Omics and Network-Based Approaches in Understanding HD Pathogenesis

Abstract

Huntington’s Disease (HD) is a rare, progressive neurodegenerative disease caused by CAG repeat expansion in the Huntingtin gene. HD is an incurable disease; therefore, there is a growing need for effective therapeutic treatments and candidate biomarkers for prognosis and diagnosis of HD. Technological advancements over the past couple of years, have led to high-throughput experiments and omics data. The use of System Bioinformatics (SB) approaches, allows for the integration of information across different -omics, this can clarify synergistic relationships across biological molecules, resulting in complex biological networks. SB and network-based approaches, are able to shed light on the potential interactions of genes, proteins, metabolites and pathways participating in HD pathogenesis and how dysregulation of these biological entities, can affect age on onset, disease severity and progression. Moreover, −omics data analysis and network-based approaches can provide better understanding how these biological molecules interact with each other and provides potential drug targets and biomarkers that can be used to treat HD or delay symptom onset; therefore, opening the door towards precision medicine. The aim of the following chapter, is to discuss the most popular -omics related to HD research, and the growing popularity of single cell analysis, repositories and software available for bulk and single cell analysis. In addition, network-based approaches regarding HD will also be mentioned.

1. Introduction

Huntington’s Disease (HD) was first described in 1842 by Charles Oscar Waters and by 1872 a description of the disease was given by George Huntington, who assessed the medical history of several generations of a family exhibiting similar symptoms [1]. In 1983, a linkage on chromosome 4 was performed and by 1993 the Huntingtin (HTT) gene was discovered [1], leading to an increase in the interest of HD. Moreover, the identification of the gene resulted in the development of new animal models, research and therapeutic treatments and drugs to treat HD [1, 2]. HD is the most common monogenic neurological diseases in the developed world [3]. It is a rare, progressive neurodegenerative disease (ND) with autosomal dominant inheritance [1, 2, 3] affecting the medium spiny neurons (MSN) of the basal ganglia. HD is characterized by motor, cognitive and behavioral impairment caused by CAG trinucleotide repeat expansion at the N-terminus of the HTT gene responsible for encoding the HTT protein [4].

HD as other NDs, including Alzheimer’s disease (AD), Parkinson’s disease (PD) and Amyotrophic lateral Sclerosis (ALS) remain incurable [5]. Due to the overwhelming nature of these diseases, high economic and social costs and lack of effective therapeutic treatments [5]. Therefore, there is an increased need for novel biomarkers, pathways, drug targets and new novel pharmacotherapies that will enable the predication, prevention, understanding of disease pathways and most importantly effective treatments for NDs such as HD [5].

In the recent decade, bioinformatics approaches have become of increasing interest especially in NDs as they enable the identification of candidate biomarkers, pathways and mechanisms implicated in disease and potential drugs and their targets [5]. Bioinformatics is a multidisciplinary field utilizing methods from statistics, data-analysis, computer science, mathematics and biology to solve complex biological questions [5]. Bioinformatics is essential for analyzing and interpreting data from high-throughput technologies (DNA- and RNA-sequencing, proteomics, metabolomics, lipidomics etc.) [5, 6]. Due to the large amount of data produced, bioinformatics-based data management, and analysis tools are needed for the extraction, analysis, interpretation, visualization and storage of the data [5]. A relatively new approach is System Bioinformatics (SB), an intersection between systems biology and bioinformatics, which focuses on the integration of information across different -omics levels [6]. The goal of SB is to elucidate synergistic relationships between numerous factors in contrast to representing them as single biological entities, resulting in complex molecular networks of interactions [5]. This allows to better understand how different biological entities interact with one another, thus providing a clear understanding of disease pathogenesis, pathways, genes, proteins, metabolite, lipids and several additional types of biological information interacting together to lead to disease development [5]. Therefore, it is not a surprise that bioinformatics methods are used increasingly in ND research [5, 6].

Cells are complex, heterogenous and show a radical variation at the individual level and recent technological advancements, have enabled cell profiling at the individual level known as single cell analysis (SCA); the analysis of this data using bioinformatics pipelines is a practical solution for researchers; shedding light in understanding diseases, pathogenic mechanisms and the identification of potential biomarkers of diseases. The following chapter will discuss the most popular -omics approaches, SC technologies, data repositories and network-based analysis in HD research.

2. Omics

The past 15 years have seen the affluent use and integration of -omics approaches that include genomics and epigenomics for DNA, transcriptomics for RNA, proteomics for proteins and post-translational modification, metabolomics for metabolites, lipidomics for lipids and several additional -omics approaches (pharmacogenomics, radiomics etc.) [7, 8]. Technological advancement has allowed for the comprehensive measurement and analysis and the global assessment of biological molecules [7, 8] of genes, proteins, metabolites, lipids and several additional omics types within biological fluids, tissues or cells [8, 9], to assist in the understanding and investigation of human health and diseases [7]. These high-throughput technologies can be applied to the biological system and disease of interest to obtain a snapshot of the underlying biology at a resolution which was not previously possible [8], leading to unprecedented ways to better understanding NDs in terms of their pathogenesis, pathways, mechanisms and the interaction between different biological entities. Omics can contribute to biomarker, drug discovery and drug repurposing resulting in novel therapeutic treatments for NDs. Omics data is analyzed using bioinformatics pipelines and tools to obtain a list of genes, proteins, metabolites or lipids from the disease of interest [10], biological interpretation using different bioinformatics approachesTable 1 (pathway enrichment analysis, protein–protein interaction networks (PPIs), protein-metabolite networks etc.) is vital to gain insight into the mechanisms, pathways and molecules affected in disease. The most common omics and the research conducted in regards to HD will be discussed below (Figure 1).

Name	Type	Omics content	Ref	Website (If Available)
Galaxy	Open access	Genomics; Transcriptomics; Proteomics; Metabolomics; Metagenomics	[11]	https://usegalaxy.org/
Limma	R-package	Transcriptomics; Proteomics; Metabolomics	[12]	https://www.bioconductor.org/packages/release/BiocViews.html#___Metabolomics
Ingenuity pathway analysis (IPA)	Commercial Software	Transcriptomics; Proteomics; Metabolomics	—	—
NextFlow	Software	Genomics; Transcriptomics; Proteomics; Metabolomics	—	https://www.nextflow.io/
EdgeR	R package	Genomics; Transcriptomics	[13, 14, 15]	—
MaxQuant & Perseus	Open access	Proteomics	[16]	https://www.maxquant.org/
OpenMS	C++ library	Proteomics; Metabolomics	[17]	https://openms.de/
MsCoreUtils	R package	Proteomics; Metabolomics	[18]	https://www.bioconductor.org/packages/release/bioc/html/MsCoreUtils.html
MetaboAnaylst	Open access & R package	Metabolomics; Lipidomics; Multi-omics	[19, 20, 21]	https://www.metaboanalyst.ca/
MetaCyc	Open access	Metabolomics	[22]	https://metacyc.org/
XCMS	Open access & R package	Metabolomics	[23, 24, 25]	https://xcmsonline.scripps.edu/landing_page.php?pgcontent=mainPage
Mzmine	R package	Metabolomics	[26]	http://mzmine.github.io/

Table 1.

