Accurate prediction of peptide structures and peptide-receptor complexes is essential for rational peptide drug development. However, the inherent conformational flexibility of short and disordered peptides presents a fundamental challenge. The AlphaFold model series, which has progressed from AlphaFold2 through AlphaFold-Multimer to AlphaFold3, has substantially advanced computational peptide structure prediction through innovations in geometric reasoning (invariant point attention) and interface-focused confidence metrics (ipTM score), achieving high accuracy for both monomeric peptide structures and multi-chain complexes. However, these models output static conformations, whereas many bioactive peptides adopt their functional conformations only upon binding-often corresponding to low-probability states that static predictions may overlook, leading to failures in virtual screening. This review synthesizes recent advances in the AlphaFold series for peptide studies and applications, discusses their current strengths in structure prediction and receptor-binding analysis, and examines the limitations in capturing conformational dynamics, transient interactions, and chemical modifications. Recent studies have suggested that integrated computational strategies that combine AlphaFold predictions with molecular dynamics simulations, free energy calculations, and ensemble sampling to enhance predictive accuracy and better represent the dynamic nature of peptide-drug interactions. These complementary approaches position AlphaFold as a central computational platform in structure-guided peptide drug design, enabling more efficient lead identification and optimization while bridging the gap between static computational predictions and the complex biophysical reality of peptide therapeutics.
Determining the structural basis of antigen recognition by antibodies and T cell receptors (TCRs) provides critical insights into effective immune targeting and can inform design of biotherapeutics and vaccines. Accurate computational modeling of antibodies and TCRs in complex with their targets poses a major challenge for predictive methods, including AlphaFold, which is generally accurate for modeling protein complexes but has shown limited success for immune recognition. In this study we assessed the performance of AlphaFold2, AlphaFold3, increased sampling protocols, and related deep learning methods for modeling antibody-protein, antibody-peptide, and TCR-peptide-major histocompatibility complex (pMHC) recognition. We show that increased sampling and AlphaFold3 generally improve performance relative to default sampling and AlphaFold2, however predictive accuracy and improvement levels varied considerably among interface classes, with antibody-peptide complexes representing a challenge despite their small antigen size. Comparing per-case success across methods showed some complementarity, indicating opportunities for increased success through model pooling approaches, for instance increasing antibody-peptide near-native success from 41% to 59%. Analysis of AlphaFold confidence scores and modeling of a noncanonical complex provided further insights into predictive performance. These results highlight considerations for predictive antibody and TCR complex modeling efforts, while revealing key distinctions among protocols, scoring, and immune complex classes.
Virtual screening has become an indispensable tool in modern structure-based drug discovery, enabling the identification of candidate molecules by computationally evaluating their potential to bind target proteins. The accuracy of such screenings critically depends on the quality of the target structures employed. Recent advances in protein structure prediction, particularly AlphaFold2, have revolutionized this field with unprecedented accuracy. However, AlphaFold2 models often exhibit limitations in local structural details, especially within binding pockets, which limit their utility for small molecule docking. In contrast, molecular dynamics simulations with accurate atomistic force fields can refine protein structures, but lack the ability to leverage the structural information provided by deep learning approaches. Here, we introduce bAIes, an integrative method that bridges this gap by combining physics-based force fields with data-driven predictions through Bayesian inference. Crucially, bAIes demonstrates a superior ability to discriminate between binders and nonbinders in virtual screening campaigns, outperforming both AlphaFold2 and molecular dynamics-refined models. By enhancing the usability of AlphaFold2 models without requiring extensive experimental or computational resources, bAIes offers a convenient solution to a longstanding challenge in structure-based drug design, potentially accelerating the early phases of drug discovery.
