Graph embedding based multi-label Zero-shot Learning.

Multi-label Zero-shot Learning (ZSL) is more reasonable and realistic than standard single-label ZSL because several objects can co-exist in a natural image in real scenarios. Intra-class feature entanglement is a significant factor influencing the alignment of visual and semantic features, resulting in the model's inability to recognize unseen samples comprehensively and completely. We observe that existing multi-label ZSL methods place a greater emphasis on attention-based refinement and decoupling of visual features, while ignoring the relationship between label semantics. Relying on label correlations to solve multi-label ZSL tasks has not been deeply studied. In this paper, we make full use of the co-occurrence relationship between category labels and build a directed weighted semantic graph based on statistics and prior knowledge, in which node features represent category semantics and weighted edges represent conditional probabilities of label co-occurrence. To guide the targeted extraction of visual features, node features and edge set weights are simultaneously updated and refined, and embedded into the visual feature extraction network from a global and local perspective. The proposed method's effectiveness was demonstrated by simulation results on two challenging multi-label ZSL benchmarks: NUS-WIDE and Open Images. In comparison to state-of-the-art models, our model achieves an absolute gain of 2.4% mAP on NUS-WIDE and 2.1% mAP on Open Images respectively.

Case report: A novel 10.8-kb deletion identified in the ß-globin gene through the long-read sequencing technology in a Chinese family with abnormal hemoglobin testing results.

Background: Thalassemia is a common inherited hemoglobin disorder caused by a deficiency of one or more globin subunits. Substitution variants and deletions in the HBB gene are the major causes of ß-thalassemia, of which large fragment deletions are rare and difficult to be detected by conventional polymerase chain reaction (PCR)-based methods. Case report: In this study, we reported a 26-year-old Han Chinese man, whose routine blood parameters were found to be abnormal. Hemoglobin testing was performed on the proband and his family members, of whom only the proband's mother had normal parameters. The comprehensive analysis of thalassemia alleles (CATSA, a long-read sequencing-based approach) was performed to identify the causative variants. We finally found a novel 10.8-kb deletion including the ß-globin (HBB) gene (Chr11:5216601-5227407, GRch38/hg38) of the proband and his father and brother, which were consistent with their hemoglobin testing results. The copy number and exact breakpoints of the deletion were confirmed by multiplex ligation-dependent probe amplification (MLPA) and gap-polymerase chain reaction (Gap-PCR) as well as Sanger sequencing, respectively. Conclusion: With this novel large deletion found in the HBB gene in China, we expand the genotype spectrum of ß-thalassemia and show the advantages of long-read sequencing (LRS) for comprehensive and precise detection of thalassemia variants.

Critical Classes and Samples Discovering for Partial Domain Adaptation.

Partial domain adaptation (PDA) attempts to learn transferable models from a large-scale labeled source domain to a small unlabeled target domain with fewer classes, which has attracted a recent surge of interest in transfer learning. Most conventional PDA approaches endeavor to design delicate source weighting schemes by leveraging target predictions to align cross-domain distributions in the shared class space. Accordingly, two crucial issues are overlooked in these methods. First, target prediction is a double-edged sword, and inaccurate predictions will result in negative transfer inevitably. Second, not all target samples have equal transferability during the adaptation; thus, "ambiguous" target data predicted with high uncertainty should be paid more attentions. In this article, we propose a critical classes and samples discovering network (CSDN) to identify the most relevant source classes and critical target samples, such that more precise cross-domain alignment in the shared label space could be enforced by co-training two diverse classifiers. Specifically, during the training process, CSDN introduces an adaptive source class weighting scheme to select the most relevant classes dynamically. Meanwhile, based on the designed target ambiguous score, CSDN emphasizes more on ambiguous target samples with larger inconsistent predictions to enable fine-grained alignment. Taking a step further, the weighting schemes in CSDN can be easily coupled with other PDA and DA methods to further boost their performance, thereby demonstrating its flexibility. Extensive experiments verify that CSDN attains excellent results compared to state of the arts on four highly competitive benchmark datasets.

