In vivo and in silico alpha-synuclein propagation dynamics: The role of genotype, epicentre, and connectivity

Reviewer #1 (Public review):

Summary:

Tullo et al. address an important and currently unresolved mechanistic question: does the prion-like spreading of alpha-synuclein (aSyn) generalize across three biological factors: host genotype, preformed-fibril (PFF) species, and disease epicentre (brain region); and can the resulting neurodegeneration be predicted computationally? Using a longitudinal design, adult M83 A53T-hemizygous mice and wild-type littermates received intrastriatal human- or mouse-PFF or PBS, with in vivo brain MRI at 7T, motor testing, survival and weight followed to 120 days post-injection. A parallel experiment seeded human-PFF or PBS into the hippocampal dentate gyrus. Atrophy was quantified by deformation-based morphometry, brain-behaviour coupling by partial least squares, and spread was simulated with a Susceptible-Infected-Removed agent-based model constrained by the Allen mouse connectome and SNCA expression. The authors conclude that aSyn-associated atrophy generalizes across genotype and fibril species but is anatomically distinct for the two epicentres, emphasizing regional vulnerability. This is a technically strong and ambitious study, reflecting a substantial and well-executed research effort.

Strengths:

This is a technically accomplished and ambitious study from a group with clear expertise in mouse neuroimaging and network modelling. The study addresses a real knowledge gap, relevant for mechanistic explorations of alpha-synucleinopathies: genotype and fibril inoculum species have rarely been compared head-to-head, and the relationship between aSyn propagation and downstream atrophy outside the striatum has been under-examined so far. The central hypothesis, that regional vulnerability constrains aSyn-associated neurodegeneration, together with the first attempt to model aSyn-induced atrophy computationally in rodents, is conceptually and methodologically valuable and translationally relevant.

The longitudinal dataset is unusually rich (687 in vivo scans), and both the data and the analysis pipeline are openly shared (OpenNeuro ds007671, Zenodo, GitHub), which is extremely important for reproducibility and of clear value to the community. The work uses a multi-modal methodology in which the same phenomenon is examined across anatomical MRI, multivariate brain-behaviour modelling, and a mechanistic simulation, and is the first to model aSyn-induced atrophy computationally in mice.

Another strength is the consistent incorporation of sex as a biological variable throughout, including sex-stratified survival, behavioural, and voxelwise atrophy analyses. This allowed the authors to identify sex differences in disease progression, notably in survival and symptom onset, while indicating that the core PFF-induced atrophy pattern was largely preserved across sexes.

The core descriptive findings, that PFF-induced atrophy and motor impairment are reproducible across genotype and fibril species, and that striatal and hippocampal seeding yield distinct anatomical signatures, are convincingly supported and represent a very valuable advance.

Weaknesses:

The strongest mechanistic and epicentre interpretations would benefit from some additional support, although the study's core findings are robust.

First, the framing centres on aSyn propagation, but the sole in-cohort readout is MRI-derived atrophy assessment; no aSyn/phospho-Ser129 pathology is shown for these animals. The atrophy-propagation link remains inferential.

Second, the computational model's fit is reported as the peak correlation across simulation time steps and is not yet benchmarked against null or baseline models, so it is somewhat difficult to determine how much the connectome and dynamics contribute beyond gene expression alone; parameter provenance is also not described in the text.

Third, the epicentre difference is well supported empirically at matched inoculum, but the computational comparison (SIR) is so far inoculum-mismatched: the striatal model was evaluated against mouse-PFF atrophy while the hippocampal model used human-PFF. A matched striatal human-PFF map is already available, so this could be reconciled without new data. The reduced hippocampal vulnerability despite higher hippocampal SNCA expression also remains unexplained.

Appraisal and impact:

The authors largely achieve their aims, and the generalization of atrophy across genotype and fibril species, together with the epicentre-specific anatomy, is well supported by a strong and openly available dataset. The more mechanistic conclusions - aSyn propagation specifically, connectome-driven vulnerability, and epicentre-determined resistance - would be strengthened where feasible by pathology validation, model benchmarking, and completing the already-available matched computational comparison, and should be interpreted with corresponding caution. Even so, the combination of a large longitudinal imaging resource, a factorial in vivo design, and the first rodent computational model of aSyn-related atrophy makes this a valuable contribution that is likely to be a useful reference and methodological template for the synucleinopathy and network-neurodegeneration communities.