Surface-Induced Dissociation native Mass Spectrometry (SID-nMS) is a tandem MS activation method that yields information on the connectivity and stoichiometry of protein complexes. While insufficient for direct structure elucidation, the data derived from SID-nMS has considerable potential to inform multimeric protein structure prediction. We hypothesized that incorporating this data into a machine-learning framework could improve multimer prediction accuracy beyond that of existing deep-learning methods. To this end, we developed SIDFold, a novel AlphaFold-based deep-learning network. SIDFold is the first AlphaFold-like network to leverage experimental data during protein complex prediction, and the first deep-learning network to utilize nMS data for structure prediction. We benchmarked SIDFold on the BETA protein set, and observed an improvement in RMSD in 138 of 227 cases including 27 targets in which the predicted structure attained near-native accuracy. We then evaluated the network on 20 proteins with experimental SID-nMS data, yielding an improved RMSD in 18 cases, with five of these cases improving to a high-accuracy complex. Finally, we tested SIDFold against a previously published SID-guided Rosetta docking method, where we saw improvement in 13 of 16 proteins. SIDFold is freely available on GitHub, with example files and commands available in the Supplementary Information.
Deep-learning methods have transformed our ability to predict the three-dimensional structures of folded proteins from sequence, and coarse-grained simulations have made it possible to study intrinsically disordered proteins at the proteome scale. More than half of human proteins, however, contain mixtures of disordered regions and one or more folded domains, and the biological function of these multi-domain proteins (MDPs) depends on the interplay between the folded and disordered regions. Here, we developed AF-CALVADOS, a coarse-grained simulation model that is informed by AlphaFold to model the dynamics of intrinsically disordered proteins and MDPs containing mixtures of folded and disordered regions. AF-CALVADOS leverages information from AlphaFold 2 to model folded regions that we then integrate with the coarse-grained CALVADOS model. Our automated framework makes it possible to perform simulations of any soluble folded or disordered protein without manually defining the folded regions, enabling scaling to the proteome level. We validate AF-CALVADOS using experimental small-angle x-ray scattering data for more than 400 proteins and find that it performs well across proteins with varying amounts of ordered and disordered regions. We demonstrate the scalability of AF-CALVADOS by performing simulations of 12,483 intracellular human proteins and make the data freely available; we envisage that large-scale simulation data generated by AF-CALVADOS can be used to benchmark or train machine learning models for flexible, MDPs. The conformational ensembles can also be used to study sequence-dynamics-function relationships at scale, and can shed light on the interplay between folded and disordered regions. We illustrate this by analyzing the disordered regions in 1487 human transcription factors. AF-CALVADOS is available as part of the CALVADOS package at: github.com/KULL-Centre/CALVADOS. Single-chain simulations with AF-CALVADOS can be run via Google Colab: colab.research.google.com/github/KULL-Centre/_2025_buelow_AF-CALVADOS/blob/main/AF_CALVADOS.ipynb.
Cellulosomes are large, surface-displayed enzyme complexes that enable anaerobic bacteria to degrade recalcitrant plant polysaccharides, yet cellulosome-expressing bacteria are thought to be rare in the human gut. Here, we show that extensive sequence divergence obscures the detection of many ruminococcal cellulosomes by conventional sequence homology-based methods. Using proteome-scale AlphaFold2 structural predictions, we uncovered a substantially expanded set of putative cellulosome-producing Ruminococcus species, including six previously unrecognized human symbionts. Structure-based clustering identifies several novel cohesin families that retain conserved folds despite extreme sequence divergence and define distinct, phylogenetically conserved cellulosome architectures. The analysis reveals R. callidus and related human symbionts encode elaborate cellulosomes that are invisible to sequence-based annotation. Similarly, R. difficilis, a human gut symbiont, has been found to possess genes for an atypical cohesin-based assembly enriched in amylases and related starch-binding proteins, which may enable this microbe to degrade resistant starches that evade digestion in the upper gastrointestinal tract. Together, these findings reveal that ruminococcal cellulosomes are far more prevalent and diverse than previously appreciated and demonstrate the power of structural proteomics to uncover deeply divergent functional systems in the gut microbiome.IMPORTANCEPlant cell wall polysaccharides are a major dietary carbon source, yet their degradation relies on rare, highly specialized microbial enzyme assemblies known as cellulosomes, which have long been considered uncommon in the human gut. Using proteome-scale structure prediction combined with experimental validation, we show that cellulosomes are far more widespread and structurally diverse in human-associated Ruminococcus species than previously appreciated. We identify multiple new cohesin families and reveal distinct cellulosome architectures likely adapted to degrade different dietary substrates. Together, these findings redefine the distribution and evolution of cellulosomes in gut microbes and demonstrate the power of structural proteomics to uncover deeply diverged biological systems.
