Sampling Conformational Ensembles of Highly Dynamic Proteins via Generative Deep Learning.

Proteins are inherently dynamic, and their conformational ensembles are functionally important in biology. Large-scale motions may govern protein structure-function relationship, and numerous transient but stable conformations of intrinsically disordered proteins (IDPs) can play a crucial role in biological function. Investigating conformational ensembles to understand regulations and disease-related aggregations of IDPs is challenging both experimentally and computationally. In this paper first an unsupervised deep learning-based model, termed Internal Coordinate Net (ICoN), is developed that learns the physical principles of conformational changes from molecular dynamics (MD) simulation data. Second, interpolating data points in the learned latent space are selected that rapidly identify novel synthetic conformations with sophisticated and large-scale sidechains and backbone arrangements. Third, with the highly dynamic amyloid-ß1-42 (Aß42) monomer, our deep learning model provided a comprehensive sampling of Aß42's conformational landscape. Analysis of these synthetic conformations revealed conformational clusters that can be used to rationalize experimental findings. Additionally, the method can identify novel conformations with important interactions in atomistic details that are not included in the training data. New synthetic conformations showed distinct sidechain rearrangements that are probed by our EPR and amino acid substitution studies. The proposed approach is highly transferable and can be used for any available data for training. The work also demonstrated the ability for deep learning to utilize learned natural atomistic motions in protein conformation sampling.

Sampling Conformational Ensembles of Highly Dynamic Proteins via Generative Deep Learning.

Proteins are inherently dynamic, and their conformational ensembles are functionally important in biology. Large-scale motions may govern protein structure-function relationship, and numerous transient but stable conformations of intrinsically disordered proteins (IDPs) can play a crucial role in biological function. Investigating conformational ensembles to understand regulations and disease-related aggregations of IDPs is challenging both experimentally and computationally. In this paper we first introduced an unsupervised deep learning-based model, termed Internal Coordinate Net (ICoN), which learns the physical principles of conformational changes from molecular dynamics (MD) simulation data. Second, we selected interpolating data points in the learned latent space that rapidly identify novel synthetic conformations with sophisticated and large-scale sidechains and backbone arrangements. Third, with the highly dynamic amyloid-ß 1-42 (Aß42) monomer, our deep learning model provided a comprehensive sampling of Aß42's conformational landscape. Analysis of these synthetic conformations revealed conformational clusters that can be used to rationalize experimental findings. Additionally, the method can identify novel conformations with important interactions in atomistic details that are not included in the training data. New synthetic conformations showed distinct sidechain rearrangements that are probed by our EPR and amino acid substitution studies. This approach is highly transferable and can be used for any available data for training. The work also demonstrated the ability for deep learning to utilize learned natural atomistic motions in protein conformation sampling.

EPR Studies of Aß42 Oligomers Indicate a Parallel In-Register ß-Sheet Structure.

Aß aggregation leads to the formation of both insoluble amyloid fibrils and soluble oligomers. Understanding the structures of Aß oligomers is important for delineating the mechanism of Aß aggregation and developing effective therapeutics. Here, we use site-directed spin labeling and electron paramagnetic resonance (EPR) spectroscopy to study Aß42 oligomers prepared by using the protocol of Aß-derived diffusible ligands. We obtained the EPR spectra of 37 Aß42 oligomer samples, each spin-labeled at a unique residue position of the Aß42 sequence. Analysis of the disordered EPR components shows that the N-terminal region has a lower local structural stability. Spin label mobility analysis reveals three structured segments at residues 9-11, 15-22, and 30-40. Intermolecular spin-spin interactions indicate a parallel in-register ß-sheet structure, with residues 34-38 forming the structural core. Residues 16-21 also adopt the parallel in-register ß-structure, albeit with weaker intermolecular packing. Our results suggest that there is a structural class of Aß oligomers that adopt fibril-like conformations.

The supersaturation perspective on the amyloid hypothesis.

