VISH-Pred: an ensemble of fine-tuned ESM models for protein toxicity prediction.

Peptide- and protein-based therapeutics are becoming a promising treatment regimen for myriad diseases. Toxicity of proteins is the primary hurdle for protein-based therapies. Thus, there is an urgent need for accurate in silico methods for determining toxic proteins to filter the pool of potential candidates. At the same time, it is imperative to precisely identify non-toxic proteins to expand the possibilities for protein-based biologics. To address this challenge, we proposed an ensemble framework, called VISH-Pred, comprising models built by fine-tuning ESM2 transformer models on a large, experimentally validated, curated dataset of protein and peptide toxicities. The primary steps in the VISH-Pred framework are to efficiently estimate protein toxicities taking just the protein sequence as input, employing an under sampling technique to handle the humongous class-imbalance in the data and learning representations from fine-tuned ESM2 protein language models which are then fed to machine learning techniques such as Lightgbm and XGBoost. The VISH-Pred framework is able to correctly identify both peptides/proteins with potential toxicity and non-toxic proteins, achieving a Matthews correlation coefficient of 0.737, 0.716 and 0.322 and F1-score of 0.759, 0.696 and 0.713 on three non-redundant blind tests, respectively, outperforming other methods by over $10\%$ on these quality metrics. Moreover, VISH-Pred achieved the best accuracy and area under receiver operating curve scores on these independent test sets, highlighting the robustness and generalization capability of the framework. By making VISH-Pred available as an easy-to-use web server, we expect it to serve as a valuable asset for future endeavors aimed at discerning the toxicity of peptides and enabling efficient protein-based therapeutics.

Peptide Toxins from Antarctica: The Nemertean Predator and Scavenger Parborlasia corrugatus (McIntosh, 1876).

Peptide toxins from marine invertebrates have found use as drugs and in biotechnological applications. Many marine habitats, however, remain underexplored for natural products, and the Southern Ocean is among them. Here, we report toxins from one of the top predators in Antarctic waters: the nemertean worm Parborlasia corrugatus (McIntosh, 1876). Transcriptome mining revealed a total of ten putative toxins with a cysteine pattern similar to that of alpha nemertides, four nemertide-beta-type sequences, and two novel full-length parborlysins. Nemertean worms express toxins in the epidermal mucus. Here, the expression was determined by liquid chromatography combined with mass spectrometry. The findings include a new type of nemertide, 8750 Da, containing eight cysteines. In addition, we report the presence of six cysteine-containing peptides. The toxicity of tissue extracts and mucus fractions was tested in an Artemia assay. Notably, significant activity was observed both in tissue and the high-molecular-weight mucus fraction, as well as in a parborlysin fraction. Membrane permeabilization experiments display the membranolytic activity of some peptides, most prominently the parborlysin fraction, with an estimated EC50 of 70 nM.

Peptide toxins that target vertebrate voltage-gated sodium channels underly the painful stings of harvester ants.

Harvester ants (genus Pogonomyrmex) are renowned for their stings which cause intense, long-lasting pain, and other neurotoxic symptoms in vertebrates. Here, we show that harvester ant venoms are relatively simple and composed largely of peptide toxins. One class of peptides is primarily responsible for the long-lasting local pain of envenomation via activation of peripheral sensory neurons. These hydrophobic, cysteine-free peptides potently modulate mammalian voltage-gated sodium (NaV) channels, reducing the voltage threshold for activation and inhibiting channel inactivation. These toxins appear to have evolved specifically to deter vertebrates.

PEGylation of Terminal Ligands as a Route to Decrease the Toxicity of Radiocontrast Re6-Clusters.

The development of novel radiocontrast agents, mainly used for the visualization of blood vessels, is still an emerging task due to the variety of side effects of conventional X-ray contrast media. Recently, we have shown that octahedral chalcogenide rhenium clusters with phosphine ligands-Na2H14[{Re6Q8}(P(C2H4COO)3)6] (Q = S, Se)-can be considered as promising X-ray contrast agents if their relatively high toxicity related to the high charge of the complexes can be overcome. To address this issue, we propose one of the most widely used methods for tuning the properties of proteins and peptides-PEGylation (PEG is polyethylene glycol). The reaction between the clusters and PEG-400 was carried out in acidic aqueous media and resulted in the binding of up to five carboxylate groups with PEG. The study of cytotoxicity against Hep-2 cells and acute toxicity in mice showed a twofold reduction in toxicity after PEGylation, demonstrating the success of the strategy chosen. Finally, the compound obtained has been used for the visualization of blood vessels of laboratory rats by angiography and computed tomography.

