Suzhou Institute of Biomedical Engineering and Technology

This study systematically evaluated the taxonomic relationship between Streptococcus taoyuanensis ST2T and Streptococcus caecimuris CLAAV18T. Comparative genomic analysis revealed a high 16S rRNA gene sequence similarity of 99.6%, with the two strains clustering closely in the 16S rRNA-based phylogenetic tree. The genetic relatedness was further validated by Multi-Locus Sequence Typing (MLST) analysis: assessments of seven conserved housekeeping genes (atpD, gapA, gyrB, GdhA, recA, dnaK, and sdhA) demonstrated complete concordance in target fragment lengths (ranging from 33 bp to 121 bp). No size polymorphisms, insertions, or deletions were detected, indicating a highly conserved core genome. At the whole-genome level, the Average Amino Acid Identity (AAI), Average Nucleotide Identity (ANI), and digital DNA-DNA hybridization (dDDH) values between the two strains were 96.8%, 95.7%, and 84.6%, respectively. These values significantly exceed the established thresholds for species delineation (AAI: 95.5%; ANI: 95%; dDDH: 70%), providing robust genomic evidence that both strains belong to the same species. Furthermore, phenotypic testing confirmed nearly identical physiological characteristics, with only minor biochemical variations. Based on the integration of phylogenetic, genomic, and phenotypic evidence, we formally propose Streptococcus taoyuanensis as a later heterotypic synonym of Streptococcus caecimuris.

Traditional knowledge distillation relies on simple MSE or KL divergence losses that fail to capture the complex distributional relationships between teacher and student model representations. We propose FlowDistill, a novel distillation framework that employs normalizing flows to model and transfer the intricate knowledge distributions from teacher to student models. Our approach introduces three key innovations: (1) Invertible Knowledge Mapping using continuous normalizing flows (CNFs) to learn bijective transformations between teacher and student representation spaces, enabling precise knowledge transfer without information loss, (2) Flow-Guided Progressive Distillation that gradually increases the complexity of knowledge transfer by learning hierarchical flow transformations from simple to complex distributions, and (3) Conditional Flow Networks that adapt knowledge transfer based on input context and task requirements. Unlike previous diffusion-based distillation methods such as DiffKD that suffer from computational overhead due to iterative denoising processes and information loss during noise addition, our flow-based approach provides exact invertible transformations with significantly reduced computational cost. Extensive experiments on ImageNet classification, COCO object detection, and Cityscapes semantic segmentation demonstrate that FlowDistill achieves superior performance with 2.1% accuracy improvement over DiffKD on ResNet-34 to ResNet-18 distillation while reducing inference time by 3.5×. Our method establishes new state-of-the-art results across multiple distillation benchmarks and provides theoretical guarantees for lossless knowledge transfer through invertible flow transformations.

Electrochemiluminescence immunoassay (ECLIA) is widely used in clinical diagnostics owing to its high sensitivity, broad dynamic range, and excellent analytical stability. However, the influence of magnetic bead deposition behavior on electrochemiluminescence (ECL) signal performance remains insufficiently characterized. In this study, a quantitative evaluation method for magnetic bead distribution uniformity on the electrode surface was established and applied to optimize fluidic parameters in an ECLIA measurement system. By combining microscopic imaging with image analysis, magnetic bead spreading behavior under different flow conditions was systematically characterized and correlated with luminescence signal intensity. Optimization of the flow rate (18.46 µL·s-1) improved bead distribution uniformity and resulted in a 26.32% increase in luminescence intensity without altering bead coverage or assay chemistry. The optimized system was further validated using thyroid-stimulating hormone (TSH) detection, showing a linear response over 0.016-120 µIU·mL-1 (R2 > 0.996) and high consistency with a commercial analyzer (R2 = 0.998) from Roche. These results demonstrate that quantitative control of magnetic bead distribution provides an effective strategy for improving ECLIA performance and offers a general optimization framework for bead-based electrochemiluminescence systems.