Development of therapeutic interventions for Alzheimer's over the past three decades has been guided by the amyloid hypothesis, which puts Aß deposition as the initiating event of a pathogenic cascade leading to dementia. In the current form, the amyloid hypothesis lacks a comprehensive framework that considers the complex nature of Aß aggregation. The explanation of how Aß deposition leads to downstream pathology, and how reducing Aß plaque load via anti-amyloid therapy can lead to improvement in cognition remains insufficient. In this perspective we integrate the concept of Aß supersaturation into the amyloid hypothesis, laying out a framework for the mechanistic understanding and therapeutic intervention of Alzheimer's disease. We discuss the important distinction between in vitro and in vivo patterns of Aß aggregation, the impact of different aggregation stages on therapeutic strategies, and how future investigations could integrate this concept in order to produce a more thorough understanding and better treatment for Alzheimer's and other amyloid-related disorders.

A protein aggregation platform that distinguishes oligomers from amyloid fibrils.

Deposition of aggregated proteins is a pathological feature in many neurodegenerative disorders such as Alzheimer's and Parkinson's. In addition to insoluble amyloid fibrils, protein aggregation leads to the formation of soluble oligomers, which are more toxic and pathogenic than fibrils. However, it is challenging to screen for inhibitors targeting oligomers due to the overlapping processes of oligomerization and fibrillization. Here we report a protein aggregation platform that uses intact and split TEM-1 ß-lactamase proteins as reporters of protein aggregation. The intact ß-lactamase fused with an amyloid protein can report the overall protein aggregation, which leads to loss of lactamase activity. On the other hand, reconstitution of active ß-lactamase from the split lactamase construct requires the formation of amyloid oligomers, making the split lactamase system sensitive to oligomerization. Using Aß, a protein that forms amyloid plaques in Alzheimer's disease, we show that the growth curves of bacterial cells expressing either intact or split lactamase-Aß fusion proteins can report changes in the Aß aggregation. The cell lysate lactamase activity assays show that the oligomer fraction accounts for 20% of total activity for the split lactamase-Aß construct, but only 3% of total activity for the intact lactamase-Aß construct, confirming the sensitivity of the split lactamase to oligomerization. The combination of the intact and split lactamase constructs allows the distinction of aggregation modulators targeting oligomerization from those targeting overall aggregation. These low-cost bacterial cell-based and biochemical assays are suitable for high-throughput screening of aggregation inhibitors targeting oligomers of various amyloid proteins.

Preparation and Fractionation of Heterogeneous Aß42 Oligomers with Different Aggregation Properties.

Deposition of amyloid-ß (Aß) aggregates in the form of amyloid plaques is a central feature of Alzheimer's disease. The end products of Aß aggregation are amyloid fibrils. Soluble Aß aggregates called oligomers are also formed either on or off the pathway of fibril formation. The amyloid fibrils from different clinical subtypes of Alzheimer's disease have been found to adopt different structures, a phenomenon called fibril polymorphism. Meanwhile, different types of Aß oligomers have also been found. Recently, it has been shown that different types of Aß42 oligomers may form fibrils of different structures, linking oligomer heterogeneity to fibril polymorphism. In this chapter, we describe methods to prepare heterogeneous Aß42 oligomers and to quantify the concentration of these oligomers at a low micromolar range using a fluorescamine method. Fractionation of these oligomers by size using ultrafiltration filters allows for the formation of Aß42 fibrils with different structural properties.

Static and dynamic disorder in Aß40 fibrils.

Deposition of Aß aggregates in the form of amyloid fibrils is a pathological hallmark of Alzheimer's disease. Understanding the structure and dynamics of Aß fibrils is important for delineating the mechanism of Aß aggregation and developing effective therapeutic strategies. Here we used site-directed spin labeling and EPR spectroscopy to study the Aß40 fibril structure and dynamics. We obtained the EPR spectra of 40 spin-labeled Aß40 fibril samples, with spin labeling coverage of the entire Aß40 sequence. Analysis of the spin exchange interaction and spin label mobility using spectral simulations suggest that the strength of spin exchange interaction is primarily determined by static disorder in the Aß40 fibrils. EPR data suggest that the entire Aß40 sequence except residue D1 is highly ordered and the two hydrophobic regions at residues 17-20 and 31-36 show the lowest static disorder. Dynamic disorder is relatively constant across all reside positions, with residues 22 and 23 having the highest dynamic disorder. Comparison of the EPR data for Aß40 and Aß42 fibrils shows overall more ordered packing interactions in Aß40 fibrils. Another noteworthy difference is the C-terminal residue, which has high static disorder in Aß42 fibrils, but is ordered in Aß40 fibrils. The higher static disorder in Aß42 fibrils may lead to increased fragmentation, monomer dissociation, and structural defects, which may contribute to increased aggregation through secondary nucleation.