This study establishes an integrated genome-to-structure-to-process framework that significantly enhances 1-hydroxyphenazine (1-OH-PHZ) biosynthesis in Pseudomonas aeruginosa KAEH25. Nanopore long-read sequencing completely resolved the 10,358 bp phenazine biosynthetic locus, confirming intact organization of core (phzB-phzG) and tailoring (phzH, phzS) genes. AlphaFold modeling and InterProScan analysis validated conserved catalytic domains across all pathway enzymes, confirming functional coherence. Plackett-Burman screening identified temperature, pH, glucose, peptone, inoculum size, and incubation time as significant production determinants, while response surface methodology optimized these factors to achieve a maximum 1-OH-PHZ titre of 24.85 µg mL⁻1 under refined conditions (37 °C, pH 6.5-7, 1% glucose, 2 g L⁻1 peptone, 10 mL inoculum, 48 h). The statistical model demonstrated exceptional explanatory power (R2 = 0.9605; Adeq Precision = 19.11). Preliminary antimicrobial assays demonstrated selective activity of 1-hydroxyphenazine (1-OH-PHZ), with inhibition zones of 20 ± 1 mm against Escherichia coli, 13 ± 0.5 mm against Salmonella enterica serovar Typhimurium, 13 ± 0.7 mm against Klebsiella pneumoniae, and 8 ± 0.3 mm against Staphylococcus aureus, while no inhibition was observed against Enterococcus faecalis or Listeria monocytogenes. Collectively, this systems-guided approach provides a transferable platform for optimizing microbial secondary metabolites and validates 1-OH-PHZ as a promising lead for antimicrobial development.
Peptide-activated G protein-coupled receptors (GPCRs) regulate physiological processes through interaction with neuropeptides and peptide hormones. Identifying endogenous peptide agonists remains challenging, as peptide-GPCR pairings often follow gene-family relationships that offer limited predictive insight for orphan GPCRs without characterized homologs. Using a dataset of experimentally validated peptide-GPCR interactions from Caenorhabditis elegans, we demonstrate that AF-multimer confidence metrics partially discriminate agonist from non-agonist complexes, with improved discrimination using AF-Multistate-derived active-state templates. Feature analysis revealed that AF-multimer's pair representations outperform single representations, with distinct subregions providing complementary signals. Leveraging these insights, we developed DeorphaNN, a graph neural network integrating active-state GPCR-peptide structural predictions, interatomic interactions, and deep learning embeddings to prioritize putative peptide agonists for experimental screening. DeorphaNN generalized across diverse species, as shown by performance on annelid and human retrospective benchmarks. Experimental validation confirmed predicted agonists for two orphan GPCRs, demonstrating its utility for accelerating peptide-GPCR deorphanization.
The Min system disassembles FtsZ-rings after septation in Bacillus subtilis and is localized to the nascent division plane and cell poles by the protein MinJ. The N-terminal region of MinJ contains transmembrane segments while the C-terminal region of MinJ contains a PDZ domain but its topology and functional domains are poorly understood. Here we empirically test MinJ topology based on a variety of transmembrane prediction models and find that the data is most consistent with Alphafold3, which predicts a 9-pass transmembrane protein with an external N-terminus and internal C-terminus. Deletion analysis indicates that all regions of the protein tested are required for function but deletion of the PDZ domain alone preserves polar localization and interaction with both MinD and DivIVA. Moreover, Alphafold predicts that transmembrane segments 6 and 7 comprise staves of an unusual transmembrane β-sheet and deletion of the putative β-sheet in the absence of MinD results in a minicell frequency that exceeds mutation of MinD alone. Bioinformatic analysis indicates that MinJ is highly conserved within Firmicutes and is co-conserved with MinD and DivIVA with which it interacts. Our data clarify the structure of MinJ and support models in which MinJ has functions in addition to restricting the activity of the Min system. Faithful positioning of the bacterial division site is important for cell growth and is coordinated by the conserved Min system. Although the Min system of Bacillus subtilis has been extensively studied, MinJ, the membrane protein that links the division inhibitor MinCD to the polar determinant DivIVA, remains the least well-understood. Here we experimentally define the membrane topology of MinJ and show that our data are most consistent with a nine-pass transmembrane architecture predicted by AlphaFold3. We further provide genetic, cell biological, and evolutionary evidence supporting that two of the staves form a highly conserved putative transmembrane β-sheet, a structure normally excluded from the plasma membrane. Our findings refine MinJ structural organization and provide a framework for understanding its conserved functions in bacterial cell division.
