Designing antimicrobial peptides using deep learning and molecular dynamic simulations.

With the emergence of multidrug-resistant bacteria, antimicrobial peptides (AMPs) offer promising options for replacing traditional antibiotics to treat bacterial infections, but discovering and designing AMPs using traditional methods is a time-consuming and costly process. Deep learning has been applied to the de novo design of AMPs and address AMP classification with high efficiency. In this study, several natural language processing models were combined to design and identify AMPs, i.e. sequence generative adversarial nets, bidirectional encoder representations from transformers and multilayer perceptron. Then, six candidate AMPs were screened by AlphaFold2 structure prediction and molecular dynamic simulations. These peptides show low homology with known AMPs and belong to a novel class of AMPs. After initial bioactivity testing, one of the peptides, A-222, showed inhibition against gram-positive and gram-negative bacteria. The structural analysis of this novel peptide A-222 obtained by nuclear magnetic resonance confirmed the presence of an alpha-helix, which was consistent with the results predicted by AlphaFold2. We then performed a structure-activity relationship study to design a new series of peptide analogs and found that the activities of these analogs could be increased by 4-8-fold against Stenotrophomonas maltophilia WH 006 and Pseudomonas aeruginosa PAO1. Overall, deep learning shows great potential in accelerating the discovery of novel AMPs and holds promise as an important tool for developing novel AMPs.

o-Vanillin binds covalently to MAL/TIRAP Lys-210 but independently inhibits TLR2.

Toll-like receptor (TLR) innate immunity signalling protects against pathogens, but excessive or prolonged signalling contributes to a range of inflammatory conditions. Structural information on the TLR cytoplasmic TIR (Toll/interleukin-1 receptor) domains and the downstream adaptor proteins can help us develop inhibitors targeting this pathway. The small molecule o-vanillin has previously been reported as an inhibitor of TLR2 signalling. To study its mechanism of action, we tested its binding to the TIR domain of the TLR adaptor MAL/TIRAP (MALTIR). We show that o-vanillin binds to MALTIR and inhibits its higher-order assembly in vitro. Using NMR approaches, we show that o-vanillin forms a covalent bond with lysine 210 of MAL. We confirm in mouse and human cells that o-vanillin inhibits TLR2 but not TLR4 signalling, independently of MAL, suggesting it may covalently modify TLR2 signalling complexes directly. Reactive aldehyde-containing small molecules such as o-vanillin may target multiple proteins in the cell.

Structural Basis of the Bivalency of the TRPV1 Agonist DkTx.

Bivalency is a prevalent natural mechanism to enhance receptor avidity. Various two-domain disulfide-rich peptides exhibiting bivalent action have been identified from animal venoms. A unique characteristic of these peptides is that they induce a pharmacological response different from that provoked by any of the constituent domains. The enhanced potency and avidity of such peptides is therefore a consequence of their domain fusion by a peptide linker. The role of the linker itself, beyond conjugation, remains unclear. Here, we investigate how the linker affects the bivalency of the capsaicin receptor (TRPV1) agonist DkTx. We recombinantly produced isotope labelled DkTx using a protein splicing approach, to solve the high-resolution solution structure of DkTx, revealing residual linker order stabilised by linker-domain interactions leading to biased domain orientations. The significance of this was studied using a combination of mutagenesis, spin relaxation studies and electrophysiology measurements. Our results reveal that disrupting the pre-organisation of the domains of DkTx is accompanied by reductions in potency and onset of avidity. Our findings support a model of pre-configured two-domain binding, in favour of the previously suggested sequential binding model. This highlights the significance of ordered elements in linker design and the natural evolution of these in bivalent toxins.

The transmembrane adapter SCIMP recruits tyrosine kinase Syk to phosphorylate Toll-like receptors to mediate selective inflammatory outputs.

