Computational Studies of Tau Protein : Implications for the Pathogenesis and Treatment of Neurodegenerative Diseases

Tau protein is the primary constituent of protein aggregates known as neurofibrillary tangles, a pathological hallmark of Alzheimer's disease (AD). Previous studies suggest that tau protein may play a contributing role in neurodegenerative diseases such as AD. Thus characterizing the structural properties of tau is critical to understanding disease pathogenesis. However, obtaining a detailed structural description of tau protein has been difficult because it belongs to a class of heteropolymers known as intrinsically disordered proteins (IDPs). Unlike most proteins, IDPs adopt many distinct conformations under physiological conditions. In spite of their disordered nature, evidence exists that such proteins may exhibit residual structural preferences. In this work, models of tau are constructed to characterize these structural preferences. We begin by performing molecular dynamics simulations to study the inherent conformational preferences of the minimal tau subsequence required for in vitro aggregation. To model residual structure in larger regions of tau, we developed a novel method called Energy-minima Mapping and Weighting (EMW). The method samples energetically favorable conformations within an IDP and uses these structures to construct ensembles that are consistent with experimental data. This method is tested on a region of another IDP, p21 Wafl/Cipl/Sdil(14 5 -164 ), for which crystal structures of substrate-bound conformations are available. Residual conformational preferences identified using EMW were found to be comparable to crystal structures from substrate-bound conformations of p21 Wafl/Cipl/Sdil( 14 5 -164 ). By applying EMW to tau, we find disease-associated forms of tau exhibit a conformational preference for extended conformations near the aggregation-initiating region. Since an increased preference for extended states may facilitate the propagation of cross-13 conformation associated with aggregated forms of tau, these results help to explain how local conformational preferences in disease-associated states can promote the formation of tau aggregates. Finally, we examine limitations of the current methods for characterizing IDPs such as tau and discuss future directions in the modeling of these proteins. A b stra ct .............................. ... .......... .. .. ................................... ....... ...... .......... ........................ 3 Chapter 1: Introduction................................ .. .............. ............... .......................... 7 Chapter 2: Finding Order within Disorder Elucidating the Structure of Proteins Associated with Neurodegenerative Disease ...................... ........... ............. ............................................... ......... 10 A bstract .................................. ..... ................ ......... ........... ..... ............ ........... .............. 10 Intro d uctio n ............. ........... .. . ....... ......................... ................................................ ................ 11 Characterizing the Structure of an Intrinsically Disordered Protein ..................................................... 14 Current Experimental Approaches to Studying Intrinsically Disordered Proteins ...... ............................ 17 Methods for Constructing Models of the Unfolded Ensemble ........................................ ..... 21 Modeling IDPs Associated with Neurodegenerative Disorders ...................... ...... 25 A m y lo id ............................................................ .... ..... ......................... .... ................................. 2 5 a -sy n u c le in ........................................ .... ... .... .......... .... ..................... . ..................................... 2 8 Tau Protein ............... .......... .......... . .. ... ... ....... ....... ... ...... ............................ 30 ID Ps as Targets fo r D rug D esign ....................................................................................................... 32 Chapter 3: Conformational Sampling with Implicit Solvent Models: Application to the PHF6 Peptide in Tau Protein 34 A b stra ct .......................................... ...... . .. ....... .. . .......... ... ...... ........................... 3 4 In tro d u ctio n ............................ ....... .............................. .... ........................................ 3 5 M e th o d s ........................................ .. .... .. .. ..... .. ........................... . ....................................... 3 9 Quenched Molecular Dynamics with Explicit Solvent ................................. .... ........................ 39 Quenched Molecular Dynamics in vacuum ........ .................. ...... ................................ 40 Quenched Molecular Dynamics Simulations with Implicit Solvent................................. .... 40 Generation of Ram achandran Plots ........................................ .......... .... .............................. 43 Generation of Minimum Pairwise Distance (MPD) Plots .................... .......... 44 Potential of Mean Force Calculations for PHF6 ....................................... 45 Calculating Vibrational Entropies ...................................... . ... .......... .......... .......................... 46 R e s u lts .................................................... .. ................................ ... ....................................... 4 7 Minimum energy conformations with explicit solvent ................................. 47 Minimum energy conformations with implicit solvent .................... ............. ... 49 Potential of M ean Force Calculations .................................... ...... ... ........................ 52 Ranking Minima from the Implicit Solvent Models............................... ......................... 54 D iscu ssio n .......................................... .. ................... .......................................... 57 Chapter 4: The Effect of a AK280 Mutation on the Unfolded State of a Microtubule-Binding Repeat in Tau 62 Abstract ................................................................................................................................................. 62 Introduction ............................................................................................................................................. 63 Results ................................................................................................................................................... 65 Discussion................................................................................................................................................77 M ethods .................................................................................................... ........................................ 84 Energy-m inim a M apping and W eighting ..................................................................... ................. 84 Identifying Locally Preserved Conform ations .............................................. ....... ........ ......... 90 Acknow ledgem ents ........................................................................................... ................................ 91 Chapter 5: M odels of K18................................................................ ................................................ 92 Introduction.............................................................................................................................................92 The Segm ent M odel ........................................................................................... ............................... 