Interaction Motif Research Articles

MucLiPred: Multi-Level Contrastive Learning for Predicting Nucleic Acid Binding Residues of Proteins.

Protein-molecule interactions play a crucial role in various biological functions, with their accurate prediction being pivotal for drug discovery and design processes. Traditional methods for predicting protein-molecule interactions are limited. Some can only predict interactions with a specific molecule, restricting their applicability, while others target multiple molecule types but fail to efficiently process diverse interaction information, leading to complexity and inefficiency. This study presents a novel deep learning model, MucLiPred, equipped with a dual contrastive learning mechanism aimed at improving the prediction of multiple molecule-protein interactions and the identification of potential molecule-binding residues. The residue-level paradigm focuses on differentiating binding from non-binding residues, illuminating detailed local interactions. The type-level paradigm, meanwhile, analyzes overarching contexts of molecule types, like DNA or RNA, ensuring that representations of identical molecule types gravitate closer in the representational space, bolstering the model's proficiency in discerning interaction motifs. This dual approach enables comprehensive multi-molecule predictions, elucidating the relationships among different molecule types and strengthening precise protein-molecule interaction predictions. Empirical evidence demonstrates MucLiPred's superiority over existing models in robustness and prediction accuracy. The integration of dual contrastive learning techniques amplifies its capability to detect potential molecule-binding residues with precision. Further optimization, separating representational and classification tasks, has markedly improved its performance. MucLiPred thus represents a significant advancement in protein-molecule interaction prediction, setting a new precedent for future research in this field.

Advances on BTB protein ubiquitination mediated plant development and stress response

The BTB (broad-complex, tramtrack, and bric-à-brac) domain is a highly conserved protein interaction motif in eukaryotes. They are widely involved in transcriptional regulation, protein degradation and other processes. Recently, an increasing number of studies have shown that these genes play important roles in plant growth and development, biotic and abiotic stress processes. Here, we summarize the advances of these proteins ubiquitination-mediated development and abiotic stress responses in plants based on the protein structure, which may facilitate the study of this type of gene in plants.

Surface Charge Can Modulate Phase Separation of Multidomain Proteins.

Phase separation has emerged as an important mechanism explaining the formation of certain biomolecular condensates. Biological phase separation is often driven by the multivalent interactions of modular protein domains. Beyond valency, the physical features of folded domains that promote phase separation are poorly understood. We used a model system─the small ubiquitin modifier (SUMO) and its peptide ligand, the SUMO interaction motif (SIM)─to examine how domain surface charge influences multivalency-driven phase separation. Phase separation of polySUMO and polySIM was altered by pH via a change in the protonation state of SUMO surface histidines. These effects were recapitulated by histidine mutations, which modulated SUMO solubility and polySUMO-polySIM phase separation in parallel and were quantitatively explained by atomistic modeling of weak interactions among proteins in the system. Thus, surface charge can tune the phase separation of multivalent proteins, suggesting a means of controlling phase separation biologically, evolutionarily, and therapeutically.

From interaction networks to interfaces, scanning intrinsically disordered regions using AlphaFold2

The revolution brought about by AlphaFold2 opens promising perspectives to unravel the complexity of protein-protein interaction networks. The analysis of interaction networks obtained from proteomics experiments does not systematically provide the delimitations of the interaction regions. This is of particular concern in the case of interactions mediated by intrinsically disordered regions, in which the interaction site is generally small. Using a dataset of protein-peptide complexes involving intrinsically disordered regions that are non-redundant with the structures used in AlphaFold2 training, we show that when using the full sequences of the proteins, AlphaFold2-Multimer only achieves 40% success rate in identifying the correct site and structure of the interface. By delineating the interaction region into fragments of decreasing size and combining different strategies for integrating evolutionary information, we manage to raise this success rate up to 90%. We obtain similar success rates using a much larger dataset of protein complexes taken from the ELM database. Beyond the correct identification of the interaction site, our study also explores specificity issues. We show the advantages and limitations of using the AlphaFold2 confidence score to discriminate between alternative binding partners, a task that can be particularly challenging in the case of small interaction motifs.