Online tools, software and R-packages for omics data analysis.

Figure 1.

Omics data types and bioinformatics based approaches.

2.1 Genomics

Since the discovery of the DNA structure by Rosalind Franklin, Frances Crick and James Watson in 1953, there has been a revolution in the field of biological sciences regarding technological advancements such as Sanger sequencing [27, 28, 29] and the discovery of the polymerase chain reaction [30] at the beginning of the 21st century. NGS became available, enabling the production of huge amounts of data along with the ability to produce highly efficient, rapid, low-cost approach along with accurate DNA sequencing [31]. Genomics is defined as the interdisciplinary field of biology focusing on genes, their structure, function, evolution, expression, mapping and genome editing [32]. The genome refers to all the genetic material present in any organism, including chromosomal and extrachromosomal DNA, coding and non-coding genes, microRNAs (miRNAs) and single nucleotide polymorphisms (SNPs) [33]. Sequencing of the human genome is important as it can provide insight into: i) gene expression profile of a specific tissue, organ or tumor, ii) human variation and how genomic alterations lead to disease development, iii) identification of genetic modifiers of disease, and iv) determine rate sensitivity to drugs in patients, based on their DNA sequences, this is known as pharmacogenomics [34]. Genomics has the ability to discover and identify genes associated with a disease; making it a vital omics component towards precision medicine. The majority of studies that have taken place regarding HD research have involved either the identification of genetic modifiers that may play a role in the early or delayed onset of HD.

2.1.1 Genomics in HD patients

Marchina et al. [35] investigated the gene expression profiles of fibroblasts of HD patients and healthy controls using microarray technology, and RT-PCR to validate the consistency of the microarray data. Analysis was carried out on nine genes, namely APC, CDC42EP2, CTNNB1, DNM1L, PLCB4, ROCK1, ROCK2, SSH1, and UBE2D3. These lists of genes, were chosen as some of these genes showed increased differential up-regulation (PLCB4, APC) or down-regulation (SSH1, CDC42EP2); while other genes were selected due to their possible involvement in HD pathology (UBE2D3, ROCK1, ROCK2, DNM1L, CTNNB1) as indicated by previous studies. The APC, PLCB4, ROCK1, and UBE2D3 genes were found to be up-regulated in HD patients in comparison to controls [35]. Some gene ontology (GO) terms identified are, i) transcription, ii) regulation of transcription, DNA-dependent, iii) cell cycle, iv) response to DNA damage stimulus, v) DNA repair, vi) ubiquitin dependent protein catabolic process and vii) DNA recombination. Based on this evidence the gene expression profiles of fibroblasts seem to be altered in HD patients compared to healthy controls [35]. Validation of the differential expressions at the protein level is needed to confirm if fibroblasts can be considered as a suitable model for the identification of HD biomarkers [35].

A study by Wright et al. [36] investigated the gene expression profiles complementing the analysis of genomic modifiers in HD. CAG repeat length is not the only contributing factor to disease onset but genetic modifiers also contribute to disease onset [36]. A genome-wide association study assessing heritable differences in genetically determined expression in diverse tissues, with genome-wide data taken from 4000 HD patients was conducted. In addition, functional validation of prioritized genes was performed in isogenic HD stem cells and post-mortem brain tissue [36]. The genes of FAN1, GPR161, PMS2 and SUMF2 are associated with age of onset in HD and showed co-localization with gene expression signals in brain tissue [36].

Other genomic studies conducted in HD patients or post-mortem brain tissue include Moss et al. [27], Li et al. [28].

2.1.2 Genomics in HD animal models

One study by Tang et al. [29] performed gene expression profiling on the R6/2 HD transgenic mice model, with varying CAG repeat lengths to reveal genes associated with HD onset and progression [29]. The R6/2 transgenic mice with >300 CAG repeats had prolonged HD progression and a longer lifespan than the parent R6/2 mice with 150 CAG repeats. However, the mechanisms regarding this phenotypic amelioration remains unknown [29].

The expression profiles of the striatum of R6/2 transgenic mice with >300 CAG repeats (R6/2Q300 transgenic mice) and R6/2 transgenic mice with 150 CAG repeats (R6/2Q150 transgenic mice) and littermate wild-type (WT) controls were performed to identify, genes playing an important role, in HD onset and progression. Both R6/2 mouse models demonstrated decreased expression while up-regulated gene expression was seen in R6/2Q300 mice [29]. The up-regulated genes play a role in ubiquitin ligase complex, cell adhesion, protein folding and establishment of protein localization. Increased gene expression for Lrsam1, Erp29, Nasp, Tap1, Rab9b and Pfdn5 was validated using qPCR. Moreover, Lrsam1 and Erp29 were significantly up-regulated in R6/2Q300 mice and may have potential neuroprotective effects in primary striatal cultures over-expressing mHTT [29]. Over-expression of Lrsam1 prevented the loss of NeuN-positive cell bodies in htt171-82Q cultures, associated with a decrease of nuclear HTT aggregates. However, over-expression of Erp29 demonstrated no significant effect in cells [29]. Therefore, prolonged HD onset and progression seen in R6/2 mice with increased CAG repeat expansions are a result of differential up-regulation in genes involved in protein localization and clearance. These genes may be possible novel therapeutic avenues in decreasing HTT aggregation toxicity and neuronal cell death, with Lrsam1 being a promising, novel candidate disease modifier [29].

Langfelder et al. [37] integrated genomics and proteomics data to define HTT CAG length networks in HD mice. To gain insight into how mHTT CAG repeat length modifies HD pathogenesis, profiling of mRNA in 600 brain and peripheral tissue samples obtained from HD knock-in mice with increased CAG repeat length was performed [37]. The study found, the repeat length dependent transcriptional signatures were notable in the striatum while the cortex and liver showed less transcriptional signatures [37]. Co-expression networks revealed 13 and 5 striatal and cortical modules, respectively, that were highly correlated with CAG repeat length and age, and are preserved in HD models. The top striatal modules suggest mutant HTT (mHTT) CAG length and age impair the MSN, while a dysregulation of genes and pathways involved in cAMP signaling, cell death, and protocadherin were seen [37]. Proteomics analysis was used to confirm, 790 genes and 5 striatal modules with CAG length-dependent dysregulation was observed at both the RNA and protein level, 22 striatal module genes were validated as modifiers of mHTT toxicities in vivo [37].

Other genomics studies in HD animal models or cells include Choudhury et al. [38] and Alcalá-Vida et al. [39].

