Huntington’s disease blood and brain show a common gene expression pattern and share an immune signature with Alzheimer’s disease

There is widespread transcriptional dysregulation in Huntington’s disease (HD) brain, but analysis is inevitably limited by advanced disease and postmortem changes. However, mutant HTT is ubiquitously expressed and acts systemically, meaning blood, which is readily available and contains cells that are dysfunctional in HD, could act as a surrogate for brain tissue. We conducted an RNA-Seq transcriptomic analysis using whole blood from two HD cohorts, and performed gene set enrichment analysis using public databases and weighted correlation network analysis modules from HD and control brain datasets. We identified dysregulated gene sets in blood that replicated in the independent cohorts, correlated with disease severity, corresponded to the most significantly dysregulated modules in the HD caudate, the most prominently affected brain region, and significantly overlapped with the transcriptional signature of HD myeloid cells. High-throughput sequencing technologies and use of gene sets likely surmounted the limitations of previously inconsistent HD blood expression studies. Our results suggest transcription is disrupted in peripheral cells in HD through mechanisms that parallel those in brain. Immune upregulation in HD overlapped with Alzheimer’s disease, suggesting a common pathogenic mechanism involving macrophage phagocytosis and microglial synaptic pruning, and raises the potential for shared therapeutic approaches.


Supplementary figures
. Network diagram of the relationship between significantly (q<0.05) upregulated gene modules (Table 4) and generic biological pathways (Table S2)  Intensity of shading indicates p-value (darker colours have lower p-values), node size indicates size of gene content, node shape indicates origin of data (modules or pathways). For clarity, biological pathways with similar gene content are grouped together, as described in Supplementary Table S18, and the shading reflects the most significant pathway in the group. Nodes are arranged such that the distance between them reflects similarity in gene content. Diagram rendered in Cytoscape.  (Table 4) and generic biological pathways (Table S3) based on shared gene membership. The thickness of the edges corresponds to the proportion of overlap from the smaller term to the larger (overlap coefficient). Intensity of shading indicates p-value (darker colours have lower p-values), node size indicates size of gene content, node shape indicates origin of data (modules or pathways). For clarity, biological pathways with similar gene content are grouped together, as described in Supplementary Table S19, and the shading reflects the most significant pathway in the group. Nodes are arranged such that the distance between them reflects similarity in gene content. Diagram rendered in Cytoscape. Table S1. Differential expression analysis in HD (premanifest and manifest combined) versus controls for the combined Track-HD and Leiden cohorts. p (diffexp) -p value for differential expression between HD and controls; q (diffexp) -q value shows correction for multiple testing in the combined dataset; Log2(FC) -log2 of the ratio of the mean counts in HD and controls.  .   Table S4. The 10 most significantly dysregulated genes (p<0.01) in up or downregulated generic pathways (q<0.05). p (Comb/Track-HD/Leiden) -p value for differential expression between HD and controls in the combined, Track-HD or Leiden datasets; Log2FC -log2 of the ratio of mean counts in HD and controls. Table S5. All significantly dysregulated genes (p<0.05) from generic pathways that were dysregulated (up or down) in HD blood. p (Comb/Track-HD/Leiden) -p value for differential expression between HD and controls in the combined, Track-HD or Leiden datasets; Log2FC -log2 of the ratio of the mean counts in HD and controls.   Table S9. All WGCNA brain expression modules significantly dysregulated (p < 0.05) in both Track-HD and Leiden datasets in HD versus control blood. HD brain modules were defined by Neueder and Bates 2 , and Control brain modules were derived from Braineac 3 or Gibbs, et al. 4 expression data. Neueder and Bates 2 module identifiers are given in brackets where available. * denotes the caudate modules that were highly positively and negatively correlated with HD in their study. HTT is part of modules 66 (CNneg1) and 3 (CBneg2). HD co-expression modules defined by Neueder and Bates 2 ; CTRL (B) -control brain co-expression modules from Braineac 3 ; CTRL (G) -control brain co-expression modules from Gibbs, et al. 4 . p (Combined/Track-HD/Leiden) -p value for differential expression between HD and controls in the combined, Track-HD or Leiden datasets; BH (HD) the Benjamini Hochberg significance value of correlation with HD in Neueder and Bates 2 brain expression analysis, corrected for multiple comparisons; Cor (HD) the direction and size of correlation of a module with HD in Neueder and Bates 2 ; CN -caudate nucleus; FC -frontal cortex;
Table S10. All nominally significantly dysregulated genes (p<0.05) from the WGCNA brain expression modules that were dysregulated (up or down) in HD blood. p (Comb/Track-HD/Leiden) -p value for differential expression between HD and controls in the combined, Track-HD or Leiden datasets; Log2FC -log2 of the ratio of the mean counts in HD and controls; HD co-expression modules defined by Neueder and Bates 2 ; CTRL (B) -control brain co-expression modules from Braineac 3 ; CTRL (G) -control brain coexpression modules from Gibbs, et al. 4 . Table S11. Module membership (kME) of genes in module 48 (CNpos2) that are dysregulated in both blood and caudate. There is a significant correlation between the dysregulation of a gene (p value) in the combined Track-HD and Leiden HD blood dataset and its kME, or degree of module membership (p = 7.6 x 10 -4 ). kME -correlation of a gene's expression profile with the module eigengene (representative of all gene expression profiles in a module). 0 implies no connection, 1 a strong positive and -1 a strong negative connection to the genes in a module. Highly connected intramodule hub genes have high kME; Log2FC (blood/caudate) -log2 of the ratio of the mean counts in HD and controls in our combined blood dataset or the caudate nucleus from Neueder and Bates 2 ; Directional p (blood/caudate) -directional p value for differential expression between HD and controls in our combined blood dataset or the caudate nucleus from Neueder and Bates 2 . Table S12. Generic pathways significantly upregulated in both HD blood and prefrontal cortex. Comparing gene expression changes in the combined Track-HD and Leiden HD blood dataset with HD prefrontal cortex from Labadorf, et al. 5 , a significant (p < 0.001) excess of generic pathways are significantly upregulated (p < 0.05) in both datasets. Blood/brain p the p value for pathway enrichment in HD relative to controls in the combined Track-HD and Leiden blood dataset (Combined) or the prefrontal cortex dataset (Labadorf).

