Using Mass Spectrometry-Based Methods to Understand Amyloid Formation and Inhibition of Alpha-Synuclein and Amyloid Beta

Abstract

Amyloid fibrils, insoluble β-sheets structures that arise from protein misfolding, are associated with several neurodegenerative disorders. Many small molecules have been investigated to prevent amyloid fibrils from forming; however, there are currently no therapeutics to combat these diseases. Mass spectrometry (MS) is proving to be effective for studying the high order structure (HOS) of aggregating proteins and for determining structural changes accompanying protein-inhibitor interactions. When combined with native MS (nMS), gas-phase ion mobility, protein footprinting, and chemical cross-linking, MS can afford regional and sometimes amino acid spatial resolution of the aggregating protein. The spatial resolution is greater than typical low-resolution spectroscopic, calorimetric, and the traditional ThT fluorescence methods used in amyloid research today. High-resolution approaches can struggle when investigating protein aggregation as the proteins exist as complex oligomeric mixtures of many sizes and several conformations or polymorphs. Thus, MS is positioned to complement both high- and low-resolution approaches to studying amyloid fibril formation and protein-inhibitor interactions.

This review covers basics in MS paired with Ion Mobility (IM), Continuous Hydrogen Deuterium Exchange (continuous HDX), Pulsed Hydrogen Deuterium Exchange (pulsed HDX), Fast Photochemical Oxidation of Proteins (FPOP) and other irreversible labeling methods, and Chemical Cross-Linking (XL). We then review the applications of these approaches to studying amyloid-prone proteins with a focus on amyloid beta (Aβ) and alpha-synuclein (aS). Another focus is the determination of protein-inhibitor interactions. The expectation is that MS will bring new insights to amyloid formation and thereby play an important role to prevent their formation.

I. INTRODUCTION

A. Amyloids

The formation of amyloid fibrils is key to several neurodegenerative disorders, including Alzheimer’s (AD) and Parkinson’s disease (PD).^1,2,3 Protein misfolding and aggregation are the first steps in the formation of amyloids, leading to insoluble, β-sheet-rich strands called amyloid fibrils.^2,3,4 These fibrils threaten the original role of the protein, whose loss of function can lead ultimately to AD and PD.

A protein can misfold due to changes in pH, temperature, interaction with lipid membranes, crowding, and post-translational modifications.² The amyloid state is reached when fibers elongate and form many β-sheet strands. The fibrils are typically long, unbranched, and sometimes found in extracellular space.^1,5 There are approximately 25 amyloid-forming proteins that have been identified and connected to serious diseases.^1,3,5

In this review, we will describe the basic fundamentals of nMS, IM-MS, continuous HDX-MS, pulsed HDX-MS, FPOP and other irreversible labeling methods, and XL-MS, and then follow this with an analysis of recent work applied to aggregating proteins (Aβ and aS) studied by each MS-based approach. Our aim is to represent different MS-based approaches including their data processing available to elucidate HOS of proteins and oligomers. We chose Aβ and aS as the two protein systems for this review because they are causative elements in two important and devastating diseases, AD and PD. Further, they are prototypal amyloids and have spawned considerable MS research to understand their aggregation. Thus, approaches to understand them will be applicable to other amyloids. Furthermore, summarizing recent Aβ and aS aggregation studies may aid in designing future work with these proteins to understand more deeply their aggregation and their interaction with potential inhibitors to obtain preventative and possibly cures for AD and PD.

Different aspects of these subjects have been reviewed by Grasso,⁶ Hu et al.,⁷ Artigues et al.,⁸ Konermann et al.,⁹ McKenzie-Coe et al.,¹⁰ James et al.,¹¹ Kumar et al.,¹² Politis and Schmidt,¹³ De Simone et al.,¹⁴ Bennett et al.,¹⁵ Williams and Pukala,¹⁶ Uversky,¹⁷ Tsutsui and Wintrode,¹⁸ Landreh et al.,¹⁹ Chavez and Bruce,²⁰ O’Reilly and Rappsilber,²¹ Piersimoni et al.,²² Chavez et al.,²³ Liu et al.,²⁴ Vallejo et al.,²⁵ and Petrotchenko and Borchers.²⁶ Although Grasso⁶ discusses several MS studies (including HDX-MS and nMS) on amyloid beta (Aβ) to determine aggregation state, interactions with other species and metals, and aggregation pathways, the authors do not cover background details for the MS approaches. Hu et al.⁷ briefly discuss the basics of nMS, IM-MS, HDX-MS, and XL-MS, and also highlight some recent research with each approach applied to aggregating proteins. Artigues et al.⁸ highlight HDX-MS and XL-MS, but do not discuss applications to aggregating proteins. Konermann et al.⁹ discuss HDX-MS and hydroxyl radical (·OH) labeling. McKenzie-Coe et al.¹⁰ extensively review hydroxyl radical labeling, including FPOP. James et al.¹¹ evaluate HDX-MS, showing its ability to elucidate protein structure. Kumar et al.¹² discuss the mechanism of protein aggregation and potential therapeutic strategies, as well as briefly describing several techniques/approaches that can be employed to study aggregation (including HDX-MS and IM-MS). Politis and Schmidt¹³ give a brief background on HDX-MS, IM-MS, and XL-MS and show an application to an aggregating protein (β2-microglobulin). De Simone et al.¹⁴ and Bennett et al.¹⁵ present a synopsis of nMS and IM-MS (as well as spectroscopic and microscopic methods) and discuss applications and drug discovery for affecting Aβ aggregation. Williams and Pukala¹⁶ describe the various types of ion mobility and emphasize applications with aggregation prone proteins, including Aβ and aS. Uversky¹⁷ reviews in detail the aggregation of aS, including some references to HDX-MS and IM-MS. Tsutsui and Wintrode¹⁸ present a general overview of HDX-MS and illustrate some applications for amyloids. Landreh et al.¹⁹ briefly discuss IM-MS and XL-MS and give a detailed review of HDX-MS. Chavez and Bruce,²⁰ O’Reilly and Rappsilber,²¹ and Piersimoni et al.²² provide backgrounds in XL-MS and discuss potential applications. Chavez et al.²³ describe the ability of HDX-MS, FPOP, XL-MS, and other labeling methods of proteins and complexes in cells. Liu et al.²⁴ extensively cover background for nMS, HDX-MS, FPOP, and XL-MS, but the focus of that review is protein footprinting without detail for aggregating proteins. Vallejo et al.²⁵ review nMS, IM-MS, HDX-MS, FPOP, XL-MS, and other approaches and highlight situations in which each method can be applied. Petrotchenko and Borchers²⁶ also emphasize HDX-MS, FPOP (along with other covalent labeling methods), XL-MS, and other methods that can be utilized with MS readout to elucidate protein structure.

Although there seems to be many reviews, not all are associated with amyloid formation, and some only briefly cover some MS-based approaches. This review attempts to describe thoroughly the basics of each MS-based approach and the applications with aggregating proteins, and we show detailed analysis for each method with specific examples for both Aβ and aS.

B. Amyloid Fibril Formation

Amyloid formation is a multistep process that begins with the development of a misfolded protein, usually in its monomeric state.^2,3,4,27 Monomers in solution may assemble with other monomers or natively structured proteins, forming higher order assemblies (dimers and trimers,⁴ Figure 1) or aggregates. Aggregates tend to be heterogeneous because they are comprised of different numbers of monomers, making it difficult to follow them during protein aggregation. Smaller aggregates that range from 2–20 mers are often termed oligomers.⁴ (Some proteins are dimers, trimers, or some small oligomer in their native state. Thus, their aggregation starts from the misfolding of the oligomeric state.) Oligomers build via two general categories of oligomer formation: off-pathway and on-pathway (Figure 1).²⁸ Although off-pathway oligomers do not lead to a final amyloid fibril, on-pathway oligomers have an appropriate structure to allow elongation of the amyloid fibril to continue.²⁸ Figure 1 shows a protofibrillar oligomer species between on-pathway oligomers and amyloid fibrils. Protofibrillar oligomers are large, on-pathway oligomers that lead to amyloid fibril formation.²⁸ Amyloid fibrils are linear aggregates with a repetitive cross-beta structure that can contain thousands of monomers.^3,4 Amyloid fibrils elongate at a much faster rate than oligomers by adding monomers to the ends of the fibril.^3,4 Wells et al.²⁹ describe the process of protein aggregation and amyloid formation in more detail, emphasizing Aβ and aS.

Figure 1.

The pathway of amyloid fibril formation. Assembly starts with monomers that assemble to form oligomers that can be either on-pathway or off-pathway. The on-pathway oligomers continue to grow to amyloid fibrils. The red and blue color schemes indicate the relative amount of β-sheet formation, where red and blue are high and low levels of β-sheet formation, respectively. Higher levels of β-sheet formation generally correlate with higher toxicity. (Reproduced with permission from reference 28.)

The formation of amyloid fibrils can lead to adverse health conditions throughout the body,³ specifically amyloid fibrils produced from Aβ and aS can cause plaque buildup in the brain.^2,3 The fibrils or plaque may prevent nearby neuron cells to signal one another, preventing proper brain function. An early theory was that the toxic species are the final amyloid fibril, and this theory seems to underpin the recent controversial FDA approval of the antibody, aducanumab that is designed to decrease Aβ plaques.³⁰ Other recent research, however, shows increasing evidence that oligomers and other early, soluble aggregates are the true toxic species, rather than the final amyloid fibrils, for several neurodegenerative disorders,^2,5,31,32,33 summarized in an opinion piece by Walsh and Selkoe.³⁴ The toxicity of these oligomers and early aggregates may arise from interactions with cellular membranes, leading to impairment of cellular processes.² The toxicity of the oligomers is likely related to their relative levels of β-sheet secondary structure²⁸ (i.e., the uncertainty in toxicity can be seen in Figure 1, where oligomers, protofibrillar oligomers, and amyloid fibrils are labeled in red). Interestingly, off-pathway oligomers also may be toxic. These oligomers likely have β-sheet secondary structure that can be stabilized by covalent bonding stemming from oxidative modifications.²⁸ We can see in Figure 1 that one off-pathway oligomer (in red) may lead to high levels of β-sheet formation and potential toxicity, while the other off-pathway oligomer gives species with low levels of β-sheet formation and a lack of toxicity (blue). Overall, there is still debate about which species is most toxic and is the major cause of neurodegenerative diseases. Additional research is needed to determine whether single or multiple species are toxic and to identify those species that are the most toxic.

C. Kinetics of Amyloid formation

Amyloid fibril formation usually displays sigmoidal growth kinetics,⁴ that can be divided into a lag, a growth, and saturation phases (the final plateau)^2,3 (Figure 2). The relatively flat lag and saturation phase correspond to an increasing number of species at similar levels of aggregation manifested by the initial nearly horizontal region.

Figure 2.

The sigmoidal curve for amyloid fibril formation. The change is usually measured as an increase in fluorescence caused by immobilization of the dye in the growing aggregates. (Reproduced with permission from reference 35.)

The lag phase starts with a protein in its native (monomeric or small oligomeric) structure. It then undergoes misfolding to form a species competent for aggregation, which initiates amyloid fibril formation. This process happens repeatedly, and the rearranged monomers start to associate and form oligomers.³⁶ These oligomers can exchange subunits or undergo conformational change. Once a nucleus is formed, the aggregation curve transitions from the lag to the growth phase. In the growth phase, fibril elongation is accelerated by an autocatalytic growth reaction that continues as shown by the upswing in the curve to the saturation phase, where a full amyloid fibril begin to form.² Once formed, there is little growth of the soluble fibril, resulting in a final plateau for the curve. The overall conversion rate of proteins into their amyloid forms is greatest during the growth phase, which is signified by the steep increase in the curve.⁴

The kinetics of amyloid formation can be changed depending on the environment of the starting solution.^2,37 For example, the entire lag phase can be avoided upon addition of pre-formed nuclei that act as fibrillar seeds to expedite formation and decrease the time needed to form the fibril.³⁶ On the other hand, addition of an inhibitor can have the opposite effect. The inhibitor can lengthen the lag phase and the overall time necessary to form the amyloid fibril. Some inhibitors have the potential to stop amyloid fibril formation completely.^2,37 The challenge is to understand these inhibitors on a molecular basis so that aggregation can be slowed in vivo, serving to keep at bay AD, PD, and other amyloid-causing disorders.

D. Amyloid Beta and Alpha-Synuclein

Two proteins receiving considerable attention in recent year are amyloid beta (Aβ) and α-synuclein (aS); both are involved in the etiology of neurodegenerative disorders.³ These are the primary aggregating proteins discussed in this review. One reason is our lab has conducted more studies of these proteins compared to other aggregation proteins, thus gaining some expertise to share with the community. The formation of the amyloid fibrils causes the Aβ and aS to lose their original function, possibly causing these diseases. Even though amyloid fibrils have been linked to these disorders, there is some dispute on what causes the disease. Evidence has shown that oligomeric species are potentially the most toxic, which may be the reason for the disorders.^5,32,33 Research focused on the regions responsible for protein aggregation and the binding sites of potential inhibitors will be discussed later, but now we will present some biomedical background on Aβ and aS.