2.2 Transcriptomics

The transcriptome is the complete set of transcripts within a cell, and their quantity, for a physiological condition or specific developmental stage [40]. Understanding the transcriptome is crucial for interpreting the functional components of the genome and revealing the molecular entities of cells and tissues and also for understanding development and disease pathogenesis [40]. The aim of transcriptomics is to: i) catalog the transcripts, including mRNAs, non-coding RNAs and small RNAs; of all species to determine the transcriptional structure of genes, regarding their start sites, 5′ and 3′ ends, alternative splicing patterns and any additional post-transcriptional modifications; and ii) quantify the fluctuating expression levels of each transcript under different conditions and during development [40]. The development of innovative high-throughput DNA sequencing methods known as RNA Sequencing (RNA-Seq) has allowed the mapping and quantification of transcriptomes. Genomics and transcriptomics are the two most popular omics used for HD research; therefore, it is vital to examine whether peripheral tissues may serve as a possible source of readily accessible biological signatures at not only the DNA and RNA level but also at the protein level and not only. There are numerous studies that have used transcriptomics to perform studies regarding either peripheral tissues, cerebrospinal fluid (CSF) or post-mortem brain tissue samples obtained from HD patients and animal models.

2.2.1 Transcriptomics in HD patients

A study by Neueder et al. [41] investigated the abnormal molecular signatures that characterize inflammation, energy metabolism and vesicle biology in human HD peripheral tissues. Specifically, skeletal muscle, adipose tissue and skin from the quadriceps femoris muscle and blood was collected from 21 pre-HD patients, 20 early motor manifest HD patients and 20 healthy controls. Furthermore, primary fibroblast and myoblast cell lines were established [41]. RNA-Seq and proteomics analysis were used to investigate the involvement of inflammation and energy metabolism in HD pathogenesis.

Initially, RNA-Seq analysis was performed on adipose and muscle tissue from the pre-HD patients (pre-HD), early motor manifest HD patients (early HD) and healthy control groups. In total, 78 and 53 genes were identified to be significantly dysregulated in adipose and skeletal muscle tissue, respectively [41]. Distinctly, only RPH3A, PAX6 and AC016582.1 were regulated similarly in the pre-HD and early HD groups in comparison to controls. Furthermore, some genes were differentially regulated in both groups [41]. TBC1D3D, was differentially regulated in muscle and adipose tissue, the gene was found to be up-regulated in the early HD group in comparison to the pre-HD group [41].

GO enrichment analysis in adipose tissue identified the terms of protein sumoylation (GO: 0016), interleukin-1 mediated signaling pathway (GO:0006470) and cellular response to organic cyclic compound (GO: 0071407) [41]. For the different comparisons, the RE1 silencing TF (REST) target gene was dysregulated only in the pre-HD group, while sumoylation was dysregulated only in the early HD group. Alteration in protein sumoylation is a well-known mechanism and has been implicated in HD pathogenesis [41].

GO analysis in muscle tissue, identified a disruption in homeostatic pathways, including anterograde trans-synaptic signaling (GO: 0098916), regulation of fatty acid oxidation (GO:0046320), regulation of glucose metabolic process (GO:0010906), negative regulation of cell communication (GO:00106458), protein dephosphorylation (GO:0006470), regulation of protein kinase B signaling (GO:0051896) and regulation of MAPK cascade (GO:0043408) particularly in the pre-HD group [41]. PAX6 is a vital regulator of assorted peripheral and central nervous system processes, it was identified to be highly and gradually up-regulated in both HD groups; therefore, suggesting a compensatory mechanism of muscle regeneration in response to mutant HTT expression [41]. Enrichment analysis of the dysregulated genes in the muscle tissue of the pre-HD group suggest that peroxisome proliferator activated receptor alpha (PPARA) acts as a regulatory protein [41].

In addition, RNA-Seq analysis was also performed on the fibroblast cell lines. Only a few genes were significantly dysregulated between pre-HD and early-HD patient groups compared to healthy controls [41]. Subsequently, GO enrichment analysis resulted in the identification of a few enriched terms, such as phosphorylation (GO:0016310) and regulation of apoptotic processes (GO:0042981) [41]. Interestingly, TBC1D3D, was found to be dysregulated not only in adipose and muscle but also in fibroblasts [41]. Furthermore, proteomics analysis was performed on the tissues obtained; however, this will be further discussed in the proteomics section of the chapter.

Miller et al. [42], performed RNA-Seq analysis on myeloid cells of HD patients to identify transcriptional dysregulation associated with proinflammatory pathway activation. Specifically, whole transcriptome analysis of primary monocytes from 30 manifesting HD patients and 31 healthy control subjects, with or without proinflammatory stimulus [42].

HD monocytes exhibit resting proinflammatory transcriptional changes, 12,599 genes were identified to be differentially expressed (DE) and analysis of said data was conducted, to determine the genes significantly altered between HD and control monocytes, whereas the stimulated and unstimulated monocytes were analyzed separately [43]. Analysis of unstimulated monocytes revealed 130 DE genes, of which 101 were up-regulated and 29 down-regulated genes in resting HD monocytes compared to healthy controls. The DE genes were associated with proinflammatory cytokines and chemokines. HD monocytes had significantly increased expression of IL6, IL12B, IL19, IL23A, CCL8, CCL19, CCL20, CXCL6 and CSF2 gene transcripts. All transcripts have a > 2-fold increase in mRNA expression [42]. In contrast, the stimulated monocytes showed little indication of differential expression between HD and controls, the genes of DNAJB13, STAC and RASEF were DE. Furthermore, each of these genes were observed to be DE in unstimulated monocytes [42]. Generally, gene expression differences between HD and controls were lower in the stimulated 116 DE genes in comparison to the unstimulated 130 DE genes in monocytes [43]. Moreover, gene set enrichment analysis (GSEA) was performed for the up- and down regulated gene expression for both the stimulated and unstimulated monocyte datasets [43]. Analysis of unstimulated monocytes revealed a total of 85 enriched gene sets, a majority of pathways relating to innate immunity, inflammatory response, cytokine production and secretion were identified. Furthermore, pathways such as NFkB and JAK/STAT signaling pathways were also identified to be significantly enriched [43]. In comparison to the down-regulated genes, 6 gene sets were identified and included significantly enriched pathways related to vacuole, lysosome and catabolic functions [42]. Interestingly, the simulated dataset did not reveal any enriched gene sets among the up-regulated genes. However, 83 gene sets were significantly enriched among the down-regulated genes, pathways relating to cholesterol homeostasis, cellular components such as cellular membrane, mitochondria and lysosomes [42].

To understand the mechanisms underlying differential gene expression in unstimulated HD monocytes, upstream regulator analysis was performed using the Ingenuity Pathway Analysis (IPA) software. A total of 155 upstream regulators were identified for the unstimulated dataset, of these 125 were predicted to be significantly activated, while 30 were predicted to be significantly inhibited. A large number of molecules associated with intracellular signaling pathways downstream of the TLR4 receptor were represented in the unstimulated group. The following data were consistent with previous studies, showing NFkB dysregulation in HD myeloid cells, the RELA and the NFkB complex were among the top most significant results ranked by z-score activation [43]. Other prominent potential regulators include NFkB1, ERK and p38 MAPKs, in addition to the transcription factor STAT3 [42]. The study suggests that transcriptional changes observed in the RNA-Seq analysis of stimulated and unstimulated HD monocytes is related to the abnormal activation of specific upstream signaling molecules responsible for driving gene expression in unstimulated HD myeloid cells [42]. Furthermore, HD myeloid cells have a proinflammatory phenotype in the absence of stimulation; consistent with the priming effect of mutant huntingtin (mHTT), whereas basal dysfunction results in an exaggerated inflammatory response once a stimulus is encountered [42]. The following data provides an understanding of mHTT pathogenesis, establishing unstimulated myeloid cells as a vital source of HD immune dysfunction, and demonstrating the importance of immunity as a potential HD treatment.

