Hemochromatosis neural archetype reveals iron disruption in motor circuits

Abstract

Our understanding of brain iron regulation and its disruption in disease is limited. Excess iron affects motor circuitry, contributing to Parkinson's disease (PD) risk. The molecular mechanisms regulating central iron levels, beyond a few well-known genes controlling peripheral iron, remain unclear. We generated scores based on the archetypal brain iron accumulation observed in magnetic resonance imaging scans of individuals with excessive dietary iron absorption and hemochromatosis risk. Genome-wide analysis revealed that this score is highly heritable, identifying loci associated with iron homeostasis, and driven by peripheral iron levels. Our score predicted gait abnormalities and showed a U-shaped relationship with PD risk, identifying individuals with threefold increased risk. These results establish a hormetic relationship between brain iron and PD risk, where central iron levels are strongly determined by genetics via peripheral iron. This framework combining forward and reverse genetics is a powerful study design to understand genomic drivers underlying high dimensional phenotypes.

Fig. 1.. Overview of study design.

Top: Hemochromatosis Brain classifier training in subsample A of UKB to differentiate controls from C282Y homozygotes from T2-weighted MRI scans, with univariate and regularized classifier weights, classifier performance (receiver operator characteristic), and feature importance by brain region. Bottom: Deploy classifier in subsample B (UKB) and ABCD Study to generate PVS capturing Hemochromatosis Brain liability. In subsample B, we then conduct a GWAS to find variants associated with this PVS liability scale. Using these GWAS results, we perform tissue and single-cell enrichment analysis, Mendelian randomization (MR), and PolyGenic Score (PGS) validation against the PVS generated from the ABCD Study. Last, we test this PVS against neurological disorders within UKB. AUC, area under the curve.

Fig. 2.. Results of GWAS on PVS of Hemochromatosis Brain in subsample B.

(A) Manhattan plot GWAS result with peaks annotated—a total of 42 loci are discovered (see supplementary data tables for full list). Red dots indicate 13 novel loci. (B) Brain cell type enrichment of GWAS signal using FUMA, and red bars indicate FDR significantly enriched cell types. (C) Iron concentration (measured by using x-ray spectrometry) of cell types in rat brain with permissions from Reinert et al. (38). *P < 0.001. (D) MR results, using GSMR, to quantify strength of causal relationship between peripheral blood iron markers and brain iron accumulation as measured by Hemochromatosis Brain PVS. Each plot shows results of conditioning on peripheral blood iron markers as exposure and PVS as outcome, and reverse direction GSMR results are shown in fig. S28. See table S5 for full MR results. Error bars in (C) and (D) represent SEs in estimates.

Fig. 3.. Logistic regression of PVS against neurological disorder diagnoses.

Inverse probability weights (IPW) for each disorder can be found in table S6. (A) PVS OR predicting each neurological disorder. (B) PVS quantile weighted regression to predict PD in subsample C, each point represents a categorical factor indicating one of four PVS quantiles (blue) or C282Y homozygosity (orange). IPW of 3.89 was used for PD cases in imaging sample (blue)—see Materials and Methods. Regression (y axis) was performed using PVS from T2-weighted, and x axis is an estimate of mean brain iron concentration using T2* imaging for each group. Error bars for both plots indicate 95% confidence intervals (CI). The Pearson correlation of PVS with estimated mean brain iron concentration was 0.41.