Graph Regulation Network for Point Cloud Segmentation.

In point cloud, some regions typically exist nodes from multiple categories, i.e., these regions have both homophilic and heterophilic nodes. However, most existing methods ignore the heterophily of edges during the aggregation of the neighborhood node features, which inevitably mixes unnecessary information of heterophilic nodes and leads to blurred boundaries of segmentation. To address this problem, we model the point cloud as a homophilic-heterophilic graph and propose a graph regulation network (GRN) to produce finer segmentation boundaries. The proposed method can adaptively adjust the propagation mechanism with the degree of neighborhood homophily. Moreover, we build a prototype feature extraction module, which is utilised to mine the homophily features of nodes from the global prototype space. Theoretically, we prove that our convolution operation can constrain the similarity of representations between nodes based on their degree of homophily. Extensive experiments on fully and weakly supervised point cloud semantic segmentation tasks demonstrate that our method achieves satisfactory performance. Especially in the case of weak supervision, that is, each sample has only 1%-10% labeled points, the proposed method has a significant improvement in segmentation performance.

Animal Models of Kidney Disease: Challenges and Perspectives.

Kidney disease is highly prevalent and affects approximately 850 million people worldwide. It is also associated with high morbidity and mortality, and current therapies are incurable and often ineffective. Animal models are indispensable for understanding the pathophysiology of various kidney diseases and for preclinically testing novel remedies. In the last two decades, rodents continue to be the most used models for imitating human kidney diseases, largely because of the increasing availability of many unique genetically modified mice. Despite many limitations and pitfalls, animal models play an essential and irreplaceable role in gaining novel insights into the mechanisms, pathologies, and therapeutic targets of kidney disease. In this review, we highlight commonly used animal models of kidney diseases by focusing on experimental AKI, CKD, and diabetic kidney disease. We briefly summarize the pathological characteristics, advantages, and drawbacks of some widely used models. Emerging animal models such as mini pig, salamander, zebrafish, and drosophila, as well as human-derived kidney organoids and kidney-on-a-chip are also discussed. Undoubtedly, careful selection and utilization of appropriate animal models is of vital importance in deciphering the mechanisms underlying nephropathies and evaluating the efficacy of new treatment options. Such studies will provide a solid foundation for future diagnosis, prevention, and treatment of human kidney diseases.

Long and Short-Range Dependency Graph Structure Learning Framework on Point Cloud.

Graph convolutional neural networks can effectively process geometric data and thus have been successfully used in point cloud data representation. However, existing graph-based methods usually adopt the K-nearest neighbor (KNN) algorithm to construct graphs, which may not be optimal for point cloud analysis tasks, owning to the solution of KNN is independent of network training. In this paper, we propose a novel graph structure learning convolutional neural network (GSLCN) for multiple point cloud analysis tasks. The fundamental concept is to propose a general graph structure learning architecture (GSL) that builds long-range and short-range dependency graphs. To learn optimal graphs that best serve to extract local features and investigate global contextual information, respectively, we integrated the GSL with the designed graph convolution operator under a unified framework. Furthermore, we design the graph structure losses with some prior knowledge to guide graph learning during network training. The main benefit is that given labels and prior knowledge are taken into account in GSLCN, providing useful supervised information to build graphs and thus facilitating the graph convolution operation for the point cloud. Experimental results on challenging benchmarks demonstrate that the proposed framework achieves excellent performance for point cloud classification, part segmentation, and semantic segmentation.

[Mechanism of heat-clearing prescriptions in alleviating type 2 diabetes mellitus:a review].

Type 2 diabetes mellitus(T2DM), a common chronic metabolic disease, is often accompanied by internal heat syndrome. Heat-clearing prescriptions are widely used to treat different heat syndromes of T2DM from the aspects of clearing stagnant heat, excess heat, damp heat, phlegm heat, and heat toxin, demonstrating remarkable effects. The mechanism of blood sugar-lowering agents has always been a hotspot of research. Recently, the basic studies of heat-clearing prescriptions from different perspectives have been increasing year by year. To clarify the mechanisms of heat-clearing prescriptions and find specific mechanisms, we systematically reviewed the basic studies of heat-clearing prescriptions commonly used for the treatment of T2DM in the past decade, intending to provide a reference for related research.