Other transcriptomic studies conducted in HD patients include Seefelder and Kochanek [44], Sinha et al. [45], Mastrokolias et al. [46, 47], Borovecki et al. [48], Lin et al. [40], Cha [49], Moily et al. [43], Mehta et al. [50], Stopa et al. [51] and Runne et al. [52].

2.2.2 Transcriptomics in HD mouse models

Jin et al. [53] investigating the miRNA and mRNA expression profiles of the cerebral cortex of N171-82Q HD mice [53]. There is growing evidence indicating that miRNAs may play a role in HD pathogenesis [53]. During disease progression significant changes of miRNAs in the cerebral cortex were also detected in the striatum of HD mice [53]. The study revealed that a significant alteration in the miR-200 miRNA family, more specifically, miR-200a and miR-200c in the cerebral cortex and striatum were seen during the early HD stages in N171-82Q mice.

A computational bioinformatics approach was used to integrate miRNA and mRNA profiles. The gene targets were identified using TargetScan, for miR-200a and miR-200C, there are 680 and 462 gene targets respectively. Some of the predicted gene targets for the miRNAs include KIF3a, NRXN1, PTPRD, TRIM2, ATXN1, KCNA2 and numerous additional targets [53]. The predicted targets play a role in regulating synaptic function, neuronal survival and neurodevelopment. The results of the study suggest that altered miR-200a and miR-200c expression may interrupt protein production involved in neuronal plasticity and survival [53]. Therefore, further investigation of the involvement of dysfunctional miRNA expression in HD is required and this may result in novel approaches for HD therapy.

Hervás-Corpión [54] investigated the early alternations of epigenetic-related transcription in HD mouse models. The gene expression profiles of the cortex, striatum, hippocampus and cerebellum of juvenile R6/1 and N171-82Q mice, were studied, the profiles consisted of tissular and neuronal-specific genes and showed significant correspondence with transcriptional alteration in HD mouse models deficient of epigenetic regulatory genes [54]. A noteworthy case was the conditional knockout of the lysine acetyltransferase CBP in post-mitotic forebrain neurons, the double knockout of the histone methyltransferases Ezh1 and Ezh2, components of the polycomb repressor complex 2 (PRC2), and the conditional mutants of the histone methyltransferases G9a (Ehmt2) and GLP (Ehmt1) [54]. It is likely that the neuronal epigenetic status is compromised prior to HD onset resulting in transcriptional dysregulation [54].

Other transcriptomic studies conducted in HD animal models include Bensale et al. [55], Dickson et al. [56], Reyes-Ortiz [57], Chaves et al. [58] and Huang et al. [59].

2.3 Proteomics

Proteomics is a new omics type that has rapidly developed especially in the fields of therapeutics and biomarkers discovery [60]. Proteomics is defined as the study of interactions, functions, composition and structure of proteins and their cellular activities. The proteome is defined as the complete set of proteins within a cell, tissue, organism or biological fluid. Proteomics allows for a better understanding of the structure and function of an organism than genomics. However, proteomics is much more cumbersome than genomics as the protein expression is altered according to environmental conditions and time. It is approximated that there are almost one million human proteins, many of which contain some form of post-translational modifications [60]. The majority of proteomics studies have been conducted in HD mouse models, there seems to be a lack of human HD proteomics studies. Proteomics can assist in the discovery of new therapies, biomarkers and better understanding of proteins and protein-complexes in disease.

2.3.1 Proteomics in HD patients

The study by Neueder et al. [49] which previously performed transcriptomic analysis as mentioned in Section 2.2.1 also performed proteomics analysis and revealed, 1347, 2671 and 4640 proteins for adipose, muscle and skin tissue respectively [49]. A progressive increase in the number of dysregulated proteins from the pre-HD to early HD stage was observed for all three tissues [49]. In adipose tissue, there was no overlap of commonly dysregulated proteins between the three comparisons. In muscle tissue, syntaxin-binding protein 3 (STXBP3) was up-regulated in both pre- and early-HD samples. Interestingly, the major change was observed in the skin samples, MOB4 was up-regulated in both pre- and early HD samples [49]. A total of 162 and 23 proteins were significantly dysregulated in the early and pre-HD stages respectively.

GO enrichment on adipose dataset (early HD vs. controls) and skin dataset (early HD vs. controls and the comparison of pre-HD vs. early HD), revealed dysregulated lipid metabolism and proteasome function in the adipose tissue dataset. While proteins related to gene expression such RNA processing and translation and amino acid metabolism are mainly affecting in the early HD samples vs. controls [49]. The predicted regulators are all genes involved in gene expression, specifically, lysine acetyltransferase 2A (KAT2A) and the androgen receptor (AR), in a complex with ataxin 7 (ATXN7) which are associated with polyglutamine diseases [49]. The GO terms for pre- and early HD in the skin datasets, included the regulation of cell survival and proliferation. The results obtained from the above study, indicate that alterations in biological signatures at the RNA and protein level point towards the direction of inflammation, energy metabolism and vesicle biology in peripheral tissues in HD. The following biological signatures may act as suitable biomarkers in clinical trials upon further validation.

A proteomics study by Fang et al. [61] integrated five sets of proteomics data profiling the CSF derived from HD affected and unaffected individuals with genomics data profiling, various human and mouse tissue, including the human HD brain [61]. According to the integrated analysis, brain specific proteins were 1.8 times more likely to be observed in CSF compared to plasma. Furthermore, brain specific proteins decrease in HD CSF compared to unaffected CSF [61].

Approximately, 81% of brain specific proteins have quantitative changes that agree with transcriptional changes seen in different HD brain regions, while the proteins identified to increase in HD CSF tend to be liver associated. The protein changes are consistent with microgliosis, astrocytosis and neurodegeneration which are known to occur in HD [61]. The most significantly over and under abundant dysregulated proteins in CSF between HD affected and unaffected individuals, include, chromogranin B (CHGB), isoform I of Sialate O-actelyesterase precursor (SIAE), isoform long of iduronate 2-sulfatase precursor (IDS), Neurexin 3-alpha (NRXN3) Endonuclease domain-containing 1 protein precursor (ENDOD1), major prion protein precursor (PRNP) and several additional proteins have a trend of decreasing with disease progression (controls > HD early > mid-HD). Some of the over abundant proteins identified are, complement component 1, q subunit, c chain precursor (C1QC), hemopexin (HPX), Triosephosphate isomerase (TPI1), Isoform M1 of Pyruvate kinase isozymes M1/M2; Isoform R-type of Pyruvate kinase isozymes R/L (PKM2/PKLR), Lysozyme C precursor (LYZ) and several other proteins, which have shown to have a trend of increasing as the disease progresses (control > HD early > mid HD) [61].

Other proteomics studies conducted in HD patients include, Chen et al. [62], Schönberger et al. [63], Sorolla et al. [64] and Chae et al. [65], McQuade et al. [66] and Dalrymple et al. [67].

