Abstract:Lipid droplets (LDs) are dynamic organelles that are ubiquitous across most organisms, including animals, plants, protists, and microorganisms. Their core consists of neutral lipids, surrounded by a phospholipid monolayer adorned with a specific set of proteins. As critical intracellular hubs of metabolic regulation, lipid droplets play essential roles in maintaining physiological homeostasis and contributing to the progression of various pathological processes. They store neutral lipids for energy production during periods of starvation or for membrane biosynthesis, and they sequester fatty acids to mitigate lipotoxicity. Clinically, dysregulation of lipid droplet function is associated with a wide range of diseases, including metabolic dysfunction-associated steatotic liver disease (MASLD), obesity, type 2 diabetes mellitus (T2DM), neurodegenerative disorders, and cancer. Research into the biological functions of lipid droplets—as dynamic organelles and their links to multiple diseases—has emerged as a cutting-edge focus in cell biology. In recent years, significant advances have been made in understanding lipid droplet biogenesis. Researchers have developed a more refined framework that elucidates how LDs are assembled in the endoplasmic reticulum (ER). Triacylglycerols and sterol esters are synthesized between the inner and outer leaflets of the ER bilayer, and when they exceed the critical nucleation concentration (CNC), they coalesce to form neutral lipid lenses. These then bud from the ER under the coordinated action of key proteins such as Seipin, fat storage-inducing transmembrane protein 2 (FIT2), and the peroxisomal membrane protein Pex30. This budding process is driven by changes in membrane curvature and surface tension, induced by the asymmetric distribution of phospholipids. Nascent lipid droplets recruit lipid-synthesizing enzymes via ER-LD bridging structures, enabling localized lipid production and surface expansion, ultimately resulting in the formation of mature LDs. Biochemical and biophysical approaches have revealed important features of this process, underscoring the critical roles of ER membrane biophysical properties and specific phospholipids. Structural biology and proteomic studies have identified key regulators—particularly Seipin and FIT2—as central players in LD biogenesis. This review systematically summarizes recent advances in the molecular mechanisms of LD biogenesis. It delves into the processes of LD nucleation, membrane budding, and expansion in eukaryotic cells, with a special focus on how core factors such as Seipin and FIT2 dynamically regulate LD morphology. In addition, it examines the mechanisms and pathways by which class I and class II proteins are targeted to LDs, compares LD biogenesis involving different neutral lipid cores, and discusses the disease relevance of specific regulatory proteins. Finally, the review outlines critical unresolved questions in the field of LD biogenesis, offering clear directions for future research and providing a comprehensive framework for deepening our understanding of LD formation and its implications for disease intervention.

Abstract:Systemic lupus erythematosus (SLE) is an autoimmune disease of unknown etiology, primarily characterized by systemic inflammation and hyperactivation of both B and T lymphocytes. Key immunological features include increased consumption of complement components, sustained overproduction of type I interferons (IFN-I), and persistent production of a broad spectrum of autoantibodies, such as anti-dsDNA antibodies. However, the use of autoantibodies as biomarkers for the early detection of SLE is associated with a high false-positive rate, suggesting that antibody characteristics evolve during disease progression. N-glycosylation is a critical post-translational modification of antibodies that significantly influences their structure and receptor-binding properties, thereby modulating biological activities and functions. In particular, glycosylation patterns affect the antibody’s affinity for Fc gamma receptors (FcγRs), subsequently regulating various antibody-mediated immune responses. Numerous studies have investigated the impact of individual monosaccharides—such as sialic acid, fucose, and N-acetylglucosamine, which constitute N-glycans—on the immunological functions of antibodies. This review systematically summarizes the aberrant immunoglobulin G (IgG) N-glycosylation patterns observed in SLE patients, with a focus on correlations between disease progression or complications and quantitative alterations in individual glycan components. We first review how different types of N-glycosylation modifications affect the biological activity and functional properties of IgG, particularly regarding the effects of specific monosaccharides—such as sialic acid, fucose, and galactose—on FcγR binding affinity and the resulting downstream immune functions. We then summarize the differential expression of IgG N-glycans and glycosyltransferase genes between SLE patients and healthy controls, and outline the associations between glycosylation changes and SLE-related pathological responses. In response to the inconsistencies and limitations in current research, we propose potential explanations from the perspectives of study methodologies, participant characteristics, and variations in N-glycan structures, aiming to provide a constructive reference for future studies. Given the close relationship between antibody glycosylation and SLE, this review highlights the potential of IgG N-glycosylation patterns as promising biomarkers for early diagnosis and disease monitoring. In terms of therapy, we discuss how IgG glycosylation can enhance the efficacy of intravenous immunoglobulin (IVIg) treatment and introduce emerging therapeutic strategies that aim to modulate endogenous IgG N-glycans as a novel glycan-based approach for SLE management. In summary, N-glycans are essential structural components of antibodies that regulate immune responses by modulating antibody-receptor interactions. Aberrant glycosylation is closely associated with the pathogenesis of autoimmune diseases, including SLE. However, due to the structural diversity of N-glycans and the complexity of glycosylation processes, the precise roles of IgG N-glycosylation in SLE pathophysiology remain incompletely understood. Moreover, therapeutic strategies targeting IgG glycosylation are still in early development and have not yet reached clinical application. Continued progress in glycan analysis technologies and other biological tools, along with interdisciplinary collaboration, will be essential for advancing this field.

