AGGRESCAN: method, application, and perspectives for drug design

Deep-GB: A novel deep learning model for globular protein prediction using CNN-BiLSTM architecture and enhanced PSSM with trisection strategy

Globular proteins (GPs) play vital roles in a wide range of biological processes, encompassing enzymatic catalysis and immune responses. Enzymes, among these globular proteins, facilitate biochemical reactions, while others, such as haemoglobin, contribute to essential physiological functions such as oxygen transport. Given the importance of these considerations, accurately identifying Globular proteins is essential. To address the need for precise GP identification, this research introduces an innovative approach that employs a hybrid-based deep learning model called Deep-GP. We generated two datasets based on primary sequences and developed a novel feature descriptor called, Consensus Sequence-based Trisection-Position Specific Scoring Matrix (CST-PSSM). The model training phase involved the application of deep learning techniques, including the bidirectional long short-term memory network (BiLSTM), gated recurrent unit (GRU), and convolutional neural network (CNN). The BiLSTM and CNN were hybridised for ensemble learning. The CST-PSSM-based ensemble model achieved the most accurate predictive outcomes, outperforming other competitive predictors across both training and testing datasets. This demonstrates the potential of harnessing deep learning for precise GB prediction as a robust tool to expedite research, streamline drug discovery, and unveil novel therapeutic targets.

Inhibition and disaggregation effect of flavonoid-derived carbonized polymer dots on protein amyloid aggregation

In this research, four water-insoluble flavonoid compounds were utilized and reacted with arginine to prepare four carbonized polymer dots with good water-solubility in a hydrothermal reactor. Structural characterization demonstrated that the prepared carbonized polymer dots were classic core-shell structure. Effect of the prepared carbonized polymer dots on protein amyloid aggregation was further investigated using hen egg white lysozyme and human lysozyme as model protein in aqueous solution. All of the prepared carbonized polymer dots could retard the amyloid aggregation of hen egg white lysozyme and human lysozyme in a dose-depended manner. All measurements displayed that the inhibition ratio of luteolin-derived carbonized polymer dots (CPDs-1) was higher than that of the other three carbonized polymer dots under the same dosage. This result may be interpreted by the highest content of phenolic hydroxyl groups on the periphery. The inhibition ratio of CPDs-1 on hen egg white lysozyme and human lysozyme reached 88 % and 83 % at the concentration of 0.5 mg/mL, respectively. CPDs-1 also could disaggregate the formed mature amyloid fibrils into short aggregates.

AGRAMP: machine learning models for predicting antimicrobial peptides against phytopathogenic bacteria

Introduction

Antimicrobial peptides (AMPs) are promising alternatives to traditional antibiotics for combating plant pathogenic bacteria in agriculture and the environment. However, identifying potent AMPs through laborious experimental assays is resource-intensive and time-consuming. To address these limitations, this study presents a bioinformatics approach utilizing machine learning models for predicting and selecting AMPs active against plant pathogenic bacteria.

Methods

N-gram representations of peptide sequences with 3-letter and 9-letter reduced amino acid alphabets were used to capture the sequence patterns and motifs that contribute to the antimicrobial activity of AMPs. A 5-fold cross-validation technique was used to train the machine learning models and to evaluate their predictive accuracy and robustness.

Results

The models were applied to predict putative AMPs encoded by intergenic regions and small open reading frames (ORFs) of the citrus genome. Approximately 7% of the 10,000-peptide dataset from the intergenic region and 7% of the 685,924-peptide dataset from the whole genome were predicted as probable AMPs. The prediction accuracy of the reported models range from 0.72 to 0.91. A subset of the predicted AMPs was selected for experimental test against Spiroplasma citri, the causative agent of citrus stubborn disease. The experimental results confirm the antimicrobial activity of the selected AMPs against the target bacterium, demonstrating the predictive capability of the machine learning models.

Discussion

Hydrophobic amino acid residues and positively charged amino acid residues are among the key features in predicting AMPs by the Random Forest Algorithm. Aggregation propensity appears to be correlated with the effectiveness of the AMPs. The described models would contribute to the development of effective AMP-based strategies for plant disease management in agricultural and environmental settings. To facilitate broader accessibility, our model is publicly available on the AGRAMP (Agricultural Ngrams Antimicrobial Peptides) server.

AB-Amy: machine learning aided amyloidogenic risk prediction of therapeutic antibody light chains

Over 120 FDA-approved antibody-based therapeutics are used to treat a variety of diseases.However, many candidates could fail because of unfavorable physicochemical properties. Light-chain amyloidosis is one form of aggregation that can lead to severe safety risks in clinical development. Therefore, screening candidates with a less amyloidosis risk at the early stage can not only save the time and cost of antibody development but also improve the safety of antibody drugs. In this study, based on the dipeptide composition of 742 amyloidogenic and 712 non-amyloidogenic antibody light chains, a support vector machine-based model, AB-Amy, was trained to predict the light-chain amyloidogenic risk. The AUC of AB-Amy reaches 0.9651. The excellent performance of AB-Amy indicates that it can be a useful tool for the in silico evaluation of the light-chain amyloidogenic risk to ensure the safety of antibody therapeutics under clinical development. A web server is freely available at http://i.uestc.edu.cn/AB-Amy/.

Exploring cryptic amyloidogenic regions in prion-like proteins from plants

Prion-like domains (PrLDs) are intrinsically disordered regions (IDRs) of low sequence complexity with a similar composition to yeast prion domains. PrLDs-containing proteins have been involved in different organisms' regulatory processes. Regions of moderate amyloid propensity within IDRs have been shown to assemble autonomously into amyloid fibrils. These sequences tend to be rich in polar amino acids and often escape from the detection of classical bioinformatics screenings that look for highly aggregation-prone hydrophobic sequence stretches. We defined them as cryptic amyloidogenic regions (CARs) and recently developed an integrated database that collects thousands of predicted CARs in IDRs. CARs seem to be evolutionary conserved among disordered regions because of their potential to stablish functional contacts with other biomolecules. Here we have focused on identifying and characterizing CARs in prion-like proteins (pCARs) from plants, a lineage that has been poorly studied in comparison with other prionomes. We confirmed the intrinsic amyloid potential for a selected pCAR from Arabidopsis thaliana and explored functional enrichments and compositional bias of pCARs in plant prion-like proteins.