The development and characterization of a CRISPR/Cas9-mediated PD-1 functional knockout rat as a tool to study idiosyncratic drug reactions.

Idiosyncratic drug reactions are rare but serious adverse drug reactions unrelated to the known therapeutic properties of the drug and manifest in only a small percentage of the treated population. Animal models play an important role in advancing mechanistic studies examining idiosyncratic drug reactions. However, to be useful, they must possess similarities to those seen clinically. Although mice currently represent the dominant mammalian genetic model, rats are advantageous in many areas of pharmacologic study where their physiology can be examined in greater detail and is more akin to that seen in humans. In the area of immunology, this includes autoimmune responses and susceptibility to diabetes, in which rats more accurately mimic disease states in humans compared with mice. For example, oral nevirapine treatment can induce an immune-mediated skin rash in humans and rats, but not in mice due to the absence of the sulfotransferase required to form reactive metabolites of nevirapine within the skin. Using CRISPR-mediated gene editing, we developed a modified line of transgenic rats in which a segment of IgG-like ectodomain containing the core PD-1 interaction motif containing the native ligand and therapeutic antibody domain in exon 2 was deleted. Removal of this region critical for mediating PD-1/PD-L1 interactions resulted in animals with an increased immune response resulting in liver injury when treated with amodiaquine.

Regulation of micro- and small-exon retention and other splicing processes by GRP20 for flower development.

Pre-mRNA splicing is crucial for gene expression and depends on the spliceosome and splicing factors. Plant exons have an average size of ~180 nucleotides and typically contain motifs for interactions with spliceosome and splicing factors. Micro exons (<51 nucleotides) are found widely in eukaryotes and in genes for plant development and environmental responses. However, little is known about transcript-specific regulation of splicing in plants and about the regulators for micro exon splicing. Here we report that glycine-rich protein 20 (GRP20) is an RNA-binding protein and required for splicing of ~2,100 genes including those functioning in flower development and/or environmental responses. Specifically, GRP20 is required for micro-exon retention in transcripts of floral homeotic genes; these micro exons are conserved across angiosperms. GRP20 is also important for small-exon (51-100 nucleotides) splicing. In addition, GRP20 is required for flower development. Furthermore, GRP20 binds to poly-purine motifs in micro and small exons and a spliceosome component; both RNA binding and spliceosome interaction are important for flower development and micro-exon retention. Our results provide new insights into the mechanisms of micro-exon retention in flower development.

An Assessment of Dispersion-Corrected DFT Methods for Modeling Nonbonded Interactions in Protein Kinase Inhibitor Complexes.

Accurate modeling of nonbonded interactions between protein kinases and their small molecule inhibitors is essential for structure-based drug design. Quantum chemical methods such as density functional theory (DFT) hold significant promise for quantifying the strengths of these key protein-ligand interactions. However, the accuracy of DFT methods can vary substantially depending on the choice of exchange-correlation functionals and associated basis sets. In this study, a comprehensive benchmarking of nine widely used DFT methods was carried out to identify an optimal approach for quantitative modeling of nonbonded interactions, balancing both accuracy and computational efficiency. From a database of 2139 kinase-inhibitor crystal structures, a diverse library of 49 nonbonded interaction motifs was extracted, encompassing CH-π, π-π stacking, cation-π, hydrogen bonding, and salt bridge interactions. The strengths of nonbonded interaction energies for all 49 motifs were calculated at the advanced CCSD(T)/CBS level of theory, which serve as references for a systematic benchmarking of BLYP, TPSS, B97, ωB97X, B3LYP, M062X, PW6B95, B2PLYP, and PWPB95 functionals with D3BJ dispersion correction alongside def2-SVP, def2-TZVP, and def2-QZVP basis sets. The RI, RIJK, and RIJCOSX approximations were used for selected functionals. It was found that the B3LYP/def2-TZVP and RIJK RI-B2PLYP/def2-QZVP methods delivered the best combination of accuracy and computational efficiency, making them well-suited for efficient modeling of nonbonded interactions responsible for molecular recognition of protein kinase inhibitors in their targets.

