapolation; it would be preferable to label it explicitly as a conjecture.","section":"Section 3.2"},{"comment":"The panel headers in Figure 4 are difficult to read because of the small font size and the use of Greek symbols in subscripts; using a clearer notation such as 'E22 psi' and 'D23 phi' would improve legibility.","section":"Figure 4"}],"recommendation":"major_revision","confidential_remarks":"The isoaspartate issue is the main correctness risk: the D23 torsional data in Table 1 and the associated steric analysis are not valid for a native A-beta fibril end unless the model is rebuilt with standard aspartate at position 23. The remaining qualitative trend (glycine low barrier, bulky residues higher barriers) is plausible, but the quantitative barrier claims need error estimates and the NEB details need to be reported. I would not accept the manuscript until the D23 problem is resolved or the conclusions are explicitly re-derived without the D23 data."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Plain take: the paper is worth a referee, but the isoAsp issue at residue 23 is real and needs fixing before the D23 numbers can support the central claim. What is new: the computed set of six twisted conformations for the Abeta(20-34) fibril-end dimer, the relative energies from DFT-D3, and the NEB barriers. The paired psi/phi comparisons within each segment are a good design; they control for broken hydrogen bonds and make the glycine outlier in G25(psi) fairly clean. The steric-hindrance explanation is reasonable for the remaining residues. Credit: the calculations are standard, no fitting, and the paper is honest about the missing solvent. Soft spots: the starting structure PDB 6OIZ is not native Abeta; the published structure contains L-isoaspartate at residue 23. The manuscript never mentions this, yet defines psi/phi on residue 23 as if it were normal Asp. The D23 barriers in Table 1 and the O-O distances are therefore possibly artifacts of the isoAsp linkage. Excluding D23 data, the sidechain-size argument still has E22, V24, S26 versus glycine, but the compelling charged-residue example is gone. Minor issues: no error bars or convergence tests, gamma-point only, one starting structure, zero-temperature barriers described as 'propensity to revert,' and Table 1 has residue-label slips (S25 vs S26, V24 vs D24). I would not cite this as quantitative, but as a qualitative mechanism it is plausible. Recommendation: send to peer review, but the authors should first rebuild the model with normal L-Asp at position 23 and recompute the D23 pathways.","headline":"Useful DFT/NEB study of twisted conformations at an Abeta fibril-end dimer, with a clear steric-hindrance argument, but the starting structure 6OIZ contains isoaspartate at residue 23, which undermines the D23 barriers.","tokens_in":722,"tokens_out":1335,"would_cite":false,"duration_ms":41845,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"At the end of an amyloid-β fibril, a peptide segment twisted off the sheet is a real local minimum, and whether it snaps back is set by the size of the sidechain at the torsion axis: bulky sidechains impose ~0.5 eV return barriers, while…","keywords":["amyloid β peptide","fibril end","twisted conformation","density functional theory","nudged elastic band","potential energy surface","steric hindrance","Alzheimer's disease"],"falsifier":"Recompute the six torsional pathways with explicit water, for example by embedding the fibril-end dimer in a QM/MM calculation, and compare the G25($\\psi$) return barrier with the bulky-sidechain barriers; if the near-zero barrier is not preserved, the steric-gating claim fails. A complementary experimental check is to look for twisted conformations at fibril ends by cryo-electron microscopy or solid-state NMR: the claim predicts that persistent twisted ends should be found at axes with bulky sidechains, not at glycine axes.","tokens_in":9951,"feed_emoji":"🧬","tokens_out":13426,"duration_ms":119168,"temperature":0.7,"pith_summary":"This paper asks what happens when the free end of an amyloid-β fibril twists: can a part of the terminal peptide lift away from its neighbor, and once lifted, does it return? Using density-functional theory on a two-peptide model of the fibril end, it shows that these twisted states are genuine local minima, always less stable than the fully stacked fibril state, and that the cost of creating them rises with the number of backbone hydrogen bonds broken and with the loss of hydrophobic contacts. The central result is about coming back: the twisted-to-fibril activation barrier is about 0.4–0.6 eV when a bulky sidechain (glutamate, aspartate, valine) sits near the torsion axis, but only about 0.05 eV for a glycine-site twist, where the sidechain is a single hydrogen atom. The paper concludes that, apart from one case stabilized by an internal hydrogen bond, the tendency of a twisted fibril end to revert is governed by the local steric hindrance around the torsion axis. This matters because fibril elongation happens at the fibril end, so the same sidechain gating could control how long the end stays open for monomer addition.","feed_headline":"Bulky sidechains trap twisted ends of amyloid fibrils","feed_subtitle":"Computer simulations find a near-zero return barrier at glycine sites but ~0.5 eV barriers where bulky sidechains sit.","key_machinery":"The central object is the twisted conformation: a local minimum in which part of the upper monomer of the fibril-end dimer is lifted from the lower monomer by torsion about one backbone single bond, while the rest remains stacked in the parallel $\\beta$-sheet. The machinery that carries the argument is the nudged elastic band (NEB) minimum-energy path, together with a count of broken backbone hydrogen bonds and a set of interatomic contact distances measured at the barrier top and at the twisted minimum. These convert the structural idea of 'sidechain near the torsional axis' into a quantitative barrier ranking.","core_discovery":"At the terminal layer of an amyloid-β fibril, rotating the backbone around a single bond—the $\\psi$ or $\\phi$ dihedral angle—can lift a stretch of the upper peptide off its neighbor while the rest remains hydrogen-bonded in the parallel $\\beta$-sheet. The paper calls these twisted conformations and shows by density-functional theory that each is a local minimum sitting above the fully stacked fibril conformation, with the endothermic energy tracking the number of broken backbone hydrogen bonds ($\\Delta