Semantic Correlation Transfer for Heterogeneous Domain Adaptation.

Heterogeneous domain adaptation (HDA) is expected to achieve effective knowledge transfer from a label-rich source domain to a heterogeneous target domain with scarce labeled data. Most prior HDA methods strive to align the cross-domain feature distributions by learning domain invariant representations without considering the intrinsic semantic correlations among categories, which inevitably results in the suboptimal adaptation performance across domains. Therefore, to address this issue, we propose a novel semantic correlation transfer (SCT) method for HDA, which not only matches the marginal and conditional distributions between domains to mitigate the large domain discrepancy, but also transfers the category correlation knowledge underlying the source domain to target by maximizing the pairwise class similarity across source and target. Technically, the domainwise and classwise centroids (prototypes) are first computed and aligned according to the feature embeddings. Then, based on the derived classwise prototypes, we leverage the cosine similarity of each two classes in both domains to transfer the supervised source semantic correlation knowledge among different categories to target effectively. As a result, the feature transferability and category discriminability can be simultaneously improved during the adaptation process. Comprehensive experiments and ablation studies on standard HDA tasks, such as text-to-image, image-to-image, and text-to-text, have demonstrated the superiority of our proposed SCT against several state-of-the-art HDA methods.

Bidirectional stochastic configuration network for regression problems.

To adapt to the reality of limited computing resources of various terminal devices in industrial applications, a randomized neural network called stochastic configuration network (SCN), which can conduct effective training without GPU, was proposed. SCN uses a supervisory random mechanism to assign its input weights and hidden biases, which makes it more stable than other randomized algorithms but also leads to time-consuming model training. To alleviate this problem, we propose a novel bidirectional SCN algorithm (BSCN) in this paper, which divides the way of adding hidden nodes into two modes: forward learning and backward learning. In the forward learning mode, BSCN still uses the supervisory mechanism to configure the parameters of the newly added nodes, which is the same as SCN. In the backward learning mode, BSCN calculates the parameters at one time based on the residual error feedback of the current model. The two learning modes are performed iteratively until the prediction error of the model reaches an acceptable level or the number of hidden nodes reaches its maximum value. This semi-random learning mechanism greatly speeds up the training efficiency of the BSCN model and significantly improves the quality of the hidden nodes. Extensive experiments on ten benchmark regression problems, two real-life air pollution prediction problems, and a classical image processing problem show that BSCN can achieve faster training speed, higher stability, and better generalization ability than SCN.

Closure of pulmonary cavity of a multidrug-resistant tuberculosis patient with catheter insertion - A case report.

The treatment of multidrug-resistant tuberculosis (MDR-TB) relies heavily on optimal chemotherapy, but interventional therapies can be adopted as adjuvant treatment to speed up illness control and increase the cure rate. We present a case of a 31-year-old MDR-TB male patient with a massive pulmonary cavity in the right lower lung cured by chemotherapy with a catheter inserted in the cavity as adjuvant treatment. This case illustrated that early interventional therapy increases the treatment success rate for pulmonary MDR-TB patients with empyema and massive cavity without the need of major invasive surgery and consequently preserve lung functions.

Receptor for Activated C Kinase 1 (RACK1) Promotes Dishevelled Protein Degradation via Autophagy and Antagonizes Wnt Signaling.

Wnt signaling plays a critical role in embryonic development, tissue homeostasis, and cancer development. Dishevelled (Dvl) is an essential and central component in Wnt signaling, and its stability and activity is tightly regulated. It has been shown that Dvl can be degraded via both the proteasome and autophagy-lysosome pathways. Here we report that receptor for activated C kinase 1 (RACK1) negatively regulates Dishevelled stability and Wnt signaling. RACK1 interacts with Dvl proteins and promotes their lysosomal degradation, and this effect is enhanced by autophagy induction. RACK1 also interacts with LC3 and enhances the association of LC3 with Dvl2, thereby leading to degradation of Dvl proteins through autophagy. These findings reveal a novel regulatory function of RACK1 in Wnt signaling by modulating Dvl stability.