Signal-regulated Unmasking of Nuclear Localization Motif in the PAS Domain Regulates the Nuclear Translocation of PASK

The ligand-regulated PAS domains are one of the most diverse signal-integrating domains found in proteins from prokaryotes to humans. By biochemically connecting cellular processes with their environment, PAS domains facilitate an appropriate cellular response. PAS domain-containing Kinase (PASK) is an evolutionarily conserved protein kinase that plays important signaling roles in mammalian stem cells to establish stem cell fate. We have shown that the nuclear translocation of PASK is stimulated by differentiation signaling cues in muscle stem cells. However, the mechanistic basis of the regulation of PASK nucleo-cytoplasmic translocation remains unknown. Here, we show that the PAS-A domain of PASK contains a putative monopartite nuclear localization sequence (NLS) motif. This NLS is inhibited in cells through intramolecular association with a short linear motif, termed the PAS Interacting Motif (PIM), found upstream of the kinase domain. This interaction serves to retain PASK in the cytosol in the absence of signaling cues. Consistent with that, we show that metabolic inputs induce PASK nuclear import, likely by disrupting this association. We suggest that a route for such linkage may occur through the PAS-A ligand binding cavity. We show that PIM recruitment and artificial ligand binding to the PAS-A domain occur at neighboring locations that could facilitate metabolic control of the PAS-PIM interaction. Thus, the intramolecular interaction in PASK integrates metabolic signaling cues for nuclear translocation and could be targeted to control the balance between self-renewal and differentiation in stem cells.

How collectively integrated are ecological communities?

Beyond abiotic conditions, do population dynamics mostly depend on a species' direct predators, preys and conspecifics? Or can indirect feedback that ripples across the whole community be equally important? Determining where ecological communities sit on the spectrum between these two characterizations requires a metric able to capture the difference between them. Here we show that the spectral radius of a community's interaction matrix provides such a metric, thus a measure of ecological collectivity, which is accessible from imperfect knowledge of biotic interactions and related to observable signatures. This measure of collectivity integrates existing approaches to complexity, interaction structure and indirect interactions. Our work thus provides an original perspective on the question of to what degree communities are more than loose collections of species or simple interaction motifs and explains when pragmatic reductionist approaches ought to suffice or fail when applied to ecological communities.

Selective inhibitors targeting Fis1/Mid51 protein-protein interactions protect against hypoxia-induced damage in cardiomyocytes.

Cardiovascular diseases (CVDs) are the most common non-communicable diseases globally. An estimated 17.9 million people died from CVDs in 2019, representing 32% of all global deaths. Mitochondria play critical roles in cellular metabolic homeostasis, cell survival, and cell death, as well as producing most of the cell's energy. Protein-protein interactions (PPIs) have a significant role in physiological and pathological processes, and aberrant PPIs are associated with various diseases, therefore they are potential drug targets for a broad range of therapeutic areas. Due to their ability to mimic natural interaction motifs and cover relatively larger interaction region, peptides are very promising as PPI inhibitors. To expedite drug discovery, computational approaches are widely used for screening potential lead compounds. Here, we developed peptides that inhibit mitochondrial fission 1 (Fis1)/mitochondrial dynamics 51kDa (Mid51) PPI to reduce the cellular damage that can lead to various human pathologies, such as CVDs. Based on a rational design approach we developed peptide inhibitors of the Fis1/Mid51 PPI. In silico and in vitro studies were done to evaluate the biological activity and molecular interactions of the peptides. Two peptides, CVP-241 and CVP-242 were identified based on low binding energy and molecular dynamics simulations. These peptides inhibit Fis1/Mid51 PPI (-1324.9kcal mol-1) in docking calculations (CVP-241, -741.3kcal mol-1, and CVP-242, -747.4kcal mol-1), as well as in vitro experimental studies Fis1/Mid51 PPI (KD 0.054µM) Fis1/Mid51 PPI + CVP-241 (KD 3.43µM), and Fis1/Mid51 PPI + CVP-242 (KD 44.58µM). Finally, these peptides have no toxicity to H9c2 cells, and they increase cell viability in cardiomyocytes (H9c2 cells). Consequently, the identified inhibitor peptides could serve as potent molecules in basic research and as leads for therapeutic development.

