Predicting mechanical properties of silk from its amino acid sequences via machine learning.

The silk fiber is increasingly being sought for its superior mechanical properties, biocompatibility, and eco-friendliness, making it promising as a base material for various applications. One of the characteristics of protein fibers, such as silk, is that their mechanical properties are significantly dependent on the amino acid sequence. Numerous studies have been conducted to determine the specific relationship between the amino acid sequence of silk and its mechanical properties. Still, the relationship between the amino acid sequence of silk and its mechanical properties is yet to be clarified. Other fields have adopted machine learning (ML) to establish a relationship between the inputs, such as the ratio of different input material compositions and the resulting mechanical properties. We have proposed a method to convert the amino acid sequence into numerical values for input and succeeded in predicting the mechanical properties of silk from its amino acid sequences. Our study sheds light on predicting mechanical properties of silk fiber from respective amino acid sequences.

Neural networks to learn protein sequence-function relationships from deep mutational scanning data.

The mapping from protein sequence to function is highly complex, making it challenging to predict how sequence changes will affect a protein's behavior and properties. We present a supervised deep learning framework to learn the sequence-function mapping from deep mutational scanning data and make predictions for new, uncharacterized sequence variants. We test multiple neural network architectures, including a graph convolutional network that incorporates protein structure, to explore how a network's internal representation affects its ability to learn the sequence-function mapping. Our supervised learning approach displays superior performance over physics-based and unsupervised prediction methods. We find that networks that capture nonlinear interactions and share parameters across sequence positions are important for learning the relationship between sequence and function. Further analysis of the trained models reveals the networks' ability to learn biologically meaningful information about protein structure and mechanism. Finally, we demonstrate the models' ability to navigate sequence space and design new proteins beyond the training set. We applied the protein G B1 domain (GB1) models to design a sequence that binds to immunoglobulin G with substantially higher affinity than wild-type GB1.

LxCxD motif of the APC/C coactivator subunit FZR1 is critical for interaction with the retinoblastoma protein.

Retinoblastoma protein (pRB) regulates cell cycle by utilizing different regions of its pocket domain for interacting with E2F family of transcription factors and with cellular and viral proteins containing an LxCxE motif. An LxCxE-like motif, LxCxD, is present in FZR1, an adaptor protein of the multi-subunit E3 ligase complex anaphase-promoting complex/cyclosome (APC/C). The APC/CFZR1 complex regulates the timely degradation of multiple cell cycle proteins for mitotic exit and maintains G1 state. We report that FZR1 interacts with pRB via its LxCxD motif. By using point mutations, we found that the cysteine residue in the FZR1 LxCxD motif is critical for direct interaction with pRb. The direct binding of the LxCxD motif of FZR1 to the pRB LxCxE binding pocket is confirmed by using human papillomavirus protein E7 as a competitor, both in vitro and in vivo. While mutation of the cysteine residue significantly disrupts FZR1 interaction with pRB, this motif does not affect FZR1 and core APC/C association. Expression of the FZR1 point mutant results in accumulation of S-phase kinase-associated protein 2 (SKP2) and Polo-like kinase 1 (PLK1), while p27Kip1 and p21Cip1 proteins are downregulated, indicating a G1 cell cycle defect. Consistently, cells containing point mutant FZR1 enter the S phase prematurely. Together our results suggest that the LxCxD motif of FZR1 is a critical determinant for the interaction between FZR1 and pRB and is important for G1 restriction.

CSConv2d: A 2-D Structural Convolution Neural Network with a Channel and Spatial Attention Mechanism for Protein-Ligand Binding Affinity Prediction.

The binding affinity of small molecules to receptor proteins is essential to drug discovery and drug repositioning. Chemical methods are often time-consuming and costly, and models for calculating the binding affinity are imperative. In this study, we propose a novel deep learning method, namely CSConv2d, for protein-ligand interactions' prediction. The proposed method is improved by a DEEPScreen model using 2-D structural representations of compounds as input. Furthermore, a channel and spatial attention mechanism (CS) is added in feature abstractions. Data experiments conducted on ChEMBLv23 datasets show that CSConv2d performs better than the original DEEPScreen model in predicting protein-ligand binding affinity, as well as some state-of-the-art DTIs (drug-target interactions) prediction methods including DeepConv-DTI, CPI-Prediction, CPI-Prediction+CS, DeepGS and DeepGS+CS. In practice, the docking results of protein (PDB ID: 5ceo) and ligand (Chemical ID: 50D) and a series of kinase inhibitors are operated to verify the robustness.