An in silico approach to study the role of epitope order in the multi-epitope-based peptide (MEBP) vaccine design

With different countries facing multiple waves, with some SARS-CoV-2 variants more deadly and virulent, the COVID-19 pandemic is becoming more dangerous by the day and the world is facing an even more dreadful extended pandemic with exponential positive cases and increasing death rates. There is an urgent need for more efficient and faster methods of vaccine development against SARS-CoV-2. Compared to experimental protocols, the opportunities to innovate are very high in immunoinformatics/in silico approaches, especially with the recent adoption of structural bioinformatics in peptide vaccine design. In recent times, multi-epitope-based peptide vaccine candidates (MEBPVCs) have shown extraordinarily high humoral and cellular responses to immunization. Most of the publications claim that respective reported MEBPVC(s) assembled using a set of in silico predicted epitopes, to be the computationally validated potent vaccine candidate(s) ready for experimental validation. However, in this article, for a given set of predicted epitopes, it is shown that the published MEBPVC is one among the many possible variants and there is high likelihood of finding more potent MEBPVCs than the published candidates. To test the same, a methodology is developed where novel MEBP variants are derived by changing the epitope order of the published MEBPVC. Further, to overcome the limitations of current qualitative methods of assessment of MEBPVC, to enable quantitative comparison and ranking for the discovery of more potent MEBPVCs, novel predictors, Percent Epitope Accessibility (PEA), Receptor specific MEBP vaccine potency (RMVP), MEBP vaccine potency (MVP) are introduced. The MEBP variants indeed showed varied MVP scores indicating varied immunogenicity. Further, the MEBP variants with IDs, SPVC_446 and SPVC_537, had the highest MVP scores indicating these variants to be more potent MEBPVCs than the published MEBPVC and hence should be preferred candidates for immediate experimental testing and validation. The method enables quicker selection and high throughput experimental validation of vaccine candidates. This study also opens the opportunity to develop new software tools for designing more potent MEBPVCs in less time.

Self-assembly and Hydrogelation Properties of Peptides Derived from Peptic Cleavage of Aggregation-prone Regions of Ovalbumin

Egg white protein hydrolysate generated with pepsin was investigated for the presence of peptides with self-assembly and hydrogelation properties. Incubation of the hydrolysates for 16 h resulted in aggregates with significantly (p < 0.05) lower free amino nitrogen and sulfhydryl contents, and higher particle diameter and surface hydrophobicity compared to the hydrolysates. LC-MS/MS analysis of the aggregates resulted in identification of 429 ovalbumin-derived peptides, among which the top-six aggregation-prone peptides IFYCPIAIM, NIFYCPIAIM, VLVNAIVFKGL, YCPIAIMSA, MMYQIGLF, and VYSFSLASRL were predicted using AGGRESCAN by analysis of the aggregation “Hot Spots”. NIFYCPIAIM had the highest thioflavin T fluorescence intensity, particle diameter (5611.3 nm), and polydispersity index (1.0) after 24 h, suggesting the formation of β-sheet structures with heterogeneous particle size distribution. Transmission electron microscopy of MMYQIGLF, and VYSFSLASRL demonstrated the most favorable peptide self-assembly, based on the formation of densely packed, intertwined fibrils. Rheological studies confirmed the viscoelastic and mechanical properties of the hydrogels, with IFYCPIAIM, NIFYCPIAIM, VLVNAIVFKGL, and VYSFSLASRL forming elastic solid hydrogels (tan δ < 1), while YCPIAIMSA and MMYQIGLF formed viscous liquid-like hydrogels (tan δ > 1). The results provide valuable insight into the influence of peptide sequence on hydrogelation and self-assembly progression, and prospects of food peptides in biomaterial applications.

Monomodular Pseudomonas aeruginosa phage JG004 lysozyme (Pae87) contains a bacterial surface-active antimicrobial peptide-like region and a possible substrate-binding subdomain

The structure of the monomodular Pseudomonas aeruginosa bacteriophage JG004 lysin Pae87 is presented and investigated in relation to repurposing its function as an antimicrobial agent. The structure with its peptidoglycan ligand revealed a possible cell-wall-binding region. A C-terminal antimicrobial peptide-like region is shown to be important for disrupting the bacterial cell wall.

Phage lysins are a source of novel antimicrobials to tackle the bacterial antibiotic-resistance crisis. The engineering of phage lysins is being explored as a game-changing technological strategy to introduce a more precise approach in the way in which antimicrobial therapy is applied. Such engineering efforts will benefit from a better understanding of lysin structure and function. In this work, the antimicrobial activity of the endolysin from Pseudomonas aeruginosa phage JG004, termed Pae87, has been characterized. This lysin had previously been identified as an antimicrobial agent candidate that is able to interact with the Gram-negative surface and disrupt it. Further evidence is provided here based on a structural and biochemical study. A high-resolution crystal structure of Pae87 complexed with a peptidoglycan fragment showed a separate substrate-binding region within the catalytic domain, 18 Å away from the catalytic site and located on the opposite side of the lysin molecule. This substrate-binding region was conserved among phylogenetically related lysins lacking an additional cell-wall-binding domain, but not among those containing such a module. Two glutamic acids were identified to be relevant for the peptidoglycan-degradation activity, although the antimicrobial activity of Pae87 was seemingly unrelated. In contrast, an antimicrobial peptide-like region within the Pae87 C-terminus, named P87, was found to be able to actively disturb the outer membrane and display antibacterial activity by itself. Therefore, an antimicrobial mechanism for Pae87 is proposed in which the P87 peptide plays the role of binding to the outer membrane and disrupting the cell-wall function, either with or without the participation of the catalytic activity of Pae87.

Bap-Independent Biofilm Formation in Staphylococcus xylosus

The biofilm associated protein (Bap) is recognised as the essential component for biofilm formation in Staphylococcus aureus V329 and has been predicted as important for other species as well. Although Bap orthologs are also present in most S. xylosus strains, their contribution to biofilm formation has not yet been demonstrated. In this study, different experimental approaches were used to elucidate the effect of Bap on biofilm formation in S. xylosus and the motif structure of two biofilm-forming S. xylosus strains TMW 2.1023 and TMW 2.1523 was compared to Bap of S. aureus V329. We found that despite an identical structural arrangement into four regions, Bap from S. xylosus differs in key factors to Bap of S. aureus, i.e., isoelectric point of aggregation prone Region B, protein homology and type of repeats. Disruption of bap had no effect on aggregation behavior of selected S. xylosus strains and biofilm formation was unaffected (TMW 2.1023) or at best slightly reduced under neutral conditions (TMW 2.1523). Further, we could not observe any typical characteristics of a S. aureus Bap-positive phenotype such as functional impairment by calcium addition and rough colony morphology on congo red agar (CRA). A dominating role of Bap in cell aggregation and biofilm formation as reported mainly for S. aureus V329 was not observed. In contrast, this work demonstrates that functions of S. aureus Bap cannot easily be extrapolated to S. xylosus Bap, which appears as non-essential for biofilm formation in this species. We therefore suggest that biofilm formation in S. xylosus follows different and multifactorial mechanisms.