Innate immune signaling by Toll-like receptors (TLRs) involves receptor phosphorylation, which helps to shape and drive key inflammatory outputs, yet our understanding of the kinases and mechanisms that mediate TLR phosphorylation is incomplete. Spleen tyrosine kinase (Syk) is a nonreceptor protein tyrosine kinase, which is known to relay adaptive and innate immune signaling, including from TLRs. However, TLRs do not contain the conserved dual immunoreceptor tyrosine-based activation motifs that typically recruit Syk to many other receptors. One possibility is that the Syk-TLR association is indirect, relying on an intermediary scaffolding protein. We previously identified a role for the palmitoylated transmembrane adapter protein SCIMP in scaffolding the Src tyrosine kinase Lyn, for TLR phosphorylation, but the role of SCIMP in mediating the interaction between Syk and TLRs has not yet been investigated. Here, we show that SCIMP recruits Syk in response to lipopolysaccharide-mediated TLR4 activation. We also show that Syk contributes to the phosphorylation of SCIMP and TLR4 to enhance their binding. Further evidence pinpoints two specific phosphorylation sites in SCIMP critical for its interaction with Syk-SH2 domains in the absence of immunoreceptor tyrosine-based activation motifs. Finally, using inhibitors and primary macrophages from SCIMP-/- mice, we confirm a functional role for SCIMP-mediated Syk interaction in modulating TLR4 phosphorylation, signaling, and cytokine outputs. In conclusion, we identify SCIMP as a novel, immune-specific Syk scaffold, which can contribute to inflammation through selective TLR-driven inflammatory responses.

Structural venomics reveals evolution of a complex venom by duplication and diversification of an ancient peptide-encoding gene.

Spiders are one of the most successful venomous animals, with more than 48,000 described species. Most spider venoms are dominated by cysteine-rich peptides with a diverse range of pharmacological activities. Some spider venoms contain thousands of unique peptides, but little is known about the mechanisms used to generate such complex chemical arsenals. We used an integrated transcriptomic, proteomic, and structural biology approach to demonstrate that the lethal Australian funnel-web spider produces 33 superfamilies of venom peptides and proteins. Twenty-six of the 33 superfamilies are disulfide-rich peptides, and we show that 15 of these are knottins that contribute >90% of the venom proteome. NMR analyses revealed that most of these disulfide-rich peptides are structurally related and range in complexity from simple to highly elaborated knottin domains, as well as double-knot toxins, that likely evolved from a single ancestral toxin gene.

Structural insights into the mechanism of inhibition of AHAS by herbicides.

Acetohydroxyacid synthase (AHAS), the first enzyme in the branched amino acid biosynthesis pathway, is present only in plants and microorganisms, and it is the target of >50 commercial herbicides. Penoxsulam (PS), which is a highly effective broad-spectrum AHAS-inhibiting herbicide, is used extensively to control weed growth in rice crops. However, the molecular basis for its inhibition of AHAS is poorly understood. This is despite the availability of structural data for all other classes of AHAS-inhibiting herbicides. Here, crystallographic data for Saccharomyces cerevisiae AHAS (2.3 Å) and Arabidopsis thaliana AHAS (2.5 Å) in complex with PS reveal the extraordinary molecular mechanisms that underpin its inhibitory activity. The structures show that inhibition of AHAS by PS triggers expulsion of two molecules of oxygen bound in the active site, releasing them as substrates for an oxygenase side reaction of the enzyme. The structures also show that PS either stabilizes the thiamin diphosphate (ThDP)-peracetate adduct, a product of this oxygenase reaction, or traps within the active site an intact molecule of peracetate in the presence of a degraded form of ThDP: thiamine aminoethenethiol diphosphate. Kinetic analysis shows that PS inhibits AHAS by a combination of events involving FAD oxidation and chemical alteration of ThDP. With the emergence of increasing levels of resistance toward front-line herbicides and the need to optimize the use of arable land, these data suggest strategies for next generation herbicide design.

Selective NaV1.1 activation rescues Dravet syndrome mice from seizures and premature death.

Dravet syndrome is a catastrophic, pharmacoresistant epileptic encephalopathy. Disease onset occurs in the first year of life, followed by developmental delay with cognitive and behavioral dysfunction and substantially elevated risk of premature death. The majority of affected individuals harbor a loss-of-function mutation in one allele of SCN1A, which encodes the voltage-gated sodium channel NaV1.1. Brain NaV1.1 is primarily localized to fast-spiking inhibitory interneurons; thus the mechanism of epileptogenesis in Dravet syndrome is hypothesized to be reduced inhibitory neurotransmission leading to brain hyperexcitability. We show that selective activation of NaV1.1 by venom peptide Hm1a restores the function of inhibitory interneurons from Dravet syndrome mice without affecting the firing of excitatory neurons. Intracerebroventricular infusion of Hm1a rescues Dravet syndrome mice from seizures and premature death. This precision medicine approach, which specifically targets the molecular deficit in Dravet syndrome, presents an opportunity for treatment of this intractable epilepsy.

