Huntington disease alters the actionable information in plasma extracellular vesicles

Dear Editor, In the present study, we demonstrate that examining extracellular vesicles (EVs) can give new insights into pathologic mechanisms in Huntington disease (HD), and might be used as biomarkers. We generated multi-omics datasets using enriched and purified EVs that were then evaluated with advanced bioinformatics and supervised machine learning to reveal HD-specific alterations in EV biology. We show that EVs from HD individuals convey specific actionable information in comparison to EVs from healthy people highlighting the biological relevance and potential use of EVs as a biomarker in clinical trials. The CAG repeat expansion mutation in the huntingtin gene (HTT) causing HD is expressed in all tissues. HD is primarily a neurodegenerative, mixed movement disorder, however, non-neuronal tissues such as liver and skeletal muscle also display various pathological changes.1 In line with this, we have recently shown that molecular signatures of inflammation, energy metabolism and vesicle biology are different in peripheral tissues of HTT mutation carriers (MTM-HD study).2 EVs are secreted by most, if not all cells. They are composed of lipids, proteins, RNAs and other small molecules and are generally categorised into three subtypes: exosomes, ectosomes and apoptotic bodies.3 EVs transfer their cargo by fusing with the recipient cell. Since all EVs are generated with material from the secreting cell, analysis of their composition potentially allows insights into the molecular state of these cells, including in the context of disease.3 Following the updated guidelines of the International Society for Extracellular Vesicles (MISEV2018),4 we established a robust, fast and highly standardised threefold purification procedure for EVs of all types from human plasma (Figure 1A). Our combined strategy markedly reduced free protein content, especially contaminants such as serum albumin, immunoglobulins or Golgi proteins in the final EV preparation (Figures 1B and S1). The final EV fraction was positive for EV marker proteins LAMP1/CD107a and ICAM1/CD54 by western blotting (Figure S1), and in proteomics data (e.g., ITGA*, ACT*, TUB*, HSP*) demonstrating successful purification of EVs. Transmission electron microscopy of purified, negatively stained EVs showed the expected cup shaped morphology (Figure S2). We then used plasma samples from the large MTM-HD study to purify EVs (control, n = 24; pre-HD, n = 22; early-HD, n = 20).2 Purification of EVs from 500 µL of plasma was sufficient to isolate enough EVs for all analyses from the same sample including for proteomics and transcriptomics. This avoids the influence of a batch effect due to different purifications. Nanoparticle tracking analysis (Figure 1C) showed that sizes of EVs were consistent with the expected mixture of exosomes and ectosomes. We neither detect larger EVs (apoptotic bodies) in any of the samples nor differences in the overall EV concentration, mean or mode diameter when comparing the three groups (Figure 1). However, we noticed a marked increase in small particles in the two HD groups (Figure 1C, inset), in particular for particles in the size range between 30 and 50 nm (Figure 1I,J). Above 60 nm, we did not detect any statistically significant group differences (Figure 1K). Next, we generated proteomics and RNAseq datasets from the purified EVs (Figure 2). Using STRING network analysis of the proteomics data, each network exhibited significantly more edges than what would be expected from a random set (Figure 2A–D). Gene ontology analysis using Enrichr5 showed a high expression of many dysregulated proteins in liver indicating that liver might be the main source of the observed changes in EV protein content and composition. Our strategy for the analysis of EV-associated RNAs was designed to identify all types of RNAs, coding and non-coding (Figures S3 and S4) including HTT itself (Figure S5). Dysregulation analysis for any of the pair-wise comparisons showed several dozen significantly dysregulated transcripts with different biotypes for both, sense and antisense mapping. This was reflected in the high confidence datasets mapping RNAs (Figure 2H–K). K-mer clustering showed three main clusters of expression patterns for the high confidence dataset for sense mapping RNAs with differential regulation between groups (Figure 3L). These RNAs included also the G3BP1 mRNA (greater than twofold upregulated in early-HD), which was previously implicated in EV biology in HD.6 Further analysis of the clusters pointed towards regulation of these genes by REST, a transcriptional repressor involved in expression of cell identity genes and epigenetics,7 which we also had identified by transcriptomic changes in adipose tissue from HTT mutation carriers.2 We hypothesised that the RNA content of EVs could be used as an HD stage biomarker. To test this hypothesis, we used the expression data from the filtered datasets to generate unsupervised, rank-based Kendall clustering of the sample relationships (Figures 3A and S6). EV RNAs RNA content was indeed able to accurately distinguish between the groups (Figure 3A). To examine whether a small subset of RNAs, as would be more practical in a clinical setting, would have similar predictive value of HD stage designation, we used a supervised machine learning approach and generated a classifier consisting of only 19 RNAs (Figure 3B) that was able to distinguish the groups with good accuracy (Figure 3C). In our final set of experiments, we generated proof-of-concept data that the purified plasma EVs retained