N_{\\mathrm{HB}}$) and the loss of hydrophobic dispersion contacts such as the F20 $\\pi$-stacking. The central quantitative finding is the return barrier: nudged elastic band paths give twisted-to-fibril barriers of roughly 0.4–0.6 eV when a bulky sidechain (E22, D23, V24) sits near the torsion axis, because the sidechain collides with atoms of the lower monomer on the way back, but only about 0.05 eV for the G25 $\\psi$ torsion, where the sidechain is a single hydrogen. In the one small-sidechain case with a high barrier, the S26 $\\phi$ torsion, an intramolecular hydrogen bond stabilizes the twisted form; setting that case aside, the paper concludes that the propensity to revert is controlled by the local steric hindrance around the torsion axis.","pith_inferences":["If the steric-gating rule is right, replacing G25 with a residue carrying a bulkier sidechain in the same dimer model should raise the near-zero return barrier into the 0.4–0.6 eV range; the paper does not test mutations, but the mechanism predicts this.","The same local-steric logic should apply to torsion axes that lift the C-terminal side of the chain; the paper states this expectation but does not compute those pathways.","In a longer fibril, the lower monomer is itself anchored to further monomers, which may stiffen the sheet and shift the barriers, so whether the dimer ranking survives in a trimer or larger model is an open test.","If twisted states are real in solution, they may appear as short-lived frayed ends in single-molecule experiments, with the paper implying that glycine-adjacent twists would be too short-lived to catch whereas bulky-sidechain twists would be the observable ones."],"forward_implications":["Twisted states at fibril ends should persist precisely where the torsion axis passes through residues with bulky sidechains, making those the sites where partially detached, open conformations can be captured.","A twist at a glycine site such as G25($\\psi$) should relax back to the full $\\beta$-sheet almost immediately, so open states at glycine positions are unlikely to survive long enough to be observed.","Because the endothermic energy grows with the number of broken backbone hydrogen bonds, twists that lift longer N-terminal stretches are both less stable and more costly, giving a concrete ranking of which torsional axes are plausible in a growing fibril.","The barrier gap—about 0.4–0.6 eV versus 0.05 eV—implies very different return time scales, so bulky-sidechain twisted ends should revert much more slowly than glycine-site twisted ends at physiological temperature.","If water does not disrupt the local packing, the same steric barrier ordering should survive in solution even though the relative stabilities of the twisted states will shift because water can hydrogen-bond the exposed backbone."],"supporting_citations":[{"why":"supplies the crystallized Aβ(20–34) atomic coordinates used to build the fibril-end dimer model.","marker":"[29]"},{"why":"provides the density-functional theory code used for all structural optimizations and energy calculations.","marker":"[30-32]"},{"why":"supplies the exchange-correlation functional used in the DFT calculations.","marker":"[33]"},{"why":"provides the dispersion correction needed to describe the van der Waals interactions that set the energy trends.","marker":"[34]"},{"why":"supplies the nudged elastic band method used to compute the potential energy surfaces and activation barriers.","marker":"[36]"},{"why":"defines the geometric criteria used to count intermolecular hydrogen bonds and determine the number broken by each torsion.","marker":"[38]"},{"why":"provides the analogy that torsional metastable states can retain an activation barrier in solution, supporting the solvent assumption.","marker":"[39,40]"}],"fun_headline_variants":["Bulky sidechains trap twisted amyloid fibril ends","Glycine sites let amyloid fibril ends revert quickly","Return barrier at amyloid end grows with sidechain size","Twisted A-beta ends: steric hindrance decides the way back","Fibril end twist reversal hinges on sidechain bulk"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The conclusions assume that water does not rearrange the local contacts around the torsion axis, so the vacuum-computed barrier heights still decide which twisted ends snap back quickly and which do not.","fun_headline_variants_meta":{"raw":{"variants":["Bulky sidechains trap twisted amyloid fibril ends","Glycine sites let amyloid fibril ends revert quickly","Return barrier at amyloid end grows with sidechain size","Twisted A-beta ends: steric hindrance decides the way back","Fibril end twist reversal hinges on sidechain bulk"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000216,"raw_usage":{"total_tokens":1476,"prompt_tokens":1031,"completion_tokens":445,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":647,"completion_tokens_details":{"reasoning_tokens":364}},"tokens_in":647,"tokens_out":445,"duration_ms":4535,"temperature":1.0,"reasoning_tokens":364,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T21:38:02.667861+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Recompute the six torsional pathways with explicit water, for example by embedding the fibril-end dimer in a QM/MM calculation, and compare the G25($\\psi$) return barrier with the bulky-sidechain barriers; if the near-zero barrier is not preserved, the steric-gating claim fails. A complementary experimental check is to look for twisted conformations at fibril ends by cryo-electron microscopy or solid-state NMR: the claim predicts that persistent twisted ends should be found at axes with bulky sidechains, not at glycine axes.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"supplies the crystallized Aβ(20–34) atomic coordinates used to build the fibril-end dimer model."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"supplies the exchange-correlation functional used in the DFT calculations."},{"cited_title":"Grimme, J","cited_arxiv_id":null,"evidence_quote":"provides the dispersion correction needed to describe the van der Waals interactions that set the energy trends."},{"cited_title":"Henkelman and H","cited_arxiv_id":null,"evidence_quote":"supplies the nudged elastic band method used to compute the potential energy surfaces and activation barriers."},{"cited_title":"Zheng, E","cited_arxiv_id":null,"evidence_quote":"defines the geometric criteria used to count intermolecular hydrogen bonds and determine the number broken by each torsion."}],"review_version":1}