GASZ self-interaction clusters mitochondria into the intermitochondrial cement for proper germ cell development.

Mitochondrial features and activities vary in a cell type- and developmental stage-dependent manner to critically impact cell function and lineage development. Particularly in male germ cells, mitochondria are uniquely clustered into intermitochondrial cement (IMC), an electron-dense granule in the cytoplasm to support proper spermatogenesis. But it remains puzzling how mitochondria assemble into such a stable structure as IMC without limiting membrane during development. Here, we showed that GASZ (germ cell-specific, ankyrin repeat, SAM and basic leucine zipper domain containing protein), a mitochondrion-localized germ cell-specific protein, self-interacted with each other to cluster mitochondria and maintain protein stability for IMC assembling. When the self-interaction of GASZ was disrupted by either deleting its critical interaction motif or using a blocking peptide, the IMC structure was destabilized, which in turn led to impaired spermatogenesis. Notably, the blocked spermatogenesis was reversible once GASZ self-interaction was recovered. Our findings thus reveal a critical mechanism by which mitochondrion-based granules are properly assembled to support germ cell development while providing an alternative strategy for developing nonhormonal male contraceptives by targeting IMC protein interactions.

Bsp1, a fungal CPI motif protein, regulates actin filament capping in endocytosis and cytokinesis.

The capping of barbed filament ends is a fundamental mechanism for actin regulation. Capping protein controls filament growth and actin turnover in cells by binding to the barbed ends of the filaments with high affinity and slow off-rate. The interaction between capping protein and actin is regulated by capping protein interaction (CPI) motif proteins. We identified a novel CPI motif protein, Bsp1, which is involved in cytokinesis and endocytosis in budding yeast. We demonstrate that Bsp1 is an actin binding protein with a high affinity for capping protein via its CPI motif. In cells, Bsp1 regulates capping protein at endocytic sites and is a major recruiter of capping protein to the cytokinetic actin ring. Lastly, we define Bsp1-related proteins as a distinct fungi-specific CPI protein group. Our results suggest that Bsp1 promotes actin filament capping by the capping protein. This study establishes Bsp1 as a new capping protein regulator and promising candidate to regulate actin networks in fungi.

The ankyrin repeat-containing protein PlANK1 mediates mycelial growth, oospore development, and virulence in Peronophythora litchii

Litchi downy blight, caused by Peronophythora litchii, is one of the most serious diseases in major litchi-producing regions worldwide. The ankyrin (ANK) repeat is one of the most common protein-protein interaction motifs found in all kingdoms of life proteins. ANK-containing proteins have been demonstrated to regulate various biological processes in animals, plants, and fungi. However, their functions in phytopathogenic oomycetes remain unknown. Here, we identified 284 non-redundant genes that encode ANK-containing proteins in P. litchii and classified them into 11 subfamilies. Among them, PlANK1 was found to be highly up-regulated in oospores and from zoospores to the infection process. Loss of PlANK1 in P. litchii resulted in impaired mycelial growth and cyst germination, accelerated zoospore encystment, and increased sensitivity to hyperosmotic stresses and Congo red. Furthermore, PlANK1 deletion mutants were defective in oospore formation and development. Inoculation assays showed that the absence of PlANK1 severely diminished the pathogen’s virulence on litchi leaf and fruit. Through transcriptome analysis and nitrogen source utilization assays, we demonstrated that PlANK1 modulates the pathogen’s nitrogen metabolism. Altogether, our findings indicate that PlANK1 is a key regulator of sexual and asexual development, and virulence in P. litchii.