The formation of amyloid fibrils of Aβ is believed to play an important role in AD.^{32,38,39,40,41} There are two main forms of Aβ: Aβ₄₀ and Aβ₄₂, both intrinsically disordered. Aβ₄₀ makes up ~90% of all Aβ species, while Aβ₄₂ makes up ~10% (see Figure 3 for the amino acid sequence of Aβ). Aβ is produced in the brain by proteolytic cleavage of the transmembrane protein amyloid precursor protein (APP) by β- and γ-secretases.³⁹ Although the physiological role of Aβ is still unknown,³⁹ Aβ_42, the minor constituent, is the main component of amyloid plaques, which are one of the distinctive features of AD.^32,38,39,40 With Aβ₄₂ being the dominant species in plaques, the last two residues at the C terminus (I and A) must play a role in aggregation. Formation of Aβ amyloid fibrils is a complicated process that involves several intermediate oligomers. The Aβ species in the amyloid plaques are typically fibrillar, β-sheet-rich aggregates.^5,32,33 The oligomers, possibly those that give rise to fibrils, may be the major neurotoxic species.^5,32,33 Gremer et al.⁴² employed cryogenic electron microscopy (cryo-EM) on Aβ₄₂ amyloid fibrils, showing a “LS” shaped structure for each subunit and protection of the C-termini at the dimer interface (Figure 4). Given there is no cure for AD, developing a more comprehensive understanding of amyloid formation and finding potential inhibitors of Aβ amyloid formation may be a profitable approach for the near future.

Figure 3.

The amino acid sequence of Aβ. Aβ₄₀ includes the first 40 residues, whereas Aβ₄₂ is the full 42 residues. The residues in green represent negatively charged residues. The residues in orange represent positively charged residues.

Figure 4.

A) The masked cryo-EM density of two subunits of the Aβ₄₂ amyloid fibril highlighting the dimer interface. B) A tilted view of the cryo-EM density showing the cross section of the two protofilaments comprising the fibril. (Reproduced with permission from reference 42.)

Similar to Aβ and AD, the aggregation and amyloid fibril formation of aS is believed to play a significant role in PD.³⁸ aS is a 140-residue protein that is largely intrinsically disordered, but has three main structural domains (Figure 5): (1) N-terminal (residues 1−60), which binds to lipids and proteins, (2) non-amyloid-β-component (NAC) (61−95), which contains aggregation prone β-sheets, and (3) the C-terminal (96–140) that contains acidic groups.²⁷ The physiological role of aS is still uncertain, but some evidence suggests that aS plays a role in neuron cell signaling. Perhaps aS plays a role in regulating synaptic vesicle release and stabilization of SNARE complexes.³⁸ When the N-terminus of aS is bound to a membrane, aS adopts an α-helical conformation. When unbound, it reverts back to its intrinsically disordered structure.³⁸ Research (to be reviewed later) shows that the NAC region is aggregation-prone and adopts a rich β-sheet content.²⁷ Li et al.⁴³ used cryo-EM to determine the structure of aS amyloid fibrils and found two dominant structures (a rod and a twister) that differ in their inter-protofilament interfaces (Figure 6).

Figure 5.

The three regions of aS and the full list of residues that comprise aS. (Reprinted (adapted) with permission from reference 27. Copyright 2018 American Chemical Society.)

Figure 6.

A) Cryo-EM and atomic illustration of the rod conformation for aS amyloid fibrils. B) Cryo-EM and atomic illustration of the twister conformation for aS amyloid fibrils. (Reproduced with permission from reference 43.)

E. Potential therapies

Currently there are only a few therapies available for fighting neurodegenerative diseases. These therapies focus on improving the symptoms of the disease, rather than its prevention.^5,44 The development of better therapeutics is essential for the millions of people suffering or will suffer from these diseases.

As mentioned earlier, inhibitors of amyloid formation increase the time needed to form amyloid fibrils and, in some cases, completely prevent amyloid fibril formation, at least in vitro. One purpose of inhibitors is to keep the protein in its native state and slow the aggregation; another is to channel aggregation to other pathways with nontoxic species. Natural products, peptides, and small molecules are front runners for inhibiting aggregation (and will be highlighted later).⁴⁵ Developing inhibitors, however, requires methods to determine the interactions between inhibitor and the target protein. Knowledge of these interactions may lead to a rational scheme to prevent protein aggregation and enable development of new and better inhibitors.⁴⁶ An effective inhibitor is not required to stop completely amyloid formation but to delay or interfere with the production of amyloid fibrils and thereby toxic oligomers,^45,47 extending the lag phase by several years, potentially beyond the lifetime of the average human.

II. METHODS FOR STUDYING AGGREGATING PROTEINS

A. Biochemistry-based Methods

Circular dichroism (CD), Thioflavin-T (ThT) fluorescence, Fourier transfer infrared spectroscopy (FTIR), electron paramagnetic resonance (EPR), small-angle X-ray scattering (SAXS), transmission electron microscopy (TEM), atomic force microscopy (AFM), and super-resolution microscopy are considered low-resolution techniques.^24,27 All have the ability to afford a global-level view of a protein/amyloid fibril. For example, CD can show the amount of β-sheet or α-helical content in a protein/amyloid fibril, but it does not display regional or atomic resolution of the fibril.²⁴ Although these methods are fast (high throughput) and well understood, their downside is they provide little regional or atomic coverage.^24,27

On the other hand, several methods can provide a high-resolution structure of the protein/amyloid of interest, reaching the atomic level.²⁴ These approaches include high-resolution X-ray crystallography,⁴⁸ cryo-EM,⁴⁹ and nuclear magnetic resonance (NMR),^50,51,52,53 particularly solid-state NMR for insoluble fibrils. X-ray crystallography and NMR have become benchmarks for determining protein HOS,²⁴ whereas cryo-EM is now emerging with the potential to change structural biology. These latter approaches can address the starting monomer and/or the final, solid-state fibril, but they reveal little about the intermediate species.²⁴ These high-resolution approaches are difficult to employ on a regular basis; examples are the inability to monitor protein structure in a heterogenous solution, the need for a crystal to apply crystallography, the requirement for a large amount of protein sample, the tedious and lengthy data analysis, and the size of the protein being monitored.²⁴ Despite their limitations, however, cryo-EM, along with X-ray crystallography and NMR, are considered the best methods for high resolution protein structure elucidation, providing atomic coordinates for many constituent atoms.²⁴ Toyama and Weissman⁵⁴ review these high-resolution techniques in more detail and illustrate some applications.

Most of these biochemistry-based methods have been applied to amyloid fibril formation and even inhibitor-protein interactions. These methods may show that if an inhibitor is effective, a difference will be seen for the protein alone and the protein incubated with an inhibitor (which may prevent amyloid fibril formation). These approaches can provide some information about the ability of the inhibitor to stop aggregation of the protein of interest, even with the disadvantages stated earlier. When first testing a new inhibitor candidate, typically a global approach (e.g., CD, ThT fluorescence, or TEM) can determine if the candidate inhibitor modifies or stops aggregation because these methods are amenable to quick data acquisition and global analysis. If the candidate hinders amyloid formation, further investigation with a higher resolution method can be planned to determine how the approach prevents aggregation. A note of caution: sometimes an inhibitor candidate may interact with the fluorescent dye instead of the protein of interest, skewing the data and creating a false positive.⁵⁵

Although there are a plethora of methods to study protein/amyloid structure, all low- and high-resolution methods have some disadvantages, motivating development of new methods of intermediate capacity.²⁴ MS can fill the gap between low and high resolution because it provides regional (and sometimes amino-acid) coverage, is fast and sensitive, can produce meaningful results for solutions that have heterogeneous oligomer composition and dynamics, yet only requires small sample amounts. Although MS is considered a mid-resolution approach, it does provide information that is nearly impossible to obtain with any other method. In the following section, we describe MS methods used for amyloids to assist the non-expert in understanding the applications that follow the approach tutorial.

B. Mass-Spectrometry based methods

MS has attracted increasing attention in bioanalytical fields, for its ability to obtain a regional/amino acid level resolution from a small sample.⁷ Another advantage of MS- based approaches is the capability to analyze one species in a heterogenous solution without altering the equilibrium.² This is a problem with the atomic level methods (i.e., X-ray crystallography, NMR, and cryo-EM), which require a purified sample, sometimes in another state of matter. MS offers high spatial resolution for solution species, usually does not change the equilibrium conditions to perform analysis, and permits the amyloid to be maintained in a relevant state.

In MS, gaseous ions must be produced from either solution or solid-state samples, and then these ions are detected according to their mass-to-charge ratios (m/z).^7,24 For a solution containing a heterogenous mixture of aggregated proteins, one protein/oligomer may be selected and separated from other ions of different m/z. If the protein/oligomer is digested before MS analysis, then the resulting peptides can be separated according to their m/z.²⁴ Tandem MS or MS/MS allows gaseous ions to be isolated with the first analyzer, then submitted to fragmentation, and the resulting fragments (product ions) are mass analyzed (to give a “product-ion” or MS/MS spectrum).^7,24,56 In a protein footprinting experiment/analysis (e.g., HDX or FPOP, which will be discussed below), tandem MS provides information on the sites of the protein/peptide that underwent modification.^24,56 This analysis can also elucidate the binding location between protein-protein and protein-small molecule interactions.

In general, MS can provide mass information of a protein and determine whether a protein complex is a monomer, dimer, or other smaller oligomer.⁷ Amyloids, where the oligomeric number is in thousands, are difficult to analyze by MS with precision because samples are highly complex and dispersed in terms of m/z signals. These complications make it difficult to focus on one ion or protein peak. There has been some success studying amyloid species with charge detection MS, a subject developed and reviewed by Jarrold.⁵⁷ In brief, charge detection MS simultaneously measures the charge and m/z of an individual ion to determine the molecular mass.⁵⁷ The molecular mass measurement is repeated multiple times, and the results are binned together, yielding a mass spectrum.⁵⁷ To obtain these results, ions pass through a detection cylinder, where a charge carried by the ions is detected by a charge-sensitive amplifier.⁵⁷ With a sufficiently long cylinder, the charge of the ion will induce charge in the detection cylinder, and the flight time of the ion through the cylinder can provide the m/z,⁵⁷ thus providing both the charge and m/z simultaneously. One specific example of charge detection MS of Aβ₄₂ is work done by Pansieri et al.,⁵⁸ who found Aβ₄₂ amyloid species ranging from 20–600 MDa.

Even with the complications from studying large aggregate species, vital information can be gained about monomers and small oligomers with MS. Although debatable, increasing evidence indicates that oligomers and other early, soluble aggregates are the true toxic species,^2,5,31,32,33 and their study is appropriate for MS. MS can be used in combination with footprinting (e.g., HDX and FPOP), chemical cross-linking (XL), and ion mobility (IM) to obtain crucial information on early aggregate species, such as locating misfolding sites, regions of oligomerization, and areas where inhibitors bind.

1. Native Mass Spectrometry (nMS) and Ion Mobility

nMS serves as a valuable approach for the detection of intact protein assemblies, and when combined with ion mobility, advances the field of protein HOS determination.^7,59 Standard ESI-MS uses a high voltage (2.5–3.0 kV), high temperatures, low pH (to facilitate protonation), and organic solvents (such as acetonitrile and methanol) for fast desolvation.^7,59,60 These conditions disrupt non-covalent bonds in proteins and oligomers and cause denaturation.^59,60 nMS also employs electrospray ionization (ESI), but it uses milder ionization, including physiological temperature and pH and volatile, ammonium-based electrolytes (e.g., ammonium acetate).^59,60 By using these conditions, the non-covalent interactions of the protein complex are preserved, at least in part, in the ionization.^59,60

nMS and IM can provide information on the shape and structure of proteins and their oligomers when studying protein aggregation and oligomerization. IM separates one or several species on the basis of the ions rotationally averaged collision cross section (CCS).^{2,14,15,16,61} For example, if there are several monomers in solution (one folded native state and another less folded or denatured) at the same m/z, IM-MS may differentiate them. The ions are admitted to an ion mobility region where they make many collisions with an inert gas to adjust their drift motion to facilitate separation before reaching the mass analyzer.^2,14,15,16 A weak electric field is applied to the ion mobility cell to ensure that the ions migrate at their characteristic drift times.^2,61 Compact ions with a small CCS undergo fewer collisions with the inert gas and pass through the ion mobility cell quickly (have a short drift time),² whereas larger ions undergo more collisions and migrate more slowly (have a longer drift time).² (There are several types of ion mobility instruments (reviewed by Cumeras et al.⁶¹), and covering them is beyond the scope of this review.)

IM-MS also allows separation of monomers, dimers, and trimers of the same m/z. In this scenario, a monomer, dimer, and trimer at the +1, +2, and +3 are separated because a greater charge state causes the ion to migrate faster through the electric field (given their CCS values are similar).