Reviewer #2 (Public review):

Summary:

This study explores risk factors for neural atrophy following alpha-synuclein injection from two complementary perspectives. First, it evaluates the effect of biological and experimental factors (genotype, alpha-synuclein species, biological sex, seeded brain region and time since injection) on the extent of neural atrophy. Second, it assesses whether regional biological features (gene expression and structural connectivity) can predict the spatial distribution of that atrophy. Using longitudinal in vivo MRI, the authors map brain volume changes over time. They relate the brain changes from striatum seeding to behavioral outcome, identifying factors associated with more severe pathology. Finally, the authors validate a previously developed in silico model for predicting brain atrophy from alpha-synuclein seeding. The model is based on the alpha-synuclein prion-like spreading hypothesis and uses local gene expression and structural connectivity to predict atrophy following the injection. They conclude that the model accurately predicts atrophy following striatal seeding but performs poorly for hippocampal seeding. They further show that structural connectivity alone is insufficient to explain the observed atrophy after striatal seeding, and that incorporating regional gene expression substantially improves model performance.

Strengths:

The authors have expanded on their previous work by systematically evaluating how multiple biological and experimental variables influence the development of brain atrophy. The use of MRI to map structural changes and the subsequent analysis is well validated by this group and enables comprehensive whole-brain quantification across a large number of experimental conditions. The evaluation of the in silico model linking regional gene expression and structural connectivity to patterns of atrophy under different experimental conditions is important for expanding our understanding of how atrophy develops in synucleinopathies.

Weaknesses:

My principal concern is that the manuscript is framed as an investigation of alpha-synuclein propagation, whereas the primary outcome measured throughout the study is a change in regional brain volume. Although atrophy is likely related to the underlying spread of pathological alpha-synuclein, the spatial distribution of alpha-synuclein pathology is not directly quantified. Conclusions regarding propagation of alpha-synuclein and the relationship with tissue loss are inferred from the performance of the in silico model in predicting atrophy. I think the manuscript could be revised to make this distinction clearer.

A second concern relates to the comparison between striatal and hippocampal seeding. A key conclusion of the manuscript is that the in silico model accurately predicts atrophy following striatal seeding but not hippocampal seeding. However, the two analyses use different experimental group comparisons (striatum: M83 Ms-PFF versus WT PBS; hippocampus: M83 Hu-PFF versus M83 PBS). It would be helpful to demonstrate that the observed difference in model performance is not attributable to these differing experimental/ control groups.

Reviewer #3 (Public review):

Summary:

This work studied the prion-like α-synuclein spreading hypothesis from the view of different host genotypes (M83 transgenic vs wild-type), fibril species (mouse vs human PFFs), and disease epicenter (striatum vs hippocampus). Major results include tracking neurodegeneration longitudinally with in vivo MRI, behavior, and survival in the same mice. Furthermore, this work sought to link atrophy patterns to structural connectivity and regional SNCA expression. Finally, the authors tested whether a connectome-based SIR spreading model could predict the atrophy in silico and generalize across seed sites.

Strengths:

(1) Same mice imaged repeatedly across four timepoints (−7, 30, 90, 120 dpi), giving true within-subject volumetric trajectories rather than cross-sectional snapshots.

(2) Investigate the atrophy pattern for striatal-vs-hippocampal seeding in PD.

Weaknesses:

(1) The hypothesis (regional vulnerability) is not novel, although the manuscript presents compelling and interesting results supporting it in Figures 2 and 3.

(2) The findings primarily establish statistical associations rather than causal mechanisms. This limitation appears inherent to the cross-cohort dataset utilized, which the authors should explicitly address in the discussion.

(3) The descriptions of the statistical analyses in Sections 2.5 and 2.6 lack sufficient detail. The authors should provide additional technical specifics to ensure reproducibility.

(4) Given that VBM was used to determine atrophy patterns, it is necessary to address how the multiple comparisons problem was handled in the statistical analysis to control for false positives.