Protective effect of quercetin on lung epithelial cell bystander-effect in the conditioned medium model of lung cancer metastasis induced by radiation.

To investigate the protective effect of Quercetin (Que) on lung epithelial cells (BEAS-2B) induced bystander effect (RIBE) after heavy ion irradiation of A549 cells. A549 cells were irradiated with 2 Gy X heavy ion rays to obtain a conditioned medium. BEAS-2B was incubated with a conditioned medium or Que. CCK-8 assay was used to screen the optimal effective concentration of Que and detect cell proliferation. Cell number was measured by cell counter and apoptosis rate was measured by flow cytometry. HMGB1 and ROS levels were measured by ELISA. Western blot was used to detect the protein expression of HMGB1, TLR4, p65, Bcl-2, Bax, Caspase3 and Cleaved Caspase3. The growth and proliferation rate of BEAS-2B decreased while the apoptosis rate increased after conditioned medium stimulation, and Que intervention inhibited this effect. The expression of HMGB1 and ROS increased after conditioned medium stimulation, and this effect was inhibited by Que intervention. In addition, the conditioned medium increased the levels of proteins of HMGB1, TLR4, p65, Bax, Caspase3 and Cleaved Caspase 3, and decreased levels of Bcl-2 protein, but Que intervention decreased the levels of HMGB1, TLR4, p65, Bax, Caspase3 and Cleaved Caspase 3proteins, and increased levels of Bcl-2 protein. The RIBE of BEAS-2B induced by irradiation of A549 is associated with HMGB1TLR4/NF-κB signaling pathway in conditioned medium inducing apoptosis by activating ROS, and Que may block RIBE-induced apoptosis by regulating HMGB1/TLR4/NF-κB pathway.

[Guiqi Yiyuan Ointment protects rat left lung from bystander effect of right lung injury induced by ~(12)C~(6+) beam].

This study observed the effects of Guiqi Yiyuan Ointment(GQYY) on the left lung subjecting to bystander effect of right lung injury induced by ~(12)C~(6+) beam in rats and decipher the underlying mechanism from NOD-like receptor protein 3(NLRP3)/apoptosis-associated speck-like protein containing a CARD(ASC)/cysteinyl aspartate specific proteinase-1(caspase-1) pathway. Wistar rats were randomized into 7 groups: blank, model, inhibitor [200 mg·kg~(-1), N-acetylcysteine(NAC)], western drug [140 mg·kg~(-1) amifostine(AMI)], and high-, medium-, and low-dose(4.8, 2.4, and 1.2 g·kg~(-1), respectively) GQYY groups. The model of bystander effect damage was established by 4 Gy ~(12)C~(6+) beam irradiation of the right lung(with the other part shielded by a lead plate). The pathological changes in the lung tissue, the level of reactive oxygen species(ROS) in the lung tissue, and the levels of superoxide dismutase(SOD) and malondialdehyde(MDA) in the serum were observed and measured in each group. Furthermore, the mRNA and protein levels of NLRP3, ASC, caspase-1, and phosphorylated nuclear factor-κB p65(p-NF-κB p65)/nuclear factor-κB p65(NF-κB p65) were determined. Compared with the blank group, the model group showed thickened alveolar wall, narrowed alveolar cavity, and presence of massive red blood cells and inflammatory infiltration in the alveolar wall and alveolar cavity. In addition, the model group showed elevated ROS levels in both left and right lungs, elevated MDA level, lowered SOD level, and up-regulated mRNA and protein levels of NLRP3, ASC, caspase-1, and p-NF-κB p65/NF-κB p65. Compared with the model group, the drug administration in all the groups reduced inflammatory cell infiltration in the lung tissue. The inhibitor group and the western drug group showed enlarged alveolar cavity, thinned interstitium, and reduced inflammation. There was a small amount of alveolar wall rupture in the high-and medium-dose GQYY groups and reduced inflammatory cell infiltration in the low dose GQYY group. Compared with the model group, drug administration lowered level of ROS in the left and right lungs, lowered the MDA level, elevated the SOD level, and down-regulated the mRNA and protein levels of NLRP3, ASC, caspase-1, and p-NF-κB p65/NF-κB p65. GQYY can effectively reduce the damage caused by radiation and bystander effect, which may be associated with the ROS-mediated NLRP3 inflammasome activation.