TRIM26 restricts Epstein-Barr virus infection in nasopharyngeal epithelial cells through K48-linked ubiquitination of HSP-90β.

The tripartite interaction motif (TRIM) family of proteins is known for their antiviral activity through different mechanisms, such as interfering with viral components, regulating immune responses, and participating in autophagy-mediated defense pathways. In this study, we investigated the role of tripartite interaction motif 26 (TRIM26), which is encoded by a major histocompatibility complex (MHC) gene, in regulating Epstein-Barr virus (EBV) infection of nasopharyngeal epithelial cells. We found that TRIM26 expression was induced upon EBV infection and that it indirectly targeted EphA2, a crucial epithelial receptor for EBV entry. Our results showed that TRIM26 interacted with heat shock protein 90-beta (HSP-90β) and promoted its polyubiquitination, which led to its degradation via the proteasome pathway. This, in turn, affected EphA2 integrity and suppressed EBV infection. These findings suggest that TRIM26 could be a valuable target for developing therapeutic interventions against EBV infection and its associated pathogenesis.

A poly-proline II helix in YadA from Yersinia enterocolitica serotype O:9 facilitates heparin binding through electrostatic interactions.

Poly-proline II helices are secondary structure motifs frequently found in ligand-binding sites. They exhibit increased flexibility and solvent exposure compared to the strongly hydrogen-bonded α-helices or β-strands and can therefore easily be misinterpreted as completely unstructured regions with an extremely high rotational freedom. Here, we show that the adhesin YadA of Yersinia enterocolitica serotype O:9 contains a poly-proline II helix interaction motif in the N-terminal region. The motif is involved in the interaction of YadAO:9 with heparin, a host glycosaminoglycan. We show that the basic residues within the N-terminal motif of YadA are required for electrostatic interactions with the sulfate groups of heparin. Biophysical methods including CD spectroscopy, solution-state NMR and SAXS all independently support the presence of a poly-proline helix allowing YadAO:9 binding to the rigid heparin. Lastly, we show that host cells deficient in sulfation of heparin and heparan sulfate are not targeted by YadAO:9 -mediated adhesion. We speculate that the YadAO:9 -heparin interaction plays an important and highly strain-specific role in the pathogenicity of Yersinia enterocolitica serotype O:9.

UNC-82/NUAK kinase is required by myosin A, but not myosin B, to assemble and function in the thick filament arms of C. elegans striated muscle.

The mechanisms that ensure proper assembly, activity, and turnover of myosin II filaments are fundamental to a diverse range of cellular processes. In Caenorhabditis elegans striated muscle, thick filaments contain two myosins that are functionally distinct and spatially segregated. Using transgenic double mutants, we demonstrate that the ability of increased myosin A expression to restore muscle structure and movement in myosin B mutants requires UNC-82/NUAK kinase activity. Myosin B function appears unaffected in the kinase-impaired unc-82(e1220) mutant: the recessive antimorphic effects on early assembly of paramyosin and myosin A in this mutant are counteracted by increased myosin B expression and exacerbated by loss of myosin B. Using chimeric myosins and motility assays, we mapped the region of myosin A that requires UNC-82 activity to a 531-amino-acid region of the coiled-coil rod. This region includes the 264-amino-acid Region 1, which is sufficient in chimeric myosins to rescue the essential filament-initiation function of myosin A, as well as two sites that interact with myosin head domains in the Interacting Heads Motif. A specific physical interaction between myosin A and UNC-82::GFP is supported by GFP labeling of ectopic myosin A filaments but not thin filaments. We hypothesize that UNC-82 regulates assembly competence of myosin A during parallel assembly in the filament arms.