92 Results ..................................................................................................................................................... 96 The Segm ent M odel ........................................................................................ .............................. 96 Energy M inim a M apping and W eighting M odels of K18 .............................................. 99 Discussion ......................................................... ...................................... 101 M ethods ................................................................................................................................................ 103 Sampling Conformations of K18 with the Segment Model ..................................... 103 Generation and Analysis of EMW Ensembles for K18..................................... 105 Chapter 6: Future W ork ........................................................... ................................................ 107 Appendix: Residual structure within the disordered C-terminal segment of p21Wafl/Cipl/Sdil and its im plications for m olecular recognition ........................................ 109 Abstract .............. .................................................................................................................. 109 Introduction...........................................................................................................................................110 Results ................................................................................................................................................. 112 Residual secondary structure in p21(145-164) detected by NMR spectroscopy ........................... 112 M odeling the unfolded state of p21(145-164) w ith M D sim ulations ............................................... 113 Helical mode of p21(145-164) binding to Ca2+-calmodulin from NMR dipolar couplings ............... 116 Discussion ............................................................................................................................................ 117 M aterials and M ethods ......................................................................................................... 119 Cloning, Protein Expression and Purification ..................................... ........ ......... 119 NM R Spectroscopy .......................... ........ ........... .... ..... ...... .............................. ............ 120 M olecular dynam ics sim ulation ......................................... .......... ...... ....... ........................ 121 Acknow ledgem ents ........ .................... ... .............. . ........................... .......... ... ............................... 131 References ........................ ......... ...... ........... .... .... .................................... 133 Chapter 1: Introduction Alzheimer's disease (AD) is a neurodegenerative disorder characterized by progressive memory loss, cognitive dysfunction, and behavioral disturbances [3]. The disease has a high prevalence, afflicting approximately 18 million people worldwide and is the most common cause of senile dementia [4]. The two pathological hallmarks of Alzheimer's disease are extracellular protein aggregates of amyloid-P (AP), known as amyloid plaques, and intracellular protein aggregates of tau protein, known as neurofibrillary tangles [5]. Much data suggests that the proteins which constitute these aggregates, A3 and tau, also play a role in disease pathogenesis [6-10]. A structural description of these proteins is required to understand the conformational transitions accompanying aggregation into potentially toxic forms and to assist in the design of therapeutics targeting these proteins [ 11]. Despite that the majority of AD research has focused on AP, much evidence suggests that tau dysfunction contributes to disease progression in AD [6, 7, 12, 13]. Tau protein also plays an important role in a related family of neurodegenerative diseases, known as tauopathies, which are neurodegenerative disorders characterized by pathological aggregation of tau [14]. Tau protein belongs to a class of heteropolymers known as intrinsically disordered proteins (IDPs) [11]. These proteins are sometimes referred to as natively unfolded proteins (NUPs) or intrinsically unstructured proteins (IUPs). In contrast to most proteins, which fold into a unique, three-dimensional structure or at least contain large regions of structure, IDPs fluctuate between many distinct conformations under physiological conditions. Presently there are no existing experimental methods to fully characterize the set of structures populated by these proteins. In this work, we combine biophysical modeling and conformational sampling approaches with published experimental measurements to characterize the structural properties of tau protein. Thus, one can obtain detailed structural insights that are not available from experiments alone. The thesis is organized as follows: Chapter 2 provides an overview of recent experimental and modeling approaches for characterizing structural properties of IDPs involved in neurodegenerative diseases. Chapter 3 discusses molecular dynamics simulations performed on a peptide corresponding to a key tau subsequence which is required for aggregation in vitro. There were two motivations for this study. First, conformational preferences of this subsequence were of inherent interest due to its importance in tau aggregation. Second, these simulations were used to evaluate implicit solvent models for the purpose of sampling conformational minima. Using these methods, we find that the aggregation-initiating sequence has an intrinsic propensity for extended conformations. Furthermore, we identified an implicit solvent potential for efficient sampling of conformational minima. In order to identify residual conformational preferences in larger regions of tau, we developed a novel semi-empirical method for constructing conformational ensembles of intrinsically disordered proteins. This protocol is discussed and used in the studies described in chapter 4 and the appendix. We initially tested the method on a region of the intrinsically disordered protein p 2 1 Wafl/C ipl /Sd i l (appendix). Unlike tau, p2 1 Waf l Cipl/di l is an intrinsically disordered protein for which crystal and NMR structures of substrate-bound subsequences exist [15, 16]. Thus, we compared structural properties described by our method against known bound conformations. This analysis showed that local conformational preferences in the unfolded state of p2 1Waf Cipl/ Sdi identified by our method were comparable to structured substrate-bound conformations. This work was performed with Veena Venkatachalam, an undergraduate student, and in collaboration with James Chou [17]. The method was then applied to both wild-type and disease-associated mutant forms of tau protein (Chapter 4). The resulting conformational ensembles describe how changes in local conformational preferences between normal and disease-associated forms of tau can contribute to differences in their propensity to aggregate. Recently, additional structural data for the microtubule-binding repeat domain of tau (referred to as K18) have been published [18, 19]. Chapter 5 discusses work to incorporate these data into models of K18. In addition, we attempted to test an alternate modeling approach which does not require fitting to experimental data. This model is based on the hypothesis that conformational preferences of sequentially long-range positions can be approximated as being independent. In Chapter 6, we discuss limitations of current experimental and modeling approaches to characterizing structure in tau (and IDPs in general) and future directions for