IM-MS can report on the size and shape of protein/amyloid structures by “filtering” them in IM-MS.² By using nMS conditions for introducing inhibitor/amyloid complexes, a comparison between an unbound protein and protein with an inhibitor can be conducted. This enables the study of their binding interactions, changes in oligomerization, and identification of oligomeric states that are avoided through interactions with an inhibitor.

Typically, small molecule inhibitors have four modes of action visible by IM-MS: negative, positive, non-specific, and colloidal (Figure 7).³⁷ A small molecule with negative inhibition on the amyloid protein does not bind to the protein, so it does not inhibit amyloid formation.³⁷ A small molecule that is a positive inhibitor binds to one or more parts of the protein to give a binomial distribution of corresponding peaks in the mass spectrum, where there is a specific number of binding sites on the protein.^37,62 A non-specific small molecule inhibitor can also bind to the protein but shows a Poisson distribution of bound peaks, where there are many potential binding sites on the protein.^37,62 A colloidal inhibitor has the ability to self-associate (i.e., the inhibitor self-associates), which will generate a range of overlapping peaks.³⁷ For therapeutic use, the best small molecule inhibitors would be positive inhibitors that do not interact with other proteins. Therefore, potential inhibitors can be identified and possibly developed for therapeutic use by using IMMS as a tool.

Figure 7.

The four modes of action for a potential small molecule inhibitor as seen by IM-MS, which are negative, positive, non-specific, and colloidal. (Reproduced with permission from reference 5.)

2. Hydrogen Deuterium Exchange (HDX)

Hydrogen−deuterium exchange combined with MS (HDX-MS) can be used to investigate protein stability, protein folding and misfolding, protein−protein interactions, and ligand binding, offering opportunities for study of amyloid proteins. HDX-MS provides structural information not only for the full protein, but also regional (peptide) and near-residue level of the protein.^11,27 It has become one of the most used covalent labeling protein footprinting approaches, despite its reversible nature.²⁴

The first steps in HDX are solubilizing the protein in H₂O and diluting with D₂O. This causes labile hydrogens on the protein to exchange with the surrounding deuterated solvent.^11,24 Hydrogens that are a part of O−H, S−H, and N−H back-bone and side-chain bonds in a protein molecule exchange with deuterium most rapidly, leading to an increase in mass of the protein, measurable with MS.^11,24 Of prime interest are the hydrogens that exchange on the backbone, although amino acid side chains also contain exchangeable hydrogens (e.g., NH₂, OH, SH). These latter hydrogens participate weakly in intramolecular hydrogen bonding, making their exchange fast and difficult to follow, but also amenable to fast back exchange following quench in H₂O.^11,24 Thus, the measured exchange is primarily of the amide N-Hs of peptide bonds (Equation 1), and those Hs exchange (and back exchange) slower during the experiment.²⁴

Equation 1.

The amide bonds exchanging hydrogen for deuterium when D2O is added are principally measure in an HDX experiment.

The rate of exchange for a given amide bond depends on the solvent accessibility, hydrogen bonding, and temperature²⁴ (Scheme 1). Local ordered structure is often stabilized by hydrogen bonding between an amide H and a carbonyl (i.e., N−H···O=C) that is part of a dynamic conformational state that can be described as a breathing motion,²⁴ a motion whereby the amide bond cycles from closed to external solvent to open and then back to closed. The rate constant at which this amide bond goes from closed to open is k_op, and the rate constant at which the amide bond goes from open to closed is k_cl (Scheme 1). While the amide bond is in the open state, deuterium exchange occurs with a rate constant, k_ch.²⁴ Each amide bond will have a unique set of k_op, k_cl, and k_ch. Together, these three rate constants constitute the overall rate constant for HDX.²⁴

Scheme 1.

The rate constants k_op, k_cl, and k_ch comprise the overall rate constant. The amide bond can go from a closed to open state, and in the open state, solvent enters the newly exposed region, allowing deuterium exchange with hydrogen. (Reprinted (adapted) with permission from reference 24. Copyright 2020 American Chemical Society.)

Typically, amide N-H that are solvent-exposed and not involved in strong hydrogen bonding with other parts of the protein will exchange rapidly. An example is an unstructured region of the protein.^24,63 For amides that are less solvent-exposed or that participate in intra-protein hydrogen bonding, the rates of HDX will be slower.^24,63 An example is an amide bond in a β-sheet formation that has considerable structure and is protected from the solvent. For proteins that form amyloids, the regional-specific exchange rates are a function of the protein oligomer structure.

There are two ways to perform HDX experiments to study amyloid fibril formation: continuous HDX and pulsed HDX, as will be described in the next two sections.

2.1. Continuous HDX-MS

At several times of exchange, the HDX is quenched by decreasing the pH of the solution to 2.5 and decreasing the temperature to 0 °C,²⁴ where exchange (including back exchange) is minimized.⁶⁴ The extent of HDX is then measured by MS for the intact protein or for constituent peptides produced by acid-insensitive protease digestion. Digestion is best achieved by flowing the protein solution through a column packed with immobilized pepsin protease, as pepsin digests under these acidic conditions.²⁴ Liquid chromatography-mass spectrometry (LC-MS) analysis then follows, in which the eluents are adjusted to acidic conditions to minimize back exchange. The MS data are analyzed by fitting the isotopic peak distribution for each peptide to obtain the mass centroid.²⁴ These centroids can be “corrected” with that of the unexchanged peptide as a function of time to determine the kinetics of HDX²⁴ (Figure 8).

Figure 8.

Workflow of continuous HDX-MS. The protein in upper right is submitted to exchange in D₂O causing amides and other active H’s to exchange to D (middle). Upon quench (pH ~ 2.5, T ~ 0 °C), the protein is proteolyzed with acid insensitive enzymes, the resulting peptides are separated by HPLC, and their mass spectra obtained. (Reprinted (adapted) with permission from reference 24. Copyright 2020 American Chemical Society.)

2.2. Pulsed HDX-MS

Pulsed HDX-MS also follows the same protocol as continuous HDX-MS, except the design separates or deconvolutes the kinetics of HDX with the kinetics of aggregation by keeping constant the time of HDX (e.g., 1 min). Pulsed HDX has been utilized for detecting protein folding and, relevant to this discussion, the formation of oligomers and amyloid fibrils.^40,65,66,67 Unlike continuous HDX, the protein is incubated in an H₂O solution in the absence of D₂O for increasing times of aggregation (t_agg in Figure 9).⁶⁸ After suitable aggregation times, D₂O is pulsed into the solution to allow HDX to occur for typically 1 min⁶⁸ (pulse time is labeled t_pulse in Figure 9) followed by quenching, online digestion, and LC-MS, as for the continuous HDX workflow.⁶⁸

Figure 9.

Schematic of pulsed HDX. Proteins in H₂O (left tubes) are submitted to variable times of aggregation (represented by t′_agg, t″_agg, and t‴_agg). At t_agg, D₂O is pulsed for a constant, short time followed by quenching, online digestion, and LC-MS.

At short t_agg, when the amount of aggregation is small, the extent of HDX is representative of the monomer and small oligomers (these times correspond to the lag phase of the sigmoidal aggregation curve in Figure 2). Specifically, the monomer and small oligomers show considerable HDX during the D₂O pulse because little HOS characteristic of an aggregate has yet formed^68,69 (see Figure 10 at short t_agg), and the corresponding mass spectral peaks shift quickly to higher m/z. At longer t_agg, however, more aggregation occurs, giving oligomers and ultimately fibrils in solution, and the corresponding peaks will be of lower m/z than at short t_agg.

Figure 10.

Comparison of short vs long t_agg in pulsed HDX for an aggregating protein (such as Aβ or aS). The native protein is in solution for a variable amount of time, t_agg, before the D₂O is pulsed into solution. At short t_agg, small amounts of aggregation occur, and the extent of HDX is large (as is the m/z shift). Longer t_agg times result in more aggregation and more secondary structure, and consequently less HDX and a smaller m/z. (Reprinted (adapted) with permission from reference 68. Copyright 2014 American Chemical Society.)

3. Fast Photochemical Oxidation of Proteins (FPOP)

FPOP is another footprinting method that informs the HOS of an amyloid. FPOP is classified as “covalent labeling”, giving irreversible modifications.⁷⁰ For FPOP, photolysis of hydrogen peroxide (H₂O₂), gives two ·OH.^71,72 These radicals then react with the side chains on the protein mainly to replace H with ·OH and increase the mass by 15.9949 Da, although other chemistries occur.⁷¹ The amount of hydroxyl radical labeling of the protein depends on the inherent reactivity of the radical with the side-chain residue and the solvent accessibility of the side chain as a function of HOS. Compared to HDX, FPOP reacts more rapidly and offers residue specificity, but it cannot label every residue (as can HDX (except Pro)), because some residues are not reactive on the short time scale of the exposure.⁷⁰ The time scale (dosage) of footprinting can even be tuned by varying the amount of scavenger (typically a single amino acid). The high reactivity of free radicals coupled with use of a scavenger enables labeling of the protein in its native or near-native state at rates faster than protein unfolding without structural perturbation produced by excess modifications.⁷⁰ FPOP has been utilized to study, in addition to amyloid formation, protein fast/slow folding, protein-ligand interactions, protein dynamics, and to identify hidden conformations.⁷⁰

A key feature of FPOP is the use of a pulsed laser to form ·OH (Figure 11). A 248 nm KrF excimer laser pulse cleaves hydrogen peroxide, at ~15 mM or 0.04%,^70,72 but has minimal effects on water and most proteins because their absorbance is low at 248 nm. Convex lenses are used to focus the laser beam onto an exposure window (2.0–3.0 mm wide) in the silica tubing perpendicular to the laser beam through which the sample is passing.^70,72 Irradiation gives a high quantum yield of hydroxyl radicals even though the absorbance of H₂O₂ is not high at 248 nm.⁷⁰ The frequency of the laser, controlled by an external pulse generator, and the flow rate of the solution are controlled to minimize multiple laser irradiation of the same solution plug.⁷⁰ A small volume flow (called “exclusion volume”, see Figure 11) is purposely excluded from laser irradiation to minimize “double shots” by the laser and leave some unreactive protein. The FPOP capillary exits into a sample collection tube, containing catalase and free methionine in buffer to eradicate leftover hydrogen peroxide and any remaining long-lived radicals from causing post-footprinting oxidation.^70,72

Figure 11.

Setup of FPOP. The KrF laser irradiates hydrogen peroxide solution through the exposure window in the FPOP capillary, creating hydroxyl radicals (~ 1 mM) that react with the side chains of the protein. (Reprinted (adapted) with permission from reference 70. Copyright 2018 American Chemical Society.)

Hydrogen peroxide and a scavenger are added to the protein solution, and the resulting solution is flowed through the capillary pushed with a syringe pump. Once the mixed solution enters the region of the exposed window of the capillary, laser-triggered photolysis of hydrogen peroxide creates hydroxyl radicals in a few nanosec. The hydroxyl radicals either react with the side chains of the protein, or with the scavenger, or themselves to revert back to hydrogen peroxide (self-quenching is shown in Scheme 2).⁷⁰

Scheme 2.

Formation of hydroxyl radicals and their self-quenching by reforming hydrogen peroxide. (Reprinted (adapted) with permission from reference 70. Copyright 2018 American Chemical Society.)

The hydroxyl radical lifetime can be controlled by the nature of the scavenger and its concentration. Without a scavenger, the hydroxyl radical lifetime is determined by the reformation of hydrogen peroxide, which requires ~100 μs.⁷⁰ In 100 μs, the protein may undergo conformational changes by unfolding and refolding to expose buried residues to the hydroxyl radicals and skew the data used to infer protein structure.⁷⁰ By adding excess scavenger, the lifetime of the hydroxyl radicals can be shortened 100-fold (to ~1 μs),^70,72 which is sufficient time for labeling the exposed residues in the native protein, unperturbed by protein unfolding/folding during the radical exposure time. This emphasizes the necessary role of the scavenger to control radical lifetime so that changes so that an unbiassed HOS of the protein can be determined.

Once the sample has undergone FPOP modification, bottom-up proteomics is usually employed for data analysis (Figure 12). The bottom-up proteomics affords regional and residue-specific structural information. Given that FPOP is an irreversible method, many different types of proteases can be used to digest exhaustively the protein.⁷⁰ This is an advantage over HDX, where digestion must be done quickly with online pepsin immobilized in a column. In fact, several digestion combinations can be employed to afford more peptides and residues and to increase structural resolution. After digestion, the peptides are analyzed via LC-MS/MS.⁷⁰ Modified peptides typically elute earlier than unmodified by reversed-phase LC, allowing for some separation of peptides with different sites of modification. From mass spectra taken over elution time, an extracted ion chromatogram is obtained and used to assign the hydroxyl radical modification at the peptide level and sometimes and with MS/MS at the residue level.⁷⁰ The labeling extent in a control vs. test experiment can pinpoint differential structure changes as a function of aggregation.