Solution structure of the TLR adaptor MAL/TIRAP reveals an intact BB loop and supports MAL Cys91 glutathionylation for signaling.

MyD88 adaptor-like (MAL) is a critical protein in innate immunity, involved in signaling by several Toll-like receptors (TLRs), key pattern recognition receptors (PRRs). Crystal structures of MAL revealed a nontypical Toll/interleukin-1 receptor (TIR)-domain fold stabilized by two disulfide bridges. We therefore undertook a structural and functional analysis of the role of reactive cysteine residues in the protein. Under reducing conditions, the cysteines do not form disulfides, but under oxidizing conditions they are highly amenable to modification. The solution structure of the reduced form of the MAL TIR domain, determined by NMR spectroscopy, reveals a remarkable structural rearrangement compared with the disulfide-bonded structure, which includes the relocation of a ß-strand and repositioning of the functionally important "BB-loop" region to a location more typical for TIR domains. Redox measurements by NMR further reveal that C91 has the highest redox potential of all cysteines in MAL. Indeed, mass spectrometry revealed that C91 undergoes glutathionylation in macrophages activated with the TLR4 ligand lipopolysaccharide (LPS). The C91A mutation limits MAL glutathionylation and acts as a dominant negative, blocking the interaction of MAL with its downstream target MyD88. The H92P mutation mimics the dominant-negative effects of the C91A mutation, presumably by preventing C91 glutathionylation. The MAL C91A and H92P mutants also display diminished degradation and interaction with interleukin-1 receptor-associated kinase 4 (IRAK4). We conclude that in the cell, MAL is not disulfide-bonded and requires glutathionylation of C91 for signaling.

Potent neuroprotection after stroke afforded by a double-knot spider-venom peptide that inhibits acid-sensing ion channel 1a.

Stroke is the second-leading cause of death worldwide, yet there are no drugs available to protect the brain from stroke-induced neuronal injury. Acid-sensing ion channel 1a (ASIC1a) is the primary acid sensor in mammalian brain and a key mediator of acidosis-induced neuronal damage following cerebral ischemia. Genetic ablation and selective pharmacologic inhibition of ASIC1a reduces neuronal death following ischemic stroke in rodents. Here, we demonstrate that Hi1a, a disulfide-rich spider venom peptide, is highly neuroprotective in a focal model of ischemic stroke. Nuclear magnetic resonance structural studies reveal that Hi1a comprises two homologous inhibitor cystine knot domains separated by a short, structurally well-defined linker. In contrast with known ASIC1a inhibitors, Hi1a incompletely inhibits ASIC1a activation in a pH-independent and slowly reversible manner. Whole-cell, macropatch, and single-channel electrophysiological recordings indicate that Hi1a binds to and stabilizes the closed state of the channel, thereby impeding the transition into a conducting state. Intracerebroventricular administration to rats of a single small dose of Hi1a (2 ng/kg) up to 8 h after stroke induction by occlusion of the middle cerebral artery markedly reduced infarct size, and this correlated with improved neurological and motor function, as well as with preservation of neuronal architecture. Thus, Hi1a is a powerful pharmacological tool for probing the role of ASIC1a in acid-mediated neuronal injury and various neurological disorders, and a promising lead for the development of therapeutics to protect the brain from ischemic injury.

A new vector coupling ligation-independent cloning with sortase a fusion for efficient cloning and one-step purification of tag-free recombinant proteins.