Anchoring Effects of Surface Chemistry on Gold Nanorods: Modulates Autophagy.

Gold nanorods (Au NRs) have been receiving extensive attention owing to their extremely attractive properties which make them suitable for various biomedical applications. Au NRs could induce nano-toxicity, but this trouble could turn into therapeutic potential through tuning the autophagy. However, the autophagy-inducing activity and mechanism of Au NRs is still unclear. Here we showed that surface chemical modification can tune the autophagy-inducing activity of Au NRs in human lung adenocarcinoma A549 cells. CTAB-coated Au NRs induce remarkable levels of autophagy activity as evidenced by LC3-II conversion and p62 degradation, while PSS- and PDDAC-coated Au NRs barely induce autophagy. More importantly, we also demonstrated that the AKT-mTOR signaling pathway was responsible for CTAB-coated Au NRs-induced autophagy. We furthermore showed that CTAB-coated Au NRs also induces autophagy in human fetal lung fibroblast MRC-5 cells in a time-dependent manner. This study unveils a previously unknown function for Au NRs in autophagy induction, and provides a new insight for designing surface modifications of Au NRs for biomedical applications.

The Wnt Signaling Antagonist Dapper1 Accelerates Dishevelled2 Degradation via Promoting Its Ubiquitination and Aggregate-induced Autophagy.

Autophagy is a regulated process that sequesters and transports cytoplasmic materials such as protein aggregates via autophagosomes to lysosomes for degradation. Dapper1 (Dpr1), an interacting protein of Dishevelled (Dvl), antagonizes Wnt signaling by promoting Dishevelled degradation via lysosomes. However, the mechanism is unclear. Here, we show that Dpr1 promotes the von Hippel-Lindau tumor suppressor (VHL)-mediated ubiquitination of Dvl2 and its autophagic degradation. Knockdown of Dpr1 decreases the interaction between Dvl2 and pVHL, resulting in reduced ubiquitination of Dvl2. Dpr1-mediated autophagic degradation of Dvl2 depends on Dvl2 aggregation. Moreover, the aggregate-prone proteins Dvl2, p62, and the huntingtin mutant Htt103Q promote autophagy in a Dpr1-dependent manner. These protein aggregates enhance the Beclin1-Vps34 interaction and Atg14L puncta formation, indicating that aggregated proteins stimulate autophagy initiation. Ubiquitination is not essential for the aggregate-induced autophagy initiation as inhibition of the ubiquitin-activation E1 enzyme activity did not block the aggregate-induced Atg14L puncta formation. Our findings suggest that Dpr1 promotes the ubiquitination of Dvl2 by pVHL and mediates the protein aggregate-elicited autophagy initiation.

pH-responsive biocompatible fluorescent polymer nanoparticles based on phenylboronic acid for intracellular imaging and drug delivery.

To address current medical challenges, there is an urgent need to develop drug delivery systems with multiple functions, such as simultaneous stimuli-responsive drug release and real-time imaging. Biocompatible polymers have great potential for constructing smart multifunctional drug-delivery systems through grafting with other functional ligands. More importantly, novel biocompatible polymers with intrinsic fluorescence emission can work as theranostic nanomedicines for real-time imaging and drug delivery. Herein, we developed a highly fluorescent nanoparticle based on a phenylboronic acid-modified poly(lactic acid)-poly(ethyleneimine)(PLA-PEI) copolymer loaded with doxorubicin (Dox) for intracellular imaging and pH-responsive drug delivery. The nanoparticles exhibited superior fluorescence properties, such as fluorescence stability, no blinking and excitation-dependent fluorescence behavior. The Dox-loaded fluorescent nanoparticles showed pH-responsive drug release and were more effective in suppressing the proliferation of MCF-7 cells. In addition, the biocompatible fluorescent nanoparticles could be used as a tool for intracellular imaging and drug delivery, and the process of endosomal escape was traced by real-time imaging. These pH-responsive and biocompatible fluorescent polymer nanoparticles, based on phenylboronic acid, are promising tools for intracellular imaging and drug delivery.