Amylin antibodies frequently display cross-reactivity with CGRP: characterization of eight amylin antibodies.

Amylin is a 37-amino acid endocrine hormone secreted from the pancreas in response to nutrient intake, acting centrally to promote meal-ending satiation. With many studies linking amylin action to the nervous system, determining the distribution or expression of amylin in the nervous system is critical. However, amylin shares sequence identity and structural homology to the related neuropeptide calcitonin gene-related peptide (CGRP). This creates challenges in identifying selective amylin antibodies that do not cross-react with CGRP, especially in neural tissues, where CGRP is densely packed into secretory vesicles. Here, we characterized eight amylin antibodies to determine their ability to detect amylin and cross-react with rat or human αCGRP, using immunoblots and preabsorption controls in rat pancreas. We observed that amylin antibodies frequently cross-reacted with αCGRP and are therefore not suitable for use in tissues that highly express CGRP. Earlier work using these antibodies should be revisited in light of our findings.

Breakdown of supersaturation barrier links protein folding to amyloid formation.

The thermodynamic hypothesis of protein folding, known as the "Anfinsen's dogma" states that the native structure of a protein represents a free energy minimum determined by the amino acid sequence. However, inconsistent with the Anfinsen's dogma, globular proteins can misfold to form amyloid fibrils, which are ordered aggregates associated with diseases such as Alzheimer's and Parkinson's diseases. Here, we present a general concept for the link between folding and misfolding. We tested the accessibility of the amyloid state for various proteins upon heating and agitation. Many of them showed Anfinsen-like reversible unfolding upon heating, but formed amyloid fibrils upon agitation at high temperatures. We show that folding and amyloid formation are separated by the supersaturation barrier of a protein. Its breakdown is required to shift the protein to the amyloid pathway. Thus, the breakdown of supersaturation links the Anfinsen's intramolecular folding universe and the intermolecular misfolding universe.

The high catalytic rate of the cold-active Vibrio alkaline phosphatase requires a hydrogen bonding network involving a large interface loop.

The role of surface loops in mediating communication through residue networks is still a relatively poorly understood part in the study of cold adaptation of enzymes, especially in terms of their quaternary interactions. Alkaline phosphatase (AP) from the psychrophilic marine bacterium Vibrio splendidus (VAP) is characterized by an analogous large surface loop in each monomer, referred to as the large loop, that hovers over the active site of the other monomer. It presumably has a role in the high catalytic efficiency of VAP which accompanies its extremely low thermal stability. Here, we designed several different variants of VAP with the aim of removing intersubunit interactions at the dimer interface. Breaking the intersubunit contacts from one residue in particular (Arg336) reduced the temperature stability of the catalytically potent conformation and caused a 40% drop in catalytic rate. The high catalytic rates of enzymes from cold-adapted organisms are often associated with increased dynamic flexibility. Comparison of the relative B-factors of the R336L crystal structure to that of the wild-type confirmed surface flexibility was increased in a loop on the opposite monomer, but not in the large loop. The increase in flexibility resulted in a reduced catalytic rate. The large loop increases the area of the interface between the subunits through its contacts and may facilitate an alternating structural cycle demanded by a half-of-sites reaction mechanism through stronger ties, as the dimer oscillates between high affinity (active) or low phosphoryl group affinity (inactive).

Exploring the sequence fitness landscape of a bridge between protein folds.