Use of hydrous ABE-glycerin-diesel microemulsions in a nonroad diesel engine - Performance and unignorable emissions.

Oversupply, extra energy consumption, and CO2 emissions from the refinery of biodiesel-derived glycerin (G) led to the consideration of its use as an alternative fuel. In this study, a nonroad diesel engine generator was employed to represent potential emissions under stringent regulated standards. G-diesel has been reported to reduce nitrogen oxides (NOx) and soot levels but increase CO and hydrocarbon emissions. A bio-producible acetone-butanol-ethanol (ABE) solution with multiple polarities was added to stabilize the glycerin and water in diesel examined in this study. A series of ABE-G-diesel blends were prepared to form the thermostable microemulsions. Four blends with small and well-dispersed bubbles were tested in the engine generator. The specific thermal efficiencies of the engine were slightly improved by using ABE-G from regular diesel due to better spray quality, longer ignition delay, and fuel-oxygen content that would enhance combustion. Meanwhile, the PM-NOx-CO emission trade-off in the previous study has been overcome by using ABE-G-diesel since the better fuel atomization and more premixed combustion were approached, as well as the lower and homogeneous in-cylinder temperature caused by water content and micro-explosion. However, the condensable particulate matter and nitro-PAHs were also observed and realized their unignorable contribution, which has not been regulated and even researched for the generators. Fortunately, the new fuels could inhibit both of them to a certain degree. Consequently, this study proposes using recyclable glycerin with a simple pretreatment mixed with ABE and diesel for greener nonroad diesel engine especially those equipped with low-grade aftertreatment.

Effect of spin labelling on the aggregation kinetics of yeast prion protein Ure2.

Amyloid formation is involved in a wide range of neurodegenerative diseases including Alzheimer's and prion diseases. Structural understanding of the amyloid is critical to delineate the mechanism of aggregation and its pathological spreading. Site-directed spin labelling has emerged as a powerful structural tool in the studies of amyloid structures and provided structural evidence for the parallel in-register ß-sheet structure for a wide range of amyloid proteins. It is generally accepted that spin labelling does not disrupt the structure of the amyloid fibrils, the end product of protein aggregation. The effect on the rate of protein aggregation, however, has not been well characterized. Here, we employed a scanning mutagenesis approach to study the effect of spin labelling on the aggregation rate of 79 spin-labelled variants of the Ure2 prion domain. The aggregation of Ure2 protein is the basis of yeast prion [URE3]. We found that all spin-labelled Ure2 mutants aggregated within the experimental timeframe of 15 to 40 h. Among the 79 spin-labelled positions, only five residue sites (N23, N27, S33, I35 and G42) showed a dramatic delay in the aggregation rate as a result of spin labelling. These positions may be important for fibril nucleation, a rate-limiting step in aggregation. Importantly, spin labelling at most of the sites had a muted effect on Ure2 aggregation kinetics, showing a general tolerance of spin labelling in protein aggregation studies.

Lipid membranes induce structural conversion from amyloid oligomers to fibrils.

Formation of amyloid oligomers and fibrils underlies the pathogenesis of a number of neurodegenerative diseases such as Alzheimer's. One mechanism of action by which Aß aggregates cause neuronal toxicity is through interactions with cellular membranes. Aß aggregates have been shown to disrupt membrane integrity via pore formation, membrane thinning, or lipid extraction. At the same time, lipid membranes also affect the rate of Aß aggregation and remodel pre-formed Aß fibrils. Here we show that Aß42 globulomers, a type of well-characterized and stable Aß oligomers, convert to amyloid fibrils in the presence of DOPC liposomes. Electron paramagnetic resonance studies show that the fibrils converted from Aß42 globulomers adopt the same structure as fibrils formed directly from monomers. Our results suggest that the interactions between Aß oligomers and cellular membranes are dynamic. By converting Aß oligomers to fibrils, the lipid membrane can reduce the membrane-disrupting activities caused by these oligomers. Modulation of Aß-membrane interactions as a therapeutic strategy should take into account the dynamic nature of these interactions.