Single-cell transcriptomic analysis highlights origin and pathological process of human endometrioid endometrial carcinoma.

Endometrial cancers are complex ecosystems composed of cells with distinct phenotypes, genotypes, and epigenetic states. Current models do not adequately reflect oncogenic origin and pathological progression in patients. Here we use single-cell RNA sequencing to profile cells from normal endometrium, atypical endometrial hyperplasia, and endometrioid endometrial cancer (EEC), which altogether represent the step-by-step development of endometrial cancer. We find that EEC originates from endometrial epithelial cells but not stromal cells, and unciliated glandular epithelium is the source of EEC. We also identify LCN2 + /SAA1/2 + cells as a featured subpopulation of endometrial tumorigenesis. Finally, the stromal niche and immune environment changes during EEC progression are described. This study elucidates the evolution of cell populations in EEC development at single-cell resolution, which would provide a direction to facilitate EEC research and diagnosis.

Modeling Human Organ Development and Diseases With Fetal Tissue-Derived Organoids.

Recent advances in human organoid technology have greatly facilitated the study of organ development and pathology. In most cases, these organoids are derived from either pluripotent stem cells or adult stem cells for the modeling of developmental events and tissue homeostasis. However, due to the lack of human fetal tissue references and research model, it is still challenging to capture early developmental changes and underlying mechanisms in human embryonic development. The establishment of fetal tissue-derived organoids in rigorous time points is necessary. Here we provide an overview of the strategies and applications of fetal tissue-derived organoids, mainly focusing on fetal organ development research, developmental defect disease modeling, and organ-organ interaction study. Discussion of the importance of fetal tissue research also highlights the prospects and challenges in this field.

GUIDE: Training Deep Graph Neural Networks via Guided Dropout Over Edges.

Graph neural networks (GNNs) have made great progress in graph-based semi-supervised learning (GSSL). However, most existing GNNs are confronted with the oversmoothing issue that limits their expressive ability. A key factor that leads to this problem is the excessive aggregation of information from other classes when updating the node representation. To alleviate this limitation, we propose an effective method called GUIded Dropout over Edges (GUIDE) for training deep GNNs. The core of the method is to reduce the influence of nodes from other classes by removing a certain number of inter-class edges. In GUIDE, we drop edges according to the edge strength, which is defined as the time an edge acts as a bridge along the shortest path between node pairs. We find that the stronger the edge strength, the more likely it is to be an inter-class edge. In this way, GUIDE can drop more inter-class edges and keep more intra-class edges. Therefore, nodes in the same community or class are more similar, whereas different classes are more separated in the embedded space. In addition, we perform some theoretical analysis of the proposed method, which explains why it is effective in alleviating the oversmoothing problem. To validate its rationality and effectiveness, we conduct experiments on six public benchmarks with different GNNs backbones. Experimental results demonstrate that GUIDE consistently outperforms state-of-the-art methods in both shallow and deep GNNs.

Cell-in-cell structure mediates in-cell killing suppressed by CD44.

Penetration of immune cells into tumor cells was believed to be immune-suppressive via cell-in-cell (CIC) mediated death of the internalized immune cells. We unexpectedly found that CIC formation largely led to the death of the host tumor cells, but not the internalized immune cells, manifesting typical features of death executed by NK cells; we named this "in-cell killing" which displays the efficacy superior to the canonical way of "kiss-killing" from outside. By profiling isogenic cells, CD44 on tumor cells was identified as a negative regulator of "in-cell killing" via inhibiting CIC formation. CD44 functions to antagonize NK cell internalization by reducing N-cadherin-mediated intercellular adhesion and by enhancing Rho GTPase-regulated cellular stiffness as well. Remarkably, antibody-mediated blockade of CD44 signaling potentiated the suppressive effects of NK cells on tumor growth associated with increased heterotypic CIC formation. Together, we identified CIC-mediated "in-cell killing" as a promising strategy for cancer immunotherapy.