Figure 12.

An example of bottom-up proteomics workflow after submitting a protein to footprinting. The native protein, upper left, is submitted to footprinting (upper center) and then to proteolysis to afford peptide and sometimes residue-level footprints for usually a control and a test experiment. (Reprinted (adapted) with permission from reference 24. Copyright 2020 American Chemical Society.)

4. Other Irreversible Labeling Methods

There are several other irreversible labeling reagents/methods that can be employed to study protein structure and aggregation. Although this review will not go into detail, these may have applications in aggregation studies. The corresponding analysis used for these other methods is similar to that employed for FPOP, where several digestion combinations can be utilized to increase protein coverage, and MS/MS analysis can yield residue specific information.

Similar to FPOP, X-ray footprinting (XFMS) utilizes hydroxyl radical labeling to determine protein HOS.^{73,74,75,76,77,78,79} At the National Synchrotron Light Source and the Advanced Light Source, photons of 3–30 keV are produced and can interact water to create hydroxyl radicals (see Scheme 3).^{73,74,75,76,77,78,79} To provide an in-depth discussion of XFMS, the Chance group reviewed the subject and described potential applications.^74,79 Some disadvantages of XFMS are the need of the synchrotron and a longer timeframe (ms) than for FPOP to produce the hydroxyl radicals.⁸⁰ In the ms timeframe, the protein may undergo conformational changes and skew the labeling results.⁷⁰

Scheme 3.

The production of hydroxyl radicals from water using XFMS.

Carbene footprinting is another labeling strategy that has been successfully employed to determine protein HOS.^81,82,83 Carbenes can be produced through the photolysis of a diazirine (Figure 15) using near-UV wavelengths (~350 nm).^83,84,85 Once formed, carbenes are able to insert into any X-H bond (where X is C, O, N, or S),⁸⁶ so it can footprint potentially all amino acids.⁸⁷ The carbene modification of the protein is in the sub-μs range, so footprinting is complete before the protein unfolds. The Gross lab successfully implemented the carbene footprinting strategy into the FPOP flow system by using a slightly modified setup described above in the FPOP section.⁸³

Figure 15.

Examples of both an NHS ester (left) and diazirine (right) cross linker. The three membered ring is photolyzed to lose N2 and yield a carbene at the diazirine site.

Another method to induce labeling via oxidization is metal-catalyzed oxidation (MCO).^{88,89,90,91,92} MCO labeling is achieved by incubating a metal-bound protein with both a reducing agent (such as ascorbate) and an oxidizing agent (such as O₂).^90,91 The metal creates reactive oxygen species, including hydroxyl radicals, through redox cycling (via Fenton-like chemistry).^90,91 The reactive oxygen species can quickly react with residues nearby the metal, resulting in either oxidation or cleavage of the protein at these sites.^{88,89,90,91,92} Oxidation or cleavage can be limited to only the residues bound to the metal when reaction is performed under appropriate conditions, thus informing on the binding site of the metal.^{88,89,90,91,92,93}

Fenton chemistry has also been utilized to produce oxidation labeling.^78,94,95 By incubating Fe(II) with hydrogen peroxide, Fe(II) becomes oxidized to form Fe(III) and then produces hydroxide and hydroxyl radicals that footprint the protein.^78,94 The addition of ascorbate to the reactions maximizes hydroxyl radical formation, as ascorbate reduces Fe(III) to Fe(II) for recycling.^78,94 Although Fenton chemistry has been successful in footprinting various protein systems,^96,97,98 there are concerns. that addition of other reagents (e.g., iron) can induce a conformational change on the protein. Further, the reaction occur continuously, not via a fast (μs) pulse, and undesired secondary reactions and protein unfolding may occur at long incubation times.^94,99

Chemical reagents can also be employed to footprint solvent-accessible surfaces of a protein.^24,72,80,100 Specific amino acid footprinting utilizes reagents that react with particular functional groups (e.g., −SH, −COOH, etc.) on protein side chains.^24,72,80,100 A few chemical reagents (see Figure 13) for footprinting are N-ethylmaleimide (NEM), which labels cysteines, glycine ethyl ester (GEE), which labels aspartic and glutamic acids, and diethylpyrocarbonate (DEPC), which labels most nucleophilic residues.^24,72,80,100 These example chemical reagents have a longer timescale of labeling (from seconds to hours) and can vary in reversibility. However, these reagents are typically considered irreversible (especially in comparison to HDX), and bottom-up proteomics can be employed for identification of the footprinted residues.^24,72,80,100

Figure 13.

Structure of chemical reagents used for specific amino acid footprinting.

For oligomer/amyloid formation, specific amino acid footprinting has successfully determined the binding interface for dimeric¹⁰¹ and tetrameric¹⁰² β−2-microglobulin and the role of Cu(II)¹⁰³ and potential inhibitors¹⁰⁴ of amyloid formation of β−2-microglobulin. The downsides to specific amino acid footprinting are that reactions with the chemical reagent have the potential to induce structural change of the protein (especially since the reactions occur on much longer timescales than ms) and labeling occurs only on a few residues.^72,105

5. Chemical Cross-linking (XL)

Chemical cross-linking (XL) is another approach to study protein-protein interactions and conformational changes in proteins.^{7,20,21,22,106} The concept of XL involves adding to a protein solution a chemical reagent that links two functional groups on protein side chains.^{7,20,21,22,106} Typical chemical reagents contain two reactive groups that are separated by a spacer to control approximately the distance that can be crosslinked. After crosslinking, the protein solution is submitted to enzymatic digestion, followed by tandem MS.^{7,20,21,22,106} From the MS analysis, peptides that are linked together by the chemical reagent can be identified, revealing the distance (~ the length of the spacer) between two regions of a protein. The distances between peptide regions are then used as constraints in computational modeling to assign a protein tertiary structure^{7,20,21,22,106} (see Figure 14 for illustration of XL workflow).

Figure 14.

General workflow and analysis of XL-MS. (Reproduced with permission from reference 106.)

Two common classes of XL reagents are N-hydroxysuccinimide (NHS) esters and diazirines (Figure 15).¹⁰⁶ NHS esters readily react with amine groups (lysine side chains and the N-terminus of the protein) to form crosslinks. These esters also display low reactivity with hydroxy groups in serines, threonines, and tyrosines.¹⁰⁶ Diazirine works differently compared to NHS esters because it is photoactivated to a highly reactive species that react to give a higher yield of crosslinks than do NHS esters. Irradiation with UV-A light produces a highly reactive carbene that can link protein side chains.¹⁰⁶ Another difference between NHS esters and diazirines is that diazirines have broader reactivity,¹⁰⁶ yielding a higher diversity of linked peptides.

One advantage of XL that it can be performed at physiological pH to interrogate native protein structure and interactions¹⁰⁶ and to allow identification of regions associated with protein-protein interactions and sites of amyloid formation. A disadvantage of XL-MS is that it only gives separation distances within a protein or protein complex when used as an stand-alone approach.¹⁰⁶ XL-MS achieves higher utility when used in combination with a molecular modeling based on distance constraints.¹⁰⁶

III. APPLICATIONS OF MS-BASED APPROACHES

The following sections focus on using nMS, IM-MS, continuous HDX, pulsed HDX, FPOP, and XL-MS to study amyloid-forming proteins, mainly Aβ and aS. Some sections provide details on prototypal studies to illustrate each method, whereas others give a comprehensive list of studies that employ each method. The outcomes are examples of identifying Aβ and aS aggregation-prone regions and locating the binding sites of inhibitors that prevent amyloid aggregate formation.

A. Native MS

1. Prototypal study of Native MS of Aβ

Lermyte et al.¹⁰⁷ used top-down nMS to study Aβ₄₂ bonding for nine metal ions. nMS preserves the native state to determine the location of the binding sites (via comparison of top-down tandem MS for metal bound vs metal free Aβ₄₂ for Na⁺, K⁺, Mg²⁺, Ca²⁺, Mn²⁺, Co²⁺, Ni²⁺, Cu²⁺, and Fe³⁺. These metal ions are physiologically relevant, and these and future findings could lead to potential therapeutics for Alzheimer’s disease.¹⁰⁷

Electron capture dissociation (ECD), infrared multiphoton dissociation (IRMPD), and collision-induced dissociation (CID) alleviate concern whether the binding site remains constant during the dissociation process¹⁰⁷ because the three activation methods reveal similar binding sites for each metal ion. The binding to Aβ₄₂ is Cu²⁺> Co²⁺ and Ni²⁺. The native mass spectra for these three metal-protein interactions are similar, giving two noteworthy peaks. The first is Aβ₄₂ at 4+ (centered around m/z 1129), where it was shown that some of the Aβ₄₂ does not bind metal ions. The other major peak is 4+ metal-Aβ₄₂ for Co²⁺, Ni²⁺, and Cu²⁺-Aβ₄₂. (Note this 4+ charge corresponds to [Aβ₄₂ + 2H + metal]⁴⁺). These metal-protein ions were then selected for top-down tandem MS; the spectra revealed that the binding site is at two histidine residues (His6, His13) near the N-terminus (see Figure 16 for the region--all details are available).¹⁰⁷ The full native mass spectra showing metal-protein interactions are also omitted, but the spectra used for Cu²⁺, Co²⁺, and Ni²⁺.)

Figure 16.

Visual representation of the regions where the nine metal cations bind to Aβ₄₂. On the right is a chart of the metal cations with the specific binding region assigned by the residue number from Aβ₄₂. (Reproduced with permission from reference 107.)

The metals Mg²⁺, Ca²⁺, Mn²⁺, Na⁺, and K⁺ interact weakly with Aβ₄₂ and bind near the C-terminus. Mg²⁺, Ca²⁺, and Mn²⁺ bind at Leu34, whereas Na⁺ and K⁺ bind near Gly38 (Figure 16). Lastly, Fe³⁺ displays very different binding in the region between Ser8 and Gly25 (Figure 16).

B. IM-MS

1. Early IM-MS studies of Aβ

Most of the early work with IM-MS on Aβ was pioneered by Bowers, who reviewed his work in 2014.¹⁰⁸ This review covers the several ways IM-MS can be utilized in general and specifically for aggregating proteins. Bowers has studied aggregating proteins in two ways: at the peptide level and at the global level. This ability of IM-MS is highlighted in a more recent review by Bleiholder and Bowers.¹⁰⁹ At the peptide level, Bowers has studied the structure of Aβ^{110,111,112,113,114} and small molecule/peptide interactions with Aβ peptides.^{115,116,117,118,119,120} We will not discuss details but will instead focus on global protein aggregation.

In one global protein aggregation study, Bowers et al.¹²¹ used IM-MS to monitor aggregate species (e.g., dimers and tetramers) and showed that Aβ₄₀ forms monomers, dimers, and tetramers, but no higher order oligomer species. Aβ₄₂ is more complex, giving hexamers and dodecamers along with the monomers, dimers, and tetramers. Mutants of both Aβ₄₀ and Aβ₄₂ (F19P and M35M(O)) show similar patterns as Aβ₄₀, where the tetramer is the highest oligomeric species. Only wild-type Aβ₄₂ produced a detectable oligomeric species that was larger than a tetramer, suggesting that higher oligomers are unstable in nMS.

Generally, as oligomerization progresses, the relative CCS per monomeric unit decreases (e.g., the tetramer has a larger CCS value than the dimer, but each monomeric unit contributes a higher CCS in the dimer compared to in the tetramer). The drift peak width was significantly narrow for Aβ₄₂ dodecamer and hexamer, suggesting mainly a single structure or conformation for the dodecamer and hexamer.

The investigators developed a model to illustrate the shape of the oligomeric species in which each monomeric unit is a sphere, and oligomers are made up of multiple spheres (Figure 17). From the model, they concluded the pathway to fibril formation differs for Aβ₄₀ and Aβ₄₂. Aβ₄₀ bypasses the hexamer and dodecamer species to form amyloid fibrils, whereas Aβ₄₂ forms these higher oligomeric species to create amyloid fibrils. Further, the Aβ₄₂ tetramer has a larger CCS than the Aβ₄₀ tetramer, indicating the Aβ₄₂ tetramer shape is more open and accommodates more monomeric additions, whereas the Aβ₄₀ tetramer is more closed and cannot bind as many monomers, possibly explaining that Aβ₄₂ creates amyloid fibrils more quickly than Aβ₄₀.

Figure 17.

The mechanism of oligomerization and fibril formation for Aβ₄₀ and Aβ₄₂. (Reproduced with permission from reference 121.)

Bowers and coworkers also utilized IM-MS to study mutant forms of Aβ, pinpointing changes in the aggregation states of the mutants.^{122,123,124,125,126,127} Interactions between Aβ and amyloid fibril inhibitors, such as PADK (Z-Phe-Ala-diazomethylketone),¹²⁸ molecular tweezer (CLR01),¹²⁹ and newly synthesized small molecules ([AC0107]¹³⁰ and ML¹³¹), can be characterized by IM-MS, and Aβ₄₀ inhibits Aβ₄₂ aggregation and fibril formation.¹³² Although Bowers has been a leader in IM-MS Aβ studies, others are following his lead to employ IM-MS to study Aβ.