Study of IsCT analogue peptide against Candida albicans and toxicity/teratogenicity in zebrafish embryos (Danio rerio).

Aim: An IsCT analogue peptide (PepM3) was designed based on structural studies of wasp mastoparans and tested against Candida albicans. Its effects on fungal cell membranes and toxicity were evaluated. Materials & methods: Antifungal activity was analyzed using a microdilution susceptibility test. Toxicity was assessed using human skin keratinocytes (HaCaT) and zebrafish embryos. Results: PepM3 demonstrated activity against C. albicans and a synergistic effect with amphotericin B. The peptide presented fungicidal action with damage to the fungal cell membrane, low toxicity in HaCat cells and was nonteratogenic in zebrafish embryos. Conclusion: Evaluating structural modifications is essential for the development of new agents with potential activity against fungal pathogens and for the reduction of toxic and teratogenic effects.

Assessing the Safety of MA-[D-Leu-4]-OB3, a Synthetic Peptide Leptin Mimetic: Two Pre-Clinical Toxicity Studies in Male and Female C57BL/6 Mice.

Although the regulatory influence of leptin on energy balance, glycemic control, immunity, reproduction, and cognition is well established, its clinical application to common obesity and its co-morbidities has been limited by impaired transport across the blood-brain barrier, and tendencies to induce adverse side effects. To circumvent these drawbacks, MA-[D-Leu-4]-OB3, a leptin-related synthetic peptide that mimics the metabolic and neurotrophic effects of leptin in mouse models of genetic and non-genetic obesity, diabetes, and cognitive dysfunction, has been developed. This report presents the results of our initial efforts to assess the safety of orally delivered MA-[D-Leu-4]-OB3. Two pre-clinical studies were done in male and female C57BL/6 mice: a short-term study with a high dose of MA-[D-Leu-4]-OB3 (50 mg/kg/100 µL/day) and a dose-response study with 3 increasing concentrations of MA-[D-Leu-4]-OB3 (16.6, 50, and 150 mg/kg/100 µL/day). Body weight, food and water intake, glucose tolerance, and episodic memory were evaluated. Once-daily cage-side clinical observations were conducted to detect any physical or behavioral indicators of toxicity. Our results indicate that all metabolic and neurologic endpoints tested were either unaffected or improved by MA-[D-Leu-4]-OB3, and no clinical indicators of toxicity were evident. Together with our previously reported efficacy data, these results provide additional evidence supporting further development of this novel synthetic peptide leptin mimetic as a first-in-class peptide drug candidate for the treatment of a number of metabolic and/or cognitive dysfunctions in humans.

Heterologous Expression of an Insecticidal Peptide Obtained from the Transcriptome of the Colombian Spider Phoneutria depilate.

Spider venoms are composed, among other substances, of peptide toxins whose selectivity for certain physiological targets has made them powerful tools for applications such as bioinsecticides, analgesics, antiarrhythmics, antibacterials, antifungals and antimalarials, among others. Bioinsecticides are an environmentally friendly alternative to conventional agrochemicals. In this paper, the primary structure of an insecticidal peptide was obtained from the venom gland transcriptome of the ctenid spider Phoneutria depilata (Transcript ID PhdNtxNav24). The peptide contains 53 amino acids, including 10 Cys residues that form 5 disulfide bonds. Using the amino acid sequence of such peptide, a synthetic gene was constructed de novo by overlapping PCRs and cloned into an expression vector. A recombinant peptide, named delta-ctenitoxin (rCtx-4), was obtained. It was expressed, folded, purified and validated using mass spectrometry (7994.61 Da). The insecticidal activity of rCtx-4 was demonstrated through intrathoracic injection in crickets (LD50 1.2 µg/g insect) and it was not toxic to mice. rCtx-4 is a potential bioinsecticide that could have a broad spectrum of applications in agriculture.

Safety Assessment of Soy Proteins and Peptides as Used in Cosmetics.

The Expert Panel for Cosmetic Ingredient Safety (Panel) assessed the safety of soy proteins and peptides, which function in cosmetics primarily as hair conditioning agents and skin-conditioning agents-miscellaneous. The Panel considered relevant data related to these ingredients. The Panel concluded that soy proteins and peptides are safe in cosmetics in the present practices of use and concentration described in this safety assessment.

Venom composition and bioactive RF-amide peptide toxins of the saddleback caterpillar, Acharia stimulea (Lepidoptera: Limacodidae).