We have developed a new ligation independent cloning (LIC) vector - pSrtA9, which can be utilized for one-step purification of recombinant proteins. The new LIC site in the pSrtA9 vector, hosts a DNA sequence centered on a SfoI restriction site and integrates a coding sequence for sortase A (SrtA) recognition. Preceding the LIC site, pSrtA9 incorporates an N-terminal 6xHis-tag and the catalytic core of SrtA from Staphylococcus aureus (SrtAΔ59). Thus, after cloning and protein expression in Escherichia coli, the resultant fusion protein comprises an N-terminal 6xHis-tag, SrtAΔ59, an L-P-E-T-G linker and the protein of interest at the C-terminus. The fusion protein can be captured onto immobilized Ni-NTA resin and any unwanted proteolysis activity of SrtA is suppressed during the purification by optimisation of solution conditions. Upon addition of Ca2+ and triglycine (Gly3), the immobilized fusion protein undergoes on-column SrtA-mediated cleavage at the T-G bond of LPETG linker to selectively release 90% of the protein of interest within 3 h when incubated at room temperature. This new pSrtA9 vector, thus, offers an efficient method for LIC of genes and a one-step purification procedure to obtain a tag-free recombinant protein, and is therefore suitable for the high-throughput proteins production.

The CC domain structure from the wheat stem rust resistance protein Sr33 challenges paradigms for dimerization in plant NLR proteins.

Plants use intracellular immunity receptors, known as nucleotide-binding oligomerization domain-like receptors (NLRs), to recognize specific pathogen effector proteins and induce immune responses. These proteins provide resistance to many of the world's most destructive plant pathogens, yet we have a limited understanding of the molecular mechanisms that lead to defense signaling. We examined the wheat NLR protein, Sr33, which is responsible for strain-specific resistance to the wheat stem rust pathogen, Puccinia graminis f. sp. tritici We present the solution structure of a coiled-coil (CC) fragment from Sr33, which adopts a four-helix bundle conformation. Unexpectedly, this structure differs from the published dimeric crystal structure of the equivalent region from the orthologous barley powdery mildew resistance protein, MLA10, but is similar to the structure of the distantly related potato NLR protein, Rx. We demonstrate that these regions are, in fact, largely monomeric and adopt similar folds in solution in all three proteins, suggesting that the CC domains from plant NLRs adopt a conserved fold. However, larger C-terminal fragments of Sr33 and MLA10 can self-associate both in vitro and in planta, and this self-association correlates with their cell death signaling activity. The minimal region of the CC domain required for both cell death signaling and self-association extends to amino acid 142, thus including 22 residues absent from previous biochemical and structural protein studies. These data suggest that self-association of the minimal CC domain is necessary for signaling but is likely to involve a different structural basis than previously suggested by the MLA10 crystallographic dimer.

Two proteins for the price of one: Structural studies of the dual-destiny protein preproalbumin with sunflower trypsin inhibitor-1.

Seed storage proteins are both an important source of nutrition for humans and essential for seedling establishment. Interestingly, unusual napin-type 2S seed storage albumin precursors in sunflowers contain a sequence that is released as a macrocyclic peptide during post-translational processing. The mechanism by which such peptides emerge from linear precursor proteins has received increased attention; however, the structural characterization of intact precursor proteins has been limited. Here, we report the 3D NMR structure of the Helianthus annuus PawS1 (preproalbumin with sunflower trypsin inhibitor-1) and provide new insights into the processing of this remarkable dual-destiny protein. In seeds, PawS1 is matured by asparaginyl endopeptidases (AEPs) into the cyclic peptide SFTI-1 (sunflower trypsin inhibitor-1) and a heterodimeric 2S albumin. The structure of PawS1 revealed that SFTI-1 and the albumin are independently folded into well-defined domains separated by a flexible linker. PawS1 was cleaved in vitro with recombinant sunflower HaAEP1 and in situ using a sunflower seed extract in a way that resembled the expected in vivo cleavages. Recombinant HaAEP1 cleaved PawS1 at multiple positions, and in situ, its flexible linker was removed, yielding fully mature heterodimeric albumin. Liberation and cyclization of SFTI-1, however, was inefficient, suggesting that specific seed conditions or components may be required for in vivo biosynthesis of SFTI-1. In summary, this study has revealed the 3D structure of a macrocyclic precursor protein and provided important mechanistic insights into the maturation of sunflower proalbumins into an albumin and a macrocyclic peptide.

Toxin structures as evolutionary tools: Using conserved 3D folds to study the evolution of rapidly evolving peptides.