Prediction of antimicrobial peptides toxicity based on their physico-chemical properties using machine learning techniques

Background

Antimicrobial peptides are promising tools to fight against ever-growing antibiotic resistance. However, despite many advantages, their toxicity to mammalian cells is a critical obstacle in clinical application and needs to be addressed.

Results

In this study, by using an up-to-date dataset, a machine learning model has been trained successfully to predict the toxicity of antimicrobial peptides. The comprehensive set of features of both physico-chemical and linguistic-based with local and global essences have undergone feature selection to identify key properties behind toxicity of antimicrobial peptides. After feature selection, the hybrid model showed the best performance with a recall of 0. 876 and a F1 score of 0. 849.

Conclusions

The obtained model can be useful in extracting AMPs with low toxicity from AMP libraries in clinical applications. On the other hand, several properties with local nature including positions of strand forming and hydrophobic residues in final selected features show that these properties are critical definer of peptide properties and should be considered in developing models for activity prediction of peptides. The executable code is available at https://git.io/JRZaT.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12859-021-04468-y.

Structural and functional swapping of amyloidogenic and antimicrobial peptides: Redefining the role of amyloidogenic propensity in disease and host defense

Peptides constitute an essential component of all organisms' protein homeostasis ranging from bacteria, plants, and animals. They have organically been evolved to perform a wide range of essential functions, including their role as neurotransmitters, antimicrobial peptides (AMPs), and hormones. AMPs are short peptides synthesized by almost all organisms, implicated in guarding the host from various microbial infections. Their inherent ability to differentiate the target microbes from the host confers them excellent prospects in fighting against microbial infections and affirming their robust therapeutic potential against numerous drug-resistant microbes. Amyloidogenic peptides (AMYs) represent another class of short peptides armed with inherent aggregation propensity and form fibrillar aggregates rich in cross β-sheet structure. They are often involved in various degenerative diseases, including Alzheimer's disease (AD), Parkinson's disease (PD), and type-2 diabetes. Although these two distinct classes of peptides (i.e., AMPs and AMYs) appear to be functionally divergent, recent studies suggest that they possess a significant degree of structural and functional reciprocity. Consistent with this, many AMPs display amphiphilic nature, and hence, they can facilitate membrane remodeling processes, such as pore formation and fusion, similar to AMYs. The mounting evidence suggests the inherent ability of AMPs to self-assemble to form amyloid-like structures. On the other hand, the demonstration of antimicrobial properties of AMYs in their monomeric conformation provides a hint about the existence of an evolutionary linkage between these two classes of peptides. The congregation of specific amino acids to form aggregation-prone regions in a protein/peptide might have served as an evolutionary reservoir from which AMPs and AMYs were consecutively evolved. The current article reviews the fundamental features of the AMPs, AMYs, and their inter-relatedness and emerging paradigm for their inter-conversion.

Machine learning analyses of antibody somatic mutations predict immunoglobulin light chain toxicity

In systemic light chain amyloidosis (AL), pathogenic monoclonal immunoglobulin light chains (LC) form toxic aggregates and amyloid fibrils in target organs. Prompt diagnosis is crucial to avoid permanent organ damage, but delayed diagnosis is common because symptoms usually appear only after strong organ involvement. Here we present LICTOR, a machine learning approach predicting LC toxicity in AL, based on the distribution of somatic mutations acquired during clonal selection. LICTOR achieves a specificity and a sensitivity of 0.82 and 0.76, respectively, with an area under the receiver operating characteristic curve (AUC) of 0.87. Tested on an independent set of 12 LCs sequences with known clinical phenotypes, LICTOR achieves a prediction accuracy of 83%. Furthermore, we are able to abolish the toxic phenotype of an LC by in silico reverting two germline-specific somatic mutations identified by LICTOR, and by experimentally assessing the loss of in vivo toxicity in a Caenorhabditis elegans model. Therefore, LICTOR represents a promising strategy for AL diagnosis and reducing high mortality rates in AL.

Aggregation Prone Regions in Antibody Sequences Raised Against Vibrio cholerae: A Bioinformatic Approach

Background:

Cholera, a diarrheal illness, causes millions of deaths worldwide due to large outbreaks. The monoclonal antibody used as therapeutic purposes of cholera is prone to be unstable due to various factors including self-aggregation.

Objectives:

In this bioinformatic analysis, we identified the aggregation prone regions (APRs) of antibody sequences of different immunogens (i.e., CTB, ZnM-CTB, ZnP-CTB, TcpA-CT-CTB, ZnM-TcpA-CT-CTB, ZnP-TcpA-CT-CTB, ZnM-TcpA, ZnP-TcpA, TcpA-CT-TcpA, ZnM-TcpACT- TcpA, ZnP-TcpA-CT-TcpA, Ogawa, Inaba and ZnM-Inaba) raised against Vibrio cholerae.

Methods:

To determine APRs in antibody sequences that were generated after immunizing Vibrio cholerae immunogens on Mus musculus, a total of 94 sequences were downloaded as FASTA format from a protein database and the algorithms such as Tango, Waltz, PASTA 2.0, and AGGRESCAN were followed to analyze probable APRs in all of the sequences.

Results:

A remarkably high number of regions in the monoclonal antibodies were identified to be APRs which could explain a cause of instability/short term protection of the anticholera vaccine.

Conclusion:

To increase the stability, it would be interesting to eliminate the APR residues from the therapeutic antibodies in such a way that the antigen-binding sites or the complementarity determining region loops involved in antigen recognition are not disrupted.

New Human Islet Amyloid Polypeptide Fragments Susceptible to Aggregation

Human Islet Amyloid Polypeptide (hIAPP) plays a key role in the pathogenesis of type II diabetes. The aim of this research was to search for new amyloidogenic fragments of hIAPP. An initial attempt to predict the amyloidogenic cores of polypeptides/proteins using five different computer programs did not provide conclusive results. Therefore, we synthesized hIAPP fragments covering the entire hormone. The fragments were assessed for their aggregation ability, using recommended methods to search for the amyloidogenic fragments of the polypeptides/proteins. It was found that fragments (18-22) H-HSSNN-OH and (33-37) H-GSNTY-NH₂ aggregate and form stable amyloid-like structures. Both of these fragments have a much higher antiproliferative activity relative to the RIN-5F cell compared to the (23-27) H-FGAIL-OH fragment widely regarded as the amyloidogenic core of amylin. The analog of (33-37) H-GSNTY-NH₂ containing a free carboxy group on the C-terminal amino acid (H-GSNTY-OH) does not have amyloidogenic properties and can therefore be considered as a potential inhibitor of amylin aggregation. Research on the use of non-aggregating amylin fragments as potential hormone aggregation inhibitors is ongoing.