Limacodidae is a family of lepidopteran insects comprising >1500 species. More than half of these species produce pain-inducing defensive venoms in the larval stage, but little is known about their venom toxins. Recently, we characterised proteinaceous toxins from the Australian limacodid caterpillar Doratifera vulnerans, but it is unknown if the venom of this species is typical of other Limacodidae. Here, we use single animal transcriptomics and venom proteomics to investigate the venom of an iconic limacodid, the North American saddleback caterpillar Acharia stimulea. We identified 65 venom polypeptides, grouped into 31 different families. Neurohormones, knottins, and homologues of the immune signaller Diedel make up the majority of A.stimulea venom, indicating strong similarities to D. vulnerans venom, despite the large geographic separation of these caterpillars. One notable difference is the presence of RF-amide peptide toxins in A. stimulea venom. Synthetic versions of one of these RF-amide toxins potently activated the human neuropeptide FF1 receptor, displayed insecticidal activity when injected into Drosophila melanogaster, and moderately inhibited larval development of the parasitic nematode Haemonchus contortus. This study provides insights into the evolution and activity of venom toxins in Limacodidae, and provides a platform for future structure-function characterisation of A.stimulea peptide toxins.

Interleukin-37 suppresses the cytotoxicity of hepatitis B virus peptides-induced CD8+ T cells in patients with acute hepatitis B.

Interleukin-37 (IL-37) is a newly identified anti-inflammatory cytokine, owning immunosuppressive activity in infectious diseases. The aim of this study was to investigate the regulatory function of IL-37 on CD8+ T cells during hepatitis B virus (HBV) infection. Eighteen acute hepatitis B (AHB) patients, thirty-nine chronic hepatitis B (CHB) patients, and twenty controls were enrolled. IL-37 concentration was measured by ELISA. IL-37 receptor subunits expressions on CD8+ T cells were assessed by flow cytometry. Purified CD8+ T cells were stimulated with HBV peptides and recombinant IL-37. Perforin and granzyme B secretion was investigated by ELISPOT. Programmed death-1 (PD-1) and cytotoxic T-lymphocyte associated protein-4 (CTLA-4) mRNA expressions were semi-quantified by real-time PCR. CD8+ T cell cytotoxicity was assessed in direct contact and indirect contact coculture with HepG2.2.15 cells. Plasma IL-37 level was down-regulated and negatively correlated with aminotransferase levels in AHB patients. There were no significant differences of IL-37 receptor subunits among AHB patients, CHB patients, and controls. Exogenous IL-37 stimulation suppressed HBV peptides-induced perforin and granzyme B secretion by CD8+ T cells in AHB patients, but not in CHB patients. Exogenous IL-37 stimulation did not affect proinflammatory cytokines secretion as well as PD-1/CTLA-4 mRNA expressions in CD8+ T cells in AHB and CHB patients. Exogenous IL-37 stimulation dampened HBV peptide-induced CD8+ T cell cytotoxicity in a cell-to-cell contact manner. The current data indicated that acute HBV infection might induce down-regulation of IL-37, which might be associated with enhanced CD8+ T cell cytotoxicity and liver damage.

Composition and toxicity of venom produced by araneophagous white-tailed spiders (Lamponidae: Lampona sp.).

Prey-specialised spiders are adapted to capture specific prey items, including dangerous prey. The venoms of specialists are often prey-specific and less complex than those of generalists, but their venom composition has not been studied in detail. Here, we investigated the venom of the prey-specialised white-tailed spiders (Lamponidae: Lampona), which utilise specialised morphological and behavioural adaptations to capture spider prey. We analysed the venom composition using proteo-transcriptomics and taxon-specific toxicity using venom bioassays. Our analysis identified 208 putative toxin sequences, comprising 103 peptides < 10 kDa and 105 proteins > 10 kDa. Most peptides belonged to one of two families characterised by scaffolds containing eight or ten cysteine residues. Toxin-like proteins showed similarity to galectins, leucine-rich repeat proteins, trypsins and neprilysins. The venom of Lampona was shown to be more potent against the preferred spider prey than against alternative cricket prey. In contrast, the venom of a related generalist was similarly potent against both prey types. These data provide insights into the molecular adaptations of venoms produced by prey-specialised spiders.

Function Prediction of Peptide Toxins with Sequence-Based Multi-Tasking PU Learning Method.