A novel flameless oxidation and in-chamber melting system coupled with advanced scrubbers for a laboratory waste plant.

This is the first study integrate the flameless oxidation (FO) and in-chamber melting (ICM) processes in a primary chamber of a laboratory waste incinerator to improve energy and emission performances. Two liquid burners created a twin-cyclonic fluid field that achieved the FO and ICM in the same chamber. The first cyclone provided a well-mixed and lower temperature FO to reduce auxiliary diesel consumption, NOx and PM emissions by 25.8%, 30.9%, and 79.2%, respectively, from the original system. The hot gases produced by FO enhance the ICM process and transformed the bottom ashes to stabler slags, in turn meeting the regulations for nonhazardous wastes. The other cyclone enhanced the drying and water-gas shift reaction in the drying zone by recirculating the CO and enthalpy from FO and ICM. Eventually, the residual CO, hydrocarbons, and H2 were sent to the secondary chamber for further oxidation. A computational fluid dynamic simulation supported the fluid field assumption posed in this study. Moreover, advanced scrubbers were employed after thermal treatments to reduce HCl and SO2 by 81.8% and 38.8% and further retarded the corrosion rate in the baghouse supporting cage by 87.7%. The precursors of condensable particulate matter were reduced by condensation and finally removed in the baghouse. Nevertheless, the emissions of the high- and mid-molecular-weight polycyclic aromatic hydrocarbons were greatly reduced by 60.8-93.1% and 80.2-99.9%, respectively. Consequently, the new system reduced annual emissions by 40.7-87.6% and operating costs by 41.5%, allowing recovery of the remodification investment in 20.5 months.

Segmental structural dynamics in Aß42 globulomers.

Aß42 aggregation plays a central role in the pathogenesis of Alzheimer's disease. In addition to the insoluble fibrils that comprise the amyloid plaques, Aß42 also forms soluble aggregates collectively called oligomers, which are more toxic and pathogenic than fibrils. Understanding the structure and dynamics of Aß42 oligomers is critical for developing effective therapeutic interventions against these oligomers. Here we studied the structural dynamics of Aß42 globulomers, a type of Aß42 oligomers prepared in the presence of sodium dodecyl sulfate, using site-directed spin labeling. Spin labels were introduced, one at a time, at all 42 residue positions of Aß42 sequence. Electron paramagnetic resonance spectra of spin-labeled samples reveal four structural segments based on site-dependent spin label mobility pattern. Segment-1 consists of residues 1-6, which have the highest mobility that is consistent with complete disorder. Segment-3 is the most immobilized region, including residues 31-34. Segment-2 and -4 have intermediate mobility and are composed of residues 7-30 and 35-42, respectively. Considering the inverse relationship between protein dynamics and stability, our results suggest that residues 31-34 are the most stable segment in Aß42 oligomers. At the same time, the EPR spectral lineshape suggests that Aß42 globulomers lack a well-packed structural core akin to that of globular proteins.

Alzheimer's Aß42 and Aß40 form mixed oligomers with direct molecular interactions.

Formation of Aß oligomers and fibrils plays a central role in the pathogenesis of Alzheimer's disease. There are two major forms of Aß in the brain: Aß42 and Aß40. Aß42 is the major component of the amyloid plaques, but the overall abundance of Aß40 is several times that of Aß42. In vitro experiments show that Aß42 and Aß40 affect each other's aggregation. In mouse models of Alzheimer's disease, overexpression of Aß40 has been shown to reduce the plaque pathology, suggesting that Aß42 and Aß40 also interact in vivo. Here we address the question of whether Aß42 and Aß40 interact with each other in the formation of oligomers using electron paramagnetic resonance (EPR) spectroscopy. When the Aß42 oligomers were formed using only spin-labeled Aß42, the dipolar interaction between spin labels that are within 20 Å range broadened the EPR spectrum and reduced its amplitude. Oligomers formed with a mixture of spin-labeled Aß42 and wild-type Aß42 gave an EPR spectrum with higher amplitude due to weakened spin-spin interactions, suggesting molecular mixing of labeled and wild-type Aß42. When spin-labeled Aß42 and wild-type Aß40 were mixed to form oligomers, the resulting EPR spectrum also showed reduced amplitude, suggesting that wild-type Aß40 can also form oligomers with spin-labeled Aß42. Therefore, our results suggest that Aß42 and Aß40 form mixed oligomers with direct molecular interactions. Our results point to the importance of investigating Aß42-Aß40 interactions in the brain for a complete understanding of Alzheimer's pathogenesis and therapeutic interventions.