Most foldable protein sequences adopt only a single native fold. Recent protein design studies have, however, created protein sequences which fold into different structures apon changes of environment, or single point mutation, the best characterized example being the switch between the folds of the GA and GB binding domains of streptococcal protein G. To obtain further insight into the design of sequences which can switch folds, we have used a computational model for the fitness landscape of a single fold, built from the observed sequence variation of protein homologues. We have recently shown that such coevolutionary models can be used to design novel foldable sequences. By appropriately combining two of these models to describe the joint fitness landscape of GA and GB, we are able to describe the propensity of a given sequence for each of the two folds. We have successfully tested the combined model against the known series of designed GA/GB hybrids. Using Monte Carlo simulations on this landscape, we are able to identify pathways of mutations connecting the two folds. In the absence of a requirement for domain stability, the most frequent paths go via sequences in which neither domain is stably folded, reminiscent of the propensity for certain intrinsically disordered proteins to fold into different structures according to context. Even if the folded state is required to be stable, we find that there is nonetheless still a wide range of sequences which are close to the transition region and therefore likely fold switches, consistent with recent estimates that fold switching may be more widespread than had been thought.

Ancestral sequence reconstruction produces thermally stable enzymes with mesophilic enzyme-like catalytic properties.

Enzymes have high catalytic efficiency and low environmental impact, and are therefore potentially useful tools for various industrial processes. Crucially, however, natural enzymes do not always have the properties required for specific processes. It may be necessary, therefore, to design, engineer, and evolve enzymes with properties that are not found in natural enzymes. In particular, the creation of enzymes that are thermally stable and catalytically active at low temperature is desirable for processes involving both high and low temperatures. In the current study, we designed two ancestral sequences of 3-isopropylmalate dehydrogenase by an ancestral sequence reconstruction technique based on a phylogenetic analysis of extant homologous amino acid sequences. Genes encoding the designed sequences were artificially synthesized and expressed in Escherichia coli. The reconstructed enzymes were found to be slightly more thermally stable than the extant thermophilic homologue from Thermus thermophilus. Moreover, they had considerably higher low-temperature catalytic activity as compared with the T. thermophilus enzyme. Detailed analyses of their temperature-dependent specific activities and kinetic properties showed that the reconstructed enzymes have catalytic properties similar to those of mesophilic homologues. Collectively, our study demonstrates that ancestral sequence reconstruction can produce a thermally stable enzyme with catalytic properties adapted to low-temperature reactions.

Insight into the protein solubility driving forces with neural attention.

Protein solubility is a key aspect for many biotechnological, biomedical and industrial processes, such as the production of active proteins and antibodies. In addition, understanding the molecular determinants of the solubility of proteins may be crucial to shed light on the molecular mechanisms of diseases caused by aggregation processes such as amyloidosis. Here we present SKADE, a novel Neural Network protein solubility predictor and we show how it can provide novel insight into the protein solubility mechanisms, thanks to its neural attention architecture. First, we show that SKADE positively compares with state of the art tools while using just the protein sequence as input. Then, thanks to the neural attention mechanism, we use SKADE to investigate the patterns learned during training and we analyse its decision process. We use this peculiarity to show that, while the attention profiles do not correlate with obvious sequence aspects such as biophysical properties of the aminoacids, they suggest that N- and C-termini are the most relevant regions for solubility prediction and are predictive for complex emergent properties such as aggregation-prone regions involved in beta-amyloidosis and contact density. Moreover, SKADE is able to identify mutations that increase or decrease the overall solubility of the protein, allowing it to be used to perform large scale in-silico mutagenesis of proteins in order to maximize their solubility.

MHC class I and II peptide homology regulates the cellular immune response.

Mammalian immune responses are initiated by "danger" signals--immutable molecular structures known as PAMPs. When detected by fixed, germline encoded receptors, pathogen-associated molecular pattern (PAMPs) subsequently inform the polarization of downstream adaptive responses depending upon identity and localization of the PAMP. Here, we report the existence of a completely novel "PAMP" that is not a molecular structure but an antigenic pattern. This pattern--the incidence of peptide epitopes with stretches of 100% sequence identity bound to both dendritic cell (DC) major histocompatibility (MHC) class I and MHC class II--strongly induces TH 1 immune polarization and activation of the cellular immune response. Inherent in the existence of this PAMP is the concomitant existence of a molecular sensor complex with the ability to scan and compare amino acid sequence identities of bound class I and II peptides. We provide substantial evidence implicating the multienzyme aminoacyl-tRNA synthetase (mARS) complex and its AIMp1 structural component as the key constituents of this complex. The results demonstrate a wholly novel mechanism by which T-helper (TH ) polarization is governed and provide critical information for the design of vaccination strategies intended to provoke cell-mediated immunity.