Peptide toxins generally have extreme pharmacological activities and provide a rich source for the discovery of drug leads. However, determining the optimal activity of a new peptide can be a long and expensive process. In this study, peptide toxins were retrieved from Uniprot; three positive-unlabeled (PU) learning schemes, adaptive basis classifier, two-step method, and PU bagging were adopted to develop models for predicting the biological function of new peptide toxins. All three schemes were embedded with 14 machine learning classifiers. The prediction results of the adaptive base classifier and the two-step method were highly consistent. The models with top comprehensive performances were further optimized by feature selection and hyperparameter tuning, and the models were validated by making predictions for 61 three-finger toxins or the external HemoPI dataset. Biological functions that can be identified by these models include cardiotoxicity, vasoactivity, lipid binding, hemolysis, neurotoxicity, postsynaptic neurotoxicity, hypotension, and cytolysis, with relatively weak predictions for hemostasis and presynaptic neurotoxicity. These models are discovery-prediction tools for active peptide toxins and are expected to accelerate the development of peptide toxins as drugs.

Investigation of Peptide Toxin Diversity in Ribbon Worms (Nemertea) Using a Transcriptomic Approach.

Nemertea is a phylum of nonsegmented worms (supraphylum: Spiralia), also known as ribbon worms. The members of this phylum contain various toxins, including peptide toxins. Here, we provide a transcriptomic analysis of peptide toxins in 14 nemertean species, including Cephalothrix cf. simula, which was sequenced in the current study. The summarized data show that the number of toxin transcripts in the studied nemerteans varied from 12 to 82. The most represented groups of toxins were enzymes and ion channel inhibitors, which, in total, reached a proportion of 72% in some species, and the least represented were pore-forming toxins and neurotoxins, the total proportion of which did not exceed 18%. The study revealed that nemerteans possess a much greater variety of toxins than previously thought and showed that these animals are a promising object for the investigation of venom diversity and evolution, and in the search for new peptide toxins.

Hybrid Hydrogels from Nongelling Polymers Using a Fibrous Peptide Hydrogelator at Low Concentrations.

Nature-made hydrogels typically combine a wide range of multiscale fibers into biological composite networks, which offer an adaptive property. Inspired by nature, we report a facile approach to construct hybrid hydrogels from a range of natural or commercially available synthetic nongelling polymers (e.g., poly(ethylene glycol), poly(acrylic acid), carboxylated cellulose nanocrystal, and sodium alginate) at a concentration as low as 0.53 wt % using a nonionic fibrous peptide hydrogelator. Through simply mixing the peptide hydrogelator with a polymer aqueous solution, stable hybrid hydrogels can be formed with the concentration of hydrogelator at ∼0.05 wt %. The gel strength of the resulting hydrogels can be effectively modulated by the concentration, molecular weight, and terminal group of the polymer. We further demonstrate that the molecular interactions between the peptide hydrogelator and the polymer are very crucial for the formation of hybrid hydrogel, which synergically induce the gelation at considerably low concentrations. A peptide hydrogelator can be easily obtained by aminolysis of alkyl-oilgo(γ-benzyl-l-glutamate) samples. Live/Dead assays indicate low cytotoxicity of the hybrid hydrogel toward HeLa cells. Combining the low-cost, scalable synthesis, and biocompatibility, the prepared peptide hydrogelator presents a potential candidate to expand the scope of polymer hydrogels for biomedical applications and also shows considerable commercial significance.

Neurotoxic and cytotoxic peptides underlie the painful stings of the tree nettle Urtica ferox.

The stinging hairs of plants from the family Urticaceae inject compounds that inflict pain to deter herbivores. The sting of the New Zealand tree nettle (Urtica ferox) is among the most painful of these and can cause systemic symptoms that can even be life-threatening; however, the molecular species effecting this response have not been elucidated. Here we reveal that two classes of peptide toxin are responsible for the symptoms of U. ferox stings: Δ-Uf1a is a cytotoxic thionin that causes pain via disruption of cell membranes, while ß/δ-Uf2a defines a new class of neurotoxin that causes pain and systemic symptoms via modulation of voltage-gated sodium (NaV) channels. We demonstrate using whole-cell patch-clamp electrophysiology experiments that ß/δ-Uf2a is a potent modulator of human NaV1.5 (EC50: 55 nM), NaV1.6 (EC50: 0.86 nM), and NaV1.7 (EC50: 208 nM), where it shifts the activation threshold to more negative potentials and slows fast inactivation. We further found that both toxin classes are widespread among members of the Urticeae tribe within Urticaceae, suggesting that they are likely to be pain-causing agents underlying the stings of other Urtica species. Comparative analysis of nettles of Urtica, and the recently described pain-causing peptides from nettles of another genus, Dendrocnide, indicates that members of tribe Urticeae have developed a diverse arsenal of pain-causing peptides.

Safety Assessment of Skin and Connective Tissue-Derived Proteins and Peptides as Used in Cosmetics.