Structure-based machine-guided mapping of amyloid sequence space reveals uncharted sequence clusters with higher solubilities

The amyloid conformation can be adopted by a variety of sequences, but the precise boundaries of amyloid sequence space are still unclear. The currently charted amyloid sequence space is strongly biased towards hydrophobic, beta-sheet prone sequences that form the core of globular proteins and by Q/N/Y rich yeast prions. Here, we took advantage of the increasing amount of high-resolution structural information on amyloid cores currently available in the protein databank to implement a machine learning approach, named Cordax (https://cordax.switchlab.org), that explores amyloid sequence beyond its current boundaries. Clustering by t-Distributed Stochastic Neighbour Embedding (t-SNE) shows how our approach resulted in an expansion away from hydrophobic amyloid sequences towards clusters of lower aliphatic content and higher charge, or regions of helical and disordered propensities. These clusters uncouple amyloid propensity from solubility representing sequence flavours compatible with surface-exposed patches in globular proteins, functional amyloids or sequences associated to liquid-liquid phase transitions.

An increasing number of amyloid structures are determined. Here, the authors present the structure-based amyloid core sequence prediction method Cordax that is based on machine learning and allows the detection of aggregation-prone regions with higher solubility, disorder and surface exposure, and furthermore predicts the structural topology, orientation and overall architecture of the resulting putative fibril core.

Aggrescan3D standalone package for structure-based prediction of protein aggregation properties

SUMMARY

Aggrescan3D (A3D) standalone is a multiplatform Python package for structure-based prediction of protein aggregation properties and rational design of protein solubility. A3D allows the re-design of protein solubility by combining structural aggregation propensity and stability predictions, as demonstrated by a recent experimental study. It also enables predicting the impact of protein conformational fluctuations on the aggregation properties. The standalone A3D version is an upgrade of the original web server implementation-it introduces a number of customizable options, automated analysis of multiple mutations and offers a flexible computational framework for merging it with other computational tools.

AVAILABILITY AND IMPLEMENTATION

A3D standalone is distributed under the MIT license, which is free for academic and non-profit users. It is implemented in Python. The A3D standalone source code, wiki with documentation and examples of use, and installation instructions for Linux, macOS and Windows are available in the A3D standalone repository at https://bitbucket.org/lcbio/aggrescan3d.

Aggrescan3D (A3D) 2.0: prediction and engineering of protein solubility

Protein aggregation is a hallmark of a growing number of human disorders and constitutes a major bottleneck in the manufacturing of therapeutic proteins. Therefore, there is a strong need of in-silico methods that can anticipate the aggregative properties of protein variants linked to disease and assist the engineering of soluble protein-based drugs. A few years ago, we developed a method for structure-based prediction of aggregation properties that takes into account the dynamic fluctuations of proteins. The method has been made available as the Aggrescan3D (A3D) web server and applied in numerous studies of protein structure-aggregation relationship. Here, we present a major update of the A3D web server to version 2.0. The new features include: extension of dynamic calculations to significantly larger and multimeric proteins, simultaneous prediction of changes in protein solubility and stability upon mutation, rapid screening for functional protein variants with improved solubility, a REST-ful service to incorporate A3D calculations in automatic pipelines, and a new, enhanced web server interface. A3D 2.0 is freely available at: http://biocomp.chem.uw.edu.pl/A3D2/

Validation of a de-immunization strategy for monoclonal antibodies using cynomolgus macaque as a surrogate for human

The immunogenicity of biotherapeutics presents a major challenge during the clinical development of new protein drugs including monoclonal antibodies. To address this, multiple humanization and de-immunization techniques that employ in silico algorithms and in vitro test systems have been proposed and implemented. However, the success of these approaches has been variable and to date, the ability of these techniques to predict immunogenicity has not been systematically tested in humans or other primates. This study tested whether antibody humanization and de-immunization strategies reduce the risk of anti-drug antibody (ADA) development using cynomolgus macaque as a surrogate for human. First human-cyno chimeric antibodies were constructed by grafting the variable domains of the adalimumab and golimumab monoclonal antibodies onto cynomolgus macaque IgG1 and Igκ constant domains followed by framework germlining to cyno to reduce the xenogenic content. Next, B and T cell epitopes and aggregation-prone regions were identified using common in silico methods to select domains with an ADA risk for additional modification. The resultant engineered antibodies had a comparable affinity for TNFα, demonstrated similar biophysical properties, and exhibited significantly reduced ADA levels in cynomolgus macaque compared with the parental antibodies, with a corresponding improvement in the pharmacokinetic profile. Notably, plasma concentrations of the engineered antibodies were quantifiable through 504 hours (chimeric) and 840 hours (germlined/de-immunized), compared with only 336 hours (adalimumab) or 336-672 hours (golimumab). The results point to the significant value in the investment in these engineering strategies as an important guide for monoclonal antibody optimization that can contribute to improved clinical outcomes.

M. Brito R. Structure and Aggregation Mechanisms in Amyloids

The aggregation of a polypeptide chain into amyloid fibrils and their accumulation and deposition into insoluble plaques and intracellular inclusions is the hallmark of several misfolding diseases known as amyloidoses. Alzheimer′s, Parkinson′s and Huntington’s diseases are some of the approximately 50 amyloid diseases described to date. The identification and characterization of the molecular species critical for amyloid formation and disease development have been the focus of intense scrutiny. Methods such as X-ray and electron diffraction, solid-state nuclear magnetic resonance spectroscopy (ssNMR) and cryo-electron microscopy (cryo-EM) have been extensively used and they have contributed to shed a new light onto the structure of amyloid, revealing a multiplicity of polymorphic structures that generally fit the cross-β amyloid motif. The development of rational therapeutic approaches against these debilitating and increasingly frequent misfolding diseases requires a thorough understanding of the molecular mechanisms underlying the amyloid cascade. Here, we review the current knowledge on amyloid fibril formation for several proteins and peptides from a kinetic and thermodynamic point of view, the structure of the molecular species involved in the amyloidogenic process, and the origin of their cytotoxicity.