Modeling Human Thyroid Development by Fetal Tissue-Derived Organoid Culture.

Euthyroidism is of profound importance for lifetime health. However, the early diagnosis or therapeutics of thyroid developmental defects has not been established, mainly due to limited understanding of human thyroid development and a lack of recapitulating research model. Herein, the authors elaborate the cell atlas and potential regulatory signaling of the evolution of heterogeneous thyrocyte population from 12 to 16 gestational weeks. Moreover, they establish a long-term culture of human fetal thyroid organoids (hFTOs) system, which retains the fetal thyroid lineages and molecular signatures, as well as the ability to generate functional human thyroid follicles post mice renal transplantation. Notably, cAMP signaling activation in hFTOs by forskolin boosts the maturation of follicle and thus thyroid hormone T4 secretion, which recapitulates the key developmental events of fetal thyroid. Employing this ex vivo system, it is found that enhanced chromatin accessibility at thyroid maturation genes (such as TPO and TG) loci permits the transcription for hormone production. This study provides the cell atlas of and an organoid model for human thyroid development, which will facilitate thyroid research and prospective medicine.

Semisupervised Laplace-Regularized Multimodality Metric Learning.

Distance metric learning, which aims at learning an appropriate metric from data automatically, plays a crucial role in the fields of pattern recognition and information retrieval. A tremendous amount of work has been devoted to metric learning in recent years, but much of the work is basically designed for training a linear and global metric with labeled samples. When data are represented with multimodal and high-dimensional features and only limited supervision information is available, these approaches are inevitably confronted with a series of critical problems: 1) naive concatenation of feature vectors can cause the curse of dimensionality in learning metrics and 2) ignorance of utilizing massive unlabeled data may lead to overfitting. To mitigate this deficiency, we develop a semisupervised Laplace-regularized multimodal metric-learning method in this work, which explores a joint formulation of multiple metrics as well as weights for learning appropriate distances: 1) it learns a global optimal distance metric on each feature space and 2) it searches the optimal combination weights of multiple features. Experimental results demonstrate both the effectiveness and efficiency of our method on retrieval and classification tasks.

Proteomics and Organoid Culture Reveal the Underlying Pathogenesis of Hashimoto's Thyroiditis.

Hashimoto's thyroiditis (HT) is an autoimmune disease, and its incidence continues to rise. Although scientists have studied this disease for many years and discovered the potential effects of various proteins in it, the specific pathogenesis is still not fully comprehended. To understand HT and translate this knowledge to clinical applications, we took the mass spectrometric analysis on thyroid tissue fine-needle puncture from HT patients and healthy people in an attempt to make a further understanding of the pathogenesis of HT. A total of 44 proteins with differential expression were identified in HT patients, and these proteins play vital roles in cell adhesion, cell metabolism, and thyroxine synthesis. Combining patient clinical trial sample information, we further compared the transient changes of gene expression regulation in HT and papillary thyroid carcinoma (PTC) samples. More importantly, we developed patient-derived HT and PTC organoids as a promising new preclinical model to verify these potential markers. Our data revealed a marked characteristic of HT organoid in upregulating chemokines that include C-C motif chemokine ligand (CCL) 2 and CCL3, which play a key role in the pathogenesis of HT. Overall, our research has enriched everyone's understanding of the pathogenesis of HT and provides a certain reference for the treatment of the disease.

p53-dependent elimination of aneuploid mitotic offspring by entosis.