Three-dimensional (3D) structures have been used to explore the evolution of proteins for decades, yet they have rarely been utilized to study the molecular evolution of peptides. Here, we highlight areas in which 3D structures can be particularly useful for studying the molecular evolution of peptide toxins. Although we focus our discussion on animal toxins, including one of the most widespread disulfide-rich peptide folds known, the inhibitor cystine knot, our conclusions should be widely applicable to studies of the evolution of disulfide-constrained peptides. We show that conserved 3D folds can be used to identify evolutionary links and test hypotheses regarding the evolutionary origin of peptides with extremely low sequence identity; construct accurate multiple sequence alignments; and better understand the evolutionary forces that drive the molecular evolution of peptides. Also watch the video abstract.

Development of a µO-Conotoxin Analogue with Improved Lipid Membrane Interactions and Potency for the Analgesic Sodium Channel NaV1.8.

The µO-conotoxins MrVIA, MrVIB, and MfVIA inhibit the voltage-gated sodium channel NaV1.8, a well described target for the treatment of pain; however, little is known about the residues or structural elements that define this activity. In this study, we determined the three-dimensional structure of MfVIA, examined its membrane binding properties, performed alanine-scanning mutagenesis, and identified residues important for its activity at human NaV1.8. A second round of mutations resulted in (E5K,E8K)MfVIA, a double mutant with greater positive surface charge and greater affinity for lipid membranes compared with MfVIA. This analogue had increased potency at NaV1.8 and was analgesic in the mouse formalin assay.

A non-uniform sampling approach enables studies of dilute and unstable proteins.

NMR spectroscopy is a powerful method in structural and functional analysis of macromolecules and has become particularly prevalent in studies of protein structure, function and dynamics. Unique to NMR spectroscopy is the relatively low constraints on sample preparation and the high level of control of sample conditions. Proteins can be studied in a wide range of buffer conditions, e.g. different pHs and variable temperatures, allowing studies of proteins under conditions that are closer to their native environment compared to other structural methods such as X-ray crystallography and electron microscopy. The key disadvantage of NMR is the relatively low sensitivity of the method, requiring either concentrated samples or very lengthy data-acquisition times. Thus, proteins that are unstable or can only be studied in dilute solutions are often considered practically unfeasible for NMR studies. Here, we describe a general method, where non-uniform sampling (NUS) allows for signal averaging to be monitored in an iterative manner, enabling efficient use of spectrometer time, ultimately leading to savings in costs associated with instrument and isotope-labelled protein use. The method requires preparation of multiple aliquots of the protein sample that are flash-frozen and thawed just before acquisition of a short NMR experiments carried out while the protein is stable (12 h in the presented case). Non-uniform sampling enables sufficient resolution to be acquired for each short experiment. Identical NMR datasets are acquired and sensitivity is monitored after each co-added spectrum is reconstructed. The procedure is repeated until sufficient signal-to-noise is obtained. We discuss how maximum entropy reconstruction is used to process the data, and propose a variation on the previously described method of automated parameter selection. We conclude that combining NUS with iterative co-addition is a general approach, and particularly powerful when applied to unstable proteins.

RNA polymerase-induced remodelling of NusA produces a pause enhancement complex.

Pausing during transcription elongation is a fundamental activity in all kingdoms of life. In bacteria, the essential protein NusA modulates transcriptional pausing, but its mechanism of action has remained enigmatic. By combining structural and functional studies we show that a helical rearrangement induced in NusA upon interaction with RNA polymerase is the key to its modulatory function. This conformational change leads to an allosteric re-positioning of conserved basic residues that could enable their interaction with an RNA pause hairpin that forms in the exit channel of the polymerase. This weak interaction would stabilize the paused complex and increases the duration of the transcriptional pause. Allosteric spatial re-positioning of regulatory elements may represent a general approach used across all taxa for modulation of transcription and protein-RNA interactions.

Determination of ligand binding modes in weak protein-ligand complexes using sparse NMR data.