Artificial intelligence (AI) has emerged as a powerful tool for solving real world problems across a wide range of industries and is increasingly being utilised by pharmaceutical companies to discover novel drug targets, biomarkers, and new drugs. Several AI-driven small molecules have entered clinical trials over the past few years, but their fate remains unknown. Currently, no commercially available compounds have been developed solely using AI approaches. In this perspective, we examine the current use of AI in drug discovery for neuropsychiatry. The pace of drug discovery in neuropsychiatric medicine has been generally sluggish, largely due to challenges such as poor pharmacological selectivity, the blood-brain barrier, and a limited understanding of disease mechanisms. AI may offer innovative solutions to these challenges. However, relative to fields such as oncology, the impact of AI on the discovery of neuropsychiatric drugs has been limited. Although novel AI-tools have been developed to overcome some of the challenges involved in neuropsychiatry drug discovery, their effectiveness has not been sufficiently evaluated. To date, innovative tools such as AlphaFold have been used to identify drug candidates for multiple neuropsychiatric conditions. AI-driven platforms have been used to study behavioural data from preclinical models to identify novel clinical candidates in clinical trials (e.g., ulotaront, phase III). It is anticipated that the availability of large-scale multi-omics data ('big data') will likely increase in the future, allowing us to gain a better understanding of gene-associated mechanisms in psychiatry. Using AI-based technologies such as AlphaFold, future pharmacological targets will be identified based on gene expression data, and large libraries of chemical compounds will be screened rapidly to identify novel drug candidates, resulting in shorter pre-clinical phase with lower costs.
Solvent accessible surface area (SASA) is widely used to describe protein stability, ligand binding, mutation effects, and protein-protein interfaces. As structural biology workloads expand to predicted-structure col-lections, trajectories, and large assemblies, SASA tools must combine reproducible calculation with high throughput, low memory use, and workflow-friendly input handling. We present zsasa, a Zig-based SASA engine with command-line and Python interfaces. zsasa implements the established Shrake-Rupley and Lee-Richards algorithms, provides exact f64/f32 modes and an optional bitmask approximation, and supports batch and trajectory workflows, compressed structure inputs, and configurable atom classification including Chemical Component Dictionary (CCD)-based radii for non-standard components. In matched Shrake-Rupley validation on 4,370 Escherichia coli AlphaFold Database structures, exact double-precision zsasa reproduced FreeSASA total SASA values to near numerical identity. In 10-thread batch benchmarks on the E. coli and 23,586-structure human AlphaFold collections, zsasa achieved a 2.94-fold speedup over a FreeSASA batch wrapper in exact f64 mode. In bitmask mode, zsasa reached up to a 9.70-fold speedup, using roughly 12.5% to 25% of the comparator peak memory. Trajectory benchmarks exceeded 1,000 frames/s at tens of megabytes of peak memory, and a 4.5-million-atom PDB stress-test file completed in less than 5 s. These results support zsasa as a practical tool for reproducible, low-memory generation of surface-derived structural features at large scale. zsasa is available under the MIT License at https://github.com/N283T/zsasa .
Astragaloside IV (AS-IV), a primary active constituent of Astragalus membranaceus, possesses diverse pharmacological properties, including anti-inflammatory, antioxidant activities. However, its direct molecular targets in poultry have not been fully elucidated. In this study, a proteome-wide target identification strategy was established for Gallus gallus, integrating AlphaFold protein structures, P2Rank-based binding pocket detection, and high-throughput reverse docking with UniDock. Following systematic screening, 15 high-confidence candidate targets were prioritized. These targets are associated with lipid metabolism, cell cycle regulation, immune modulation and oxidative stress defense, etc. These findings bridge the gap between empirical observations of AS-IV efficacy and mechanistic understanding at the molecular level. Consequently, this study will accelerate the evidence-based integration of AS-IV into poultry health management strategies.
