Tau interactome and RNA binding proteins in neurodegenerative diseases

Abstract

Pathological tau aggregation is a primary neuropathological feature of many neurodegenerative diseases. Intriguingly, despite the common presence of tau aggregates in these diseases the affected brain regions, clinical symptoms, and morphology, conformation, and isoform ratio present in tau aggregates varies widely. The tau-mediated disease mechanisms that drive neurodegenerative disease are still unknown. Tau interactome studies are critically important for understanding tauopathy. They reveal the interacting partners that define disease pathways, and the tau interactions present in neuropathological aggregates provide potential insight into the cellular environment and protein interactions present during pathological tau aggregation. Here we provide a combined analysis of 12 tau interactome studies of human brain tissue, human cell culture models and rodent models of disease. Together, these studies identified 2084 proteins that interact with tau in human tissue and 1152 proteins that interact with tau in rodent models of disease. Our combined analysis of the tau interactome revealed consistent enrichment of interactions between tau and proteins involved in RNA binding, ribosome, and proteasome function. Comparison of human and rodent tau interactome studies revealed substantial differences between the two species. We also performed a second analysis to identify the tau interacting proteins that are enriched in neurons containing granulovacuolar degeneration or neurofibrillary tangle pathology. These results revealed a timed dysregulation of tau interactions as pathology develops. RNA binding proteins, particularly HNRNPs, emerged as early disease-associated tau interactors and therefore may have an important role in driving tau pathology.

Keywords: Alzheimer’s disease; Frontotemporal dementia; HNRNP; Interactome; Neurofibrillary tangle; Protein–protein interactions; RNA binding proteins; Tau; Tauopathy.

Fig. 1

Summary of PTMs present on tau in AD and primary tauopathies. A PTMs observed in AD mapped to full length tau and B in primary tauopathies; AGD, CBD, GGT, PiD, PSP. Alternatively spliced regions are coded light orange = 1 N, dark orange = 2 N and light green = 3R, dark green 4R and purple for the proline rich region. Compiled from [1, 10, 29, 30, 36, 50, 58, 78, 92, 94, 113, 115]. Please note that while this compilation of tau PTMs was completed to the best of our knowledge, this is an evolving field and may not be comprehensive

Fig. 2

Tau interactome analyses reveal consistent tau-protein interactions across human brain and cell culture models. A PPI network of human tau interacting proteins consistent in > 3 of 7 studies. Network node colours are clustered by k-means set with 8 groups. String network is limited to physical interactions only. Clusters of interacting proteins are manually annotated (see Supplementary Table 2 for annotated proteins). B PPI network of rodent model tau interacting proteins. Network node colours are clustered by k-means set with 8 groups. String network is limited to physical interactions only. Clusters of interacting proteins are manually annotated (see Supplementary Table 3 for annotated proteins). C Zoomed in view of PPI network of RNA-binding proteins in human tissue D GO Molecular function enrichments of human and rodent models generated in R with the package clusterProfiler v4.2.2. Top 8 processes are listed. The background gene set for rodent and human studies included all proteins detected in IP studies (see Supplementary Table 4 for GO molecular function annotations). All PPI networks were generated with the STRING database (v11.5, https://string-db.org/)

Fig. 3

Analysis of proteins that interact with pTau and are altered in NFT or GVD containing neurons. A Venn diagram showing the overlap between pTau interacting proteins of Drummond et al. [30] and proteins with a trended change in expression in NFT or GVD containing neurons [52]. This trend was defined as a fold change in expression > 0.1 either up or down as p-values were not supplied with this dataset only FDR and most other proteomics studies fail to report FDR. B Enrichment plot summarising GO cellular compartment terms enriched for each dataset generated in R with clsuterProfiler v4.2.2. Nodes are divided into pie charts that show the proportion of genes shared for that category. Nodes are connected by shared genes to cluster terms into broader functions. Only the top 50 terms are shown. Clusters are annotated based on the majority of component pathways (Supplementary Table 6). C PPI network of proteins that are pTau interactors and altered in GVD and D NFT containing neurons. PPI networks were generated with the STRING database (v11.5, https://string-db.org/). String nodes are coloured by selected STRING annotation for those proteins (RNA binding and proteolysis). STRING network included physical and regulatory interactions