Arabidopsis Voltage-Dependent Anion Channels (VDACs): Overlapping and Specific Functions in Mitochondria.

Voltage-dependent anion channels (VDACs) are essential components of the mitochondrial outer membrane. VDACs are involved in the exchange of numerous ions and molecules, from ATP to larger molecules such as tRNAs, and are supposed to adjust exchanges in response to cell signals and stresses. Four major VDACs have been identified in Arabidopsis thaliana. The goal of this study was to explore the specific functions of these proteins, in particular, in tRNA import into mitochondria and stress response. The main results were: (i) VDACs appeared to differentially interact with tRNAs, and VDAC4 could be the major tRNA channel on the outer membrane, (ii) a VDAC3 mRNA isoform was found induced by different stresses, suggesting that VDAC3 might be specifically involved in early steps of stress response and (iii) an analysis of vdac3 and vdac1 mutant lines showed that VDAC3 and VDAC1 shared some, but not all functions. In conclusion, this work brings new knowledge on VDACs, which do not appear as interchangeable pores of the outer membrane and each VDAC has its own specificity.

Crystal Structure of Keratin 1/10(C401A) 2B Heterodimer Demonstrates a Proclivity for the C-Terminus of Helix 2B to Form Higher Order Molecular Contacts.

We previously determined the crystal structure of the wild-type keratin 1/10 helix 2B heterodimer at 3.3 Å resolution. We proposed that the resolution of the diffraction data was limited due to the crystal packing effect from keratin 10 (K10) residue Cys401. Cys401K10 formed a disulfide-linkage with Cys401 from another K1/10 heterodimer, creating an "X-shaped" structure and a loose crystal packing arrangement. We hypothesized that mutation of Cys401K10 to alanine would eliminate the disulfide-linkage and improve crystal packing thereby increasing resolution of diffraction and enabling a more accurate side chain electron density map. Indeed, when a K10 Cys401Ala 2B mutant was paired with its native keratin 1 (K1) 2B heterodimer partner its x-ray crystal structure was determined at 2.07 Å resolution; the structure does not contain a disulfide linkage. Superposition of the K1/K10(Cys401Ala) 2B structure onto the wild-type K1/10 2B heterodimer structure had a root-mean-square-deviation of 1.88 Å; the variability in the atomic positions reflects the dynamic motion expected in this filamentous coiled-coil complex. The electrostatic, hydrophobic, and contour features of the molecular surface are similar to the lower resolution wild-type structure. We postulated that elimination of the disulfide linkage in the K1/K10(Cys401Ala) 2B structure could allow for the 2B heterodimers to bind/pack in the A22 tetramer configuration associated with mature keratin intermediate filament assembly. Analysis of the crystal packing revealed a half-staggered anti-parallel tetrameric complex of 2B heterodimers; however, their register is not consistent with models of the A22 mode of tetrameric alignment or prior biochemical cross-linking studies.

Development and Differentiation in Monobodies Based on the Fibronectin Type 3 Domain.

As a non-antibody scaffold, monobodies based on the fibronectin type III (FN3) domain overcome antibody size and complexity while maintaining analogous binding loops. However, antibodies and their derivatives remain the gold standard for the design of new therapeutics. In response, clinical-stage therapeutic proteins based on the FN3 domain are beginning to use native fibronectin function as a point of differentiation. The small and simple structure of monomeric monobodies confers increased tissue distribution and reduced half-life, whilst the absence of disulphide bonds improves stability in cytosolic environments. Where multi-specificity is challenging with an antibody format that is prone to mis-pairing between chains, multiple FN3 domains in the fibronectin assembly already interact with a large number of molecules. As such, multiple monobodies engineered for interaction with therapeutic targets are being combined in a similar beads-on-a-string assembly which improves both efficacy and pharmacokinetics. Furthermore, full length fibronectin is able to fold into multiple conformations as part of its natural function and a greater understanding of how mechanical forces allow for the transition between states will lead to advanced applications that truly differentiate the FN3 domain as a therapeutic scaffold.