Abstract:Thrombospondin 4 (THBS4; TSP4), a crucial component of the extracellular matrix (ECM), serves as an important regulator of tissue homeostasis and various pathophysiological processes. As a member of the evolutionarily conserved thrombospondin family, THBS4 is a multidomain adhesive glycoprotein characterized by six distinct structural domains that mediate its diverse biological functions. Through dynamic interactions with various ECM components, THBS4 plays pivotal roles in cell adhesion, proliferation, inflammation regulation, and tissue remodeling, establishing it as a key modulator of microenvironmental organization. The transcription and translation of THBS4 gene, as well as the activity of the THBS4 protein, are tightly regulated by multiple signaling pathways and extracellular cues. Positive regulators of THBS4 include transforming growth factor-β (TGF-β), interferon-γ (IFNγ), granulocyte-macrophage colony-stimulating factor (GM-CSF), bone morphogenetic proteins (BMP12/13), and other regulatory factors (such as B4GALNT1, ITGA2/ITGB1, PDGFRβ, etc.), which upregulate THBS4 at the mRNA and/or protein level. Conversely, oxidized low-density lipoprotein (OXLDL) acts as a potent negative regulator of THBS4. This intricate regulatory network ensures precise spatial and temporal control of THBS4 expression in response to diverse physiological and pathological stimuli. Functionally, THBS4 acts as a critical signaling hub, influencing multiple downstream pathways essential for cellular behavior and tissue homeostasis. The best-characterized pathways include: (1) the PI3K/AKT/mTOR axis, which THBS4 modulates through both direct and indirect interactions with integrins and growth factor receptors; (2) Wnt/β-catenin signaling, where THBS4 functions as either an activator or inhibitor depending on the cellular context; (3) the suppression of DBET/TRIM69, contributing to its diverse regulatory roles. These signaling connections position THBS4 as a master regulator of cellular responses to microenvironmental changes. Substantial evidence links aberrant THBS4 expression to a range of pathological conditions, including neoplastic diseases, cardiovascular disorders, fibrotic conditions, neurodegenerative diseases, musculoskeletal disorders, and atopic dermatitis. In cancer biology, THBS4 exhibits context-dependent roles, functioning either as a tumor suppressor or promoter depending on the tumor type and microenvironment. In the cardiovascular system, THBS4 contributes to both adaptive remodeling and maladaptive fibrotic responses. Its involvement in fibrotic diseases arises from its ability to regulate ECM deposition and turnover. The diagnostic and therapeutic potential of THBS4 is particularly promising in oncology and cardiovascular medicine. As a biomarker, THBS4 expression patterns correlate significantly with disease progression and patient outcomes. Therapeutically, targeting THBS4-mediated pathways offers novel opportunities for precision medicine approaches, including anti-fibrotic therapies, modulation of the tumor microenvironment, and enhancement of tissue repair. This comprehensive review systematically explores three key aspects of THBS4research (1) The fundamental biological functions of THBS4 in ECM organization; (2) its mechanistic involvement in various disease pathologies; (3) its emerging potential as both a diagnostic biomarker and therapeutic target. By integrating recent insights from molecular studies, animal models, and clinical investigations, this review provides a framework for understanding the multifaceted roles of THBS4 in health and disease. The synthesis of current knowledge highlights critical research gaps and future directions for exploring THBS4-targeted interventions across multiple disease contexts. Given its unique position at the intersection of ECM biology and cellular signaling, THBS4 represents a promising frontier for the development of novel diagnostic tools and therapeutic strategies in precision medicine.

Abstract:Alzheimer’s disease (AD), a progressive neurodegenerative disorder and the leading cause of dementia in the elderly, is characterized by severe cognitive decline, loss of daily living abilities, and neuropsychiatric symptoms. This condition imposes a substantial burden on patients, families, and society. Despite extensive research efforts, the complex pathogenesis of AD, particularly the early mechanisms underlying cognitive dysfunction, remains incompletely understood, posing significant challenges for timely diagnosis and effective therapeutic intervention. Among the various cellular components implicated in AD, GABAergic interneurons have emerged as critical players in the pathological cascade, playing a pivotal role in maintaining neural network integrity and function in key brain regions affected by the disease. GABAergic interneurons represent a heterogeneous population of inhibitory neurons essential for sustaining neural network homeostasis. They achieve this by precisely modulating rhythmic oscillatory activity (e.g., theta and gamma oscillations), which are crucial for cognitive processes such as learning and memory. These interneurons synthesize and release the inhibitory neurotransmitter GABA, exerting potent control over excitatory pyramidal neurons through intricate local circuits. Their primary mechanism involves synaptic inhibition, thereby modulating the excitability and synchrony of neural populations. Emerging evidence highlights the significant involvement of GABAergic interneuron dysfunction in AD pathogenesis. Contrary to earlier assumptions of their resistance to the disease, specific subtypes exhibit vulnerability or altered function early in the disease process. Critically, this impairment is not merely a consequence but appears to be a key driver of network hyperexcitability, a hallmark feature of AD models and potentially a core mechanism underlying cognitive deficits. For instance, parvalbumin-positive (PV⁺) interneurons display biphasic alterations in activity. Both suppressing early hyperactivity or enhancing late activity can rescue cognitive deficits, underscoring their causal role. Somatostatin-positive (SST⁺) neurons are highly sensitive to amyloid beta protein (Aβ) dysfunction. Their functional impairment drives AD progression via a dual pathway: compensatory hyperexcitability promotes Aβ generation, while released SST-14 forms toxic oligomers with Aβ, collectively accelerating neuronal loss and amyloid deposition, forming a vicious cycle. Vasoactive intestinal peptide-positive (VIP⁺) neurons, although potentially spared in number early in the disease, exhibit altered firing properties (e.g., broader spikes, lower frequency), contributing to network dysfunction (e.g., in CA1). Furthermore, VIP release induced by 40 Hz sensory stimulation (GENUS) enhances glymphatic clearance of Aβ, demonstrating a direct link between VIP neuron function and modulation of amyloid pathology. Given their central role in network stability and their demonstrable dysfunction in AD, GABAergic interneurons represent promising therapeutic targets. Current research primarily explores three approaches: increasing interneuron numbers (e.g., improving cortical PV⁺ interneuron counts and behavior in APP/PS1 mice with the antidepressant citalopram; transplanting stem cells differentiated into functional GABAergic neurons to enhance cognition), enhancing neuronal activity (e.g., using low-dose levetiracetam or targeted activation of specific molecules to boost PV⁺ interneuron excitability, restoring neural network γ-oscillations and memory; non-invasive neuromodulation techniques like 40 Hz repetitive transcranial magnetic stimulation (rTMS), GENUS, and minimally invasive electroacupuncture to improve inhibitory regulation, promote memory, and reduce Aβ), and direct GABA system intervention (clinical and animal studies reveal reduced GABA levels in AD-affected brain regions; early GABA supplementation improves cognition in APP/PS1 mice, suggesting a therapeutic time window). Collectively, these findings establish GABAergic interneuron intervention as a foundational rationale and distinct pathway for AD therapy. In conclusion, GABAergic interneurons, particularly the PV⁺, SST⁺, and VIP⁺ subtypes, play critical and subtype-specific roles in the initiation and progression of AD pathology. Their dysfunction significantly contributes to network hyperexcitability, oscillatory deficits, and cognitive decline. Understanding the heterogeneity in their vulnerability and response mechanisms provides crucial insights into AD pathogenesis. Targeting these interneurons through pharmacological, neuromodulatory, or cellular approaches offers promising avenues for developing novel, potentially disease-modifying therapies.

