山西医科大学

Objective: Early diagnosis and timely treatment of eyelid tumors are of great significance for their prognosis. However, the diagnosis of eyelid tumors is quite challenging due to their special location and diverse morphological features, especially for primary hospitals and novice doctors with limited diagnostic and treatment experience. Methods: This article proposes a two-stage deep learning algorithm for intelligent diagnosis of eyelid masses, named YOLO-EyeTriNet, which is used for the automatic diagnosis of eyelid tumors. The first step is to construct the MDF-Res-YOLO detection network, which is then used to check for the presence of masses in patients' facial images. For regions with masses, precise localization and cropping are performed to obtain local images of the masses, thus eliminating the interference of noise around the eyelids. The second step is to apply EyeTriNet to automatically classify the located masses into benign tumors, malignant tumors, and inflammatory lesions, ultimately achieving intelligent three-category diagnosis of eyelid tumors. Results: In the first lesion detection stage, the accuracy reached 0.9887, the mAP50 was 0.9428 (95 % CI: 0.931-0.954), and the F1-score was 0.9043. In the second lesion classification stage, the accuracy was 0.9368 (95 % CI: 0.915-0.958), with a precision of 0.9061, an F1-score of 0.9061, and a recall of 0.9068. Statistical analysis demonstrated that the proposed model achieved statistically significant improvements in core evaluation metrics compared with comparative models (P < 0.05) Conclusion: This study provides an effective artificial intelligence-assisted solution for primary ophthalmologists to identify and diagnose eyelid masses at an early stage. The proposed method can be applied by primary ophthalmologists, ophthalmic nursing staff, and even patients for the detection and monitoring of the occurrence and progression of eyelid tumors.

The filamentous fungus Trichoderma reesei represents a highly promising chassis cell for the efficient synthesis of both protein and non-protein metabolites. Despite the availability of diverse genetic manipulation tools for T. reesei, the engineering of chassis cells frequently requires coordinated regulation of multiple genes, a process that remains laborious and time-consuming with current methods. Here, we designed a modular and inducible artificial gene cluster (AGC) by repurposing functional DNA elements from the quinic acid (QA)-responsive gene cluster in T. reesei. This AGC is capable of expressing up to six functional genes for the assembly of genetic circuits. Under culture conditions utilizing a mixed carbon source of glycerol and QA, both robust cellular vegetative growth and activation of the gene cluster are compatible. Moreover, we identified qai1 as a gene encoding a putative transcriptional activator that governs the expression of the qai gene cluster. Notably, its overexpression alone is sufficient to bypass QA-sensing induction. Leveraging this insight, we engineered a QA-independent, constitutively expressed version of the AGC. As a proof of concept, we reconstructed the biosynthetic pathways for ilicicolin H and erythritol in T. reesei using the above QA-dependent and QA-independent AGCs, successfully achieving the synthesis of ilicicolin H and enhanced erythritol production in the T. reesei chassis cells. In summary, our study establishes novel genetic manipulation tools for engineering T. reesei chassis cells, allowing for rapid, multiplexed gene expression regulation to facilitate metabolite biosynthesis.

Direct conversion of coal-derived syngas to ethanol represents a sustainable route to relieve the dependence on grain fermentation and petroleum-based ethylene hydration, yet catalyst development remains the critical challenge. In this study, binary ZnAl catalysts with varying Zn/Al molar ratios were fabricated through a fully liquid-phase method, and their structure-performance relationship for CO hydrogenation was systematically explored. The Zn/Al ratio effectively modulates catalyst phase composition, specific surface area, surface acidity, oxygen vacancy concentration, and surface chemical states. Notably, the catalyst with Zn/Al = 1.25 exhibits the optimal catalytic performance under mild conditions (280 degrees C, 4.0 MPa), achieving 4.8% CO conversion, 21.3% total alcohol selectivity (C-mol%), and ethanol proportion of 62.15% in total alcohols. This performance is distinctly superior to most reported Zn-based catalysts under milder reaction parameters. The rapid deactivation of the Zn/Al = 0.5 catalyst is ascribed to the reduction of surface hydroxyls, the weakened CO adsorption capacity, and the suppressed formation of formate and methyl intermediates. Based on these findings, a feasible reaction pathway involving CO adsorption, intermediate conversion, and C-C coupling to ethanol is proposed. This work offers a valuable reference for constructing non-Cu ZnAl catalysts with high efficiency for syngas-toethanol conversion.

Spinal cord injury (SCI) is characterized by persistent neuroinflammation, neuronal loss, and limited tissue repair. Extracellular vesicles derived from human umbilical cord mesenchymal stromal cells (hUC-MSC-EVs) represent a promising cell-free therapeutic approach, although strategies for improving their bioactivity and regulating their cargo composition remain incompletely defined. Here, a gelatin microcarrier-based three-dimensional (3D) culture system was used to generate hUC-MSC-EVs. Compared with EVs generated under conventional two-dimensional (2D) culture conditions, 3D EVs reduced lesion severity, improved functional recovery, and promoted reparative-like changes in ionized calcium-binding adaptor molecule 1(Iba1)-positive microglia/macrophages in a rat SCI model. Transcriptomic, tandem-reporter, and pharmacological analyses indicated that intact autophagic processing contributes to the reparative-like phenotypic effects of 3D EVs in microglial cells. Comparative microRNA (miRNA) profiling identified miRNA-99a-5p as one of the enriched candidates, and gain- and loss-of-function experiments supported its contribution to autophagy-related and phenotype-associated responses in recipient microglia. The 3D culture condition was accompanied by increased nuclear localization of Yes-associated protein (YAP) and transcriptional coactivator with PDZ-binding motif (TAZ), increased expression of Y-box binding protein 1 (YBX1), TEA domain transcription factor 4 (TEAD4) occupancy at selected YBX1 promoter regions, and higher EV-associated miR-99a-5p abundance. RNA immunoprecipitation (RIP) analysis further demonstrated an association between YBX1 and miR-99a-5p. Overall, these findings indicate that 3D culture is associated with EV cargo remodeling and improved therapeutic effects after SCI, while miR-99a-5p represents one functional EV-associated cargo contributing to selected microglial responses.