Dapper1 promotes autophagy by enhancing the Beclin1-Vps34-Atg14L complex formation.

Autophagy is an intracellular degradation process to clear up aggregated proteins or aged and damaged organelles. The Beclin1-Vps34-Atg14L complex is essential for autophagosome formation. However, how the complex formation is regulated is unclear. Here, we show that Dapper1 (Dpr1) acts as a critical regulator of the Beclin1-Vps34-Atg14L complex to promote autophagy. Dpr1 ablation in the central nervous system results in motor coordination defect and accumulation of p62 and ubiquitinated proteins. Dpr1 increases autophagosome formation as indicated by elevated puncta formation of LC3, Atg14L and DFCP1 (Double FYVE-containing protein 1). Conversely, loss of Dpr1 impairs LC3 lipidation and causes p62/SQSTM1 accumulation. Dpr1 directly interacts with Beclin1 and Atg14L and enhances the Beclin1-Vps34 interaction and Vps34 activity. Together, our findings suggest that Dpr1 enhances the Atg14L-Beclin1-Vps34 complex formation to drive autophagy.

A fluorescent probe with restricted intramolecular rotation-induced emission for label-free detection of mercury ions.

Formation of T-Hg(2+)-T complexes changes the configuration of a single-stranded DNA, leading to enhanced fluorescence of an anchored cyanine-based probe that displays restricted intramolecular rotation (RIR)-induced emission. This label-free system can be used as a sensor for mercury ions with a detection limit of 4 nM.

Highly Sensitive Simultaneous Detection of Mercury and Copper Ions by Ultrasmall Fluorescent DNA-Ag Nanoclusters.

Fluorescent metal nanoclusters (NCs) have given rise to a new class of fluorescent nanomaterials for the detection of heavy metals. Here, we design a simple, rapid and highly sensitive sensing nanosystem for the detection of Hg2+ and Cu2+ based on fluorescence quenching of ultrasmall DNA-Ag NCs. The fluorescence intensity of DNA-Ag NCs was selectively quenched by Hg2+ and Cu2+, and the limit of detection (LOD) was found to be 5 nM and 10 nM, respectively. The technique was renewable employment by EDTA addition and successfully applied to detection of Hg2+ and Cu2+ in domestic water samples. The quantum yield (QY) of DNA-Ag NCs was significantly higher to ~30% compared to traditional water-soluble fluorescent metal NCs. The DNA-Ag NC detection system make it potentially suitable for detecting Hg2+ and Cu2+ and monitoring water quality in a wide range of samples regulated under the Environmental Protection Agency.

Nanostructural systems developed with positive charge generation to drug delivery.

The surface charge of a nanostructure plays a critical role in modulating blood circulation time, nanostructure-cell interaction, and intracellular events. It is unfavorable to have positive charges on the nanostructure surface before arriving at the disease site because positively charged nanostructures interact strongly with blood components, resulting in rapid clearance from the blood, and suboptimal targeted accumulation at the tumor site. Once at the tumor site, however, the positive charge on the nanostructure surface accelerates uptake by tumor cells and promotes the release of payloads from the lysosomes to the cytosol or nucleus inside cells. Thus, the ideal nanocarrier systems for drug delivery would maintain a neutral or negatively charged surface during blood circulation but would then generate a positive surface charge after accumulation at the tumor site or inside the cancer cells. This Progress Report focuses on the design and application of various neutral or negatively charged nanostructures that can generate a positive charge in response to the tumor microenvironment or an external stimulus.

Induction of apoptosis through ER stress and TP53 in MCF-7 cells by the nanoparticle [Gd@C82(OH)22]n: A systems biology study.