Prolines Affect the Nucleation Phase of Amyloid Fibrillation Reaction; Mutational Analysis of Human Stefin B

Proline residues play a prominent role in protein folding and aggregation. We investigated the influence of single prolines and their combination on oligomerization and the amyloid fibrillation reaction of human stefin B (stB). The proline mutants influenced the distribution of oligomers between monomers, dimers, and tetramers as shown by the size-exclusion chromatography. Only P74S showed higher oligomers, reminiscent of the molten globule reported previously for the P74S of stB-Y31 variant. The proline mutants also inhibited to various degree the amyloid fibrillation reaction. At 30 and 37 °C, inhibition was complete for the P74S single mutant, two double mutants (P6L P74S and P74S P79S), and for the triple mutant P6L P11S P74S. At 30 °C the single mutant P6L completely inhibited the reaction, while P11S and P79S formed amyloid fibrils with a prolonged lag phase. P36D did not show a lag phase, reminiscent of a downhill polymerization model. At 37 °C in addition to P36D, P11S, and P79S, P6L and P11S P74S also started to fibrillate; however, the yield of the fibrils was much lower than that of the wild-type protein as judged by transmission electron microscopy. Thus, Pro 74 cis/trans isomerization proves to be the key event, acting as a switch toward an amyloid transition. Using our previous model of nucleation and growth, we simulated the kinetics of all the mutants that exhibited sigmoidal fibrillation curves. To our surprise, the nucleation phase was most affected by Pro cis/trans isomerism, rather than the fibril elongation phase.

The fitness cost and benefit of phase-separated protein deposits

Phase separation of soluble proteins into insoluble deposits is associated with numerous diseases. However, protein deposits can also function as membrane-less compartments for many cellular processes. What are the fitness costs and benefits of forming such deposits in different conditions? Using a model protein that phase-separates into deposits, we distinguish and quantify the fitness contribution due to the loss or gain of protein function and deposit formation in yeast. The environmental condition and the cellular demand for the protein function emerge as key determinants of fitness. Protein deposit formation can influence cell-to-cell variation in free protein abundance between individuals of a cell population (i.e., gene expression noise). This results in variable manifestation of protein function and a continuous range of phenotypes in a cell population, favoring survival of some individuals in certain environments. Thus, protein deposit formation by phase separation might be a mechanism to sense protein concentration in cells and to generate phenotypic variability. The selectable phenotypic variability, previously described for prions, could be a general property of proteins that can form phase-separated assemblies and may influence cell fitness.

Bacterial inclusion bodies are industrially exploitable amyloids

Understanding the structure, functionalities and biology of functional amyloids is an issue of emerging interest. Inclusion bodies, namely protein clusters formed in recombinant bacteria during protein production processes, have emerged as unanticipated, highly tunable models for the scrutiny of the physiology and architecture of functional amyloids. Based on an amyloidal skeleton combined with varying amounts of native or native-like protein forms, bacterial inclusion bodies exhibit an unusual arrangement that confers mechanical stability, biological activity and conditional protein release, being thus exploitable as versatile biomaterials. The applicability of inclusion bodies in biotechnology as enriched sources of protein and reusable catalysts, and in biomedicine as biocompatible topographies, nanopills or mimetics of endocrine secretory granules has been largely validated. Beyond these uses, the dissection of how recombinant bacteria manage the aggregation of functional protein species into structures of highly variable complexity offers insights about unsuspected connections between protein quality (conformational status compatible with functionality) and cell physiology.

Screening Protein Aggregation in Cells Using Fluorescent Labels Coupled to Flow Cytometry

Protein misfolding and aggregation into amyloid conformations have been described to underlie the onset of several human neurodegenerative diseases. Although a large number of biophysical approaches are available to study amyloids in vitro, we still need robust methods to address their self-assembly in living cells. In this context, simple cellular models, like bacteria and yeast, expressing recombinant amyloidogenic proteins are emerging as convenient systems for studying the formation of protein inclusions, their toxicity, propagation, and interactions. We describe here a simple and fast flow cytometry method able to detect intracellular inclusions, as well as to analyze the distribution of the amyloidogenic protein of interest in intact cells. Using specific fluorescent amyloid-dyes, such as thioflavin-S and ProteoStat, or the fusion of fluorescent molecules, such as GFP, the technique can be applied in the quantification of intracellular amyloid content, for the screening of antiamyloidogenic compounds, and to test epigenetic or environmental conditions able to modulate amyloid deposition in vivo.

Amyloid cores in prion domains: Key regulators for prion conformational conversion

Despite the significant efforts devoted to decipher the particular protein features that encode for a prion or prion-like behavior, they are still poorly understood. The well-characterized yeast prions constitute an ideal model system to address this question, because, in these proteins, the prion activity can be univocally assigned to a specific region of their sequence, known as the prion forming domain (PFD). These PFDs are intrinsically disordered, relatively long and, in many cases, of low complexity, being enriched in glutamine/asparagine residues. Computational analyses have identified a significant number of proteins having similar domains in the human proteome. The compositional bias of these regions plays an important role in the transition of the prions to the amyloid state. However, it is difficult to explain how composition alone can account for the formation of specific contacts that position correctly PFDs and provide the enthalpic force to compensate for the large entropic cost of immobilizing these domains in the initial assemblies. We have hypothesized that short, sequence-specific, amyloid cores embedded in PFDs can perform these functions and, accordingly, act as preferential nucleation centers in both spontaneous and seeded aggregation. We have shown that the implementation of this concept in a prediction algorithm allows to score the prion propensities of putative PFDs with high accuracy. Recently, we have provided experimental evidence for the existence of such amyloid cores in the PFDs of Sup35, Ure2, Swi1, and Mot3 yeast prions. The fibrils formed by these short stretches may recognize and promote the aggregation of the complete proteins inside cells, being thus a promising tool for targeted protein inactivation.

Mining databases for protein aggregation: a review

Protein aggregation is an active area of research in recent decades, since it is the most common and troubling indication of protein instability. Understanding the mechanisms governing protein aggregation and amyloidogenesis is a key component to the aetiology and pathogenesis of many devastating disorders, including Alzheimer's disease or type 2 diabetes. Protein aggregation data are currently found "scattered" in an increasing number of repositories, since advances in computational biology greatly influence this field of research. This review exploits the various resources of aggregation data and attempts to distinguish and analyze the biological knowledge they contain, by introducing protein-based, fragment-based and disease-based repositories, related to aggregation. In order to gain a broad overview of the available repositories, a novel comprehensive network maps and visualizes the current association between aggregation databases and other important databases and/or tools and discusses the beneficial role of community annotation. The need for unification of aggregation databases in a common platform is also addressed.

