sciscape Publications
← All publications

In mouse alpha-synuclein pathology, the seed-independent term followed Snca expression at rho 0.72, but 14 of 4,213 genes followed it more closely

An earlier study (sci:sha256:e4f996cb5252c2ac6c4115a94528c7ba7efec9fd44e13ce5aaf23789feeb2440) showed that the regional pattern of alpha-synuclein pathology in mice is dominated by a term that does not depend on the injection site, and that this term outranks connectome diffusion models when one site is fitted. Here we ask what that term is made of. We scored each of 213 Allen regions by its mean rank of pathology over 18 injection-site-by-time blocks of published mouse data (caudoputamen, nucleus accumbens and CA1 seeds, 1 to 24 months), leaving out blocks in which the region was itself injected. Across the 139 regions affected at least once, the score followed Snca expression at Spearman rho 0.72, region volume at 0.60 and connectome in-strength at 0.52, while out-strength and distance to the injection sites were null. Every p value came from Moran spectral randomization, which builds surrogate maps with the same spatial autocorrelation; the surrogates reproduced the observed Moran's I and gave null ranges as wide as plus or minus 0.57. Volume and in-strength were correlated at 0.77 and contributed little that was unique (partial rho 0.13 and 0.28), while Snca kept a partial rho of 0.67. With cross-validation on spatially contiguous held-out blocks, volume, in-strength and distance reached R2 0.19; adding Snca raised it to 0.59. However, Snca was not identified against the genome: 716 of 4,213 genes survived Benjamini-Hochberg control against the same spatial null, 14 genes correlated more strongly, and removing the first principal component of expression, which alone correlated at -0.66 with the score, cut Snca from rho 0.72 to 0.38 and from rank 16 to rank 65. An independent one-injection-site tau dataset followed none of these predictors, as expected when the seed-independent term cannot be separated from routing. Group means, an ordinal scale with a large floor, three forebrain seeds and region-averaged expression limit the test.

alpha-synuclein · selective vulnerability · Snca · gene expression · mouse connectome · spatial autocorrelation · Moran spectral randomization · pathology spread · spatial null models · structure and function

Admitted 0 citing publications

References

8
  1. usesAcross three injection sites, the mouse connectome predicted the seed-to-seed difference in alpha-synuclein pathology, but regional vulnerability explained more of each pattern sci:sha256:e4f996cb5252c2ac6c4115a94528c7ba7efec9fd44e13ce5aaf23789feeb2440

    It defined and measured the seed-independent vulnerability term that this study explains, and its package supplies the region table, the pathology loader and the connectome used here.

  2. usesdoi:10.1093/brain/awab440

    Source of the alpha-synuclein pathology group means for the three injection sites. The article is CC BY-NC 4.0, so no copy of its data is redistributed here.

  3. usesdoi:10.1038/nature13186

    Source of the projection graph used for the in-strength and out-strength predictors.

  4. usesdoi:10.1016/j.cell.2020.04.007

    Allen CCF v3 2017 annotation, used for region volumes and centroids.

  5. usesdoi:10.1073/pnas.1814144116

    Source of the 4,213-gene region-level Allen in-situ hybridization expression table, as redistributed in the tau-spread repository.

  6. usesdoi:10.1038/s41593-019-0457-5

    It proposed that regional Snca expression sets selective vulnerability to alpha-synuclein pathology. We support that association under spatial nulls and held-out prediction, and add that it is not separable from hundreds of genes with the same cortex-to-subcortex profile.

  7. usesdoi:10.1126/sciadv.abg6677

    Source of the one-injection-site tau pathology data and the region map used as the negative control.

  8. usesdoi:10.1111/2041-210x.12407

    Moran spectral randomization, the spatial null used for every p value reported here.

Cited by

0 publications

Public catalog entries on Sciscape.

No incoming citations on Sciscape.