Adeno-associated virus (AAV), a 4.7kb single-stranded DNA virus, is widely used as a gene therapy vector, but its limited packaging capacity poses challenges for delivering large genes, always resulting in truncation during packaging. Quantifying truncation is difficult because both strands of the plasmid can be packaged from the 3' end. In this article, we aim to first produce single-polarity AAV and then explore its truncation pattern. To address this, we modified one of the ITRs and created an oversized self-complementary AAV, which functions as a single-polarity vector to some extent. Using this modified backbone, we generated a series of oversized single-polarity AAV (spAAV) vectors of varying lengths and sequences, analyzing DNA truncation patterns via quantitative PCR (qPCR). The results show that as the distance from the 3'-ITR increased, less DNA was detected. At 3000bp from the 3'-ITR, 70% of the genomic DNA remained; this dropped to 50% at 4000bp, 20% at 4500bp, and almost none beyond 5000bp. Additionally, reporter gene expression significantly decreased when the expression cassette extended to 4.5kb compared to 2.7kb under identical in vitro conditions. Our results show that DNA will be truncated far earlier before 4.5kb during the packaging of very large genomes. This study provides important insights into the truncation patterns of AAV genomes, which is crucial for optimizing AAV vector design in gene therapy.

Fibroblast growth and sterilization are two critical factors for tissue repair, particularly in infected chronic wounds. Electrical stimulation improves the efficiency of tissue repair by accelerating the migration and proliferation of fibroblasts, but preventing infection in the tissues typically requires the use of antibiotics. In this work, a self-powered nanodevice is fabricated for producing electrical stimulations for antibiotic-free disinfection. It integrates a triboelectric nanogenerator (TENG) and a PLA-Au-PPY microneedle (MN) array. The TENG generates an output voltage of up to 6 V, which promotes the proliferation of NIH-3T3 fibroblasts by about three times, demonstrating strong cell aggregation capability. The electrical stimulation also doubles the antibacterial efficiency of the PPY layer by improving the charge transfer between PPY molecules and Escherichia coli. Therefore, the designed nanodevice presents promising potential for accelerating infected tissue repair and regeneration by promoting cell viability, accelerating tissue recovery, and offering antibiotic-free sterilization.

.SignificanceIn mesoscopic imaging research in neuroscience, achieving high spatial resolution optical imaging across the entire field of view (FOV) remains critical. This directly determines whether researchers can precisely analyze the large-scale dynamic activities of neural circuits at the single-cell or even subcellular level. Consistent optical quality throughout the entire imaging FOV is essential to accurately capture the spatiotemporal patterns of neural activity across brain regions, thereby providing a powerful tool for understanding the circuit mechanisms underlying cognition, behavior, and disease at cellular and subcellular resolution in vivo.AimThis study aims to develop a technology that extends the imaging FOV in a two-photon mesoscope while enhancing the optical quality across the entire FOV in vivo. The key point is to establish a robust method that can significantly extend the FOV beyond what the micro/mesoscope objective had been originally designed for, yet maintain the original resolution specifications. As such, the value of the method also extends beyond improving just one mesoscope, which we use as a demo in this study.ApproachThis study introduces an innovative approach that combines block scanning with adaptive optical (AO) correction through a bioinspired honeycomb-based cellular multipoint adaptive technology (CMAT) to achieve mesoscopic two-photon imaging. This system enables unprecedented large-FOV, high-resolution imaging by dividing an imaging area into subregions, each pre-optimized with deformable mirror (DM) compensation while applying real-time dynamic wavefront correction during scanning. Furthermore, we have designed multiple user-defined sub-region scanning functions. Each sub-region automatically loads the aberration correction compensation values from the nearest reference point relative to its center, thereby ensuring optimal optical performance for every individual sub-region. The robustness of this technology has been systematically verified across multiple neural circuit observation scenarios using transgenic mouse models, demonstrating its capability for reliable single-cell resolution imaging across extensive brain regions.ResultsComprehensive evaluation using standard samples and transgenic mouse models demonstrated that the CMAT significantly enhances the imaging performance of the two-photon mesoscope. This technique extends the effective two-photon imaging FOV from to while markedly improving the optical quality in the peripheral regions. High resolution was maintained at (lateral) and (axial) in the central area, with edge regions achieving improved resolutions of (lateral) and (axial). Quantitative analysis confirmed that multipoint AO not only enhances image contrast and optical resolution but also substantially increases the signal-to-noise ratio (SNR) in imaging. This work delivers a pivotal technical advance for large-scale functional imaging of neural circuits.ConclusionCMAT significantly extends the effective FOV and enhances the optical quality of the two-photon mesoscope system.

HLA-B*35:293 differs from HLA-B*35:01:01:01 by two nucleotide substitutions in codons 81 and 84 in exon 2.