Table S13. Gene co-expression modules significantly upregulated in both HD blood and prefrontal cortex.
Comparison of gene expression changes in the combined Track-HD and Leiden HD blood dataset with HD prefrontal cortex from Labadorf, et al. 5 . HD brain modules were defined by Neueder and Bates 2 , and Control brain modules were generated from Braineac 3

Table S14. Generic pathways significantly downregulated in both HD blood and prefrontal cortex.
Comparing gene expression changes in the combined Track-HD and Leiden HD blood dataset with HD prefrontal cortex from Labadorf, et al. (35), a significant (p = 0.028) excess of generic pathways are significantly downregulated (p < 0.05) in both datasets. Blood/brain p -the p value for pathway enrichment in HD relative to controls in the combined Track-HD and Leiden blood dataset (Combined) or the prefrontal cortex dataset (Labadorf). Table S15. Gene co-expression modules significantly downregulated in both HD blood and prefrontal cortex. Comparison of gene expression changes in the combined Track-HD and Leiden HD blood dataset with HD prefrontal cortex from Labadorf, et al. 5 . HD brain modules were defined by Neueder and Bates 2 , and Control brain modules were generated from Braineac 3 and Gibbs, et al. 4 . Neueder and Bates 2 module identifiers are given in brackets where available. CN -caudate nucleus; FC -frontal cortex; FC BA4 -BA4 region of the frontal cortex; FC BA9 -BA9 region of the frontal cortex; CB -cerebellum; TCTX -temporal cortex; Blood/brain p the p value for module enrichment in HD relative to controls in the combined Track-HD and Leiden blood dataset (combined) or the prefrontal cortex dataset (Labadorf). ; Cor (HD) the direction and size of correlation of a module with HD in Neueder and Bates 2 ; p (HD) -the p value for module enrichment in HD in Neueder and Bates 2 .

Table S16. Correlation between gene expression and TMS in gene positive Track-HD subjects. p (corr-TMS)
-p value for correlation between expression and TMS; q (corr-TMS) -q value shows correction for multiple testing of genes; Log2(FC) -the change in log2 (expression) per unit increase of TMS.
Table S17. Enrichment of up or downregulated pathways from HD vs. control blood (Table S2)  Table S18. Enrichment of negatively correlated pathways from HD vs. control blood (Table S3)  Table S19. Enrichment of modules from HD vs control blood (Table S9) Table S22. Co-expression modules from Zhang, et al. 7 late-onset Alzheimer's disease (LOAD) brain expression dataset. Several modules, including yellow that was the most significantly differentially connected in LOAD, show enrichment for upregulation in the HD blood expression dataset. Rank (Zhang)modules ranked for significance of differential connectivity with LOAD in Zhang, et al. 7 ; p (Combined/Track-HD/Leiden) -p value for enrichment of the module between HD and controls in our HD blood expression dataset; q (Comb) the false discovery rate estimate given by the q-value.