Over the past 75 years, especially the recent quarter century, bioinformatics has undergone a profound transformation, evolving from a specialized field of command-line tools to a cornerstone of modern biomedical and life sciences. We trace this journey through distinct technological eras driven by exponential biological data growth and parallel computational advances. The genomic revolution established foundational sequence analysis tools and was rapidly followed by the next-generation sequencing era, when unprecedented data volumes shifted the bottleneck from generation to analysis. This drove the development of web servers, cloud platforms, and containerized workflows to address scalability, accessibility, and reproducibility challenges. We now stand in the artificial intelligence (AI)-driven era, where deep learning (e.g. AlphaFold series) and large language models reshape structural biology, multi-modal data integration, and how researchers interact with tools through natural language prompting. This review highlights a recurring pattern as each technological era lowers barriers to entry, it simultaneously introduces new questions about transparency, trust, and rigor. By framing the popular rise of AI within this historical context, we provide a critical roadmap for navigating the cultural, ethical, and technical crossroads facing the next generation of bioinformatics.
Cone snails (Conus spp.) produce complex venoms rich in conotoxins, a diverse group of cysteine-rich peptides with high specificity toward ion channels, receptors, and transporters, making them valuable candidates for drug discovery. Despite the pharmacological potential of cone snail venoms, the venom composition of Conus inscriptus remains largely unexplored. In this study, we present the first comprehensive venom gland transcriptome of C. inscriptus collected from the southwest coast of India. High-throughput Illumina sequencing generated 179.6 million paired-end reads, which were assembled into 259,828 transcripts and 75,366 predicted coding sequences (CDS). Functional annotation revealed enrichment of genes involved in cellular processes, metabolism, protein processing, and signal transduction, reflecting the active biosynthetic nature of the venom gland. A total of 6,066 putative conotoxin genes were identified, of which 4,921 were classified into 23 recognised conotoxin superfamilies. The A, O1, and M superfamilies were the most abundant. Additionally, 1,145 transcripts were assigned to conflict groups due to overlapping superfamily characteristics. Analysis of conflict-associated transcripts revealed remarkable cystine framework diversity, including several previously unreported cysteine-rich architectures containing up to 20 cysteine residues. Structural characterisation using AlphaFold and FoldSeek identified both conserved proteins and a large proportion of highly novel proteins lacking recognisable structural homologs. Many of these proteins exhibited high intrinsic disorder, suggesting the presence of previously undescribed peptide scaffolds and lineage-specific venom components. Overall, the transcriptome of C. inscriptus reveals an extensive and previously undocumented repertoire of conotoxins and structurally unique proteins. These findings provide new insights into cone snail venom evolution and establish C. inscriptus as a promising source of novel bioactive peptides with potential applications in marine biotechnology, neuropharmacology, and peptide-based drug development.
Microbes produce bioactive secondary metabolites as toxins, pigments, or virulence factors. These specialized compounds are produced by nonribosomal peptide synthetases (NRPS), polyketide synthases (PKS), or hybrid NRPS/PKS pathways. The genes encoding NRPS and PKS reside in biosynthetic gene clusters (BGCs), some of which have no identified metabolite associated with them. Characterization of these orphan BGCs could provide insights into potential bioactive compounds that have yet to be discovered. Here, we characterize PA1216, a putative methyltransferase embedded within an NRPS BGC in Pseudomonas aeruginosa strain PAO1. We cloned, expressed, and purified PA1216, and developed an optimized differential scanning fluorimetry assay to measure its thermal stability, demonstrating concentration-dependent stabilization in the presence of established methyltransferase cofactors and inhibitors. We then adapted this assay for high-throughput screening of potential PA1216 substrates, identifying destabilizing compounds, including glycyl-glycine dipeptides, amino esters with aromatic or basic side chains, and N-Boc-protected amino acids. In contrast, sodium salts of organic acids stabilized PA1216. Lastly, we employed AlphaFold to construct a predictive model, revealing that PA1216 contains a Rossmann-like fold and a glycine-rich loop, typical of class I methyltransferases, and we corroborated these secondary structural elements using circular dichroism spectroscopy. Overall, these studies illuminate PA1216 function and establish a platform for characterizing cryptic gene clusters within secondary metabolic pathways.