Abstract:TFIIB-related factor 1 (BRF1) is an important transcription factor. It specifically regulates the transcription of RNA polymerase III-dependent genes (RNA Pol III genes). The products of these genes are some small non-coding RNAs, including transfer RNAs (tRNAs) and 5S ribosomal RNAs (5S rRNA). The transcription levels of tRNAs and 5S rRNA vary with changes in intracellular BRF1 amounts. tRNAs and 5S rRNA play a crucial role in determining protein synthesis. Studies have demonstrated that dysregulation of tRNAs and 5S rRNA is closely related to cell growth, proliferation, transformation, and even tumorigenesis. BRF1 is a key factor determining the generation of tRNAs and 5S rRNA. Increasing BRF1 expression enhances cell proliferation and transformation, promoting tumor development. In contrast, repressing BRF1 activity decreases the rates of cell proliferation and transformation, and inhibits tumor growth. High levels of BRF1 are found in the samples of patients suffering from hepatocellular carcinoma, breast cancer, gastric carcinoma, lung cancer, prostate carcinoma, and other cancers. It indicates that high levels of BRF1 are closely related to the occurrence of human cancer and may be a common landmark of tumors. But there is discrepancy in the regulatory mechanisms and signaling pathways of BRF1 overexpression in different cancers. In general, high levels of BRF1 in patients suffering from cancer show short survival period and poor prognosis. However, there is one exception, namely breast cancer. Approximate 80% of cases of breast cancer are estrogen receptor-positive (ER+) and 20% are ER-. The cases with high levels of BRF1 reveal longer survival period and better prognosis after they accepted the hormone treatment by Tamoxifen (Tam), compared to the cases with low level BRF1. It seems like a contradiction. Most of the cases with high levels of BRF1 belong to ER+ status. Tam has been used to treat ER+ cases of breast cancer after diagnosis and surgery. Thus, hormone therapy, such as Tam, is more effective on these patients. This is because, on one hand, that Tam competes with E2 (17β-estradiol) to bind to estrogen receptor α (ERα), but does not dissociate to occupy the receptors, blocking E2 binding to this receptor and inhibiting its biological effects. On other hand, Tam can inhibit the expression of BRF1, leading to a decline of intracellular BRF1 levels. Therefore, the actual levels of BRF1 are lower in the patients with ER+ breast cancer. It appears the prognosis of the high BRF1 expression cases better than that of the low BRF1 expression cases. Myocardial hypertrophy manifests magnification of cardiomyocyte volume rather than number increasing in the postnatal heart. Myocardial hypertrophy is a critical risk factor underlying cardiovascular diseases. No matter how myocardial hypertrophy occur, it will ultimately lead to myocardial dysfunction and heart failure. Hypertrophic growth of cardiomyocytes requires a large amount of protein synthesis to meet its needs of cardiomyocyte growth. Animal models and cell experiments have shown that myocardial hypertrophy stimulates a significant increase in BRF1 expression and transcription of tRNAs and 5S rRNA. Interestingly, elevated levels of BRF1 are found in the myocardium tissues of patients with myocardial hypertrophy. These studies demonstrate that BRF1 indeed plays a critical role in myocardial hypertrophy. In summary, high levels of BRF1 are found in patients suffering from different cancers and myocardial hypertrophy. It implies that BRF1 is a promising biological target of cancer and cardiomyopathy. BRF1 is expected to become a common biomarker for early diagnosis and prognostic observation of different human cancers. It is also an important biomarker for the diagnosis and treatment of cardiomyopathy. BRF1 not only holds an important position in the field of basic medical research but also has great prospects for translational medicine. In the present article, we summarize the progress on studies of BRF1 expressions in cancer and cardiomyopathy, proposes future research directions. It is a new research area. Here, we emphasize the significancy of BRF overexpression in the two huge diseases of human, cancer and cardiomyopathy to raise people"s attention to this field.

Abstract:RNA modifications constitute a crucial class of post-transcriptional chemical alterations that profoundly influence RNA stability and translational efficiency, thereby shaping cellular protein expression profiles. These diverse chemical marks are ubiquitously involved in key biological processes, including cell proliferation, differentiation, apoptosis, and metastatic potential, and they exert precise regulatory control over these functions. A major advance in the field is the recognition that RNA modifications do not act in isolation. Instead, they participate in complex, dynamic interactions—through synergistic enhancement, antagonism, competitive binding, and functional crosstalk—forming what is now termed the “RNA modification interactome” or “RNA modification interaction network.” The formation and functional operation of this interactome rely on a multilayered regulatory framework orchestrated by RNA-modifying enzymes—commonly referred to as “writers,” “erasers,” and “readers.” These enzymes exhibit hierarchical organization within signaling cascades, often functioning in upstream-downstream sequences and converging at critical regulatory nodes. Their integration is further mediated through shared regulatory elements or the assembly into multi-enzyme complexes. This intricate enzymatic network directly governs and shapes the interdependent relationships among various RNA modifications. This review systematically elucidates the molecular mechanisms underlying both direct and indirect interactions between RNA modifications. Building upon this foundation, we introduce novel quantitative assessment frameworks and predictive disease models designed to leverage these interaction patterns. Importantly, studies across multiple disease contexts have identified core downstream signaling axes driven by specific constellations of interacting RNA modifications. These findings not only deepen our understanding of how RNA modification crosstalk contributes to disease initiation and progression, but also highlight its translational potential. This potential is exemplified by the discovery of diagnostic biomarkers based on interaction signatures and the development of therapeutic strategies targeting pathogenic modification networks. Together, these insights provide a conceptual framework for understanding the dynamic and multidimensional regulatory roles of RNA modifications in cellular systems. In conclusion, the emerging concept of RNA modification crosstalk reveals the extraordinary complexity of post-transcriptional regulation and opens new research avenues. It offers critical insights into the central question of how RNA-modifying enzymes achieve substrate specificity—determining which nucleotides within specific RNA transcripts are selectively modified during defined developmental or pathological stages. Decoding these specificity determinants, shaped in large part by the modification interactome, is essential for fully understanding the biological and pathological significance of the epitranscriptome.

Abstract:Dormant tumor cells constitute a population of cancer cells that reside in a non-proliferative or low-proliferative state, typically arrested in the G0/G1 phase and exhibiting minimal mitotic activity. These cells are commonly observed across multiple cancer types, including breast, lung, and ovarian cancers, and represent a central cellular component of minimal residual disease (MRD) following surgical resection of the primary tumor. Dormant cells are closely associated with long-term clinical latency and late-stage relapse. Due to their quiescent nature, dormant cells are intrinsically resistant to conventional therapies—such as chemotherapy and radiotherapy—that preferentially target rapidly dividing cells. In addition, they display enhanced anti-apoptotic capacity and immune evasion, rendering them particularly difficult to eradicate. More critically, in response to microenvironmental changes or activation of specific signaling pathways, dormant cells can re-enter the cell cycle and initiate metastatic outgrowth or tumor recurrence. This ability to escape dormancy underscores their clinical threat and positions their effective detection and elimination as a major challenge in contemporary cancer treatment. Hypoxia, a hallmark of the solid tumor microenvironment, has been widely recognized as a potent inducer of tumor cell dormancy. However, the molecular mechanisms by which tumor cells sense and respond to hypoxic stress—initiating the transition into dormancy—remain poorly defined. In particular, the lack of a systems-level understanding of the dynamic and multifactorial regulatory landscape has impeded the identification of actionable targets and constrained the development of effective therapeutic strategies. Accumulating evidence indicates that hypoxia-induced dormancy tumor cells are accompanied by a suite of adaptive phenotypes, including cell cycle arrest, global suppression of protein synthesis, metabolic reprogramming, autophagy activation, resistance to apoptosis, immune evasion, and therapy tolerance. These changes are orchestrated by multiple converging signaling pathways—such as PI3K-AKT-mTOR, Ras-Raf-MEK-ERK, and AMPK—that together constitute a highly dynamic and interconnected regulatory network. While individual pathways have been studied in depth, most investigations remain reductionist and fail to capture the temporal progression and network-level coordination underlying dormancy transitions. Systems biology offers a powerful framework to address this complexity. By integrating high-throughput multi-omics data—such as transcriptomics and proteomics—researchers can reconstruct global regulatory networks encompassing the key signaling axes involved in dormancy regulation. These networks facilitate the identification of core regulatory modules and elucidate functional interactions among key effectors. When combined with dynamic modeling approaches—such as ordinary differential equations—these frameworks enable the simulation of temporal behaviors of critical signaling nodes, including phosphorylated AMPK (p-AMPK), phosphorylated S6 (p-S6), and the p38/ERK activity ratio, providing insights into how their dynamic changes govern transitions between proliferation and dormancy. Beyond mapping trajectories from proliferation to dormancy and from shallow to deep dormancy, such dynamic regulatory models support topological analyses to identify central hubs and molecular switches. Key factors—such as NR2F1, mTORC1, ULK1, HIF-1α, and DYRK1A—have emerged as pivotal nodes within these networks and represent promising therapeutic targets. Constructing an integrative, systems-level regulatory framework—anchored in multi-pathway coordination, omics-layer integration, and dynamic modeling—is thus essential for decoding the architecture and progression of tumor dormancy. Such a framework not only advances mechanistic understanding but also lays the foundation for precision therapies targeting dormant tumor cells during the MRD phase, addressing a critical unmet need in cancer management.