A novel RNF125 variant associated with Tenorio syndrome alters ubiquitin chain binding.

A key signalling pathway required for clearance of viruses from host cells relies on the receptor protein, retinoic acid-inducible gene I (RIG-I). The activity of RIG-I is tightly controlled, and once bound to viral dsRNA, addition of lysine 63-linked ubiquitin chains activates signalling. Meanwhile, the addition of lysine 48-linked ubiquitin chains to RIG-I is required to terminate signalling when the infection has been resolved. Really interesting new gene (RING) finger protein 125 (RNF125) is the E3 ligase responsible for addition of the ubiquitin chains that terminate signalling, with disruption of its function associated with Tenorio syndrome. Here we describe a novel RNF125 gene variant in an individual with clinical symptoms including intellectual disability, macrocephaly and congenital heart disease, consistent with Tenorio syndrome. The newly identified Tenorio syndrome-associated variant [(NM_017831.4):c.670G>C p.Glu224Gln] is the first to be found in the ubiquitin interaction motif (UIM) of RNF125. While the E3 ligase activity of this RNF125 variant is retained, it has an impaired ability to interact with lysine 63-linked ubiquitin chains. The function of the UIM in RNF125 is uncertain; however, this study suggests that the UIM binds lysine 63-linked ubiquitin chains, and that this interaction is required for the normal function of RNF125.

The Why and How of Ultrasmall Nanoparticles.

ConspectusIn this Account, we describe our research into ultrasmall nanoparticles, including their unique properties, and outline some of the new opportunities they offer. We will summarize our perspective on the current state of the field and highlight what we see as key questions that remain to be solved. First, there are several nanostructure size-scale regimes, with qualitatively distinct functional biological attributes. Broadly generalized, larger particles (e.g., larger than 300 nm) tend to be more efficiently swept away by the first line of the immune system (for example macrophages). In the "middle-sized" regime (20-300 nm), nanoparticle surfaces and shapes can be recognized by energy-dependent cellular reorganizations, then organized locally in a spatial and temporally coherent way. That energy is gated and made available by specific cellular recognition processes. The relationship between particle surface design, endogenously derived nonspecific biomolecular corona, and architectural features recognized by the cell is complex and only purposefully and very precisely designed nanoparticle architectures are able to navigate to specific targets. At sufficiently small sizes (<10 nm including the ligand shell, associated with a core diameter of a few nm at most) we enter the "quasi-molecular regime" in which the endogenous biomolecular environment exchanges so rapidly with the ultrasmall particle surface that larger scale cellular and immune recognition events are often greatly simplified. As an example, ultrasmall particles can penetrate cellular and biological barriers within tissue architectures via passive diffusion, in much the same way as small molecule drugs do. An intriguing question arises: what happens at the interface of cellular recognition and ultrasmall quasi-molecular size regimes? Succinctly put, ultrasmall conjugates can evade defense mechanisms driven by larger scale cellular nanoscale recognition, enabling them to flexibly exploit molecular interaction motifs to interact with specific targets. Numerous advances in control of architecture that take advantage of these phenomena have taken place or are underway. For instance, syntheses can now be sufficiently controlled that it is possible to make nanoparticles of a few hundreds of atoms or metalloid clusters of several tens of atoms that can be characterized by single crystal X-ray structure analysis. While the synthesis of atomically precise clusters in organic solvents presents challenges, water-based syntheses of ultrasmall nanoparticles can be upscaled and lead to well-defined particle populations. The surface of ultrasmall nanoparticles can be covalently modified with a wide variety of ligands to control the interactions of these particles with biosystems, as well as drugs and fluorophores. And, in contrast to larger particles, many advanced molecular analytical and separation tools can be applied to understand their structure. For example, NMR spectroscopy allows us to obtain a detailed image of the particle surface and the attached ligands. These are considerable advantages that allow further elaboration of the level of architectural control and characterization of the ultrasmall structures required to access novel functional regimes and outcomes. The ultrasmall nanoparticle regime has a unique status and provides a potentially very interesting direction for development.