Usage of a dataset of NMR resolved protein structures to test aggregation versus solubility prediction algorithms

There has been an increased interest in computational methods for amyloid and (or) aggregate prediction, due to the prevalence of these aggregates in numerous diseases and their recently discovered functional importance. To evaluate these methods, several datasets have been compiled. Typically, aggregation-prone regions of proteins, which form aggregates or amyloids in vivo, are more than 15 residues long and intrinsically disordered. However, the number of such experimentally established amyloid forming and non-forming sequences are limited, not exceeding one hundred entries in existing databases. In this work, we parsed all available NMR-resolved protein structures from the PDB and assembled a new, sevenfold larger, dataset of unfolded sequences, soluble at high concentrations. We proposed to use these sequences as a negative set for evaluating methods for predicting aggregation in vivo. We also present the results of benchmarking cutting edge tools for the prediction of aggregation versus solubility propensity.

Protein aggregation into insoluble deposits protects from oxidative stress

Protein misfolding and aggregation have been associated with the onset of neurodegenerative disorders. Recent studies demonstrate that the aggregation process can result in a high diversity of protein conformational states, however the identity of the specific species responsible for the cellular damage is still unclear. Here, we use yeast as a model to systematically analyse the intracellular effect of expressing 21 variants of the amyloid-ß-peptide, engineered to cover a continuous range of intrinsic aggregation propensities. We demonstrate the existence of a striking negative correlation between the aggregation propensity of a given variant and the oxidative stress it elicits. Interestingly, each variant generates a specific distribution of protein assemblies in the cell. This allowed us to identify the aggregated species that remain diffusely distributed in the cytosol and are unable to coalesce into large protein inclusions as those causing the highest levels of oxidative damage. Overall, our results indicate that the formation of large insoluble aggregates may act as a protective mechanism to avoid cellular oxidative stress.

Engineered antibody CH2 domains binding to nucleolin: Isolation, characterization and improvement of aggregation

Smaller recombinant antibody fragments are now emerging as alternatives of conventional antibodies. Especially, immunoglobulin (Ig) constant CH2 domain and engineered CH2 with improved stability are promising as scaffolds for selection of specific binders to various antigens. We constructed a yeast display library based on an engineered human IgG1 CH2 scaffold with diversified loop regions. A group of CH2 binders were isolated from this yeast display library by panning against nucleolin, which is a tumor-associated antigen involved in cell proliferation, tumor cell growth and angiogenesis. Out of 20 mutants, we selected 3 clones exhibiting relatively high affinities to nucleolin on yeasts. However, recombinant CH2 mutants aggregated when they were expressed. To find the mechanism of the aggregation, we employed computational prediction approaches through structural homology models of CH2 binders. The analysis of potential aggregation prone regions (APRs) and solvent accessible surface areas (ASAs) indicated two hydrophobic residues, Val264 and Leu309, in the β-sheet, in which replacement of both charged residues led to significant decrease of the protein aggregation. The newly identified CH2 binders could be improved to use as candidate therapeutics or research reagents in the future.

Characterization of Amyloid Cores in Prion Domains

Amyloids consist of repetitions of a specific polypeptide chain in a regular cross-β-sheet conformation. Amyloid propensity is largely determined by the protein sequence, the aggregation process being nucleated by specific and short segments. Prions are special amyloids that become self-perpetuating after aggregation. Prions are responsible for neuropathology in mammals, but they can also be functional, as in yeast prions. The conversion of these last proteins to the prion state is driven by prion forming domains (PFDs), which are generally large, intrinsically disordered, enriched in glutamines/asparagines and depleted in hydrophobic residues. The self-assembly of PFDs has been thought to rely mostly on their particular amino acid composition, rather than on their sequence. Instead, we have recently proposed that specific amyloid-prone sequences within PFDs might be key to their prion behaviour. Here, we demonstrate experimentally the existence of these amyloid stretches inside the PFDs of the canonical Sup35, Swi1, Mot3 and Ure2 prions. These sequences self-assemble efficiently into highly ordered amyloid fibrils, that are functionally competent, being able to promote the PFD amyloid conversion in vitro and in vivo. Computational analyses indicate that these kind of amyloid stretches may act as typical nucleating signals in a number of different prion domains.

Understanding and predicting protein misfolding and aggregation: Insights from proteomics

Protein misfolding and aggregation are being found to be associated with an increasing number of human diseases and premature aging, either because they promote a loss of protein function or, more frequently, because the aggregated species gain a toxic activity. Despite potentially harmful, aggregation seems to be a generic property of polypeptide chains and aggregation-prone protein sequences seem to be ubiquitous, which, counterintuitively, suggests that they serve evolutionary conserved functions. The in vitro study of individual aggregation reactions of a large number of proteins has provided important insights on the structural and sequential determinants of this process. However, it is clear that understanding the role played by protein aggregation and its regulation in health and disease at the cellular, developmental, and evolutionary levels require more global approaches. The use of model organisms and their proteomic analysis hold the power to provide answers to such issues. In the present review, we address how, initially, computational large-scale analysis and, more recently, experimental proteomics are helping us to rationalize how, why and when proteins aggregate, as well as to decipher the strategies organisms have developed to control proteins aggregation propensities.

Data on correlation between Aβ42 structural aggregation propensity and toxicity in bacteria

Protein aggregation and amyloid formation is a hallmark of an increasing number of human disorders. Because protein aggregation is deleterious for the cell physiology and results in a decrease in overall cell fitness, it is thought that natural selection acts to purify aggregating proteins during evolution. This data article contains complementary figures and results related to the research article entitled “Selection against toxic aggregation-prone protein sequences in bacteria” (Navarro et al., 2014) [1]. Here, we used the AGGRESCAN3D (A3D) server, a novel in house predictor that forecasts protein aggregation properties in protein structures to illustrate a striking correlation between the structure-based predictions of aggregation propensities for Alzheimer’s Aβ42 peptide variants and their previously reported deleterious effects in bacteria.

Cryptic Amyloidogenic Elements in the 3' UTRs of Neurofilament Genes Trigger Axonal Neuropathy

Abnormal protein aggregation is observed in an expanding number of neurodegenerative diseases. Here, we describe a mechanism for intracellular toxic protein aggregation induced by an unusual mutation event in families affected by axonal neuropathy. These families carry distinct frameshift variants in NEFH (neurofilament heavy), leading to a loss of the terminating codon and translation of the 3' UTR into an extra 40 amino acids. In silico aggregation prediction suggested the terminal 20 residues of the altered NEFH to be amyloidogenic, which we confirmed experimentally by serial deletion analysis. The presence of this amyloidogenic motif fused to NEFH caused prominent and toxic protein aggregates in transfected cells and disrupted motor neurons in zebrafish. We identified a similar aggregation-inducing mechanism in NEFL (neurofilament light) and FUS (fused in sarcoma), in which mutations are known to cause aggregation in Charcot-Marie-Tooth disease and amyotrophic lateral sclerosis, respectively. In summary, we present a protein-aggregation-triggering mechanism that should be taken into consideration during the evaluation of stop-loss variants.