Abstract:T7 RNA polymerase (T7 RNAP) is one of the simplest known RNA polymerases. Its unique structural features make it a critical model for studying the mechanisms of RNA synthesis. This review systematically examines the static crystal structure of T7 RNAP, beginning with an in-depth examination of its characteristic “thumb”, “palm”, and “finger” domains, which form the classic “right-hand-like” architecture. By detailing these structural elements, this review establishes a foundation for understanding the overall organization of T7 RNAP. This review systematically maps the functional roles of secondary structural elements and their subdomains in transcriptional catalysis, progressively elucidating the fundamental relationships between structure and function. Further, the intrinsic flexibility of T7 RNAP and its applications in research are also discussed. Additionally, the review presents the structural diagrams of the enzyme at different stages of the transcription process, and through these diagrams, it provides a detailed description of the complete transcription process of T7 RNAP. By integrating structural dynamics and kinetics analyses, the review constructs a comprehensive framework that bridges static structure to dynamic processes. Despite its advantages, T7 RNAP has a notable limitation: it generates double-stranded RNA (dsRNA) as a byproduct. The presence of dsRNA not only compromises the purity of mRNA products but also elicits nonspecific immune responses, which pose significant challenges for biotechnological and therapeutic applications. The review provides a detailed exploration of the mechanisms underlying dsRNA formation during T7 RNAP catalysis, reviews current strategies to mitigate this issue, and highlights recent progress in the field. A key focus is the semi-rational design of T7 RNAP mutants engineered to minimize dsRNA generation and enhance catalytic performance. Beyond its role in transcription, T7 RNAP exhibits rapid development and extensive application in fields, including gene editing, biosensing, and mRNA vaccines. This review systematically examines the structure-function relationships of T7 RNAP, elucidates the mechanisms of dsRNA formation, and discusses engineering strategies to optimize its performance. It further explores the engineering optimization and functional expansion of T7 RNAP. Furthermore, this review also addresses the pressing issues that currently need resolution, discusses the major challenges in the practical application of T7 RNAP, and provides an outlook on potential future research directions. In summary, this review provides a comprehensive analysis of T7 RNAP, ranging from its structural architecture to cutting-edge applications. We systematically examine: (1) the characteristic right-hand domains (thumb, palm, fingers) that define its minimalistic structure; (2) the structure-function relationships underlying transcriptional catalysis; and (3) the dynamic transitions during the complete transcription cycle. While highlighting T7 RNAP’s versatility in gene editing, biosensing, and mRNA vaccine production, we critically address its major limitation—dsRNA byproduct formation—and evaluate engineering solutions including semi-rationally designed mutants. By synthesizing current knowledge and identifying key challenges, this work aims to provide novel insights for the development and application of T7 RNAP and to foster further thought and progress in related fields.

Abstract:Cardiovascular disease (CVD) remains one of the leading causes of mortality among adults globally, with continuously rising morbidity and mortality rates. Metabolic disorders are closely linked to various cardiovascular diseases and play a critical role in their pathogenesis and progression, involving multifaceted mechanisms such as altered substrate utilization, mitochondrial structural and functional dysfunction, and impaired ATP synthesis and transport. In recent years, the potential role of peroxisome proliferator-activated receptors (PPARs) in cardiovascular diseases has garnered significant attention, particularly peroxisome proliferator-activated receptor alpha (PPARα), which is recognized as a highly promising therapeutic target for CVD. PPARα regulates cardiovascular physiological and pathological processes through fatty acid metabolism. As a ligand-activated receptor within the nuclear hormone receptor family, PPARα is highly expressed in multiple organs, including skeletal muscle, liver, intestine, kidney, and heart, where it governs the metabolism of diverse substrates. Functioning as a key transcription factor in maintaining metabolic homeostasis and catalyzing or regulating biochemical reactions, PPARα exerts its cardioprotective effects through multiple pathways: modulating lipid metabolism, participating in cardiac energy metabolism, enhancing insulin sensitivity, suppressing inflammatory responses, improving vascular endothelial function, and inhibiting smooth muscle cell proliferation and migration. These mechanisms collectively reduce the risk of cardiovascular disease development. Thus, PPARα plays a pivotal role in various pathological processes via mechanisms such as lipid metabolism regulation, anti-inflammatory actions, and anti-apoptotic effects. PPARα is activated by binding to natural or synthetic lipophilic ligands, including endogenous fatty acids and their derivatives (e.g., linoleic acid, oleic acid, and arachidonic acid) as well as synthetic peroxisome proliferators. Upon ligand binding, PPARα activates the nuclear receptor retinoid X receptor (RXR), forming a PPARα-RXR heterodimer. This heterodimer, in conjunction with coactivators, undergoes further activation and subsequently binds to peroxisome proliferator response elements (PPREs), thereby regulating the transcription of target genes critical for lipid and glucose homeostasis. Key genes include fatty acid translocase (FAT/CD36), diacylglycerol acyltransferase (DGAT), carnitine palmitoyltransferase I (CPT1), and glucose transporter (GLUT), which are primarily involved in fatty acid uptake, storage, oxidation, and glucose utilization processes. Advancing research on PPARα as a therapeutic target for cardiovascular diseases has underscored its growing clinical significance. Currently, PPARα activators/agonists, such as fibrates (e.g., fenofibrate and bezafibrate) and thiazolidinediones, have been extensively studied in clinical trials for CVD prevention. Traditional PPARα agonists, including fenofibrate and bezafibrate, are widely used in clinical practice to treat hypertriglyceridemia and low high-density lipoprotein cholesterol (HDL-C) levels. These fibrates enhance fatty acid metabolism in the liver and skeletal muscle by activating PPARα, and their cardioprotective effects have been validated in numerous clinical studies. Recent research highlights that fibrates improve insulin resistance, regulate lipid metabolism, correct energy metabolism imbalances, and inhibit the proliferation and migration of vascular smooth muscle and endothelial cells, thereby ameliorating pathological remodeling of the cardiovascular system and reducing blood pressure. Given the substantial attention to PPARα-targeted interventions in both basic research and clinical applications, activating PPARα may serve as a key therapeutic strategy for managing cardiovascular conditions such as myocardial hypertrophy, atherosclerosis, ischemic cardiomyopathy, myocardial infarction, diabetic cardiomyopathy, and heart failure. This review comprehensively examines the regulatory roles of PPARα in cardiovascular diseases and evaluates its clinical application value, aiming to provide a theoretical foundation for further development and utilization of PPARα-related therapies in CVD treatment.