2.3.2 Proteomics in mouse models of HD patients

A previous study by Agrawal and Fox [68] performed targeted proteomics analysis to investigate novel proteomic alterations in brain mitochondria to reveal insights into mitochondrial dysfunction in HD mouse models. Mitochondrial dysfunction is one of contributing pathophysiological mechanisms in HD. The following study used R6/2 and YAC128 HD mouse models to represent different HD progression rates to pinpoint HD brain mitochondrial proteomic signatures. Cerebral cortical mitochondrial of HD and WT littermates were compared by 2D SDS–PAGE electrophoresis and MALDI-TOF/TOF mass spectrometry (MS) [68].

Proteomic analyses concluded 17 and 12 DE proteins in 12-week R6/2- and 15-month YAC128 HD mice, respectively compared to controls. The proteins of peroxiredoxin 3, stress–70, DJ–1, isocitrate dehydrogenase [NAD] α subunit and ATP synthase subunit D were DE in both HD mouse models. Pathway analysis was performed using PANTHER [68], the GO molecular function terms obtained for the DE mitochondrial proteins are i) catalytic activity (GO:0003824), binding (GO:0005488) and antioxidant activity (GO:0016209); most GO biological process proteins belonged to metabolic (GO:0008152) and cellular process (GO:0009987) [68]. While for YAC128 mice, the molecular functions of the DE mitochondrial proteins were catalytic activity (GO:0003824) and binding (GO: 0005488); for GO biological processes most of the proteins also belonged to the metabolic (GO:0008152) and cellular process (GO:0009987) and biogenesis (GO:0071840). The results identify a proteomic signature of HD mitochondria in mouse models that includes previously unrecognized proteins [68].

One study by Choudhary et al. [38] performed differential proteomics and genomic profiling of mouse striatal cell model of HD vs. healthy controls. Transcriptional dysregulation is one of the pathogenic mechanisms contributing to HD. Various studies have identified altered gene expression in HD patient brains and animal models. 2D SDS-PAGE/MALDI-MS coupled with 2D-DIGE and real time PCR on an array of genes concentrated on HD pathways to identified altered proteins and gene expression in STHdhQ111/HdhQ111 compared to STHdhQ7/HdhQ7 (wild-type). Seventy-six proteins were annotated in HD cells while 31 proteins were DE by 2D-DIGE. The bioinformatics tool GeneCodis3 [38] was used to perform pathway analysis using KEGG and GO biological terms. The pathways included, unfolded protein binding (GO:0051082), negative regulation of neuron apoptosis (GO:0006916), response to superoxide’s (GO:0006950) and several other pathways. The PCR experiments showed altered gene expression of 47 genes. Altogether, 77 genes/proteins were identified in HD cell lines with potential relevance to HD biology [38].

Other proteomic studies conducted in HD animal models and cells include Mees et al. [69], Liu et al. [70], Perluigi et al. [71], Deschepper et al. [72], Culver et al. [73], Cozzolino et al. [74], Vodicka et al. [75], Sap et al. [76], Ratovitski et al. [77] and Zabel et al. [78].

2.4 Metabolomics

Traditionally, a small number of metabolites have been used for the diagnosis of complex metabolic diseases and monogenic complex diseases such as inborn errors of metabolism [79]. Metabolomics is defined as the comprehensive measurement of metabolites within a biological tissue, cell, organ or fluid (CSF, urine, plasma or serum) [79]. The metabolome is the number of metabolites in an organism. Metabolomics is an emerging technology that holds promise for the development and practice of precision medicine. In addition, metabolomics is able to provide detailed characterization of metabolic phenotypes and can enable precision medicine at numerous levels such as characterization of metabolites underlying disease, discovery of new therapeutic targets, biomarker discovery which may be used for prognosis, diagnosis or monitor activity of therapeutics [79]. Similarly with proteomics, the majority of studies in metabolomics and HD are conducted in HD mouse models.

2.4.1 Metabolomics in HD patients

One study by Rosas et al. [80] investigated the plasma metabolome of HD. The study consisted of targeted metabolomics analysis using the plasma obtained from 52 pre-HD, 102 early symptomatic HD and 140 controls [80].

The pathways altered include i) tryptophan, ii) tyrosine, iii) purine, iv) methionine, v) antioxidant pathways and numerous pathways relating to energetic and oxidative stress derived from the gut microbiome. The tryptophan, tyrosine and purine pathways were altered in prodromal and early HD stages. The selective dysregulation of a good few pathways and the increased regulation of other pathways suggest complex alterations in the feedback controls of underlying genes, proteins, enzymes and metabolites. In addition, multivariate statistical modeling demonstrated mutually distinct metabolomics profiles, suggesting that the process determining onset was likely distinct from processes determining progression [80]. Surprisingly, controls, pre-HD and early HD plasma metabolomes were mutually distinct rather than differing, suggesting varying influences during the prodromal and symptomatic disease stages [80].

Numerous metabolites differentiating the control from the pre-HD and early HD metabolomes, are linked to the gut microbiome, suggesting mHTT favors a distinct microbiome. The systemic effects of HD on the gut microbiome may possibly influence energy homeostasis, vitamin and mineral supply, metabolites and neuroimmune functions while impacting HD expression [80]. The gut microbiome derived metabolites were differentiated in the pre-HD metabolome, while the symptomatic HD metabolome was mostly influenced by metabolites likely reflecting mHTT toxicity and neurodegeneration [80]. The study suggests that the pre-HD metabolome is influenced more by the gut microbiome than the HD metabolome, possibly due to the increasing effects of mHTT toxicity and neurodegeneration. The understanding of these complex alterations is a delicate balance between the metabolome and gut microbiome in HD, and they relate to disease onset, severity, progression and phenotypic variability in HD are vital questions for future research and clinical studies in HD [80].

Herman et al. [81] investigated the metabolic alterations in tyrosine and phenylalanine pathways in the CSF of HD patients. The study consisted of 13 pre-HD mHTT carriers and 13 symptomatic HD patients and 42 controls. The comparison of symptomatic HD patient’s vs. controls, identified 24 metabolites to be significantly dysregulated, this included the metabolites of Lumichrome, Xanthine, O-succinyl-homoserine, N-acetylproline, Phenylacetate, Isoleucine, L-DOPA, Leucine, Corticosterone, Ophthalmate, Tyrosine, Valine, Salicylate, Phenylalanine and others. Pathway analysis disclosed 5 biochemical pathways affected in symptomatic HD vs. controls namely i) aminoacyl-tRNA biosynthesis; ii) phenylalanine metabolism; iii) valine, leucine and isoleucine biosynthesis; iv) valine, leucine, isoleucine degradation and v) purine metabolism. Phenylalanine metabolism was highly affected in symptomatic HD patients.