The mating-type locus of the miso and soy sauce yeast Zygosaccharomyces sp. contains a gene encoding the transcription factor Mata2 (hereafter ZygoMata2), which has a DNA-binding domain containing a high mobility group (HMG)-box not found in the baker's yeast Saccharomyces cerevisiae. To investigate the role of ZygoMata2 in mating-type a expression, we constructed mutant strains of ZygoMATa2 using CRISPR-Cas9. The resulting mutants showed reduced expression of ZygoSTE6, which is specific to mating-type a, and did not mate, suggesting that ZygoMata2 regulates the expression of mating-type a in Zygosaccharomyces sp. Expression of ZygoSTE6 was observed when the HMG-box of ZygoMata2 was exchanged with that of Mat1-Mc, which regulates mating-type expression in Schizosaccharomyces pombe, indicating that ZygoMata2 recognizes its target DNA sequence via the HMG-box. Substitution of the predicted upstream activation sequence (UAS) in ZygoSTE6 with the UAS of a mating-type a-specific gene from another yeast species led to mating-type a-specific gene expression in transformed Zygosaccharomyces sp. yeast, indicating that ZygoMata2 can recognize the UAS of mating-type a-specific genes of other species. We speculated that the flexibility in the binding sequences of ZygoMata2 is due to synergistic or cooperative binding with the DNA-binding protein ZygoMcm1. This hypothesis was supported by replacement of the UAS of ZygoSTE6 with the complete palindrome sequence, P(PAL), and by the results of protein and DNA structure prediction using AlphaFold. Collectively, our study indicates that ZygoMata2 regulates mating-type a-specific gene expression via its HMG-box, which shows flexible recognition of the binding sequence of other HMG-box transcription factors.
Cryptosporidium parvum is a major cause of severe diarrheal disease, yet therapeutic options remain limited. This study integrated host-parasite transcriptomic profiling, GC-MS phytochemical profiling, computational screening, and in vitro experimental validation to evaluate anti-Cryptosporidium phytochemicals from Garcinia mangostana extract (GME). Host epithelial transcriptional responses (GSE2077) during C. parvum infection were analyzed to map infection-associated host reprogramming. A curated library of 30 identified phytochemicals was computationally docked against prioritized parasite-relevant targets: calcium-dependent protein kinase 1 (CDPK1), C. parvum 1-phosphatidylinositol 4-kinase [PI(4)K; cgd8_4500/UniProt Q5CVD3, AlphaFold model AF-Q5CVD3-F1], and cleavage and polyadenylation specificity factor 3 (CPSF3). The anti-Cryptosporidium efficacy of GME was subsequently evaluated in vitro by assessing C. parvum oocyst count and viability over 48 h. Transcriptomic analysis suggested a biphasic pattern of host reprogramming involving inflammatory signaling, sterol metabolism, Wnt signaling, and cell-cycle regulation. GC-MS analysis identified sterols and fatty acid derivatives, including stigmasterol, campesterol, and ethyl iso-allocholate. Molecular docking against the C. parvum PI(4)K model predicted campesterol (NP028) and ethyl iso-allocholate (NP029) as leading PI(4)K-oriented candidates, with mean affinities of - 7.97 ± 0.59 and - 7.72 ± 0.61 kcal/mol, respectively, while NP008 showed a predicted multi-target interaction profile, particularly against CPSF3. In vitro assays showed a significant, concentration- and time-dependent reduction in C. parvum oocyst viability, with an LC50 of 830 µg/mL after 48 h. Overall, the parasite-specific docking and oocysticidal data support GME phytochemicals as promising anti-Cryptosporidium candidates and generate testable hypotheses for CDPK1, PI(4)K, or CPSF3 involvement, while direct intracellular infection assays and recombinant enzyme inhibition studies are required to confirm target engagement.