Deducing the functional characteristics of the human selenoprotein SELK from the structural properties of its intrinsically disordered C-terminal domain

The intrinsically disordered proteins (IDPs) cannot be described by a single structural representation but, due to their high structural fluctuation, through conformational ensembles. Certainly, molecular dynamics (MD) simulations represent a useful tool to study their different conformations capturing the conformational distribution. Our group is focusing on the structural characterization of proteins belonging to the seleno-proteome due to their involvement in cancer. They present disordered domains central for their biological function, and, in particular, SELK is a single-pass transmembrane protein that resides in the endoplasmic reticulum membrane (ER) with a C-terminal domain exposed to the cytoplasm that is known to interact with different components of the endoplasmic reticulum associated to the protein degradation (ERAD) pathway. This protein is found to be up-expressed in hepatocellular carcinoma and in other cancers. In this work we performed a detailed analysis of the C-terminal domain sequence of SELK and discovered that it is characterized by many prolines, and four negatively and eleven positively charged residues, which are crucial for its biological activity. This region can be considered as a weak polyelectrolyte and, specifically, a polycation, with high disordered propensity and different phosphorylation sites dislocated along the sequence. Then, we modeled its three-dimensional structure by performing MD simulations in water at neutral pH to analyze the structural stability as well as to identify the presence of HUB residues that play a key structural role as evidenced by the residue-residue interaction network analysis. Through this approach, we demonstrate that the C-terminal domain of SELK (i) presents a poor content of regular secondary structure elements, (ii) is dynamically stabilized by a network of intra-molecular H-bonds and H-bonds with water molecules, (iii) is highly fluctuating and, therefore, can be described only through a conformational ensemble, where we evidenced a distribution of equilibrium conformers which continuously inter-change their conformations. Finally to verify the specific role played by the negative charges, we also performed MD simulations at acidic pH. Overall, all the obtained results evidenced that SELK has the dynamic structural features to be defined as a HUB protein able to interact with multiple members. Therefore, considering the possible role that this protein can have in cancer development and progression, it can represent a target for drug design studies.

In vivo amyloid aggregation kinetics tracked by time-lapse confocal microscopy in real-time

Amyloid polymerization underlies an increasing number of human diseases. Despite this process having been studied extensively in vitro, aggregation is a difficult process to track in vivo due to methodological limitations and the slow kinetics of aggregation reactions in cells and tissues. Herein we exploit the amyloid properties of the inclusions bodies (IBs) formed by amyloidogenic proteins in bacteria to address the kinetics of in vivo amyloid aggregation. To this aim we used time-lapse confocal microscopy and a fusion of the amyloid-beta peptide (A β42) with a fluorescent reporter. This strategy allowed us to follow the intracellular kinetics of amyloid-like aggregation in real-time and to discriminate between variants exhibiting different in vivo aggregation propensity. Overall, the approach opens the possibility to assess the impact of point mutations as well as potential anti-aggregation drugs in the process of amyloid formation in living cells.

Triton X-100 enhances the solubility and secretion ratio of aggregation-prone pullulanase produced in Escherichia coli

The pullulanase from Bacillus deramificans is an industrially useful starch-debranching enzyme that is difficult to produce in large quantities. In this study, B. deramificans pullulanase was found to be an aggregation-prone protein that can be solubilized from the insoluble fraction by surfactants in vitro. Studying the effects of various surfactants on pullulanase production in Escherichia coli in shake flasks revealed that optimal pullulanase production could be obtained by adding 0.5% Triton X-100 during the later period of fermentation. A modified fed-batch fermentation strategy was then applied to the production of pullulanase in a 3-L fermentor. When supplemented with 0.5% Triton X-100 at 40 h, the maximal extracellular pullulanase production and secretion ratio were 812.4 U mL(-1) and 86.0%, which were 46.2- and 47.8-fold that of the control, respectively.

Aggregation propensity of neuronal receptors: potential implications in neurodegenerative disorders

Misfolding and aggregation of proteins in tissues is linked to the onset of a diverse set of human neurodegenerative disorders, including Alzheimer's and Parkinson's diseases. In these pathologies proteins usually aggregate into highly ordered and β-sheet enriched amyloid fibrils. However, the formation of these toxic structures is not restricted to a reduced set of polypeptides but rather an intrinsic property of proteins. This suggests that the number of proteins involved in conformational disorders might be much larger than previously thought. The propensity of a protein to form amyloid assemblies is imprinted in its sequence and can be read using computational approaches. Here, we exploit four of these algorithms to analyze the presence of aggregation-prone regions in the sequence and structure of the extracellular domains of several neuroreceptors, with the idea of identifying patches that can interact anomalously with other aggregation-prone molecules such as the amyloid-β peptide or promote their self-assembly. The number of amyloidogenic regions in these domains is rather low but they are significantly exposed to solvent and therefore are suitable for interactions. We find a significant overlap between aggregation-prone regions and receptors interfaces and/or ligand-binding sites, which illustrates an unavoidable competition between the formation of functional native interactions and that of dangerous amyloid-like contacts leading to disease.

The prion-like RNA-processing protein HNRPDL forms inherently toxic amyloid-like inclusion bodies in bacteria

Background

The formation of protein inclusions is connected to the onset of many human diseases. Human RNA binding proteins containing intrinsically disordered regions with an amino acid composition resembling those of yeast prion domains, like TDP-43 or FUS, are being found to aggregate in different neurodegenerative disorders. The structure of the intracellular inclusions formed by these proteins is still unclear and whether these deposits have an amyloid nature or not is a matter of debate. Recently, the aggregation of TDP-43 has been modelled in bacteria, showing that TDP-43 inclusion bodies (IBs) are amorphous but intrinsically neurotoxic. This observation raises the question of whether it is indeed the lack of an ordered structure in these human prion-like protein aggregates the underlying cause of their toxicity in different pathological states.