Abstract:Metaflammation is a crucial mechanism in the onset and advancement of metabolic disorders, primarily defined by the activation of immune cells and increased concentrations of pro-inflammatory substances. The function of brain-derived neurotrophic factor (BDNF) in modulating immune and metabolic processes has garnered heightened interest, as BDNF suppresses glial cell activation and orchestrates inflammatory responses in the central nervous system via its receptor tyrosine kinase receptor B (TrkB), while also diminishing local inflammation in peripheral tissues by influencing macrophage polarization. Exercise, as a non-pharmacological intervention, is extensively employed to enhance metabolic disorders. A crucial mechanism underlying its efficacy is the significant induction of BDNF expression in central (hypothalamus, hippocampus, prefrontal cortex, and brainstem) and peripheral (liver, adipose tissue, intestines, and skeletal muscle) tissues and organs. This induction subsequently regulates inflammatory responses, ameliorates metabolic conditions, and decelerates disease progression. Consequently, BDNF is considered a pivotal molecule in the motor-metabolic regulation axis. Despite prior suggestions that BDNF may have a role in the regulation of exercise-induced inflammation, systematic data remains inadequate. Since that time, the field continues to lack structured descriptions and conversations pertinent to it. As exercise physiology research has advanced, the academic community has increasingly recognized that exercise is a multifaceted activity regulated by various systems, with its effects contingent upon the interplay of elements such as type, intensity, and frequency of exercise. Consequently, it is imperative to transcend the prior study paradigm that concentrated solely on localized effects and singular mechanisms and transition towards a comprehensive understanding of the systemic advantages of exercise. A multitude of investigations has validated that exercise confers health advantages for individuals with metabolic disorders, encompassing youngsters, adolescents, middle-aged individuals, and older persons, and typically enhances health via BDNF secretion. However, exercise is a double-edged sword; the relationship between exercise and health is not linearly positive. Insufficient exercise is ineffective, while excessive exercise can be detrimental to health. Consequently, it is crucial to scientifically develop exercise prescriptions, define appropriate exercise loads, and optimize health benefits to regulate bodily metabolism. BDNF mitigates metaflammation via many pathways during exercise. Initially, BDNF suppresses pro-inflammatory factors and facilitates the production of anti-inflammatory factors by modulating bidirectional transmission between neural and immune cells, therefore diminishing the inflammatory response. Secondly, exercise stimulates the PI3K/Akt, AMPK, and other signaling pathways via BDNF, enhancing insulin sensitivity, reducing lipotoxicity, and fostering mitochondrial production, so further optimizing the body’s metabolic condition. Moreover, exercise-induced BDNF contributes to the attenuation of systemic inflammation by collaborating with several organs, enhancing hepatic antioxidant capacity, regulating immunological response, and optimizing “gut-brain” axis functionality. These processes underscore the efficacy of exercise as a non-pharmacological intervention for enhancing anti-inflammatory and metabolic health. Despite substantial experimental evidence demonstrating the efficacy of exercise in mitigating inflammation and enhancing BDNF levels, numerous limitations persist in the existing studies. Primarily, the majority of studies have concentrated on molecular biology and lack causal experimental evidence that explicitly confirms BDNF as a crucial mediator in the exercise regulation of metaflammation. Furthermore, the outcomes of current molecular investigations are inadequately applicable to clinical practice, and a definitive pathway of “exercise-BDNF-metaflammation” remains unestablished. Moreover, the existing research methodology, reliant on animal models or limited human subject samples, constrains the broad dissemination of the findings. Future research should progressively transition from investigating isolated and localized pathways to a comprehensive multilevel and multidimensional framework that incorporates systems biology and exercise physiology. Practically, there is an immediate necessity to undertake extensive, double-blind, randomized controlled longitudinal human studies utilizing multi-omics technologies (e.g., transcriptomics, proteomics, and metabolomics) to investigate the principal signaling pathways of BDNF-mediated metaflammation and to elucidate the causal relationships and molecular mechanisms involved. Establishing a more comprehensive scientific evidence system aims to furnish a robust theoretical framework and practical guidance for the mechanistic interpretation, clinical application, and pharmaceutical development of exercise in the prevention and treatment of metabolic diseases.

Abstract:Parkinson’s disease (PD), the second most common neurodegenerative disease after Alzheimer’s disease, manifests a variety of motor symptoms, such as bradykinesia, resting tremor, rigidity, postural balance disorder, and also presents non-motor symptoms, including cognitive decline, depression, constipation, and sleep disorders. Currently, treatment for PD primarily encompasses pharmacological interventions, with levodopa being the first-line therapy, and non-pharmacological approaches such as deep brain stimulation (DBS). However, both approaches exhibit therapeutic limitations, with potential adverse reactions emerging from long-term use. Levodopa is associated with dyskinesia, while DBS may lead to mental confusion, cognitive decline, and depression. Exercise, as an effective adjuvant strategy for drug treatment of PD, can significantly improve PD motor disorders. Recently, studies have found that the mechanisms of exercise improving PD motor symptoms are associated with exerkines. Exerkine refers to signalling moieties secreted in response to acute and/or chronic exercise. This review mainly summarizes the improvement of PD motor disorders by various exerkines and the underlying mechanisms. Firstly, exercise can trigger the secretion of brain-derived neurotrophic factor (BDNF) and glial cell line-derived neurotrophic factor (GDNF) in the substantia nigra (SN) and the striatum, potentially improving PD. Recent evidence has suggested that both BDNF and GDNF could improve motor symptoms of PD via restoring the number of dopaminergic neurons in the SN and striatum, increasing striatal dopamine contents, and reducing α-synuclein (α-syn) accumulation in the SN. In addition, BDNF also alleviates motor symptoms of PD by enhancing long-term potentiation and increasing the spine density of spiny projection neurons in the striatum, while GDNF by inhibiting neuroinflammation in the SN via suppressing the activation of microglia, reducing interleukin-1β (IL-1β) and tumor necrosis factor-α (TNF-α) expressions, reducing the phosphorylation of inhibitor of nuclear factor kappa Bα (IκBα), and increasing the anti-inflammatory factors IL-10 and transforming growth factor-β (TGF-β). Secondly, exercise, a main trigger for irisin secretion from skeletal muscle, can improve PD motor symptoms by stimulating the irisin/adenosine monophosphate-activated protein kinase (AMPK)/Sirtuin-1 (SIRT1) pathway. Specifically, irisin alleviates motor symptoms in PD through multiple mechanisms, including inhibiting excessive mitochondrial fission by reducing the expressions of dynamin-related protein 1 (Drp1) and mitochondrial fission protein 1 (Fis1), alleviating the apoptosis of dopaminergic neurons by increasing B-cell lymphoma 2 (Bcl-2) expression and reducing Bcl-2-associated X protein (Bax) and caspase 3 expressions, and restoring the number of dopaminergic neurons. Thirdly, new biomarkers of PD (cathepsin B and Fetuin-A) also play roles in PD development. Cathepsin B can promote the clearance of pathogenic α-syn in PD by enhancing the function of lysosomes, including strengthening the lysosomal degradation capacity, elevating the transport rate, and increasing the activity of lysosomal glucocerebrosidase (GCase). Fetuin-A has been demonstrated to improve PD by restoring the number and the morphology of Purkinje cells, which are the only efferent neurons in the cerebellar cortex and play an important role in maintaining motor coordination. This review aims to facilitate a deep understanding of the mechanism by which exercise improves PD motor symptoms and provide a theoretical basis for promotion of exercise in PD.