Comparing symptomatic HD patient’s vs. pre-HD patients, 28 metabolites were revealed to be significantly altered. Some of the metabolites were, L-DOPA, Xanthine, Ophthalmate, Creatinine, Tyrosine, 5-hydroxytryptophan, Adenosine, Phenylalanine, Phenylacetate, Thyroxine, Glutarylcarnitine, O-succinyl-homoserine, Adenine, Isoleucine, Aldosterone/Cortisone and numerous other metabolites. Fourteen of these were able to distinguish symptomatic HD patient’s from controls. Univariate analysis illustrated 11 metabolites were dysregulated (L-DOPA, xanthine, ophthalmate, creatinine, tyrosine, 5-hydroxytryptophan, adenosine and phenylalanine) remained significant after correcting for multiple comparison testing. Furthermore, 4-acetamidobutanoate and S-adenosylhomocysteine had been corrected for age dependence. There was a notable longitudinal decrease in O-acetylcarnitine in the symptomatic HD patients. Significantly altered abundances of Ophthalmate, Phenylalanine, 4-quinolinecarboxylic and N,N,N-trimethyl lysine were observed in six pre-HD patients. Pathway analysis, illustrated 8 biological pathways of: i) aminoacyl-tRNA biosynthesis; ii) phenylalanine, tyrosine and tryptophan biosynthesis; iii) valine, leucine and isoleucine biosynthesis; iv) tyrosine metabolism; v) nitrogen metabolism; vi) valine, leucine and isoleucine degradation; vii) phenylalanine metabolism and viii) purine metabolism. Tyrosine and phenylalanine metabolism pathways were significantly altered between symptomatic and pre-HD patients.

To investigate the effects of disease progression, the association between CSF concentration of altered metabolites and measure of disease severity were researched in all mHTT carriers and to a five-year risk onset of developing HD in pre-HD patients. Hierarchical clustering revealed that tyrosine metabolism, including tyrosine, thyroxine, L-DOPA and dopamine, was significantly dysregulated in symptomatic vs. pre-HD patients. These metabolites displayed moderate to strong associations of measures to disease severity and symptoms. Furthermore, Thyroxine and Dopamine were also correlated with the five-year risk of onset in pre-HD patients. Phenylalanine and purine metabolism were also significantly altered, but associated with decreased disease severity. Decreased levels of Lumichrome were frequent in mHTT carriers and concentrations correlated with five-year risk of HD onset in pre-HD carriers. Biochemical profiling illustrates that the CSF metabolome may be used to characterize molecular pathogenesis in HD, and may be vital for future development of HD therapies.

Other metabolomics studies conducted in HD patients include, McGarry et al. [82], Mastrokolias et al. [47], Cheng et al. [83], Graham [84, 85] and Patassini et al. [86].

2.4.2 Metabolomics in HD animal models

Targeted metabolic profiling on plasma of a pre-HD transgenic sheep model in order to identify potential candidate biomarkers was investigated by Skene et al. [87].

One hundred and thirty metabolites were obtained, Alanine, Arginine, Citrulline, Glutamine, Glutamate, Glycine, Histidine, Isoleucine, Phenylalanine, Proline, Tryptophan, Valine, Creatine, Kynurenine, Serotonin and several additional metabolites [87]. Citrulline and Arginine showed significantly increased levels in HD compared to control sheep, both are involved in the urea cycle, although Ornithine was also identified and is part of the urea cycle, no significant difference between HD vs. healthy controls was seen. Citrulline demonstrated the most significant change in transgenic HD sheep. Regarding the 20 amino acids measured, 10 had shown significantly decreased concentration in HD sheep, the branched amino acids (BCAA) of Valine, Leucine and Isoleucine showed the most notable effect, which have been previously identified as potential biomarkers. This was followed by Threonine, Tyrosine, Methionine, Alanine, Asparagine, Phenylalanine and Glutamine. Significant alterations in respect to genotype were distinguished in 89/130 identified metabolites, including Sphingolipids, Biogenic amines, amino acids and Urea. A significant increase in Urea, Arginine, Citrulline, asymmetric and symmetric dimethylarginine and a decrease in Sphingolipids [87].

Quantitative enrichment analysis using MetaboAnalyst [19, 20, 21], identified the top five metabolite pathway-associated metabolite sets of: i) aspartate metabolism; ii) arginine and proline metabolism; iii) valine, leucine and isoleucine degradation; iv) fatty acid metabolism and v) urea cycle. The urea cycle and nitric oxide pathways become dysregulated during the early HD stages. Additionally, logistic prediction modeling identified 8 potential biomarkers (Citrulline, Valine, PC aa C40:4, PC aa C36:5, lysoPC a C17:0, SM (OH) C24:1, Threonine, Tetradecenoylcarnitine (C14:1)). In HD sheep, the metabolites of Citrulline, PC aa C36:5 and lysoPC a C17:0 were increased significantly while Valine, PC aa C40:4, SM (OH) C24:1, Threonine and Tetradecenoylcarnitine (C14:1) were significantly decreased compared to control sheep [87]. The degree of sensitivity, using minimally invasive methods, puts forward a novel approach for monitoring disease progression in HD patients.

Andersen et al. [88] investigated the branched amino acids (BCAA) and their metabolism in the cerebral cortex of a R6/2 HD mouse model using metabolomics analysis [88]. Deficiencies in cerebral energy and neurotransmitter are suggested to play a role in neuronal dysfunction in HD. The BCAAs of Valine, Leucine and Isoleucine are vital in cerebral nitrogen homeostasis, neurotransmitter recycling and are utilized as energy substrates in the tricarboxylic acid (TCA) cycle [88]. Decreased levels of BCAAs in HD haven been previously validated in several studies. However, it remains unclear how cerebral BCAA metabolism is regulated in HD.

Isolated cerebral cortical and striatal slices of controls and R6/2 mice were incubated with labeled Leucine and Isoleucine; however, no differences in Leucine or Isoleucine concentration were shown between R6/2 and control striatal or cerebral cortical slices [88]. Suggesting that the cellular uptake of these BCAAs likely remains unaffected in the R6/2 slices compared to controls. Metabolism of Leucine and Isoleucine, entering oxidative metabolism as acetyl CoA, was preserved in R6/2 mice. However, metabolism of Isoleucine, entering the TCA cycle as Succinyl CoA, was increased in the cerebral cortical and striatal slices of R6/2 mice; therefore, suggesting a rise in metabolic influx via the replenishing of Oxaloacetate in the citric acid cycle. Enzyme expression in the BCAA metabolism was assessed, enzymes related to BCAA metabolism displayed an increased expression in the R6/3 brain, particularly enzymes related to Isoleucine metabolism [88]. This indicates that the capacity for cerebral BCAA metabolism, primarily Isoleucine, is heightened in R6/2 brain tissue indicative of alterations in cerebral BCAA homeostasis.

Other metabolomics studies conducted in HD animal models include Chang et al. [89], Chaves et al. [58], Bertrand et al. [90], Verwaest et al. [91], Hashimoto et al. [92], Kumar et al. [93] and Tsang et al. [94].

