Structure prediction of ordered domains of Huntingtin

An abnormal polyglutamine (polyQ) expansion in the N-terminal fragment of the Huntingtin (Htt) protein leads to the devastating neurodegenerative disorder Huntington’s disease (HD). Htt is ubiquitous both at the tissue and subcellular levels and is fundamental for the survival and functioning of the brain neurons that predominantly degenerate in HD. In the hope of finding a cure for HD, there has been intense research aimed at understanding the molecular mechanisms underlying the deleterious effects of the presence of the abnormal polyQ expansion in Htt (Saudou and Humbert, 2016). It is a protein of 3144 amino acids and its high molecular weight prevents crystals production and X-ray diffraction studies to determine its three-dimensional structure. Further, Htt doesn’t display significant sequence similarity with other proteins of known structure, making difficult the study of its structure-function relationships (Zuccato et al., 2010). Therefore the main motivation of this study was that of obtaining structural information on Htt using ab initio/threading protein structure prediction techniques, information which could provide new insight into its functions and the mechanisms through which the mutation leads to the neurodegeneration. Ordered and disordered regions of Htt were predicted using Foldindex (Prilusky et al., 2005), GlobPlot (Linding et al., 2003), Predict Protein (Yachdav et al., 2014), Anchor (Dosztányi et al., 2009), InterPro (Mitchell et al., 2014), Smart (Letunic et al., 2014), HMMER (Finn et al., 2011), Espritz (Potenza et al., 2014) and MobiDB (Walsh et al., 2012). Structure prediction was performed through the use of I-TASSER (Roy et al., 2010), Rosetta (Kaufmann et al., 2010) and Robetta (Kim et al., 2004). The Dali server (Holm and Rosenström, 2010) has been used to infer structural similarities and hypothesize a possible function for each Htt domain. Using a consensus between different order/disorder prediction methods, Htt sequence has been divided in five regions predicted to form ordered domains (Hunt1-5). Then a structure prediction was performed for each ordered domain, obtaining the structural models depicted in Figure 1. Four out of five models are characterized by a prevalence of α-helical structure, in agreement with the predicted presence in Htt of Heat repeats. A structural homolog has been identified for each model of Htt ordered domains. Interestingly, Hunt1, 2 and 5 models display structural similarity with the subunit A of protein Phosphatase 2, the most important serin/threonine phosphatase involved in many essential aspects of cellular function (Xu et al., 2006). Hunt3 model displays structural similarity with the karyopherin Kap121p while Hunt4 model with poly U polymerase Cid1. To our knowledge, this work represents the first attempt to predict the structural and functional features of all Htt domains and the results obtained may represent a starting point for future experimental studies.