2. Application studies with IM-MS of Aβ

Many studies address the capability of IM-MS to study the aggregation of Aβ. Kloniecki et al.¹³³ used IM-MS to discover that many oligomers of Aβ₄₀ have two conformations (a compact and extended conformation), corresponding to on-vs-off pathways. Sitkiewicz et al.¹³⁴ also found that many oligomers of Aβ₄₀ exist in two conformations and that metal-ion binding stabilizes the compact form. The investigators found a key salt bridge between D23 and K28 that stabilizes the compact conformation, and residue G25 was deemed necessary for the formation of the extended conformation. Iurascu et al.¹³⁵ also showed two conformational states for Aβ₄₀ during aggregation, and residue Met35 can form oxidized products. Beck et al.,¹³⁶ by researching the effect of metals, free organic radicals, and small compounds with Aβ, found the reactivity and a reduced toxicity of Aβ with these molecules. Choi et al.¹³⁷ studied the interaction of human serum albumin (HSA) with Aβ showing HSA delays Aβ aggregation, transports Aβ across a cell membrane, and acts as a metal chelator for Cu²⁺ and Zn²⁺. They were also able to determine that HSA binds one monomeric Aβ in its major groove.

Pujol-Pina et al.¹³⁸ demonstrated that IM-MS provides accurate information on the structure of Aβ oligomers (such as dimers and trimers), whereas SDS-PAGE can potentially give inaccurate results. SDS-PAGE was employed on a mixture of isolated dimers and trimers to show pentamers and hexamers, which were proposed to result from artifactual oligomerization due to SDS-PAGE. Österlund et al.,¹³⁹ using detergent micelles to study Aβ₄₀ and Aβ₄₂, showed that ion mobility is consistent with Aβ₄₂ forming hexamers in the shape of β-barrel, whereas Aβ₄₀ produces lower amounts of oligomers. The authors speculate that the β-barrel structure affects the toxicity of Aβ₄₂ oligomers in AD. Lu et al.¹⁴⁰ used carbene footprinting in combination with IM-MS to elucidate inhibitor-Aβ peptide interactions sites, obtaining sub residue-level resolution for aggregation sites (discussed more in the “other irreversible labeling methods” section). Irie et al.¹⁴¹ synthesized three mutant dimer models of Aβ₄₀ and used IM-MS to investigate the high oligomeric states for each model. The investigators also found through using these mutations that the hydrophobic core at the C-terminus initiates formation of toxic oligomers via MTT assays, instead of the β-sheet formation within the middle region of Aβ. Sitkiewicz et al.¹⁴² cross-linked Aβ dimers and trimers with tyrosine cross-links and determined that the cross-linked structures are more compact and preferentially form amyloid fibrils.

Fulcher et al.¹⁴³ used a specific high-field asymmetric waveform ion mobility spectrometry (FAIMS) on whole tissue samples from Alzheimer’s patients and observed several proteoforms of intact Aβ (including Aβ₄₂) without employing fractionation or purification. These IM-MS results greatly expand our understanding of Aβ aggregation and show multiple conformations of Aβ oligomers with key residues responsible for particular conformations. Future research will elucidate on-and-off pathway oligomers, toxicity of each species, and ways to inhibit fibril formation.

3. Prototypal study with IM-MS of Aβ with inhibitors

IM-MS was employed by Young et al.⁵ to investigate potential Aβ₄₀ aggregation inhibitors, including 20 small-molecule inhibitors (e.g., resveratrol and chloronaphthoquinine-tryptophan (Cl-NQTrp)) (Figure 18). The small molecules were added to monomeric Aβ₄₀, and the binding interactions were evaluated via nMS and drift time spectra. Only two showed a positive inhibition; one (compound 3) was derived from resveratrol, whereas the other (compound 16) was derived from Cl-NQTrp. nMS and drift time show the small-molecule binding to the 3+ monomer leads to a binomial distribution of bound species (Figure 19). nMS and TEM show that bonding of compounds 3 and 16 eliminates higher order oligomers and inhibits amyloid fibril formation. Besides discovering two new small molecule inhibitors of Aβ₄₀, Young et al. demonstrated the ability of IM-MS to screen small molecules and determine their mode of inhibition, showing a possible and perhaps important application for therapeutics development.

Figure 18.

The chemical structure of the two positive inhibitor compounds along with resveratrol and Cl-NQTrp. (Reproduced with permission from reference 5.)

Figure 19.

Mass spectrum, drift time, and TEM of Aβ with compound 3 and 16 that inhibit Aβ₄₀ amyloid fibril formation. In the mass spectrum, 1 = monomer, 2 = dimer, 1 + L = monomer with one small inhibitor attached, and 1 + 2L= monomer with two inhibitors attached. In the drift spectrum, the yellow dots correspond to small molecules binding the monomer. (Reproduced with permission from reference 5.)

4. Other IM-MS applications of Aβ with inhibitors

Many compounds have undergone similar analysis to show inhibition of Aβ₄₀ amyloid fibril formation. Inhibitors include small molecules (sclerotiorin,¹⁴⁴ 3-sulfopropanoic acid (3-SPA),¹⁴⁵ tramiprosate,¹⁴⁶ EGCG,^37,147,148 and newly designed compounds¹⁴⁹), peptides (Zn-porphyrin-peptide conjugate,¹⁵⁰ dipeptide Phe-Leu (FL),¹⁵¹ SEN304,¹⁵² OR02,¹⁵³ and newly designed peptides¹⁵⁴), chaperones (C-Phycocyanin¹⁵⁵), molecular tweezer (CLR01¹⁵³), and antioxidants (crocus-derived compounds¹⁵⁶). Sanders et al.¹⁵⁷ studied the effects of EGCG and resveratrol on Aβ and aS aggregation in a phospholipid bilayer, revealing that membranes decrease the efficacy of these inhibitors. The binding EGCG and resveratrol to Aβ and aS is reduced in the presence of a lipid membrane system, which investigators attributed to the partitioning environment of the membrane. Meinen et al.¹⁵⁸ discovered that SERF accelerates the early stages of fibril formation, thus enhancing aggregation.

5. Applications with IM-MS of aS

IM-MS has also been utilized to understand the conformational changes of aS during aggregation. Vlad et al.¹⁵⁹ by studying the aggregation of aS with IM-MS, discovered a highly aggregating C-terminal fragment (residues 72–140). They¹⁶⁰ later discovered different conformational states of oligomeric species of both Aβ and aS and found many truncated and proteolytic products corresponding to heterogeneous oligomers. Illes-Toth et al.¹⁶¹ studied oligomeric aS to show that lower-order oligomers (e.g., dimers and trimers) are very heterogenous in structure; however, higher-order oligomers were more homogenous in structure, forming ring-like configurations. The investigators hypothesize that the ring-like configuration may contribute to intracellular seeding of aS and transfer pathology between cells.

Bernstein et al.¹⁶² found more compact structures at lower pH, suggesting that pH changes the aggregation of aS. Frimpong et al.¹⁶³ found aS to form four conformational states upon changes in pH and alcohol content, where a highly compact aS structure is seen at all pHs but was favored in acid conditions (agreeing with the findings from Bernstein et al.¹⁶²). They also discovered a highly structured dimer in acidic solutions, which is believed to be present at the beginning of aS aggregation. Phillips et al.,¹⁶⁴ using XL-IM-MS to preserve the structure of aS oligomers in solution instead of the gas phase, demonstrated that compact, extended, and unfolded conformations are involved in aS aggregation. The investigators also studied the effect of pH on aS aggregation and found that the middle region of aS is structured, whereas the N and C termini are flexible, showing slightly different results than those of Vlad et al.¹⁵⁹

Beveridge et al.¹⁶⁵ showed that aS has many charge states and multiple conformations associated with oligomeric species, in comparison with apolipoprotein C-II (another IDP) that only has a few charge states. (This work also employed continuous HDX-MS to show that HDX occurs rapidly with aS because it is an IDP.) Beveridge et al.¹⁶⁶ developed a model using ESI-IM-MS on 20 proteins (including aS) to distinguish the most compact and most extended conformations of a protein; the model is accurate in evaluating ordered and disordered conformations. Lermyte et al.¹⁶⁷ showed two distinct protomers of aS that are detectable by IM-MS and show different properties depending on solution conditions. These studies demonstrate the importance of solution/buffer conditions on the aggregation of aS, as well as the shape of aS aggregate species (compact vs extended) owing to the surrounding environment. This research brings more understanding the of aS aggregation, and the efforts may assist the design of drugs to inhibit amyloid fibril formation.

6. Application studies with IM-MS of aS and inhibitors

Like studies of Aβ, many IM-MS investigations have focused on discovering small molecule inhibitors of aS amyloid formation. Liu et al.^168,169 showed that EGCG inhibits A53T aS (the disease-associated mutation) amyloid formation, and gallic acid (GA) binds to the extended conformation of aS and inhibits aggregation. They¹⁷⁰ later discussed the ability of IM-MS to screen inhibitors for A53T aS amyloid formation. This type of screening allows for the identification of inhibitors that work at the early stages of aggregation and can potentially prevent toxic species from appearing. Illes-Toth et al.,¹⁷¹ by studying the interactions between dopamine (DA) and aS, discovered that DA inhibits amyloid formation and binds only the extended conformation of aS. Konijnenberg et al.¹⁷² found that EGCG binds to a compact conformation, whereas DA binds to an extended conformation of aS (agreeing with Illes-Toth et al.¹⁷¹). IM-MS also shows that these inhibitors exert different structural effects on aS upon binding, where EGCG binding causes aS compaction and DA binding leads to extension of aS.

Ponzini et al.¹⁷³ by using EGCG and DA to oxidize aS, compared the oxidized versus wild type aS conformational transitions, showing lower amounts of secondary structure for oxidized aS. Reduced secondary structure of oxidized aS makes it less prone to fibrilization in comparison to wild type aS. Jovcevski et al.¹⁷⁴ showed that 2′,3′,4′-trihydroxyflavone is effective at preventing aS fibril elongation and restores the aS monomer conformation during fibril formation. Das et al.¹⁷⁵ monitored A53T aS aggregation in the presence of seven compounds (2′,3′,4′-trihydroxyflavone, honokiol, punicalagin, myricetin, transilitin, and two novel compounds) and showed that each compound inhibits amyloid formation. After 48 h incubation with each compound, the native state of A53T aS was persevered, showing these inhibitors operate during the beginning stages of aggregation. Grabenauer et al.¹⁷⁶ showed that spermine promotes formation of compact aS and accelerates the rate of aggregation.

In addition to this use in inhibitor research, IM-MS showed that metal-binding and membrane interactions affect the structure of aS. Mason et al.¹⁷⁷ demonstrated that wild type aS and H50Q aS are unable to bind copper when acetylation occurs at the N-terminus, and that lack of binding prevents copper from inducing aggregation. Wongkongkathep¹⁷⁸ using top-down nMS to locate the binding sites of cobalt and manganese, demonstrated that metal binding induces compact conformations of aS, possibly accelerating aggregation. Han et al.¹⁷⁹ determined that multiple calcium ions can bind aS causing an acceleration of amyloid fibril formation. Investigators hypothesize that Ca²⁺ binds to aS at the C-terminal end, which provokes a conformational change that exposes the NAC region, causing aggregation to occur there. Moon et al.¹⁸⁰ studied the structure of aS when bound to alkali, alkaline earth, transition, and other metal ions and found that the charge of the metal ion and the binding site affect the structure, often leading to more compact aS. These and other studies indicate a trend that metal binding induces a conformational compaction on aS. Along with metal-binding, aS interactions with membranes cause structural changes that are revealed by IM-MS. Moon et al.¹⁸¹ continued research with calcium ions and aS, now in a micelle environment, and attributed unique conformations that form with aS binding to detergents. Investigators found that compact aS conformations preferentially bound to the detergents, and binding causes aS to be compacted further. Lee et al.¹⁸² using IM-MS, discovered that the polarity of the membrane surface influences the helix folding of aS, and the hydrophobic NAC region penetrates into neutral membranes to form favorable hydrophobic interactions, potentially preventing aggregation.

In summary, IM-MS has been a productive contributor to understanding aggregation and of the multiple conformational states taken by both Aβ and aS. IM-MS appears to be useful in understanding interactions with inhibitors that prevent aggregation of Aβ and aS in vitro. For example, when adding a membrane-like environment or metal complexes, Aβ and aS can take on different conformations, and the efficacy of the inhibitors decrease. These findings may contribute to the development of novel and more potent inhibitors for amyloid formation, especially in a membrane environment (or more in vivo-like system), but it remains to be seen if inhibitors identified by IM-MS are effective in vivo.