We describe a general approach to determine the binding pose of small molecules in weakly bound protein-ligand complexes by deriving distance constraints between the ligand and methyl groups from all methyl-containing residues of the protein. We demonstrate that using a single sample, which can be prepared without the use of expensive precursors, it is possible to generate high-resolution data rapidly and obtain the resonance assignments of Ile, Leu, Val, Ala and Thr methyl groups using triple resonance scalar correlation data. The same sample may be used to obtain Met εCH3 assignments using NOESY-based methods, although the superior sensitivity of NOESY using [U-13C,15N]-labeled protein makes the use of this second sample more efficient. We describe a structural model for a weakly binding ligand bound to its target protein, DsbA, derived from intermolecular methyl-to-ligand nuclear Overhauser enhancements, and demonstrate that the ability to assign all methyl resonances in the spectrum is essential to derive an accurate model of the structure. Once the methyl assignments have been obtained, this approach provides a rapid means to generate structural models for weakly bound protein-ligand complexes. Such weak complexes are often found at the beginning of programs of fragment based drug design and can be challenging to characterize using X-ray crystallography.

Inhibition of the norepinephrine transporter by χ-conotoxin dendrimers.

Peptide dendrimers are a novel class of macromolecules of emerging interest with the potential of delayed renal clearance due to their molecular size and enhanced activity due to the multivalency effect. In this work, an active analogue of the disulfide-rich χ-conotoxin χ-MrIA (χ-MrIA), a norepinephrine reuptake (norepinephrine transporter) inhibitor, was grafted onto a polylysine dendron. Dendron decoration was achieved by employing copper-catalyzed alkyne-azide cycloaddition with azido-PEG chain-modified χ-MrIA analogues, leading to homogenous 4-mer and 8-mer χ-MrIA dendrimers with molecular weights ranging from 8 to 22 kDa. These dendrimers were investigated for their impact on peptide secondary structure, in vitro functional activity, and potential anti-allodynia in vivo. NMR studies showed that the χ-MrIA tertiary structure was maintained in the χ-MrIA dendrimers. In a functional norepinephrine transporter reuptake assay, χ-MrIA dendrimers showed slightly increased potency relative to the azido-PEGylated χ-MrIA analogues with similar potency to the parent peptide. In contrast to χ-MrIA, no anti-allodynic action was observed when the χ-MrIA dendrimers were administered intrathecally in a rat model of neuropathic pain, suggesting that the larger dendrimer structures are unable to diffuse through the spinal column tissue and reach the norepinephrine transporter. Copyright © 2016 European Peptide Society and John Wiley & Sons, Ltd.

Correction: Sparse sampling methods in multidimensional NMR.

Correction for 'Sparse sampling methods in multidimensional NMR' by Mehdi Mobli et al., Phys. Chem. Chem. Phys., 2012, 14, 10835-10843.

Solution structure of the RNA-binding cold-shock domain of the Chlamydomonas reinhardtii NAB1 protein and insights into RNA recognition.

Light-harvesting complex (LHC) proteins are among the most abundant proteins on Earth and play critical roles in photosynthesis, both in light capture and in photoprotective mechanisms. The Chlamydomonas reinhardtii nucleic acid-binding protein 1 (NAB1) is a negative regulator of LHC protein translation. Its N-terminal cold-shock domain (CSD) binds to a 13-nt element [CSD consensus sequence (CSDCS)] found in the mRNA of specific LHC proteins associated with Photosystem II (PSII), an interaction which regulates LHC expression and, consequently, PSII-associated antenna size, structure and function. In the present study, we elucidated the solution structure of the NAB1 CSD as determined by heteronuclear NMR. The CSD adopts a characteristic five-stranded anti parallel ß-barrel fold. Upon addition of CSDCS RNA, a large number of NMR chemical shift perturbations were observed, corresponding primarily to surface-exposed residues within the highly conserved ß2- and ß3-strands in the canonical RNA-binding region, but also to residues on ß-strand 5 extending the positive surface patch and the overall RNA-binding site. Additional chemical shift perturbations that accompanied RNA binding involved buried residues, suggesting that transcript recognition is accompanied by conformational change. Our results indicate that NAB1 associates with RNA transcripts through a mechanism involving its CSD that is conserved with mechanisms of sequence-specific nucleic acid recognition employed by ancestrally related bacterial cold-shock proteins (CSPs).