Entosis was proposed to promote aneuploidy and genome instability by cell-in-cell mediated engulfment in tumor cells. We reported here, in epithelial cells, that entosis coupled with mitotic arrest functions to counteract genome instability by targeting aneuploid mitotic progenies for engulfment and elimination. We found that the formation of cell-in-cell structures associated with prolonged mitosis, which was sufficient to induce entosis. This process was controlled by the tumor suppressor p53 (wild-type) that upregulates Rnd3 expression in response to DNA damages associated with prolonged metaphase. Rnd3-compartmentalized RhoA activities accumulated during prolonged metaphase to drive cell-in-cell formation. Remarkably, this prolonged mitosis-induced entosis selectively targets non-diploid progenies for internalization, blockade of which increased aneuploidy. Thus, our work uncovered a heretofore unrecognized mechanism of mitotic surveillance for entosis, which eliminates newly born abnormal daughter cells in a p53-dependent way, implicating in the maintenance of genome integrity.

Generation of liver bipotential organoids with a small-molecule cocktail.

Understanding the mechanism of how cholangiocytes (liver ductal cells) are activated upon liver injury and specified to hepatocytes would permit liver regenerative medicine. Here we achieved long-term in vitro expansion of mouse liver organoids by modulating signaling pathways with a combination of three small-molecule compounds. CHIR-99021, blebbistatin, and forskolin together maintained the liver organoids in bipotential stage with both cholangiocyte- and hepatocyte-specific gene expression profiles and enhanced capacity for further hepatocyte differentiation. By employing a chemical approach, we demonstrated that Wnt/ß-catenin, NMII-Rac, and PKA-ERK are core signaling pathways essential and sufficient for mouse liver progenitor expansion. Moreover, the advanced small-molecule culture of bipotential organoids facilitates the ex vivo investigation of liver cell fate determination and the application of organoids in liver regenerative medicine.

Gene manipulation in liver ductal organoids by optimized recombinant adeno-associated virus vectors.

Understanding the mechanism of how liver ductal cells (cholangiocytes) differentiate into hepatocytes would permit liver-regenerative medicine. Emerging liver ductal organoids provide an ex vivo system to investigate cholangiocyte-to-hepatocyte differentiation. However, as current gene manipulation methods require organoid dissociation into single cells and have only low efficiency, it is difficult to dissect specific gene functions in these organoids. Here we developed the adeno-associated virus (AAV) vector AAV-DJ as a powerful tool to transduce mouse and human liver ductal organoids. Via AAV-DJ-mediated up- or down-regulation of target genes, we successfully manipulated cholangiocyte-to-hepatocyte differentiation. We induced differentiation by overexpressing the hepatocyte-specifying regulator hepatocyte nuclear factor 4α (HNF4α) and blocked differentiation by stimulating Notch signaling or interfering with Smad signaling. Further screening for transcriptional factors critical for cholangiocyte-to-hepatocyte differentiation identified HOP homeobox (HOPX), T-box 15 (TBX15), and transcription factor CP2-like 1 (TFCP2L1) as master regulators. We conclude that this highly efficient and convenient gene manipulation system we developed could facilitate investigation into genes involved in cell lineage transitions and enable application of engineered organoids in regenerative medicine.

Weighted Graph Embedding-Based Metric Learning for Kinship Verification.

Given a group photograph, it is interesting and useful to judge whether the characters in it share specific kinship relation, such as father-daughter, father-son, mother-daughter, or mother-son. Recently, facial image-based kinship verification has attracted wide attention in computer vision. Some metric learning algorithms have been developed for improving kinship verification. However, most of the existing algorithms ignore fusing multiple feature representations and utilizing kernel techniques. In this paper, we develop a novel weighted graph embedding-based metric learning (WGEML) framework for kinship verification. Inspired by the fact that family members usually show high similarity in facial features like eyes, noses, and mouths, despite their diversity, we jointly learn multiple metrics by constructing an intrinsic graph and two penalty graphs to characterize the intraclass compactness and interclass separability for each feature representation, respectively, so that both the consistency and complementarity among multiple features can be fully exploited. Meanwhile, combination weights are determined through a weighted graph embedding framework. Furthermore, we present a kernelized version of WGEML to tackle nonlinear problems. Experimental results demonstrate both the effectiveness and efficiency of our proposed methods.