3. Single cell omics

Organisms and complex tissues are formed by a heterogeneity of cells undergoing cell division, proliferation, differentiation during various physiological states such as development and adulthood [95]. The fate of each cell is intrinsically determined and is influenced both by external factors and by cell–cell interactions [95]. Various processes taking place within individual cells are a result of complex interactions between chromatin, transcripts, proteins, metabolites, lipids and other biological entities [95, 96]. To identify these activity dependent regulatory processes the majority of approaches used include transcriptomics, proteomics, metabolomics or lipidomics methods on tissues, cells and biological fluids [95]. Although the bulk approaches are useful, they have a drawback, as they average the information derived from thousands to millions of cells, leading to masking of cell specific features or features involved in developmental processes [95]. There are some studies of single cell RNA-Seq analysis in HD [97, 98, 99].

Over the past few years, the development of methodological approaches for single cell analysis has greatly increased and addressed the drawback of bulk analysis [95]. Single cell analysis (SCA) has recently been included into multi-omics strategies, which brings together concurrent biological information from different molecular modalities and their relationship with individual cells [95]. Furthermore, analyzing cells individually at a higher resolution results in a more accurate representation of cell–cell variations in comparison to bulk analysis measurements [95]. SCA approaches include: i) single cell genomics, ii) single cell epigenomics iii) single cell transcriptomics, iv) single cell proteomics and v) single cell metabolomics. To fully grasp and understand cellular complexity and specificity of cells, tissues or biological fluid microenvironments in physiological or disease conditions, it is important to measure molecular signatures at the single cell resolution [100]. Benefits of SCA include: i) improvements in experimental design and data analysis for single cells for a disease of interest, ii) targeting of specific cell populations therefore elucidating signaling pathways and networks, iii) cell-to cell communication variation for understanding disease onset, progression and therapeutic response, iv) differentiate normal cells and comprised cells at various developmental stages, v) identify cells and their distinctive susceptibilities, vi) clarify neural communication in unprecedented detail, resulting in new strategies to understand and treat NDs [100]. SCA can provide insight for potential biomarkers, therapeutic targets, pathways and mechanisms in disease involvement [100]. Tables 2 and 3 indicate the databases and tools used for analysis of SCA, while Table 4 illustrates software and databases for cell-to-cell communication.

Name	Type	Omics content	Ref	Website (If Available)
Gene expression omnibus	Repository	Single cell RNA-Seq	[101]	https://www.ncbi.nlm.nih.gov/geo/
Single cell epression atlas	Repository	Single cell RNA-Seq	[102]	https://www.ebi.ac.uk/gxa/home
Slavov laboratory/Quantitative biology	Repository/Database	Single cell Proteomics	[103]	https://slavovlab.net/data_webs.htm
Proteome exchange	Repository/Database	Single cell Proteomics	[104, 105]	https://www.proteomexchange.org/
Metabolomics workbench	Repository/Database	Single cell metabolomics	[106]	https://www.metabolomicsworkbench.org/data/index.php

Table 2.

Single cell repositories and databases.

Name	Type	Omics content	Ref	Website (If Available)
Seurat	R package	Single cell RNA-Seq	[107, 108, 109]	https://satijalab.org/seurat/
Galaxy	Open access	Single cell RNA-Seq	[11]	https://usegalaxy.org/
scpdata	R package	Single cell Proteomics	—	https://github.com/UCLouvain-CBIO/scpdata

Table 3.

Online tools, software and R-packages for single cell omics analysis.

Name	Type	Omics content	Ref	Website (If Available)
CellChatDB	Open access & R package	Single cell RNA-Seq	[110]	http://www.cellchat.org/
NICHES	R package	Single cell RNA-Seq	[111]	https://github.com/msraredon/NICHES
CITEdb	Open access & R package	Single cell RNA-Seq	[112]	https://citedb.cn/#/index
CellPhoneDB	Open access	Single cell RNA-Seq	[113]	https://www.cellphonedb.org/

Table 4.

Software and databases for cell-to-cell communication.

4. Bioinformatics

In recent years, technological advancements have made great progress in understanding the genetics, transcripts, proteins and metabolites seen in disease, resulting in an explosion of big data, opening endless scientific possibilities [114]. However, this vast amount of data generated has its own challenges related to: i) data storage, ii) software to process such large amounts of data and ii) analysis and biological interpretation of data. In this circumstance, bioinformatics and computational biology have sought to overcome these challenges in big data generation and analysis [114]. Bioinformatics is a multi-interdisciplinary approach combining mathematics, physics, biology and computer science, it is defined as the application of computational methods and tools for the organization, analysis, understanding, visualization and storage of information associated with biological entities [114]. The development of high-throughput technologies such as NGS, RNA-Seq and liquid chromatography-mass spectrometry (LC–MS) and the analysis of data using bioinformatic approaches has opened a host of new possibilities including but not limited to, gene expression studies, methylation patterns, epigenetic markers, proteins, metabolites, lipids and others [114]. In the recent decade, bioinformatics approaches have become of increasing interest especially in NDs for novel biomarkers and drug discovery and pathways [6]. Specifically, a multi-interdisciplinary approach such as SB can enhance the contribution of computational therapeutics and diagnostics for NDs, hence providing a stepping stone towards precision medicine.

In the following section, we look into the tools and databases available for.

-omics in regards to obtaining data and analysis of said data. Moreover, we will discuss network-based approaches used in HD research.

4.1 Omics databases and tools

The databases for obtaining publicly available data of different omics data at the single and bulk -omics level is constantly changing with new data becoming available every day. In addition, the analysis of such data is increasingly important, and various tools, software and R-packages are also developed to assist researchers in the analysis of their data. Table 5 includes repositories and databases where omics data can be found, not only specific for HD but also for other NDs, as well various diseases. Table 1 illustrates the tools, software and packages mainly in R, used for analysis of -omics data. The tables contain some of the most popular databases, repositories, software, tools and R-packages used, however there are numerous others resources available which have not been listed.

Name	Type	Omics content	Ref	Website (If Available)
Gene expression omnibus	Repository	Genomics, Epigenomics, Transcriptomics,	[101]	https://www.ncbi.nlm.nih.gov/geo/
Expression atlas	Repository	Genomics	[102, 115]	https://www.ebi.ac.uk/gxa/home
Array express	Repository/Database	Genomics	[116]	https://www.ebi.ac.uk/biostudies/arrayexpress
European variation archive	Repository/Database	Genomics	[117]	https://www.ebi.ac.uk/eva/?Home
Ensembl	Database	Genomics	[118]	http://www.ensembl.org/index.html
Database of genomic variants archive	Database	Genomics	—	https://www.ebi.ac.uk/dgva/
European genome-phenome archive	Repository/Database	Genomics	—	https://ega-archive.org/
Answer ALS	Repository	Genomics Transcriptomics, Proteomics, Metabolomics	[119]	https://dataportal.answerals.org/home
Huntington’s disease in high definition (HDinHD)	Database	Genomics	—	https://www.hdinhd.org/
Enroll HD	Repository/Database	Genomics Transcriptomics, Proteomics	[120]	https://www.enroll-hd.org/
Parkinson’s progression markers initiative	Repository/Database	Genomics Transcriptomics, Proteomics, Metabolomics	[121]	https://www.ppmi-info.org/
Alzheimer’s disease neuroimaging initiative	Repository/Database	Genetics; Radiomics	[122]	https://adni.loni.usc.edu/
PRIDE	Database	Proteomics	[123]	https://www.ebi.ac.uk/pride/
Proteome exchange	Repository/Database	Proteomics	[104, 105]	https://www.proteomexchange.org/
Japan proteome standard	Repository/Database	Proteomics	[124]	https://jpostdb.org/
Mass spectrometry interactive virtual environment	Repository/Database	Proteomics	—	https://massive.ucsd.edu/ProteoSAFe/static/massive.jsp
ProteomicsDB	Repository/Database	Proteomics	[125]	https://www.proteomicsdb.org/
Human peptide atlas	Repository/Database	Proteomics	[126]	https://peptideatlas.org/
Metabolomics workbench	Repository/Database	Metabolomics	[106]	https://www.metabolomicsworkbench.org/data/index.php
MetaboLights	Repository/Database	Metabolomics	[127]	https://www.ebi.ac.uk/metabolights/index
Metabolome exchange	Repository/Database	Metabolomics	—	http://www.metabolomexchange.org/site/
Human metabolome database	Database	Metabolomics	[128]	https://hmdb.ca/

Table 5.