Light‐controlled molecular tweezers capture specific amyloid oligomers

AbstractAmyloid‐β peptide (Aβ) oligomers, characteristic symptom of Alzheimer's disease (AD), have been identified as the most neurotoxic species and significant contributors to neurodegeneration in AD. However, due to their transient and heterogeneous nature, the high‐resolution structures and exact pathogenic processes of Aβ oligomers are currently unknown. Using light‐controlled molecular tweezers (LMTs), we describe a method for precisely capturing specific Aβ oligomers produced from synthetic Aβ and AD animal models. Light irradiation can activate LMTs, which are composed of two Aβ‐targeting pentapeptides (KLVFF) motifs and a rigid azobenzene (azo) derivative, to form a tweezer‐like cis configuration that preferentially binds to specific oligomers matching the space of the tweezers via multivalent interactions of KLVFF motifs with the oligomers. Surprisingly, cis‐LMTs can immobilize the captured oligomers in transgenic Caenorhabditis elegans under light irradiation. The LMTs may serve as spatiotemporally controllable molecular tools to extract specific native oligomers for the structure and function studies via reversible photoisomerization, which would improve the understanding of the toxic mechanisms of Aβ oligomers and development of oligomer‐targeted diagnosis and therapy.

SUMOylation-triggered ALIX activation modulates extracellular vesicles circTLCD4-RWDD3 to promote lymphatic metastasis of non-small cell lung cancer

Lymph node (LN) metastasis is one of the predominant metastatic routes of non-small cell lung cancer (NSCLC) and is considered as a leading cause for the unsatisfactory prognosis of patients. Although lymphangiogenesis is well-recognized as a crucial process in mediating LN metastasis, the regulatory mechanism involving lymphangiogenesis and LN metastasis in NSCLC remains unclear. In this study, we employed high-throughput sequencing to identify a novel circular RNA (circRNA), circTLCD4-RWDD3, which was significantly upregulated in extracellular vesicles (EVs) from LN metastatic NSCLC and was positively associated with deteriorated OS and DFS of patients with NSCLC from multicenter clinical cohort. Downregulating the expression of EV-packaged circTLCD4-RWDD3 inhibited lymphangiogenesis and LN metastasis of NSCLC both in vitro and in vivo. Mechanically, circTLCD4-RWDD3 physically interacted with hnRNPA2B1 and mediated the SUMO2 modification at K108 residue of hnRNPA2B1 by upregulating UBC9. Subsequently, circTLCD4-RWDD3-induced SUMOylated hnRNPA2B1 was recognized by the SUMO interaction motif (SIM) of ALIX and activated ALIX to recruit ESCRT-III, thereby facilitating the sorting of circTLCD4-RWDD3 into NSCLC cell-derived EVs. Moreover, EV-packaged circTLCD4-RWDD3 was internalized by lymphatic endothelial cells to activate the transcription of PROX1, resulting in the lymphangiogenesis and LN metastasis of NSCLC. Importantly, blocking EV-mediated transmission of circTLCD4-RWDD3 via mutating SIM in ALIX or K108 residue of hnRNPA2B1 inhibited the lymphangiogenesis and LN metastasis of NSCLC in vivo. Our findings reveal a precise mechanism underlying SUMOylated hnRNPA2B1-induced EV packaging of circTLCD4-RWDD3 in facilitating LN metastasis of NSCLC, suggesting that EV-packaged circTLCD4-RWDD3 could be a potential therapeutic target against LN metastatic NSCLC.