The Expert Panel for Cosmetic Ingredient Safety (Panel) reviewed the safety of 19 skin and connective tissue-derived proteins and peptides, which are reported to function mainly as skin and/or hair conditioning agents in cosmetics. The Panel reviewed the relevant data provided and concluded that these ingredients are safe in the present practices of use and concentration described in this safety assessment.

GM-Pep: A High Efficiency Strategy to De Novo Design Functional Peptide Sequences.

Although peptides are regarded as ideal therapeutic agents, only a small proportion of the marketed drugs are peptides. In the past decade, pharmacists have paid great attention to the development of peptide therapeutics. Except a few approved chemically/rationally designed peptides, most attempts failed due to unsatisfactory efficacy or safety. Luckily, computation methods, such as artificial intelligence, have been utilized to accelerate the discovery of therapeutic peptides by predicting the activity, toxicity, and absorption, distribution, metabolism, and excretion of polypeptides. Usually, a specific biological activity of a peptide could be accurately determined by an interest-oriented binary classification constructed of a positive set and another un-experimentally validated negative set regardless of other characteristics, which suggests that it could be challenging to realize the comprehensive evaluation of the research object in the early stage of drug research and development. Herein, we proposed an integrated method (GM-Pep) that contained a conditional variational autoencoder model (CVAE) and a positive sample training multiclassifier (Deep-Multiclassifier) to effectively generate a single bioactive peptide sequence without toxicity and referential side effects. The results showed that our Deep-Multiclassifier model gave a sequence accuracy of up to 96.41% [toxicity (94.48%), antifungal (96.58%), antihypertensive (97.18%), and antibacterial (96.91%), respectively]. The properties of Deep-Multiclassifier and CVAE were validated through 12 first synthesized antibacterial peptides or compared to random peptides. The source code and data sets are available at https://github.com/TimothyChen225/GM-Pep.

Safety Assessment of Plant-Derived Proteins and Peptides as Used in Cosmetics.

The Expert Panel for Cosmetic Ingredient Safety (Panel) reviewed the safety of 19 plant-derived proteins and peptides, which function mainly as skin and/or hair conditioning agents in personal care products. The Panel concluded that 18 plant-derived proteins and peptides are safe as used in the present practices of use and concentration as described in this safety assessment, while the data on Hydrolyzed Maple Sycamore Protein are insufficient to determine safety.

Integrated skin sensitization assessment based on OECD methods (II): Hazard and potency by combining kinetic peptide reactivity and the "2 out of 3" Defined Approach.

Depending on regulatory requirements, the skin sensitization risk for new chemicals with potential consumer skin contact must be assessed by experimental testing by (i) binary hazard assessment to identify sensitizers, (ii) subclassification of sensitizers according to the Global Harmonized System (GHS), and (iii) derivation of a point of departure (PoD) for risk assessment. The Organisation for Economic Co-operation and Development (OECD) recently published a test guideline incorporating the "2 out of 3" defined approach (2o3 DA) for skin sensitization hazard assessment and added the kinetic direct peptide reactivity assay (kDPRA) as a stand-alone test guideline method for GHS subclassification. The 2o3 DA requires that at least two in vitro tests are conducted. The cell-based tests and the kDPRA generate, next to a binary outcome with a fixed threshold, continuous concentration-response data, which can be used in quantitative regression models to derive a PoD. The sequence of testing for the 2o3 DA is flexible. Here we compare different testing sequences and how they can be combined with kDPRA data to provide a PoD in parallel to hazard identification (hazard ID) and GHS subclassification. A set of 188 chemicals with available in vitro data was evaluated for the final PoD using these dif-ferent testing sequences. The results indicate that testing can start with DPRA / kDPRA and either of the cell-based assays, and that testing can stop after two congruent tests without major impact on the final PoD for chemicals within the applica-bility domain of the kDPRA.

Peptide mediated, enhanced toxicity of a bacterial pesticidal protein against southern green stink bug.

The damage caused by stink bugs that feed on agricultural crops accounts for such significant losses that transgenic plant resistance to stink bugs would be highly desirable. As the level of toxicity of the Bacillus thuringiensis-derived, ETX/Mtx2 pesticidal protein Mpp83Aa1 is insufficient for practical use against the southern green stink bug Nezara viridula, we employed two disparate approaches to isolate peptides NvBP1 and ABP5 that bind to specific proteins (alpha amylase and aminopeptidase N respectively) on the surface of the N. viridula gut. Incorporation of these peptides into Mpp83Aa1 provided artificial anchors resulting in increased gut binding, and enhanced toxicity. These peptide-modified pesticidal proteins with increased toxicity provide a key advance for potential future use against N. viridula when delivered by transgenic plants to mitigate economic loss associated with this important pest.