Abstract:The pathogenesis of neurodegenerative diseases (NDDs) is fundamentally linked to complex and profound alterations in metabolic networks within the brain, which exhibit marked spatial heterogeneity. While conventional bulk metabolomics is powerful for detecting global metabolic shifts, it inherently lacks spatial resolution. This methodological limitation hampers the ability to interrogate critical metabolic dysregulation within discrete anatomical brain regions and specific cellular microenvironments, thereby constraining a deeper understanding of the core pathological mechanisms that initiate and drive NDDs. To address this critical gap, spatial metabolomics, with mass spectrometry imaging (MSI) at its core, has emerged as a transformative approach. It uniquely overcomes the limitations of bulk methods by enabling high-resolution, simultaneous detection and precise localization of hundreds to thousands of endogenous molecules—including primary metabolites, complex lipids, neurotransmitters, neuropeptides, and essential metal ions—directly in situ from tissue sections. This powerful capability offers an unprecedented spatial perspective for investigating the intricate and heterogeneous chemical landscape of NDD pathology, opening new avenues for discovery. Accordingly, this review provides a comprehensive overview of the field, beginning with a discussion of the technical features, optimal application scenarios, and current limitations of major MSI platforms. These include the widely adopted matrix-assisted laser desorption/ionization (MALDI)-MSI, the ultra-high-resolution technique of secondary ion mass spectrometry (SIMS)-MSI, and the ambient ionization method of desorption electrospray ionization (DESI)-MSI, along with other emerging technologies. We then highlight the pivotal applications of spatial metabolomics in NDD research, particularly its role in elucidating the profound chemical heterogeneity within distinct pathological microenvironments. These applications include mapping unique molecular signatures around amyloid β-protein (Aβ) plaques, uncovering the metabolic consequences of neurofibrillary tangles composed of hyperphosphorylated tau protein, and characterizing the lipid and metabolite composition of Lewy bodies. Moreover, we examine how spatial metabolomics contributes to constructing detailed metabolic vulnerability maps across the brain, shedding light on the biochemical factors that render certain neuronal populations and anatomical regions selectively susceptible to degeneration while others remain resilient. Looking beyond current applications, we explore the immense potential of integrating spatial metabolomics with other advanced research methodologies. This includes its combination with three-dimensional brain organoid models to recapitulate disease-relevant metabolic processes, its linkage with multi-organ axis studies to investigate how systemic metabolic health influences neurodegeneration, and its convergence with single-cell and subcellular analyses to achieve unprecedented molecular resolution. In conclusion, this review not only summarizes the current state and critical role of spatial metabolomics in NDD research but also offers a forward-looking perspective on its transformative potential. We envision its continued impact in advancing our fundamental understanding of NDDs and accelerating translation into clinical practice—from the discovery of novel biomarkers for early diagnosis to the development of high-throughput drug screening platforms and the realization of precision medicine for individuals affected by these devastating disorders.

Abstract:Objective As the central hub of the classical visual pathway, the primary visual cortex not only encodes and processes visual information but also establishes dense neural circuit connections with higher-order cognitive brain regions. Numerous studies have shown that 40 Hz flicker stimulation can induce γ oscillations in the brain and significantly improve learning and cognitive impairments in patients with neurodegenerative diseases. Moreover, flickering light phenomena naturally occur in daily environments. Given that the primary visual cortex serves as the brain’s first cortical hub for receiving visual input, it is essential to comprehensively understand how non-invasive light flicker stimulation modulates its information processing mechanisms. This study systematically investigates the effects of non-invasive light flicker stimulation at different frequencies on the functional properties of neurons in the primary visual cortex of adult mice, aiming to uncover how such stimulation modulates this region and, consequently, affects overall brain function.Methods Three groups of adult mice (approximately 12 weeks old) were exposed to light flicker stimulation at frequencies of 20 Hz, 40 Hz, and 60 Hz, respectively, for a duration of two months. A control group was exposed to the same light intensity without flickering. Following the stimulation period, in vivo multi-channel electrophysiological recordings were conducted. During these recordings, anesthetized mice were presented with various types of moving sinusoidal light gratings to assess the effects of different flicker frequencies on the functional properties of neurons in the primary visual cortex.Results The experimental results demonstrate that two months of light flicker stimulation at 20 Hz, 40 Hz, and 60 Hz enhances the orientation tuning capabilities of neurons in the primary visual cortex. Specifically, 40 Hz and 60 Hz stimulation improved contrast sensitivity, whereas 20 Hz had no significant effect. Further analysis revealed that all three frequencies reduced neuronal response variability (as measured by the Fano factor), increased the signal-to-noise ratio, and decreased noise correlation (r_sc) between neurons.Conclusion Non-invasive light flicker stimulation enhances orientation tuning (e.g., orientation bias index) and contrast sensitivity (e.g., contrast threshold and C₅₀) in neurons of the primary visual cortex. This enhancement is likely due to improved information processing efficiency, characterized by reduced neuronal variability and increased signal-to-noise ratio. These findings suggest that the primary visual cortex can achieve precise and efficient information encoding in complex lighting environments by selectively adapting to different flicker frequencies and optimizing receptive field properties. This study provides new experimental evidence on how various types of light flicker influence visual perception and offers insights into the mechanisms through which specific frequencies enhance brain function.