Distinguishing the Effect on the Rate and Yield of Aß42 Aggregation by Green Tea Polyphenol EGCG.

Deposition of Aß42 aggregates in the form of amyloid plaques is a pathological hallmark of Alzheimer's disease. A desired avenue of intervention is the inhibition of Aß42 aggregation. Epigallocatechin gallate (EGCG), the main polyphenol in green tea, has been generally considered an inhibitor of Aß aggregation. However, previous experiments focused on the reduction of the amount of Aß42 aggregates, while the effect of EGCG on the rate of Aß42 aggregation was not critically analyzed. Here we performed an experimental evaluation of Aß42 aggregation kinetics in the absence and presence of EGCG at a wide range of concentrations. We found that EGCG reduced thioflavin T fluorescence in an EGCG concentration-dependent manner, suggesting that EGCG reduced the amount of Aß42 fibrils. The effect of EGCG on the rate of Aß42 aggregation appears to be bimodal. We found that higher EGCG-to-Aß42 ratios promoted the rate of Aß42 aggregation, while lower EGCG-to-Aß42 ratios inhibited the aggregation rate. To confirm that the reduction of thioflavin T fluorescence is due to the lowered aggregate quantity, but not due to perturbation of thioflavin T binding to Aß42 fibrils, we probed the effect of EGCG on Aß42 aggregation using site-directed spin labeling. Electron paramagnetic resonance of spin-labeled Aß42 aggregates suggests that high EGCG-to-Aß42 ratios led to a greatly reduced amount of Aß42 fibrils, and these aggregates adopt similar structures as the fibrils in the no-EGCG sample. Potential implications of this work in designing prevention or therapeutic strategies using EGCG are discussed.

Spin Label Scanning Reveals Likely Locations of ß-Strands in the Amyloid Fibrils of the Ure2 Prion Domain.

In yeast, the formation of Ure2 fibrils underlies the prion state [URE3], in which the yeast loses the ability to distinguish good nitrogen sources from bad ones. The Ure2 prion domain is both necessary and sufficient for the formation of amyloid fibrils. Understanding the structure of Ure2 fibrils is important for understanding the propagation not only of the [URE3] prion but also of other yeast prions whose prion domains share similar features, such as the enrichment of asparagine and glutamine residues. Here, we report a structural study of the amyloid fibrils formed by the Ure2 prion domain using site-directed spin labeling and electron paramagnetic resonance (EPR) spectroscopy. We completed a spin label scanning of all the residue positions between 2 and 80 of the Ure2 prion domain. The EPR data show that the Ure2 fibril core consists of residues 8-68 and adopts a parallel in-register ß-sheet structure. Most of the residues show strong spin-exchange interactions, suggesting that there are only short turns and no long loops in the fibril core. Based on the strength of spin-exchange interactions, we determined the likely locations of the ß-strands. EPR data also show that the C-terminal region of the Ure2 prion domain is more disordered than the N-terminal region. The roles of hydrophobic and charged residues are analyzed. Overall, the structure of Ure2 fibrils appears to involve a balance of stabilizing interactions, such as asparagine ladders, and destabilizing interactions, such as stacking of charged residues.

Polymorphic Aß42 fibrils adopt similar secondary structure but differ in cross-strand side chain stacking interactions within the same ß-sheet.