Incorporation of the Nonproteinogenic Amino Acid ß-Methylamino-alanine Affects Amyloid ß Fibril Properties and Toxicity.

The nonproteinogenic amino acid ß-methylamino alarelevant example for environmental hazards are nonnine (BMAA) is a neurotoxin and represents a potential risk factor for neurodegenerative diseases. Despite intense research over the last years, the pathological mechanism of BMAA is still unclear. One of the main open questions is whether BMAA can be misincorporated into proteins, especially as a substitute for serine, and whether this has structural and functional consequences for the afflicted proteins leading to early onset neurodegeneration. In this study, we hypothesize that BMAA was indeed incorporated into Aß40 molecules and study the structural and dynamical consequences of such misincorporation along with the effect such mutated Aß40 peptides have on neuronal cells. We used the synthetic ß-amyloid peptide (Aß40), a known key player in the development of Alzheimer's disease, to incorporate BMAA substitutions at three different positions in the peptide sequence: Ser8BMAA at the peptide's N-terminus, Phe19BMAA in the hydrophobic core region, and S26BMAA in the flexible turn region of Aß40 fibrils. We performed a set of biophysical experiments including fluorescence, circular dichroism, solid-state NMR spectroscopy, transmission electron microscopy, and X-ray diffraction to investigate structural and functional aspects of the mutated peptides compared to wildtype Aß40. All variants showed high structural tolerance to BMAA misincorporation. In contrast, the cellular response and neuronal survival were affected in a mutation site-specific manner. As a consequence, we can state from the physicochemical point of view that, if BMAA was misincorporated into proteins, it could indeed represent a risk factor that could potentially play a role in neurodegeneration. Further research addressing the role of BMAA, especially its protein-associated form, should be performed to obtain a better understanding of neurodegenerative diseases and to develop new therapeutic strategies.

Epistatic contributions promote the unification of incompatible models of neutral molecular evolution.

We introduce a model of amino acid sequence evolution that accounts for the statistical behavior of real sequences induced by epistatic interactions. We base the model dynamics on parameters derived from multiple sequence alignments analyzed by using direct coupling analysis methodology. Known statistical properties such as overdispersion, heterotachy, and gamma-distributed rate-across-sites are shown to be emergent properties of this model while being consistent with neutral evolution theory, thereby unifying observations from previously disjointed evolutionary models of sequences. The relationship between site restriction and heterotachy is characterized by tracking the effective alphabet dynamics of sites. We also observe an evolutionary Stokes shift in the fitness of sequences that have undergone evolution under our simulation. By analyzing the structural information of some proteins, we corroborate that the strongest Stokes shifts derive from sites that physically interact in networks near biochemically important regions. Perspectives on the implementation of our model in the context of the molecular clock are discussed.

Structure and function of crocodilian hemoglobins and allosteric regulation by chloride, ATP, and CO2.

Hemoglobins (Hbs) of crocodilians are reportedly characterized by unique mechanisms of allosteric regulatory control, but there are conflicting reports regarding the importance of different effectors, such as chloride ions, organic phosphates, and CO2. Progress in understanding the unusual properties of crocodilian Hbs has also been hindered by a dearth of structural information. Here, we present the first comparative analysis of blood properties and Hb structure and function in a phylogenetically diverse set of crocodilian species. We examine mechanisms of allosteric regulation in the Hbs of 13 crocodilian species belonging to the families Crocodylidae and Alligatoridae. We also report new amino acid sequences for the α- and ß-globins of these taxa, which, in combination with structural analyses, provide insights into molecular mechanisms of allosteric regulation. All crocodilian Hbs exhibited a remarkably strong sensitivity to CO2, which would permit effective O2 unloading to tissues in response to an increase in metabolism during intense activity and diving. Although the Hbs of all crocodilians exhibit similar intrinsic O2-affinities, there is considerable variation in sensitivity to Cl- ions and ATP, which appears to be at least partly attributable to variation in the extent of NH2-terminal acetylation. Whereas chloride appears to be a potent allosteric effector of all crocodile Hbs, ATP has a strong, chloride-independent effect on Hb-O2 affinity only in caimans. Modeling suggests that allosteric ATP binding has a somewhat different structural basis in crocodilian and mammalian Hbs.