C. Continuous HDX

1. Early Applications with Continuous HDX

One of the main contributors to early continuous HDX-MS of amyloid proteins was Wetzel et al., who focused on Aβ₄₀ aggregation. Their early continuous HDX-MS studies reveal that the rate of HDX varies with the regions of Aβ₄₀ oligomeric structures.¹⁸³ Some protofibrils are approximately 40% protected from HDX, whereas amyloid fibrils exhibit 60% protection.¹⁸⁴ To pinpoint regions of Aβ₄₀ that contribute to amyloid formation, they used on-line digestion to illustrate that the C- and N-termini undergo high exchange, whereas the middle region undergoes less exchange.¹⁸⁵ These data imply that the middle region is first to become structured and that structure expands once amyloid fibrils form. Wetzel and coworkers¹⁸⁶ also studied the interaction of calmidazolium chloride (CDC) with Aβ₄₀, showing that it assists Aβ₄₀ to exist in a protofibril-like state, instead of forming amyloid fibrils. Their early results on methodology and outcomes of these continuous HDX-MS studies were reviewed,¹⁸⁷ along with their findings with Aβ₄₀.¹⁸⁸ Additional work from Wetzel and coworkers leads to a new method for back exchange correction on continuous HDX-MS of Aβ₄₀,¹⁸⁹ as well as a triaxial probe for efficient on-line digestion of Aβ₄₀.¹⁹⁰

A more recent study revealed five distinct on-pathway oligomeric structures of Aβ₄₀.¹⁹¹ The investigators found that cross-β structure and side chain packing outside the β-sheet contribute to higher stability of the oligomeric species. They studied the aggregation of Aβ₄₂ and Aβ₄₃¹⁹² and report that Aβ₄₃ aggregates more slowly than Aβ₄₂, and Aβ₄₂ aggregates more quickly alone than as a mixture of both Aβ₄₂ and Aβ₄₃. They also found that Aβ₄₂ fibrils act as seeds for Aβ₄₃ aggregation, whereas Aβ₄₃ fibrils are not seeds for Aβ₄₂ aggregation. Wetzel led the way in applying continuous HDX-MS to amyloids, and others followed and applied this approach to discover different structural properties of Aβ aggregating species, as covered in the next section.

2. Prototypal studies with continuous HDX of Aβ

Pan et al.¹⁹³ used millisecond-continuous HDX-MS to study the formation of small oligomers of Aβ₄₀. Basic conditions (pH 9.3) were used to induce aggregation of Aβ₄₀, and the results showed a rapid conversion from monomers to oligomers via EX1-like exchange (Figure 20). This EX1-like exchange clearly displays two populations of species, where one is a monomer (corresponding to the peak at the higher m/z due to more HDX), and the other is a small oligomeric species (corresponding to the peak of lower m/z that shows less HDX). The small oligomeric species is likely to be heterogenous, consisting of dimers and tetramers. Using electron capture dissociation to obtain MS/MS data with minimal H/D scrambling, the investigators found an Aβ₄₀ deuteration pattern for residues L17-M35 to be consistent with a β–turn−β motif (Figure 21). The N-terminus, due to hydrogen bonding, stabilizes the structure, whereas the C-terminus (residues 35–40) displays low protection. Size exclusion chromatography reveals that four monomers form a tetramer. That, combined with previous results from Bowers et al.¹²¹ (Figure 17), suggest the tetramer forms in a β-barrel arrangement (Figure 21) and gives perspective on the structure and formation of early oligomeric species of Aβ₄₀.

Figure 20.

Evidence for EX1-like HDX from the mass distribution of Aβ₄₀ of 5+ after HDX labeling, using times of (A) 50 ms, (B) 100 ms, (C) 500 ms, (D) 1 s, and (E) 8 s. (Reprinted (adapted) with permission from reference 193. Copyright 2012 American Chemical Society.)

Figure 21.

(A) Illustration of the β−loop−β secondary structure for individual Aβ₄₀ monomers (residues 1−16 and 36−40 are omitted). (B) Illustration of an Aβ₄₀ tetramer, showing monomers stacked on top of one another. (C) Illustration of a β-barrel tetramer, a structure supported by other literature. (Reprinted (adapted) with permission from reference 193. Copyright 2012 American Chemical Society.)

3. Application studies with continuous HDX of Aβ

Several others also utilized continuous HDX-MS to study Aβ. Wang et al.,¹⁹⁴ by studying the effect of solvent conditions on Aβ structure, found early oligomeric formation is solvent dependent. Investigators also determined that the N-terminus is not protected in HDX during aggregation, whereas the C-terminus and middle region of Aβ undergo protection, owing to oligomeric formation in these regions. Kraus et al.¹⁹⁵ demonstrated that the middle region of Aβ₄₂ forms a β-sheet structured oligomer because it exhibits a lower exchange rate, signifying this region is responsible for aggregation. Sanchez et al.¹⁹⁶ showed that Aβ monomers dissociate and reassociate on the amyloid fibril, suggesting implications this could have on toxicity. They showed that the rate constant for dissociation off the amyloid fibril is much greater for Aβ₄₀ compared to Aβ₄₂. Zhang et al.¹⁹⁷ determined the solvent accessibility of multiple oligomeric states of Aβ₄₀ at the peptide level, showing residues 20–34 and 35–40 initiate aggregation in low order oligomers (consistent with the work of Wang et al.¹⁹⁴). Qi et al.¹⁹⁸ measured the HDX of Aβ₄₀ oligomeric species without any purification that could change the distribution of aggregates. These investigators also hypothesize that oligomeric species are the most toxic, comparing their result with those of Patel et al.,¹⁹⁹ who used SY5Y cells to asses neurotoxicity of Aβ₄₀ over the course of aggregation. The same experimental conditions were used to show that oligomeric species at the same aggregation time have the lowest cell viability.

Stelzer et al.²⁰⁰ demonstrated that the mutation T43I on APP affects the cleavage to form Aβ and disrupts the hinge region that is key for dimerization. Pan et al.²⁰¹ found similar deuterium uptake with oligomeric and amyloid fibril species of Aβ₄₂, and also showed that oligomers and fibril species have a similar β-loop-β secondary structure motif (similar to findings from their earlier work¹⁹³); however, it remains unknown if this motif is in the toxic aggregate. Przygonska et al.²⁰² employed IM along with HDX-MS to reveal that the N-terminal side-chains stabilize the oligomers of Aβ₄₀, suggesting the N-terminus may be necessary to form larger oligomers and fibrils. The investigators speculate that the N-terminal residues play key roles in the toxicity of Aβ₄₀ aggregate.

Other continuous HDX-MS experiments show that membranes affect the aggregation of Aβ, revealing that Aβ may take on unique conformations in their presence.^203,204 These studies suggest that the middle and C-terminal regions of Aβ are responsible for the formation of oligomeric species and amyloid fibrils, whereas the N-terminus assists β-sheet structure formation in these regions. Further research with Aβ in different environments, including the presence of membranes and in the presence of various buffers, will bring further understanding of the importance of different conformations and oligomers for Aβ aggregation. Early studies suggest the oligomeric species seem to be toxic. Continuing research on the formation of oligomeric species should bring insights that will assist drug design to inhibit aggregation and formation of toxic species of Aβ.

4. Prototypal study with continuous HDX of aS

Continuous HDX-MS was implemented by Mysling et al.²⁰⁵ to delineate regions of HOS in aS oligomers. After aggregation, monomers and oligomers were separated by gel filtration, and continuous HDX-MS was utilized to probe the separated aggregated species. Peptides from oligomeric species displayed either strong protection (a clear bimodal distribution with no overlapping peaks), moderate protection (a bimodal distribution with overlapping peaks), or no protection (a unimodal peak distribution) (Figure 22). Mysling et al. used these data to create a heat map that illustrates areas of protection for aS oligomers (Figure 23). The heat map reveals that the C-terminus of aS oligomers (residues 90–140) is unstructured, showing rapid HDX at all times. Three regions display strong protection: 4–17, 39–54, and 70–89, located near the N-terminus and NAC regions, suggesting they play a large role in oligomer formation. Two regions, comprised of residues 18–38 and 55–76, display moderate protection. These peptides come from regions located in the N-terminus and NAC region, and investigators suggest that these regions play some role in oligomer formation. There appears to be a consensus that the NAC region forms structure during aggregation, as indicated in this work and the IM-MS studies with Vlad et al.¹⁵⁹ and Phillips et al.¹⁶⁴ However, these three studies show some discrepancies the analysis of the N- and C-termini.

Figure 22.

Examples of strong, moderate, and no protection as determined by MS HDX data. Peptide 5–17 shows a bimodal distribution with no overlapping peaks, a characteristic of a strongly protected peptide. Peptide 18–38 shows a bimodal distribution with overlapping peaks, representing a moderately protected peptide. Peptide 125–140 shows a unimodal peak, showing a peptide with no protection. (Reprinted (adapted) with permission from reference 205. Copyright 2013 American Chemical Society.)

Figure 23.

A heat map of the oligomeric aS species showing regions of high, medium, and no protection on aS oligomers and delineating those regions for which aS oligomer is undergoing HOS change and forming aggregates. (Reprinted (adapted) with permission from reference 205. Copyright 2013 American Chemical Society.)

5. Application studies with continuous HDX of aS

Several investigators utilized continuous HDX-MS to study the structure of aS under different conditions. Landureau et al.²⁰⁶ studied the exchange of three oligomeric forms of aS, finding discrepancies in the N-terminal region. Those investigators suggest that three polymorphs of aS have different binding to other ligands and proteins. Stephens et al.²⁰⁷ showed that lyophilizing and freezing aS affects the structure of the monomeric form, revealing structure at the C-terminus. In an assembled amyloid, Del Mar et al.²⁰⁸ showed that the N- and C-termini undergo high HDX, whereas the middle (residues 39–101) is strongly protected from exchange, suggesting the middle segment induces aggregation and contains a chain fold the induces protection.

Paslawski et al.²⁰⁹ studied wild-type aS, along with three mutants, reporting two different oligomers for each aS species. These two oligomers continue on to make an amyloid fibril or an amorphous cluster. Stephens et al.²¹⁰ examined the effect of calcium binding to aS to show that the N-terminus becomes more exposed owing to C-terminus binding to the calcium, causing aS to be more susceptible to aggregation, an observation that agrees with the findings of Han et al.¹⁷⁹ Potential therapeutic designs could target the exposed N-terminal region, in hopes of preventing further aggregation.

Lee et al.¹⁸² used continuous HDX-MS (and IM-MS) to study the interactions between aS and lipid vesicles, attributing α-helical formation of aS monomers to the presence of lipids. Oganesyan et al.²¹¹ utilized nanodiscs with different headgroups to examine the structure of aS, and found that HDX is affected by the charge of the headgroup (zwitterionic vs negative). DMPC nanodiscs had little effect on aS conformation, whereas POPG nanodiscs change the structure of aS, potentially accelerating aggregation. Together, these studies bring understanding to the structure of aS, but more work is needed to understand fully the dynamics and conformational states of aS during aggregation and to understand potential aggregating environments, such as those that include interaction with a nanodisc or with various salts.

D. Pulsed HDX

1. Prototypal studies with pulsed HDX of Aβ

Continuous HDX of an aggregating protein may be difficult to interpret because the outcome is a convolution of the kinetics of HDX and of aggregation. Pulsed HDX is an option that reports principally protein aggregation kinetics. Zhang et al.⁴⁰ used pulsed HDX to determine the various aggregation stages of both Aβ₄₀ and Aβ₄₂ at the peptide level. For Aβ₄₀, each region showed a constant protection level for all times (1 min to 48 h), indicating no conformational changes or aggregation. For Aβ₄₂, however, each constituent peptide showed an increase in protection that can be modeled by a modified sigmoidal curve (Figure 24 solid lines). Although the HDX-MS data display a bimodal distribution, the protection values are centroids for the whole distribution. For a better representation of the protection percentage, binomial peaks should be compared instead of the centroid of all forms. Nevertheless, we can see the first phase of aggregation shows a rapid increase in the protection of Aβ₄₂, corresponding to monomer/small oligomer formation. Then a first plateau occurs where the protection is constant for a brief time period, signifying that larger oligomers are forming. A second rapid increase in protection ensues, thought to be larger oligomers binding together. Lastly, a second plateau is reached, signifying that the system has achieved near equilibrium among all species. From the overall centroids, the authors concluded (like others have such as Pan et al.¹⁹³, Wang et al.¹⁹⁴, Kraus et al.¹⁹⁵, and Zhang et al.¹⁹⁷) that the middle region of Aβ₄₂ aggregates first and displays the highest protection increase, suggesting this region plays a dominant role in aggregation.

Figure 24.

Pulsed HDX results for three peptic peptides: A) peptide 1–19, B) peptide 20–35, and C) peptide 36–42. The solid line represents Aβ₄₂ in the absence of Cu²⁺, and the dashed line represents Aβ₄₂ in the presence of Cu²⁺. (Reproduced with permission from reference 40.)