Formation of polymorphic amyloid fibrils is a common feature in neurodegenerative diseases involving protein aggregation. In Alzheimer's disease, different fibril structures may be associated with different clinical sub-types. Structural basis of fibril polymorphism is thus important for understanding the role of amyloid fibrils in the pathogenesis and progression of these diseases. Here we studied two types of Aß42 fibrils prepared under quiescent and agitated conditions. Quiescent Aß42 fibrils adopt a long and twisted morphology, while agitated fibrils are short and straight, forming large bundles via lateral association. EPR studies of these two types of Aß42 fibrils show that the secondary structure is similar in both fibril polymorphs. At the same time, agitated Aß42 fibrils show stronger interactions between spin labels across the full range of the Aß42 sequence, suggesting a more tightly packed structure. Our data suggest that cross-strand side chain packing interactions within the same ß-sheet may play a critical role in the formation of polymorphic fibrils.

Aß42 fibril formation from predominantly oligomeric samples suggests a link between oligomer heterogeneity and fibril polymorphism.

Amyloid-ß (Aß) oligomers play a central role in the pathogenesis of Alzheimer's disease. Oligomers of different sizes, morphology and structures have been reported in both in vivo and in vitro studies, but there is a general lack of understanding about where to place these oligomers in the overall process of Aß aggregation and fibrillization. Here, we show that Aß42 spontaneously forms oligomers with a wide range of sizes in the same sample. These Aß42 samples contain predominantly oligomers, and they quickly form fibrils upon incubation at 37°C. When fractionated using ultrafiltration filters, the samples enriched with smaller oligomers form fibrils at a faster rate than the samples enriched with larger oligomers, with both a shorter lag time and faster fibril growth rate. This observation is independent of Aß42 batches and hexafluoroisopropanol treatment. Furthermore, the fibrils formed by the samples enriched with larger oligomers are more readily solubilized by epigallocatechin gallate, a main catechin component of green tea. These results suggest that the fibrils formed by larger oligomers may adopt a different structure from fibrils formed by smaller oligomers, pointing to a link between oligomer heterogeneity and fibril polymorphism.

Site-specific structural order in Alzheimer's Aß42 fibrils.

Deposition of amyloid fibrils is a pathological hallmark of Alzheimer's disease. Aß42 is the major protein whose aggregation leads to the formation of these fibrils. Understanding the detailed structure of Aß42 fibrils is of particular importance for delineating the mechanism of Aß42 aggregation and developing specific amyloid-targeting drugs. Here, we use site-directed spin labelling and electron paramagnetic resonance spectroscopy to study the site-specific structural order at each and every residue position in Aß42 fibrils. Strong interactions between spin labels indicate highly ordered protein backbone at the labelling site, while weak interactions suggest disordered local structure. Our results show that Aß42 consists of five ß-strands (residues 2-7, 10-13, 17-20, 31-36, 39-41), three turns (residues 7-8, 14-16, 37-38) and one ordered loop (residues 21-30). Spin labels introduced at ß-strand sites show strong spin-spin interactions, while spin labels at turn or loop sites show weak interactions. However, residues 24, 25 and 28 also show strong interactions between spin labels, suggesting that the loop 21-30 is partly ordered. In the context of recent structural work using solid-state NMR and cryoEM, the site-specific structural order revealed in this study provides a different perspective on backbone and side chain dynamics of Aß42 fibrils.

Key Residues for the Formation of Aß42 Amyloid Fibrils.

Formation of amyloid fibrils by Aß42 protein is a pathological hallmark of Alzheimer's disease. Aß42 fibrillization is a nucleation-dependent polymerization process, in which nucleation is the rate-limiting step. Structural knowledge of the fibril nucleus is important to understand the molecular mechanism of Aß aggregation and is also critical for successful modulation of the fibrillization process. Here, we used a scanning mutagenesis approach to study the role of each residue position in Aß42 fibrillization kinetics. The side chain we used to replace the native residue is a nitroxide spin label called R1, which was introduced using site-directed spin labeling. In this systematic study, all residue positions of Aß42 sequence were studied, and we identified six key residues for the Aß42 fibril formation: H14, E22, D23, G33, G37, and G38. Our results suggest that charges at positions 22 and 23 and backbone flexibilities at positions 33, 37, and 38 play key roles in Aß42 fibrillization kinetics. Our results also suggest that the formation of a ß-strand at residues 15-21 is an important feature in Aß42 fibril nucleus. In overall evaluation of all of the mutational effects on fibrillization kinetics, we found that the thioflavin T fluorescence at the aggregation plateau is a poor indicator of aggregation rates.