The nanoparticle gadolinium endohedral metallofullerenol [Gd@C82(OH)22]n is a new candidate for cancer treatment with low toxicity. However, its anti-cancer mechanisms remain mostly unknown. In this study, we took a systems biology view of the gene expression profiles of human breast cancer cells (MCF-7) and human umbilical vein endothelial cells (ECV304) treated with and without [Gd@C82(OH)22]n, respectively, measured by the Agilent Gene Chip G4112F. To properly analyze these data, we modified a suit of statistical methods we developed. For the first time we applied the sub-sub normalization to Agilent two-color microarrays. Instead of a simple linear regression, we proposed to use a one-knot SPLINE model in the sub-sub normalization to account for nonlinear spatial effects. The parameters estimated by least trimmed squares- and S-estimators show similar normalization results. We made several kinds of inferences by integrating the expression profiles with the bioinformatic knowledge in KEGG pathways, Gene Ontology, JASPAR, and TRANSFAC. In the transcriptional inference, we proposed the BASE2.0 method to infer a transcription factor's up-regulation and down-regulation activities separately. Overall, [Gd@C82(OH)22]n induces more differentiation in MCF-7 cells than in ECV304 cells, particularly in the reduction of protein processing such as protein glucosylation, folding, targeting, exporting, and transporting. Among the KEGG pathways, the ErbB signaling pathway is up-regulated, whereas protein processing in endoplasmic reticulum (ER) is down-regulated. CHOP, a key pro-apoptotic gene downstream of the ER stress pathway, increases to nine folds in MCF-7 cells after treatment. These findings indicate that ER stress may be one important factor that induces apoptosis in MCF-7 cells after [Gd@C82(OH)22]n treatment. The expression profiles of genes associated with ER stress and apoptosis are statistically consistent with other profiles reported in the literature, such as those of HEK293T and MCF-7 cells induced by the miR-23a∼27a∼24-2 cluster. Furthermore, one of the inferred regulatory mechanisms comprises the apoptosis network centered around TP53, whose effective regulation of apoptosis is somehow reestablished after [Gd@C82(OH)22]n treatment. These results elucidate the application and development of [Gd@C82(OH)22]n and other fullerene derivates.

Enhanced endosomal/lysosomal escape by distearoyl phosphoethanolamine-polycarboxybetaine lipid for systemic delivery of siRNA.

Cationic liposome based siRNA delivery system has improved the efficiencies of siRNA. However, cationic liposomes are prone to be rapidly cleared by the reticuloendothelial system (RES). Although modification of cationic liposomes with polyethylene glycol (PEG) could prolong circulation lifetime, PEG significantly inhibits siRNA entrapment efficiency, cellular uptake and endosomal/lysosomal escape process, resulting in low gene silencing efficiency of siRNA. In this study, we report the synthesis of zwitterionic polycarboxybetaine (PCB) based distearoyl phosphoethanolamine-polycarboxybetaine (DSPE-PCB) lipid for cationic liposome modification. The DSPE-PCB20 cationic liposome/siRNA complexes (lipoplexes) show an excellent stability in serum medium. The siRNA encapsulation efficiency of DSPE-PCB20 lipoplexes could reach 92% at N/P ratio of 20/1, but only 73% for DSPE-PEG lipoplexes. The zeta potential of DSPE-PCB20 lipoplexes is 8.19±0.53mV at pH 7.4, and increases to 24.6±0.87mV when the pH value is decreased to 4.5, which promotes the endosomal/lysosomal escape of siRNA. The DSPE-PCB20 modification could enhance the silencing efficiency of siRNA by approximately 20% over the DSPE-PEG 2000 lipoplexes at the same N/P ratio in vitro. Furthermore, DSPE-PCB20 lipoplexes could efficiently mediate the down-regulation of Apolipoprotein B (ApoB) mRNA in the liver and consequently decrease the total cholesterol in the serum in vivo, suggesting therapeutic potentials for siRNA delivery in hypercholesterolemia-related diseases.

Ultrabright and multicolorful fluorescence of amphiphilic polyethyleneimine polymer dots for efficiently combined imaging and therapy.