Abstract:Objective Traditional Chinese medicine (TCM) constitutes a valuable cultural heritage and an important source of antitumor compounds. Poria (Poria cocos (Schw.) Wolf), the dried sclerotium of a polyporaceae fungus, was first documented in Shennong’s Classic of Materia Medica and has been used therapeutically and dietarily in China for millennia. Traditionally recognized for its diuretic, spleen-tonifying, and sedative properties, modern pharmacological studies confirm that Poria exhibits antioxidant, anti-inflammatory, antibacterial, and antitumor activities. Pachymic acid (PA; a triterpenoid with the chemical structure 3β-acetyloxy-16α-hydroxy-lanosta-8,24(31)-dien-21-oic acid), isolated from Poria, is a principal bioactive constituent. Emerging evidence indicates PA exerts antitumor effects through multiple mechanisms, though these remain incompletely characterized. Neuroblastoma (NB), a highly malignant pediatric extracranial solid tumor accounting for 15% of childhood cancer deaths, urgently requires safer therapeutics due to the limitations of current treatments. Although PA shows multi-mechanistic antitumor potential, its efficacy against NB remains uncharacterized. This study systematically investigated the potential molecular targets and mechanisms underlying the anti-NB effects of PA by integrating network pharmacology-based target prediction with experimental validation of multi-target interactions through molecular docking, dynamic simulations, and in vitro assays, aimed to establish a novel perspective on PA’s antitumor activity and explore its potential clinical implications for NB treatment by integrating computational predictions with biological assays.Methods This study employed network pharmacology to identify potential targets of PA in NB, followed by validation using molecular docking, molecular dynamics (MD) simulations, MM/PBSA free energy analysis, RT-qPCR and Western blot experiments. Network pharmacology analysis included target screening via TCMSP, GeneCards, DisGeNET, SwissTargetPrediction, SuperPred, and PharmMapper. Subsequently, potential targets were predicted by intersecting the results from these databases via Venn analysis. Following target prediction, topological analysis was performed to identify key targets using Cytoscape software. Molecular docking was conducted using AutoDock Vina, with the binding pocket defined based on crystal structures. MD simulations were performed for 100 ns using GROMACS, and RMSD, RMSF, SASA, and hydrogen bonding dynamics were analyzed. MM/PBSA calculations were carried out to estimate the binding free energy of each protein-ligand complex. In vitro validation included RT-qPCR and Western blot, with GAPDH used as an internal control.Results The CCK-8 assay demonstrated a concentration-dependent inhibitory effect of PA on NB cell viability. GO analysis suggested that the anti-NB activity of PA might involve cellular response to chemical stress, vesicle lumen, and protein tyrosine kinase activity. KEGG pathway enrichment analysis suggested that the anti-NB activity of PA might involve the PI3K/AKT, MAPK, and Ras signaling pathways. Molecular docking and MD simulations revealed stable binding interactions between PA and the core target proteins AKT1, EGFR, SRC, and HSP90AA1. RT-qPCR and Western blot analyses further confirmed that PA treatment significantly decreased the mRNA and protein expression of AKT1, EGFR, and SRC while increasing the HSP90AA1 mRNA and protein levels.Conclusion It was suggested that PA may exert its anti-NB effects by inhibiting AKT1, EGFR, and SRC expression, potentially modulating the PI3K/AKT signaling pathway. These findings provide crucial evidence supporting PA’s development as a therapeutic candidate for NB.

Abstract:Objective Tobacco-related diseases remain one of the leading preventable public health challenges worldwide and are among the primary causes of premature death. In recent years, accumulating evidence has supported the classification of nicotine addiction as a chronic brain disease, profoundly affecting both brain structure and function. Despite the urgency, effective diagnostic methods for smoking addiction remain lacking, posing significant challenges for early intervention and treatment. To address this issue and gain deeper insights into the neural mechanisms underlying nicotine dependence, this study proposes a novel graph neural network framework, termed TI-GNN. This model leverages functional magnetic resonance imaging (fMRI) data to identify complex and subtle abnormalities in brain connectivity patterns associated with smoking addiction.Methods The study utilizes fMRI data to construct functional connectivity matrices that represent interaction patterns among brain regions. These matrices are interpreted as graphs, where brain regions are nodes and the strength of functional connectivity between them serves as edges. The proposed TI-GNN model integrates a Transformer module to effectively capture global interactions across the entire brain network, enabling a comprehensive understanding of high-level connectivity patterns. Additionally, a spatial attention mechanism is employed to selectively focus on informative inter-regional connections while filtering out irrelevant or noisy features. This design enhances the model’s ability to learn meaningful neural representations crucial for classification tasks. A key innovation of TI-GNN lies in its built-in causal interpretation module, which aims to infer directional and potentially causal relationships among brain regions. This not only improves predictive performance but also enhances model interpretability—an essential attribute for clinical applications. The identification of causal links provides valuable insights into the neuropathological basis of addiction and contributes to the development of biologically plausible and trustworthy diagnostic tools.Results Experimental results demonstrate that the TI-GNN model achieves superior classification performance on the smoking addiction dataset, outperforming several state-of-the-art baseline models. Specifically, TI-GNN attains an accuracy of 0.91, an F1-score of 0.91, and a Matthews correlation coefficient (MCC) of 0.83, indicating strong robustness and reliability. Beyond performance metrics, TI-GNN identifies critical abnormal connectivity patterns in several brain regions implicated in addiction. Notably, it highlights dysregulations in the amygdala and the anterior cingulate cortex, consistent with prior clinical and neuroimaging findings. These regions are well known for their roles in emotional regulation, reward processing, and impulse control—functions that are frequently disrupted in nicotine dependence.Conclusion The TI-GNN framework offers a powerful and interpretable tool for the objective diagnosis of smoking addiction. By integrating advanced graph learning techniques with causal inference capabilities, the model not only achieves high diagnostic accuracy but also elucidates the neurobiological underpinnings of addiction. The identification of specific abnormal brain networks and their causal interactions deepens our understanding of addiction pathophysiology and lays the groundwork for developing targeted intervention strategies and personalized treatment approaches in the future.

Abstract:Objective The proteome of biological evidence contains rich genetic information, namely single amino acid polymorphisms (SAPs) in protein sequences. However, due to the lack of efficient and convenient analysis tools, the application of SAP in public security still faces many challenges. This paper aims to meet the application requirements of SAP analysis for forensic biological evidence’s proteome data.Methods The software is divided into three modules. First, based on a built-in database of common non-synonymous single nucleotide polymorphisms (nsSNPs) and SAPs in East Asian populations, the software integrates and annotates newly identified exonic nsSNPs as SAPs, thereby constructing a customized SAP protein sequence database. It then utilizes a pre-installed search engine—either pFind or MaxQuant—to perform analysis and output SAP typing results, identifying both reference and variant types, along with their corresponding imputed nsSNPs. Finally, SAPTyper compares the proteome-based typing results with the individual’s exome-derived nsSNP profile and outputs the comparison report.Results SAPTyper accepts proteomic DDA mass spectrometry raw data (DDA acquisition mode) and exome sequencing results of nsSNPs as input and outputs the report of SAPs result. The pFind and Maxquant search engines were used to test the proteome data of 2 hair shafts of 2 individuals, and both obtained SAP results. It was found that the results of the Maxquant search engine were slightly less than those of pFind. This result shows that SAPTyper can achieve SAP fingding function. Moreover, the pFind search engine was used to test the proteome data of 3 hair shafts from 1 European person and 1 African person in the literature. Among the sites fully matched by the literature method, sites detected by SAPTyper are also included; for semi-matching sites, that is, nsSNPs are heterozygous, both literature method and SAPTyper method had the risk of missing detection for one type of the allele. Comparing the analysis results of SAPTyper with the SAP test results reported in the literature, it was found that some imputed nsSNP sites identified by the literature method but not detected by SAPTyper had a MAF of less than 0.1% in East Asian populations, and therefore they were not included in the common nsSNP database of East Asian populations constructed by this software. Since the database construction of this software is based on the genetic variation information of East Asian populations, it is currently unable to effectively identify representative unique common variation sites in European or African populations, but it can still identify SAP sites shared by these populations and East Asian populations.Conclusion An automated SAP analysis algorithm was developed for East Asian populations, and the software named SAPTyper was developed. This software provides a convenient and efficient analysis tool for the research and application of forensic proteomic SAP and has important application prospects in individual identification and phenotypic inference based on SAP.