their ability to be taken up by other cells (Figure 4A,B). We showed that the EVs carry actionable information resulting in alterations of the host cell's transcriptome and an EV specific, but genotype independent response. Additionally, we showed a different regulation of genes depending on the genetic origin of the EVs with EVs from early-HD clearly inducing a distinct response in comparison to EVs from healthy individuals (Figure 4C–E). While some limitations remain, for example, co-purified lipoproteins, which pose a challenge in EV-related research and are present to some extent in our EV preparations (independent of genotype), our approach has uncovered previously unidentified HD-related biological changes: the appearance of small particles in the HD samples, potentially corresponding to exomeres.8 A key question concerns the origin of EVs. We deliberately did not enrich subpopulations of EVs in order to retain the maximum information in this first comprehensive analysis. In our proteomics dataset, we detected L1CAM and N1CAM, both of which are potential markers for EVs of neuronal origin. However, expression levels were quite low as none of the label free quantification (LFQ) values for these two proteins passed quality control. We did not detect ATP1A3, which recently was proposed as another neuronal marker for EVs.9 In summary, we think that only a very small subset of the EVs we purify is of neuronal origin. In contrast, the ontology analysis points towards liver as the main source of the changes. Liver also exhibits a high degree of somatic CAG repeat instability, a key driver of the events that lead to HD manifestations. The liver therefore might be in a biologically more advanced HD state similar to neuronal populations.10 Taken together, our RNA data indicate that peripheral EVs are indeed an attractive source of easily accessible biomarkers since HTT sequences and/or potentially HTT-derived small RNAs are associated with EVs. Moreover, the classifier we developed can distinguish HTT mutation carriers at a molecular level with very high sensitivity in the prodromal stages of HD before the onset of diagnostic motor signs, that is, in the population most likely to participate in future clinical trials aiming at disease modification. Our data also show that EVs can in principle transfer information from one cell type to another. This information is different for EVs from HTT mutation carriers indicating that mutant HTT impacts cell-to-cell communication. Our current data are supportive of a peripheral HD phenotype, ranging from the tissue molecular signatures we have previously reported to plasma EVs. This emphasises the relevance of the expression of HTT in non-neuronal cells, and the effects the mutation causing HD can have on them. It also offers an opportunity to harness such signatures in the form of EVs as biomarkers in clinical trials. First and foremost, we thank all participants of the MTM-HD study for their time and their willingness to undergo tissue biopsies and donation of biofluid samples. The authors also thank the High Performance and Cloud Computing Group at the Zentrum für Datenverarbeitung of the University of Tübingen, the state of Baden-Württemberg through bwHPC and the German Research Foundation (DFG) through grant no. INST 37/935-1 FUGG, the Zentrum für Quantitative Biologie (QBiC) and the Core Unit Mass Spectrometry and Proteomics (Ulm University) for excellent technical expertise and help. This research study was partly supported by the Ministerium für Wissenschaft, Forschung und Kunst Baden-Württemberg (grant BioDATEN), BW Science Data Center (A.N.), Deutsche Huntington-Hilfe e.V. (A.N.), German Research Foundation—DFG (grant NE 2372/1-1) (A.N.) and European Huntington's Disease Network Seed Fund 871 (A.N.). CUMP is partly funded by SFB 1074 (German Research Foundation—DFG). A.N. acted as a consultant for Triplet Therapeutics, Inc. during the time of the study. In the past 2 years, through the offices of UCL Consultants Ltd., a wholly owned subsidiary of University College London. S.J.T. has undertaken consultancy services for Alnylam Pharmaceuticals Inc., Atalanta Pharmaceuticals, F. Hoffmann-La Roche Ltd., Genentech, Guidepoint, Horama, Locanobio, LoQus23 Therapeutics Ltd., Novartis Pharma, PTC Therapeutics, Sanofi, Spark Therapeutics, Takeda Pharmaceuticals Ltd., Triplet Therapeutics, University College Irvine and Vertex Pharmaceuticals Incorporated. G.B.L. has provided consulting services, advisory board functions, clinical trial services and/or lectures for Acadia Pharmaceuticals, Affiris, Allergan, Alnylam, Amarin, AOP Orphan Pharmaceuticals AG, Bayer Pharma AG, Boehringer-Ingelheim, CHDI-Foundation, Deutsche Huntington-Hilfe, Desitin, Genentech, Genzyme, GlaxoSmithKline, F. Hoffmann-LaRoche, Ipsen, ISIS Pharma (IONIS), Lilly, Lundbeck, Medesis, Medivation, Medtronic, NeuraMetrix, Neurosearch Inc., Novartis, Pfizer, Prana Biotechnology, Prilenia, PTC Therapeutics, Raptor, Remix Therapeutics, Rhône-Poulenc Rorer, Roche Pharma AG Deutschland, Sage Therapeutics, Sanofi-Aventis, Sangamo/Shire, Siena Biotech, Takeda, Temmler Pharma GmbH, Teva, Triplet Therapeutics, Trophos, UniQure and Wave Life Sciences. B.v.E. is a part-time employee of NatIgGs GmbH. All other authors declare they have no conflicts of interest. The local ethics committees at Ulm University and University CollegeLondon approved collection and analysis of the specimens (Ulm: 265-12; London:12/LO/1565), and written informed consent was obtained from each participant. All experimental methods comply with the Helsinki Declaration. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.