The effect of Cu²⁺ on Aβ₄₂ aggregation (Figure 24 dashed lines) shows that Cu²⁺ stabilizes the soluble oligomers and prevents amyloid fibril formation. This demonstrates that pulsed HDX-MS can determine regions of aggregation and assess potential inhibitors.

Serra-Vidal et al.⁶⁸ examined the relationship between neuronal cell death and the state of aggregation for Aβ₄₀ and Aβ₄₂. A global MS analysis of both Aβ₄₀ and Aβ₄₂ revealed three distinct populations: low protected monomers/oligomers, protected oligomers, and amyloid fibrils. The low protected monomers/oligomers are found at early times of aggregation, followed by the protected oligomers, and lastly amyloid fibrils. The data also show that amyloid fibrils form more rapidly for Aβ₄₂ (11 days) than for Aβ₄₀ (24 days).

To determine the neurotoxicity of each oligomeric species, the same samples used for pulsed HDX-MS were placed with hippocampal neuronal cultures containing 30,000 neuron cells per sample. The level of neuronal survival was measured using an MTT assay (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide assay). The percent neuronal viability was graphed over time (Figure 25). For both for Aβ₄₀ and Aβ₄₂, the neurons show a higher survival early on, but the viability decreases near the middle of the time study, and then increases to approximately the original viability at later time points. This strongly suggests that protected oligomers formed prior to large species are the most toxic to the neuron cells. These findings match results from Irie et al.¹⁴¹, who also found oligomeric species to be the most toxic.

Figure 25.

The percentage of neuronal viability over time for (A) Aβ₄₀ and (B) Aβ₄₂. When these data are combined with the pulsed HDX-MS results, the percent viability is found to be lowest for the intermediate, protected oligomer species, indicating that protected oligomers are the most toxic species for neuron cells during Aβ₄₀ and Aβ₄₂ aggregation. (Reprinted (adapted) with permission from reference 68. Copyright 2014 American Chemical Society.)

2. Prototypal study with pulsed HDX of aS

Illes-Toth et al.²⁷ used pulsed HDX-MS to investigate the various aggregation stages of aS and gather regional information involved in amyloid formation. Data for the peptides from the NAC region (residues 55–76) show a strong increase in protection as aggregation occurs (Figure 26). The HDX appears as a bimodal distribution (as seen with Mysling et al.²⁰⁵ and Del Mar et al.²⁰⁸, and with Zhang et al.⁴⁰ for pulsed HDX on Aβ) that represents two main species in solution: species 1 (an aggregated species) and species 2 (possibly native). As t_agg increases, the concentration of species 1 increases with respect to species 2, indicating formation of large oligomers. The NAC region is the dominant region contributing to aggregate formation, whereas the N-terminus plays a small role, and the C-terminus plays a nondetectable part in amyloid formation. These results on the NAC region agree with continuous HDX work from Mysling et al.²⁰⁵ and Del Mar et al.²⁰⁸ and IM-MS work from Vlad et al.¹⁵⁹ and Phillips et al.¹⁶⁴

Figure 26.

Pulsed HDX-MS over aggregation time with aS. Species 1 (the aggregated species in the red column) increases with aggregation time relative to species 2 (the native state of aS in the green column). (Reprinted (adapted) with permission from reference 27. Copyright 2018 American Chemical Society.)

The kinetics of formation of species 1 and the disappearance of species 2 show that the relative concentrations of the two species are represented by typical sigmoidal curves that show a standard amyloid lag, growth, and saturation phases.²⁷

3. Application studies with pulsed HDX of Aβ and aS

Additional research employed pulsed HDX-MS to understand the aggregation of Aβ and aS. Illes-Toth et al.²¹² studied several mutant forms of Aβ₄₂ (i.e., Arctic (E22G), Iowa (D23N), Italian (E22K), Dutch (E22Q), Japanese (D7N), English (D6R), and Flemish (A21G) variants). The mutants all show varying bimodal distributions as aggregation proceeds. The bimodal distributions seen in the data indicate a protected and exposed species, and further analysis reveals that the central and C-terminus are responsible for the increase in protection. Cho et al.²¹³ demonstrated the ability of a compound (L1, see Figure 27) to reduce neurotoxicity on neuroblastoma N2A cells from Cu²⁺-Aβ₄₂ species, showing different oligomers forming between L1- Cu²⁺-Aβ₄₂ and Cu²⁺-Aβ₄₂. Kumar et al.²¹⁴ used pulsed HDX-MS to show that aS forms a ribbon-like fibril in the presence of ThT, reducing the heterogeneity of aS fibrils. Pulsed HDX-MS may be a powerful approach to elucidate regions of HOS and effects of inhibitors. Further research with this approach has the potential to bring new understanding of aggregation and to assist discovery of drug/inhibitor therapies.

Figure 27.

Structure of compound L1 that reduces neurotoxicity during Aβ₄₂ aggregation.

E. Fast Photochemical Oxidation of Proteins (FPOP)

FPOP paired with MS was used by Li et al.²¹⁵ to follow aggregation and solvent accessibility of Aβ₄₂ at the global, peptide, and residue levels for some amino acids. At the global level, Aβ₄₂ initially undergoes extensive modification because the protein is intrinsically disordered (Figure 28A). As oligomers and higher aggregates form, the modification on Aβ₄₂ decreases (Figure 28B, C). Toward the final stages of aggregation, modification of Aβ₄₂ decreases to a very low level (Figure 28D), signifying that most residue side chains are no longer exposed, and production of amyloid fibrils is underway. A plot of percent modification over time (Figure 28E) shows more detail than ThT analysis; multiple aggregation transitions (Scheme 4) are seen and their assignments are speculative. The A→B transition likely corresponds to Aβ₄₂ monomers forming dimers/small oligomers that exhibit reduced solvent accessibility and correspondingly a decrease in modification. The B→C transition may involve nucleation of the dimers/small oligomers to create paranuclei-like “seeds” for future aggregation, as seen as a plateau in modification. The C→D transition shows a decrease in the modification likely because the “seeds” have reached a large enough concentration to undergo an autocatalytic reaction to form larger oligomers. The D→E transition may represent another nucleation stage, again reaching a plateau in percent modified. The E→F transition may be formation of amyloid fibrils from the nucleated large oligomers, decreasing the solvent accessibility of side chains and the percent modification. The relative concentration of each Aβ₄₂ aggregate species over time (Figure 28F) provides kinetics of formation of monomers, dimers, and larger oligomers.

Figure 28.

FPOP and kinetic modeling describe the aggregation of Aβ₄₂ at the intact protein, peptide, and amino-acid residue levels. (A-D) Mass spectra displaying the extent of modification of the full protein (5+ charge state) with respect to time. (E) Kinetic model of the percent modification on the full protein at the 5+ charge state with respect to time. The points on the graph represent raw data, whereas the solid line represents a model fit based on two autocatalytic reactions. (F) The concentration of each Aβ₄₂ aggregate species (M = monomer; D = dimer/small oligomers; D* = protofibrils; and D** = amyloid fibril) with respect to time. (G) Aggregation curve for the peptide 1–15 representing N-terminus. (H) Aggregation curve for the peptide 28–42 representing C-terminus. (I) Aggregation curve for the residue H6. (J) Aggregation curve for residues F19/F20. (Reprinted (adapted) with permission from reference 70. Copyright 2018 American Chemical Society.)

Scheme 4.

Proposed kinetic scheme for Aβ₄₂ aggregation as a function of time. FPOP is able to provide additional information on the intermediate species from stage D to F that is not seen by other analytical approaches. (Reprinted (adapted) with permission from reference 215. Copyright 2016 American Chemical Society.)

Aβ₄₂ aggregation at the regional level was determined for three main peptides formed in digestion. The modification for the N-terminal peptide (residues 1–15) stayed relatively high for all time points (Figure 28G), demonstrating that this region does not participate in aggregation. The peptides from the mid region of Aβ₄₂ (residues 16–27) and the C-terminus (residues 28–42) show large increases in protection over aggregation time (Figure 28H), indicating these regions are responsible for the aggregation of Aβ₄₂ (as also seen in pulsed HDX work from Zhang et al.⁴⁰ and continuous HDX work from Pan et al.¹⁹³, Wang et al.¹⁹⁴, Kraus et al.¹⁹⁵, and Zhang et al.¹⁹⁷).

Additionally, the modification at the amino acid level was characterized for some residues including H6 and F19/20. H6 displays nearly constant FPOP modification over time (Figure 28I), showing that it is not involved in aggregation and remains solvent accessible. Because F19/F20 undergo a large decrease in modification during aggregation (Figure 28J), these residues are in part responsible for the formation of aggregates.⁷⁰ Overall, this study displays the ability of FPOP to deliver vital information on the aggregation of Aβ₄₂, and similar experimentation can monitor small molecule binding. In comparison with pulsed HDX from Zhang et al.,⁴⁰ FPOP reveals unprecedented additional stages of aggregation (transition from D→F) and the capability to achieve amino acid residue resolution is an advantage of FPOP over HDX-MS. Furthermore, FPOP can be expanded to follow aggregation in more complex milieu, including cells. We are now employing FPOP for studying the effects of inhibitors on aggregation.

F. Other Irreversible Labeling Methods

X-ray footprinting has been utilized by Klinger et al.²¹⁶ to examine the solvent accessible side chains of prefibrillar and fibrillar Aβ₄₀. Investigators use protection factors, where higher protection factors correspond to less footprinting, to display the change in hydroxyl radical labeling for 16 residues during aggregation. Their footprinting is consistent with some core-filament solid-state NMR models of fibrillar Aβ₄₀, and the labeling indicates two ordered segments (residues 1–8 and 25–29) that are undefined by solid-state NMR. The authors find that residues in the middle and C-terminal region increase in protection as fibrils are formed (similar to the conclusion of Zhang et al.⁴⁰, Pan et al.¹⁹³, Wang et al.¹⁹⁴, Kraus et al.¹⁹⁵, and Zhang et al.¹⁹⁷, where the middle and C-terminal regions are aggregation prone).

Lu et al.¹⁴⁰ established a sub residue footprinting strategy by employing carbene labeling that can label different atoms in identical residues, producing peptides where the same residue is labeled, but the sites of the label differ. The sub residue footprinted isomeric peptides can be identified using LC-IM-MS. For one portion of this study, a segment of Aβ (¹⁶KLVFFA²¹) was labeled and yielded three isomers with modification on Lys16. The investigators identified the three sites, which are at the N-terminal, the −CH₂- side chain, and ε-NH₂. They then discovered that the N-terminal and ε-NH₂ increase in modification during aggregation and are therefore more exposed during aggregation, while the −CH₂- side chain stays relatively constant. These results are consistent with X-ray crystallography, where a steric zipper is formed between the hydrophobic residues 17–21 and exposes the ε-NH₂ on Lys16. This work displays the effectiveness of sub residue footprinting and its ability to determine how specific residues are affected as a function of aggregation.

Metal-catalyzed oxidation was employed by Kowalik-Jankowska et al.²¹⁷ to study the oxidation of N-terminal peptides of aS (containing two methionine residues) with Cu(II). With a 1:4 ratio of aS:hydrogen peroxide (no Cu(II) in solution), both methionine residues were oxidized. When Cu(II) is added to the solution with hydrogen peroxide and aS peptide, further oxidation occurs on methionine residues, creating methionine sulfones. Along with this oxidation of methionine, the aS peptides also undergo fragmentation in the presence of Cu(II) between residues Lys10-Ala11 and Lys12-Glu13, indicating that the Lys residues are likely binding sites for Cu(II).

G. Cross linking-MS (XL-MS)

1. Prototypal study with XL-MS of Aβ

XL-MS was utilized by Du et al.²¹⁸ to study the interactions of Aβ₄₀ with transthyretin (TTR). TTR, a homotetramer protein found in blood and cerebrospinal fluid, is believed to interact with Aβ₄₀.²¹⁸ BS³ crosslinking (Figure 15) revealed regions of close proximity of TTR and Aβ₄₀. Tandem MS shows that the A strand and EF helix of TTR are vital regions for the interaction with Aβ₄₀. The TTR A strand (residues 10–21) is cross-linked at K15 to an Aβ₄₀ peptide (residues 17–40) (Scheme 5A), whereas the TTR EF helix (residues 71–80) at K76 interacts with the Aβ₄₀ peptide (residues 17–40) (Scheme 5B). The investigators visually represented the potential binding between TTR and Aβ₄₀ (Figure 29). Although binding sites between TTR and Aβ₄₀ were elucidated with XL-MS, the data are not as detailed or simple to interpret as those from FPOP and HDX. Further, it is not clear that this interaction has relevance in vivo.

Scheme 5.

Structures showing crosslinking of TTR-Aβ₄₀ peptides. A) TTR A strand (residues 10–21) cross-linked with residues 17–40 on Aβ₄₀. B) TTR EF helix (residues 71–80) cross-linked with residues 17–40 on Aβ₄₀. (Reprinted (adapted) with permission from reference 218. Copyright 2010 American Chemical Society.)

Figure 29.