Abstract:Objective This study proposes a novel single-cell protein localization method based on a class perception graph convolutional network (CP-GCN) to overcome several critical challenges in protein microscopic image analysis, including the scarcity of cell-level annotations, inadequate feature extraction, and the difficulty in achieving precise protein localization within individual cells. The methodology involves multiple innovative components designed to enhance both feature extraction and localization accuracy.Methods First, a class perception module (CPM) is developed to effectively capture and distinguish semantic features across different subcellular categories, enabling more discriminative feature representation. Building upon this, the CP-GCN network is designed to explore global features of subcellular proteins in multicellular environments. This network incorporates a category feature-aware module to extract protein semantic features aligned with label dimensions and establishes a subcellular relationship mining module to model correlations between different subcellular structures. By doing so, it generates co-occurrence embedding features that encode spatial and contextual relationships among subcellular locations, thereby improving feature representation. To further refine localization, a multi-scale feature analysis approach is employed using the K-means clustering algorithm, which classifies multi-scale features within each subcellular category and generates multi-cell class activation maps (CAMs). These CAMs highlight discriminative regions associated with specific subcellular locations, facilitating more accurate protein localization. Additionally, a pseudo-label generation strategy is introduced to address the lack of annotated single-cell data. This strategy segments multicellular images into single-cell images and assigns reliable pseudo-labels based on the CAM-predicted regions, ensuring high-quality training data for single-cell analysis. Under a transfer learning framework, the model is trained to achieve precise single-cell-level protein localization, leveraging both the extracted features and pseudo-labels for robust performance.Results Experimental validation on multiple single-cell test datasets demonstrates that the proposed method significantly outperforms existing approaches in terms of robustness and localization accuracy. Specifically, on the Kaggle 2021 dataset, the method achieves superior mean average precision (mAP) metrics across 18 subcellular categories, highlighting its effectiveness in diverse protein localization tasks. Visualization of the generated CAM results further confirms the model’s capability to accurately localize subcellular proteins within individual cells, even in complex multicellular environments.Conclusion The integration of the CP-GCN network with a pseudo-labeling strategy enables the proposed method to effectively capture heterogeneous cellular features in protein images and achieve precise single-cell protein localization. This advancement not only addresses key limitations in current protein image analysis but also provides a scalable and accurate solution for subcellular protein studies, with potential applications in biomedical research and diagnostic imaging. The success of this method underscores the importance of combining advanced deep learning architectures with innovative training strategies to overcome data scarcity and improve localization performance in biological image analysis. Future work could explore the extension of this framework to other types of microscopic imaging and its application in large-scale protein interaction studies.

Abstract:Objective Transfer RNA-derived fragments (tRFs) are a recently characterized and rapidly expanding class of small non-coding RNAs, typically ranging from 13 to 50 nucleotides in length. They are derived from mature or precursor tRNA molecules through specific cleavage events and have been implicated in a wide range of cellular processes. Increasing evidence indicates that tRFs play important regulatory roles in gene expression, primarily by interacting with target messenger RNAs (mRNAs) to induce transcript degradation, in a manner partially analogous to microRNAs (miRNAs). However, despite their emerging biological relevance and potential roles in disease mechanisms, there remains a significant lack of computational tools capable of systematically predicting the interaction landscape between tRFs and their target mRNAs. Existing databases often rely on limited interaction features and lack the flexibility to accommodate novel or user-defined tRF sequences. The primary goal of this study was to develop a machine learning based prediction algorithm that enables high-throughput, accurate identification of tRF:mRNA binding events, thereby facilitating the functional analysis of tRF regulatory networks.Methods We began by assembling a manually curated dataset of 38 687 experimentally verified tRF:mRNA interaction pairs and extracting seven biologically informed features for each pair: (1) AU content of the binding site, (2) site pairing status, (3) binding region location, (4) number of binding sites per mRNA, (5) length of the longest consecutive complementary stretch, (6) total binding region length, and (7) seed sequence complementarity. Using this dataset and feature set, we trained 4 distinct machine learning classifiers—logistic regression, random forest, decision tree, and a multilayer perceptron (MLP)—to compare their ability to discriminate true interactions from non-interactions. Each model’s performance was evaluated using overall accuracy, receiver operating characteristic (ROC) curves, and the corresponding area under the ROC curve (AUC). The MLP consistently achieved the highest AUC among the four, and was therefore selected as the backbone of our prediction framework, which we named tRF Prospect. For biological validation, we retrieved 3 high-throughput RNA-seq datasets from the gene expression omnibus (GEO) in which individual tRFs were overexpressed: AS-tDR-007333 (GSE184690), tRF-3004b (GSE197091), and tRF-20-S998LO9D (GSE208381). Differential expression analysis of each dataset identified genes downregulated upon tRF overexpression, which we designated as putative targets. We then compared the predictions generated by tRF Prospect against those from three established tools—tRFTar, tRForest, and tRFTarget—by quantifying the number of predicted targets for each tRF and assessing concordance with the experimentally derived gene sets.Results The proposed algorithm achieved high predictive accuracy, with an AUC of 0.934. Functional validation was conducted using transcriptome-wide RNA-seq datasets from cells overexpressing specific tRFs, confirming the model’s ability to accurately predict biologically relevant downregulation of mRNA targets. When benchmarked against established tools such as tRFTar, tRForest, and tRFTarget, tRF Prospect consistently demonstrated superior performance, both in terms of predictive precision and sensitivity, as well as in identifying a higher number of true-positive interactions. Moreover, unlike static databases that are limited to precomputed results, tRF Prospect supports real-time prediction for any user-defined tRF sequence, enhancing its applicability in exploratory and hypothesis-driven research.Conclusion This study introduces tRF Prospect as a powerful and flexible computational tool for investigating tRF:mRNA interactions. By leveraging the predictive strength of deep learning and incorporating a broad spectrum of interaction-relevant features, it addresses key limitations of existing platforms. Specifically, tRF Prospect: (1) expands the range of detectable tRF and target types; (2) improves prediction accuracy through multilayer perceptron model; and (3) allows for dynamic, user-driven analysis beyond database constraints. Although the current version emphasizes miRNA-like repression mechanisms and faces challenges in accurately capturing 5"UTR-associated binding events, it nonetheless provides a critical foundation for future studies aiming to unravel the complex roles of tRFs in gene regulation, cellular function, and disease pathogenesis.