Omics repositories and databases.

4.2 Network-based approaches

In all living organisms, most cellular components exert their functions via interactions with other components, the entirety of these interactions represents the human interactome [129]. The cells and their response to changes in their environment involve coordinated activity of mRNAs, proteins, metabolites and lipids [130]. The fundamentals of proper cellular function are molecular networks connecting these components to process extra-cellular environmental signals driving dynamic cellular responses [130]. Network-based approaches aim to systematically integrate measurements obtained from high-throughput experiments to gain insight of the cellular functions undergoing changes resulting in disease [130]. Systematic integration of varying biological entities is essential to identify molecular networks controlling normal and disease states, and in time, predict complex phenotypes from molecular markers [130]. Network-based approaches in human disease can lead to multiple biological and clinical applications, over the past decade, there has been an exceptional increase in human-specific molecular interaction data, resulting in a greater understanding of how different biological entities interact with one another, in biological networks and how they play a role in human diseases [129]. Examples of molecular networks are i) PPI networks, ii) metabolic networks, iii) regulatory networks, iv) RNA networks, v) cell–cell interaction networks and vi) multi-omics networks [129]. Few studies have been performed in regards to network and bioinformatic based approaches on HD. Some of these studies will be discussed below.

Chandrasekaran and Bonchev [131], performed network analysis on post-mortem tissue of the cerebellum, frontal cortex and caudate nucleus of HD patients. The microarray dataset, GSE3790, has 44 and 36 HD patients and controls respectively [131]. The dataset consists of both Affymetrix GeneChip Human Genome HG-U133A and B [131]. The seed genes are referred to as the significantly differentially expressed genes (SDEGs) [131]. In the GSE3790 HG-U133A, 617 overlapping seed genes were found between the four sets of SDEGs, while in GSE3790 HG-U133B, 351 seed genes were found [131]. Altogether, 925 seed genes were identified; however, network evaluation was performed only for the SDEGs with a p-value <0.01, with this new cut-off threshold, 531 seed genes were identified and underwent network analysis [131].

Pathway enrichment was performed using Pathway Studio 9.0 software along with the molecular interaction database ResNet 9.0 [131], to construct direct interaction, shortest-path and miRNA regulation networks. Gene prioritizing approaches based on network topological measures, high node connectivity, centrality and guilt by association were applied, based on this approach 19 novel genes were found CEBPA, CDK1, CX3CL1, EGR1, E2F1, ERBB2, LRP1, HSP90AA1 and ZNF148; these genes may be of particular interest to undergo experimental validation [131]. The seed genes underwent GO enrichment analysis using DAVID [131], while the IPA software and Pathway Enrichment Analysis in Pathway studio was used to explore the canonical pathways affected in HD [131]. The pathways identified with DAVID are i) neuron development (GO:0048666), ii) neuron differentiation (GO:0030182), iii) regulation of glucose import (GO:0046324), iv) neuron projection (GO:0043005), v) regulation of lipid catabolic process (GO: 0050994) and various additional GO terms. The KEGG pathways from DAVID are i) HD (hsa: 05016), ii) MAPK signaling (hsa:04010), iii) ErbB signaling (hsa:04012), iv) Alzheimer’s Disease (hsa: 05010), v) Amyotrophic Lateral Sclerosis (hsa: 05014) and numerous other pathways [131]. The miRNA regulatory network analysis, found miR-124, mir-135a, miR-141, miR-182 and miR-19a to be the top five scoring miRNA within the network [131]. The SDEGs, miRNA and pathways obtained, in combination with experimental validation can shed light onto possible genes, miRNAs and mechanisms affected in HD, which can lead to targeted therapeutic strategies.

Other network and bioinformatics-based approaches conducted in HD include Sneha et al. [132], Pirhaji et al. [133], Pradhan et al. [134], Xiang et al. [135], Zaho et al. [136], Fu et al. [137], Shirasaki et al. [138], Christodoulou [139, 140, 141], and Onisiforou [142].

5. Discussion and conclusion

HD remains one of the most debilitating and incurable ND, as currently there is a lack of effective treatments, although clinical trials have been conducted using different strategies such as gene silencing to decrease mHTT protein production [1, 2, 3, 5]. However, these attempts were unsuccessful as in some cases, no difference between the drug and placebo groups or no symptom improvement was observed. Therefore, there is a need for effective drugs to be identified and tested and used as pharmacotherapies for HD. Furthermore, a better understanding of how different biological molecules (genes, proteins, metabolites, lipids etc.) interact with not only each other but also with disease pathways, and possibly drugs is vital to understand how these interactions are affected in HD as it will allow to shed light on why some therapeutic treatments are not effective [5]. Dysregulation, in any specific cell type, biological molecule or pathway can influence the entire biological system leading to disease progression [7].

In the past decade technological advancements has led to high-throughput experiments resulting in -omics data and their analysis. In addition, there is a growing interest in SC-omics, allowing analysis of individual cells at a higher resolution, allowing for the accurate representation of cell–cell variation in tissues [95]. Bioinformatics, specifically network-based approaches, can be applied to investigate and understand in depth the relationship between the different biological molecules and their interaction with the HTT protein, each other and within pathways [5]. The advancement of bioinformatics has led to the progress and identification of novel candidate biomarkers and drugs, affected pathway and mechanisms and genes, proteins and metabolites affected in a disease state compared to controls [5, 6]. In addition, there has been an exponential increase in the number of repositories and database available for both bulk and SC analysis for genomics, transcriptomics, proteomics and metabolomics [95, 100]. In addition, this has resulted in numerous bioinformatics tools, software and R-packages, which have been developed to assist researchers in omics data analysis, in order to have meaningful biological interpretation of the data to make proper conclusions that will possibly lead to the discovery of potential biomarkers and drugs resulting in better prognosis, diagnosis and drug sensitivity of patients [100, 114]. The contribution of SB can provide a stepping stone towards precision medicine and potentially address the absence of HD treatments.

Conflict of interest

The authors declare no conflict of interest.

Additional information

The Cyprus Institute of Neurology and Genetics is a full member of the european reference network for rare neurological diseases (ERN-RND).