TTR (represented in pink) potential interaction with Aβ₄₀ (represented in yellow). The Lys residues that were found to crosslink with Aβ₄₀ are highlighted. (Reproduced with permission from reference 218.)

2. Application studies with XL-MS of Aβ

For completeness, we describe briefly other efforts to characterize Aβ oligomerization by cross linking. XL-MS of Aβ was reported by Cline et al.,²¹⁹ who discovered that crosslinking Aβ oligomeric species with 1,5-difluoro-2,4-dinitrobenzene (DFDNB) stabilizes the oligomers and prevents fibril formation. The oligomeric species, here stabilized by DFDNB, are toxic because they cause memory dysfunction in mice. Vazquez de la Torre et al.²²⁰ used AD patient samples to uncover a crosslinked Aβ peptide [Aβ(6–16)]₂ using a dityrosine cross-linker. This research shows evidence of Aβ dimer formation for AD patients, and the discovery suggests that inhibitors can be developed to prevent or slow down this dimerization.

Preston et al.²²¹ used Aβ_16–22 as a model peptide to study the relationship of covalent cross-links and noncovalent structure. Work by Schmid et al.²²² shows that tissue transglutaminase (TGase) induces dimerization Aβ, causing a structural change in Aβ monomers. The investigators also suggest that TGase may cause post-translational modifications of Aβ (specifically deamination of Q15), potentially seeding aggregation. Moore et al.²²³ assessed three cross-linkers with Aβ to determine the location and effect of the cross-link on the secondary structure, finding crosslinking between residues 15 and 28 of Aβ to be most abundant. Research by Atwood et al.²²⁴ shows that Cu²⁺ induces a dityrosine cross-link of Aβ and potentially induces aggregate formation. Egnaczyk et al.²²⁵ used a photoreactive Aβ₄₀ (substituted L-p-benzoylphenylalanine for F4 of Aβ) to find a intermolecular cross-link between the side chains of residues 4 and 35, suggesting the formation of an antiparallel β-sheet. XL-MS has identified several potential regions of both intramolecular and intermolecular interactions, as well as molecules that change these interactions. Future research with XL-MS should continue to elucidate regions of interaction among Aβ monomers during the formation of oligomeric species and show how different ligands may alter the conformation of Aβ.

3. Prototypal study with XL-MS of aS

The structure and dynamics of aS aggregation was also investigated by Fonseca-Ornelas et al.,²²⁶ who used XL-MS to delineate the interactions and effects of tertrapyrrole phthalocyanine tetrasulfonate (PcTS) on aS aggregation. The investigators employed two crosslinkers (tris-biridylruthenium (Ru(bpy)₃²⁺) and BS³) and discovered PcTS causes off-path oligomerization of α-helical aS. Many intermolecular crosslinks were found when PcTS was present with aS (Figure 30). This study shows that small molecules can inhibit fibril formation and that XL-MS can identify sites of intermolecular interaction in oligomers.

Figure 30.

The intermolecular interactions between aS oligomers when PcTS is present. The orange lines represent BS³ crosslinking oligomeric aS, whereas the yellow lines represent Ru(bpy)₃²⁺ crosslinking oligomeric aS. (Reproduced with permission from reference 226.)

4. Applications with XL-MS of aS

Many studies demonstrated XL-MS as a tool to determine regions of close contact and interaction of aS. Abeyawardhane et al.²²⁷ determined that Cu⁺ binds to N-terminally acetylated aS, causing both intermolecular dityrosine cross-linking in the fibril core region and intramolecular cross-linking of the C-terminal region. They also showed that the H50Q substitution impacts the crosslinking and aggregation of aS. These results are contradictory to those of Mason et al.¹⁷⁷; however, Mason et al. studied the interaction of aS with Cu²⁺ using IM-MS, while Abeyawardhane et al. used Cu⁺ and XL-MS. Werner-Allen et al.²²⁸ examined 3,4-dihydroxyphenylacetaldehyde’s ability to cross-link aS, showing that the di-dicatechol isoindole lysine linkage exists for aS dimers. Mukherjee et al.,²²⁹ by identifying dityrosine cross-links in aS under oxidative conditions, proposed a dimer structure for aS and presented a MS fragmentation pattern of dityrosine cross-linked Aβ_1–16 using several different activation methods. The investigators hypothesize the dityrosine cross-links can potentially serve as biomarkers for PD, as oxidative conditions are common.

Brodie et al.²³⁰ used a short photo-reactive cross-linker (2,4,6-triazido-1,3,5-triazine (TATA)) to identify 10 different crosslinks of the misfolded monomer aS, where the majority of the crosslinks (~70%) are with hydrophobic residues (alanine, valine, proline, glycine, methionine, and tyrosine). They presented a model for the molten-globule form of aS, suggesting that aS could adopt this structure at the initial stage of aggregation. Al-Hilaly et al.²³¹ found a dityrosine crosslink that yields a homo dimer of aS; the dimer potentially aids in the formation of amyloid fibrils. Work by Phillips et al.¹⁶⁴ (which is also highlighted in the IM-MS applications section) used XL-IM-MS of aS to discover three conformations of a structured middle region of aS. Brodie et al.²³² developed several short-range, photoreactive crosslinkers for aS and examined them by nMS to reveal several informative crosslinks that are not in agreement with a micelle-bound NMR aS structure. This study shows that aS is likely to take different structures in-solution vs in a micelle environment (as also illustrated by Lee et al.¹⁸² and Oganesyan et al.²¹¹ via continuous HDX with aS in a membrane-like environment).

Schmid et al.²³³ showed that TGase induces a intramolecular cross-link between glutamine and lysine residues in aS, where the cross-link causes formation of spherical aggregates that do not undergo amyloid fibrilization. This work also highlights Q79 and Q109 on aS as key interaction sites with TGase, and that mutating these sites can disrupt the interactions between aS and TGase. Continued work using XL-MS should provide information on aS aggregation and on potential drug therapies that inhibit those interactions that cause amyloid fibril formation. XL-MS has been able to identify several key residues of aS that may be involved in aggregation, as well as discovering multiple structural conformations of aS aggregate species. Different chemical environments (e.g., those with ligands or micelles present) appear to have an effect on the structure of aS that is detectable with XL-MS.

IV. DEVELOPMENT OF THERAPEUTICS

MS analysis has shown that several inhibitors have been successful in preventing amyloid fibril formation of Aβ and aS. Amyloid fibril inhibitors have been discussed throughout the review, but this section describes the types/categories of inhibitors and highlights developments of novel inhibitors that can pass the blood-brain barrier to interact with Aβ and aS in the brain.

A review by Doig et al.⁴⁵ states that oligomeric species are the best targets for drug discovery. Because there are many different oligomer species that display different levels of toxicity, it is difficult to develop a drug that targets all of them. Therefore, using multiple drugs/inhibitors may be the best way to combat neurodegenerative diseases. Doig et al.⁴⁵ and DaSilva et al.²³⁴ highlight the classes of inhibitors, including D-peptides, retro-inverso (RI) peptides, N-methyl peptides, molecular tweezers, polyphenols, and quinone derivatives. Each class of inhibitor has the potential to prevent amyloid formation, possibly in a unique way, and the use of ion mobility, footprinting, and chemical cross-linking can further elucidate these interactions. The challenge is to find an inhibitor that works in vivo.

One inhibitor category that has been studied extensively is polyphenols, which seem to reduce toxicity and amyloid formation for both Aβ₄₂ and aS. Polyphenols are natural products from turmeric, grapes, green tea, and red wine.⁴⁵ Some examples of promising polyphenol inhibitors, at least in vitro, are curcumin,²³⁵ resveratrol,²³⁶ e-viniferin glucoside (EVG),²³⁷ gallic acid,^238,239 and epigallocatechin gallate (EGCG),²⁴⁰ some of which were already discussed in this review. Understanding the binding of these polyphenols can lead to novel derivatives for amyloid prevention and also provide insights on aggregation.

Chen et al.²³⁸ examined amide derivatives of gallic acid, a compound that prevents protein aggregation and reduces toxicity for neuron cells.²³⁹ Gallic acid, however, is hydrophilic with a −logP value of 0.42. To pass the blood-brain barrier and interact with Aβ₄₂ and aS, molecules typically have a -logP value between 1 and 5,²⁴¹ so gallic acid is unlikely to pass the blood-brain barrier. Novel derivatives of gallic acid, however, have -logP values between 1 and 5. Some derivatives exhibit similar and stronger inhibitor effects of aS aggregation, compared to gallic acid, as monitored by using ThT fluorescence assays. Although these derivatives have potential for therapeutic use, a deeper understanding can be obtained by using MS-based structural tools to elucidate binding mechanisms. The development of novel derivatives that can pass the blood-brain barrier and modify aggregation are possible preventatives rather than cures for neurodegenerative diseases.

Further investigation is still required to answer many questions on aggregating proteins, including the number of species and polymorphs involved in aggregation, identification of toxic species and polymorphs, and ways to prevent these species from forming. MS offers the ability to answer these questions when paired with ion mobility, footprinting, and chemical cross-linking. Understanding the fundamentals of aggregation will be vital in developing effective therapeutics. From in vitro studies, controlling the aggregation pathway with small molecules or other proteins appears to be a viable option in preventing amyloid fibril formation and treating protein aggregation-caused diseases. Effective inhibitors do not have to fully prevent aggregation from occurring, but rather delay the onset of toxic species and production of amyloid fibrils.^45,47 By lengthening the lag phase of amyloid fibril formation, several years may be added to a human life.

When designing inhibitors, it is important to be mindful of the biological system that is being targeted. Specifically for neurodegenerative diseases, one key factor for an inhibitor is the correct -logP value to cross the blood-brain barrier and be effective in vivo. Another factor for therapeutics is to consider small molecules consumed in diet. In regions where AD and PD are relatively low, diet may affect the onset of these diseases.²⁴² Additionally a metabolite of a natural product may serve as an inhibitor. One specific example of this is the small molecule curcumin, which has been shown to inhibit Aβ amyloid formation.²³⁵ Curcumin is a primary constituent of turmeric (a spicy in curry) and is believed to lower the frequency of AD in the Asian Indian population.²³⁴ Metabolites of curcumin may be more effective inhibitors than the parent compound.

Performing aggregation experiments in more biologically relevant media will also help understand these diseases. MS methods can detect the various species and polymorphs for aggregation in biologically relevant media and assess the effectiveness of putative inhibitors. The efficacy of a small molecule or protein inhibitor may change when studied in different, more complex media (especially if that media is a better mimic for in vivo). Out of these studies can come correlations between in vitro and in vivo studies, and such correlations can expedite the development of successful therapeutics for aggregation-caused diseases.

V. CONCLUSION

Amyloid fibril formation may be a major cause for many neurodegenerative disorders including AD and PD.^1,2 These proteins misfold and associate to form amyloid fibrils;^2,4,27 the small, soluble oligomers appear to play a more important role in the disease that the plaques themselves. Understanding the process of protein aggregation and interaction with inhibitors should lead to drug therapies that prevent neurodegenerative diseases.⁴⁶

There are many different approaches to study amyloid fibril formation and amyloid protein-inhibitor interactions. Low-resolution techniques provide a full view of the protein but lack information at the regional and residue levels.^24,27 Although high-resolution techniques provide atomic level resolution of the protein structure, they possess several drawbacks that make them difficult to employ for amyloids. MS-based methods (ion mobility, footprinting, and chemical cross-linking) can provide vital information on protein structure with good throughput and follow aggregation for the highly polydisperse aggregates and protofibrils, where both low- and high-resolution methods struggle. The MS methods are well-suited for differential structural studies, where the properties of the aggregating protein can be compared with the monomer or in the presence of an inhibitor. Although MS approaches are slower than low resolution methods, they immediately produce suitable data without the care, commitment, and expense needed for X-ray crystallography, solution and solid-state NMR, and cryo-EM studies. Several MS approaches (nMS, IM-MS, continuous HDX-MS, pulsed HDX-MS, FPOP and other irreversible labeling methods, and XL-MS) now show the capability for studying amyloid formation.

This review provides background on each MS-based approach by means of brief tutorials for analytical chemists, biologists, and medical personnel and then identifies those studies that have used MS-based approaches, emphasizing their advantages to study Aβ and aS aggregation. These MS approaches provide comparative data to show how high order structures change in aggregation and in interactions with putative inhibitors of amyloid formation. This review also shows the outcomes of these approaches for understanding amyloid formation and for developing drug therapies to combat neurodegenerative disorders.

We expect future work will develop novel inhibitors of amyloid formation, provide understanding of how they affect aggregation, characterize their efficacy in membrane-like environments, and show the role of hydrophobicity to allow passage of the blood-brain barrier and function as inhibitors. MS approaches can be utilized also to determine the location and binding mechanism of the inhibitor and to understand biological mechanism of aggregation and inhibition for other neurodegenerative diseases. Ultimately, they may be able to deal with the heterogeneity of amyloid species and identify those that cause toxicity.

Biographies