Microglia-mediated neuroinflammation and oxidative stress are pivotal drivers of secondary injury following traumatic brain injury (TBI). While neddylation governs essential cellular functions, its specific contribution to microglial activation and TBI pathology remains poorly understood. We integrated bulk microglial RNA sequencing profiles with single-cell RNA sequencing (scRNA-seq) datasets from TBI mouse brains. To assess therapeutic potential, we employed a controlled cortical impact mouse model and treated animals with the neddylation inhibitor MLN4924. The role of microglia was validated using microglia-depleted mice. Mechanistically, a combinatorial approach utilizing AlphaFold 3 molecular docking predictions, quantitative proteomics, and immunoprecipitation-mass spectrometry was performed to identify molecular targets. We revealed a specific and robust up-regulation of neddylation exclusively within microglial clusters. Pharmacological inhibition of neddylation using MLN4924 significantly ameliorated neurological deficits, attenuated brain edema, and preserved blood-brain barrier integrity. Crucially, these neuroprotective benefits were abrogated in microglia-depleted mice, pinpointing microglia as the primary cellular target. We identified the glutamate-cysteine ligase modifier subunit (GCLM) as a novel substrate of the CUL3-KLHL12 E3 ligase complex. MLN4924 inhibits CUL3 neddylation, thereby impeding the CUL3-KLHL12-mediated ubiquitination and degradation of GCLM. Consequently, GCLM stabilization restores intracellular glutathione synthesis, effectively scavenging reactive oxygen species and mitigating neuroinflammation. Our findings characterize the Neddylation-CUL3-KLHL12-GCLM axis as a critical regulator of microglial redox homeostasis and highlight this pathway as a promising therapeutic target for TBI intervention.
Amyloid oligomers are considered to be the most toxic species in neurodegenerative diseases, such as Alzheimer's, and Parkinson's, among others. A major challenge in studying amyloid oligomers has been their heterogeneous character, coupled with a lack of methods that can probe targeted, localized structural information. In this paper, we show that in detergent micelles, the oligomeric state of amyloid cylindrin peptide K11 V increases from a hexamer to an octamer. This reveals a detergent-like toxicity mechanism of the K11 V peptide, in which the octamer-form extracts lipids from biological membranes. The findings are confirmed by Alphafold simulations, showing that the octamer has a β-barrel structure, with a pore stabilized by using the unbranched hydrophobic tails of the detergent molecules. The experimental results are obtained by a unique electron spin resonance (ESR)-based methodology that can be applied to amyloidogenic proteins to uncover inaccessible amyloid oligomer structures. The process involves (1) stabilization of the minimal amyloid oligomer domain, using a cross-linking bifunctional nitroxide spin label, combined with (2) an oligomer stabilizing mutation, (3) DEER spectroscopy, which can be performed in a native environment, and (4) T1-edited DEER measurements, which can validate structure in samples with heterogeneous populations, combined with (5) a model-free distance reconstruction approach that can analyze distance distributions with inverted distance components, ideal for analysis of T1-edited DEER.
The autoimmune regulator (AIRE) is expressed in medullary thymic epithelial cells (mTECs) and is crucial for generating an immunocompetent T cell repertoire during central tolerance. Loss of AIRE function causes autoimmune polyglandular syndrome type 1 (APS-1), also known as autoimmune polyendocrinopathy-candidiasis-ectodermal dystrophy (APECED). Intracellular proteins are regulated by ubiquitination pathways. E3 ubiquitin ligases are specific proteins that confer selectivity to polyubiquitination, tagging proteins for proteasomal degradation while also performing other functions. In mTECs, peptides generated via the class I antigen-processing pathway are presented by HLA-I molecules to CD8 thymocytes to eliminate self-reactive T-cell precursors. AIRE induces promiscuous gene expression in mTECs, but little is known about the regulatory mechanisms of AIRE. Previously, we and others demonstrated that AIRE is an apoptosis inductor. Here, we demonstrate that AIRE increases the intracellular levels of SIAH-interacting protein (SIP), an adaptor protein of the SIAH E3 ubiquitin ligase family. We also show that AIRE interacts with SIAH1 in the human thymus. While AIRE contains two putative SIAH-interacting motifs, it interacts with SIAH proteins only through the sequence spanning residues 119-125. AlphaFold modeling indicated that the interaction between AIRE and SIAH1 is highly analogous to that observed between SIP and SIAH1, suggesting that both proteins compete for SIAH1 binding. The CARD domain and the SIAH-interacting motif is required to AIRE-mediated apoptosis. AIRE co-localized with SIAH proteins in the cytoplasm of HEK-293 cells. Finally, SIAH1 ubiquitinated AIRE, targeting it for proteasomal degradation. Collectively, these findings reveal a novel pathway for the degradation and regulation of AIRE.