Atypical structural tendencies among low-complexity domains in the Protein Data Bank proteome.

A variety of studies have suggested that low-complexity domains (LCDs) tend to be intrinsically disordered and are relatively rare within structured proteins in the Protein Data Bank (PDB). Although LCDs are often treated as a single class, we previously found that LCDs enriched in different amino acids can exhibit substantial differences in protein metabolism and function. Therefore, we wondered whether the structural conformations of LCDs are likewise dependent on which specific amino acids are enriched within each LCD. Here, we directly examined relationships between enrichment of individual amino acids and secondary structure tendencies across the entire PDB proteome. Secondary structure tendencies varied as a function of the identity of the amino acid enriched and its degree of enrichment. Furthermore, divergence in secondary structure profiles often occurred for LCDs enriched in physicochemically similar amino acids (e.g. valine vs. leucine), indicating that LCDs composed of related amino acids can have distinct secondary structure tendencies. Comparison of LCD secondary structure tendencies with numerous pre-existing secondary structure propensity scales resulted in relatively poor correlations for certain types of LCDs, indicating that these scales may not capture secondary structure tendencies as sequence complexity decreases. Collectively, these observations provide a highly resolved view of structural tendencies among LCDs parsed by the nature and magnitude of single amino acid enrichment.

Modulation of adiponectin receptors AdipoR1 and AdipoR2 by phage display-derived peptides in in vitro and in vivo models.

Type 2 diabetes (T2D) is often linked to metabolic syndrome, which assembles various risk factors related to obesity. Plasma levels of adiponectin are decreased in T2D and obese subjects. Aiming to develop a peptide able to bind adiponectin receptors and modulate their signalling pathways, a 12-amino acid sequence homologous in AdipoR1/R2 has been targeted by phage display with a linear 12-mer peptide library. The selected peptide P17 recognises AdipoR1/R2 expressed by skeletal muscle, liver and pancreatic islets. In HepaRG and C2C12 cells, P17 induced the activation of AMPK (AMPKα-pT172) and the expression of succinate dehydrogenase and glucokinase; no cytotoxic effects were observed on HepaRG cells. In db/db mice, P17 promoted body weight and glycaemia stabilisation, decreased plasma triglycerides to the range of healthy mice and increased adiponectin (in high fat-fed mice) and insulin (in chow-fed mice) levels. It restored to the range of healthy mice the tissue levels and subcellular distribution of AdipoR1/R2, AMPKα-pT172 and PPARα-pS12. In liver, P17 reduced steatosis and apoptosis. The docking of P17 to AdipoR is reminiscent of the binding mechanism of adiponectin. To conclude, we have developed an AdipoR1/AdipoR2-targeted peptide that modulates adiponectin signalling pathways and has therapeutic relevance for T2D and obesity associated pathologies.

Mapping the proximity interaction network of the Rho-family GTPases reveals signalling pathways and regulatory mechanisms.

Guanine nucleotide exchange factors (RhoGEFs) and GTPase-activating proteins (RhoGAPs) coordinate the activation state of the Rho family of GTPases for binding to effectors. Here, we exploited proximity-dependent biotinylation to systematically define the Rho family proximity interaction network from 28 baits to produce 9,939 high-confidence proximity interactions in two cell lines. Exploiting the nucleotide states of Rho GTPases, we revealed the landscape of interactions with RhoGEFs and RhoGAPs. We systematically defined effectors of Rho proteins to reveal candidates for classical and atypical Rho proteins. We used optogenetics to demonstrate that KIAA0355 (termed GARRE here) is a RAC1 interactor. A functional screen of RHOG candidate effectors identified PLEKHG3 as a promoter of Rac-mediated membrane ruffling downstream of RHOG. We identified that active RHOA binds the kinase SLK in Drosophila and mammalian cells to promote Ezrin-Radixin-Moesin phosphorylation. Our proximity interactions data pave the way for dissecting additional Rho signalling pathways, and the approaches described here are applicable to the Ras family.