Results

Here we characterize the IBs formed by the human prion-like RNA-processing protein HNRPDL. HNRPDL is linked to the development of limb-girdle muscular dystrophy 1G and shares domain architecture with TDP-43. We show that HNRPDL IBs display characteristic amyloid hallmarks, since these aggregates bind to amyloid dyes in vitro and inside the cell, they are enriched in intermolecular β-sheet conformation and contain inner amyloid-like fibrillar structure. In addition, despite their ordered structure, HNRPDL IBs are highly neurotoxic.

Conclusions

Our results suggest that at least some of the disorders caused by the aggregation of human prion-like proteins would rely on the formation of classical amyloid assemblies rather than being caused by amorphous aggregates. They also illustrate the power of microbial cell factories to model amyloid aggregation.

Electronic supplementary material

The online version of this article (doi:10.1186/s12934-015-0284-7) contains supplementary material, which is available to authorized users.

Association between foldability and aggregation propensity in small disulfide-rich proteins

AIMS

Disulfide-rich domains (DRDs) are small proteins whose native structure is stabilized by the presence of covalent disulfide bonds. These domains are versatile and can perform a wide range of functions. Many of these domains readily unfold on disulfide bond reduction, suggesting that in the absence of covalent bonding they might display significant disorder.

RESULTS

Here, we analyzed the degree of disorder in 97 domains representative of the different DRDs families and demonstrate that, in terms of sequence, many of them can be classified as intrinsically disordered proteins (IDPs) or contain predicted disordered regions. The analysis of the aggregation propensity of these domains indicates that, similar to IDPs, their sequences are more soluble and have less aggregating regions than those of other globular domains, suggesting that they might have evolved to avoid aggregation after protein synthesis and before they can attain its compact and covalently linked native structure.

INNOVATION AND CONCLUSION

DRDs, which resemble IDPs in the reduced state and become globular when their disulfide bonds are formed, illustrate the link between protein folding and aggregation propensities and how these two properties cannot be easily dissociated, determining the main traits of the folding routes followed by these small proteins to attain their native oxidized states.

Inclusion bodies: not that bad…

The formation of inclusion bodies (IBs) constitute a frequent event during the production of heterologous proteins in bacterial hosts. Although the mechanisms leading to their formation are not completely understood, empirical data have been exploited trying to predict the aggregation propensity of specific proteins while a great number of strategies have been developed to avoid the generation of IBs. However, in many cases, the formation of such aggregates can be considered an advantage for basic research as for protein production. In this review, we focus on this positive side of IBs formation in bacteria. We present a compilation on recent advances on the understanding of IBs formation and their utilization as a model to understand protein aggregation and to explore strategies to control this process. We include recent information about their composition and structure, their use as an attractive approach to produce low cost proteins and other promising applications in Biomedicine.

N-terminal hydrophobic amino acids of activating transcription factor 5 (ATF5) protein confer interleukin 1β (IL-1β)-induced stabilization

Activating transcription factor 5 (ATF5) is a stress-response transcription factor that responds to amino acid limitation and exposure to cadmium chloride (CdCl2) and sodium arsenite (NaAsO2). The N-terminal amino acids contribute to the destabilization of the ATF5 protein in steady-state conditions and serve as a stabilization domain in the stress response after CdCl2 or NaAsO2 exposure. In this study, we show that interleukin 1β (IL-1β), a proinflammatory cytokine, increases the expression of ATF5 protein in HepG2 hepatoma cells in part by stabilizing the ATF5 protein. The N-terminal domain rich in hydrophobic amino acids that is predicted to form a hydrophobic network was responsible for destabilization in steady-state conditions and served as an IL-1β response domain. Furthermore, IL-1β increased the translational efficiency of ATF5 mRNA via the 5' UTRα and phosphorylation of the eukaryotic translation initiation factor 2α (eIF2α). ATF5 knockdown in HepG2 cells up-regulated the IL-1β-induced expression of the serum amyloid A 1 (SAA1) and SAA2 genes. Our results show that the N-terminal hydrophobic amino acids play an important role in the regulation of ATF5 protein expression in IL-1β-mediated immune response and that ATF5 is a negative regulator for IL-1β-induced expression of SAA1 and SAA2 in HepG2 cells.

On the role of aggregation prone regions in protein evolution, stability, and enzymatic catalysis: insights from diverse analyses

The various roles that aggregation prone regions (APRs) are capable of playing in proteins are investigated here via comprehensive analyses of multiple non-redundant datasets containing randomly generated amino acid sequences, monomeric proteins, intrinsically disordered proteins (IDPs) and catalytic residues. Results from this study indicate that the aggregation propensities of monomeric protein sequences have been minimized compared to random sequences with uniform and natural amino acid compositions, as observed by a lower average aggregation propensity and fewer APRs that are shorter in length and more often punctuated by gate-keeper residues. However, evidence for evolutionary selective pressure to disrupt these sequence regions among homologous proteins is inconsistent. APRs are less conserved than average sequence identity among closely related homologues (≥80% sequence identity with a parent) but APRs are more conserved than average sequence identity among homologues that have at least 50% sequence identity with a parent. Structural analyses of APRs indicate that APRs are three times more likely to contain ordered versus disordered residues and that APRs frequently contribute more towards stabilizing proteins than equal length segments from the same protein. Catalytic residues and APRs were also found to be in structural contact significantly more often than expected by random chance. Our findings suggest that proteins have evolved by optimizing their risk of aggregation for cellular environments by both minimizing aggregation prone regions and by conserving those that are important for folding and function. In many cases, these sequence optimizations are insufficient to develop recombinant proteins into commercial products. Rational design strategies aimed at improving protein solubility for biotechnological purposes should carefully evaluate the contributions made by candidate APRs, targeted for disruption, towards protein structure and activity.

Biotechnology requires the large-scale expression, yield, and storage of recombinant proteins. Each step in protein production has the potential to cause aggregation as proteins, not evolved to exist outside the cell, endure the various steps involved in commercial manufacturing processes. Mechanistic studies into protein aggregation have revealed that certain sequence regions contribute more to the aggregation propensity of a protein than other sequence regions do. Efforts to disrupt these regions have thus far indicated that rational sequence engineering is a useful technique to reduce the aggregation of biotechnologically relevant proteins. To improve our ability to rationally engineer proteins with enhanced expression, solubility, and shelf-life we conducted extensive analyses of aggregation prone regions (APRs) within protein sequences to characterize the various roles these regions play in proteins. Findings from this work indicate that protein sequences have evolved by minimizing their aggregation propensities. However, we also found that many APRs are conserved in protein families and are essential to maintain protein stability and function. Therefore, the contributions that APRs, targeted for disruption, make towards protein stability and function should be carefully evaluated when improving protein solubility via rational design.