Multifunctional nanoparticles as theranostic tools hold great potential for its unique and efficient way to visualize the process of disease treatment. However, the toxicity of conventional fluorescent labels and difficulty of functionalization limit their widespread use. Recently, a number of amino-rich polymers have demonstrated high luminescent fluorescence but rarely showed potential for in vivo imaging due to their blue fluorescence. Here, a general route has been found to construct polymer-based multifunctional nanoparticles for combined imaging and drug delivering. The weak fluorescent polyethyleneimine (PEI) has been conjugated with hydrophobic polylactide as the amphiphilic PEI for construction of nanoparticles which showed bright and multicolor fluorescence with high drug loading capacity. The paclitaxel-loaded nanoparticles showed significant therapy effect in contrast to the free paclitaxel. Meanwhile, fluorescence imaging of the nanoparticles showed accumulation around tumor. These results demonstrate a new type of polymer-based multifunctional nanoparticles for imaging-guided drug delivery.

Metallofullerene nanoparticles promote osteogenic differentiation of bone marrow stromal cells through BMP signaling pathway.

Although endohedral metallofullerenol [Gd@C(82)(OH)(22)](n) nanoparticles have anti-tumor efficiency and mostly deposit in the bones of mice, how these nanoparticles act in bone marrow stromal cells (MSCs) remains largely unknown. Herein, we observed that [Gd@C(82)(OH)(22)](n) nanoparticles facilitated the differentiation of MSCs toward osteoblasts, as evidenced by the enhancement of alkaline phosphatase (ALP) activity and mineralized nodule formation upon [Gd@C(82)(OH)(22)](n) nanoparticle treatment. Mechanistically, the effect of [Gd@C(82)(OH)(22)](n) nanoparticles on ALP activity was inhibited by the addition of noggin as an inhibitor of the BMP signaling pathway. Moreover, the in vivo results of the ovariectomized rats further indicated that [Gd@C(82)(OH)(22)](n) nanoparticles effectively improved bone density and prevented osteoporosis.

Metallofullerol nanoparticles with low toxicity inhibit tumor growth by induction of G0/G1 arrest.

AIMS: [Gd@C(82)(OH)(22)](n) is a new type of nanoparticle with potent antineoplastic activity and low toxicity compared with traditional drugs. In this study, we explored, for the first time, the effect of [Gd@C(82)(OH)(22)](n) on the cell cycle using human breast cancer MCF-7 and human umbilical vein endothelial ECV304 cell lines by flow cytometry. METHODS: Cell viability was assessed through CCK-8 assay, and MCF-7 tumor-bearing mice were examined after 2 weeks of treatment with [Gd@C(82)(OH)(22)](n). Cell cycle-related gene expression was detected by microarray and confirmed by real-time PCR and RNAi. RESULTS: Cell viability studies confirmed that [Gd@C(82)(OH)(22)](n) inhibits breast cancer effectively with very low toxicity. Flow cytometric data and microarray results reveal that [Gd@C(82)(OH)(22)](n) mediates G0/G1 arrest in both cell lines by regulating the expression of several genes, such as cyclin D2, cyclin E and CDK4, among others, in the related cell cycle. CONCLUSION: Results further demonstrated that [Gd@C(82)(OH)(22)](n) could inhibit tumor growth by inducing tumor cell and vein endothelial cell G0/G1 arrest, which may explain the low toxicity of [Gd@C(82)(OH)(22)](n).

Biomedical nanomaterials for imaging-guided cancer therapy.

To date, even though various kinds of nanomaterials have been evaluated over the years in order to develop effective cancer therapy, there is still significant challenges in the improvement of the capabilities of nano-carriers. Developing a new theranostic nanomedicine platform for imaging-guided, visualized cancer therapy is currently a promising way to enhance therapeutic efficiency and reduce side effects. Firstly, conventional imaging technologies are reviewed with their advantages and disadvantages, respectively. Then, advanced biomedical materials for multimodal imaging are illustrated in detail, including representative examples for various dual-modalities and triple-modalities. Besides conventional cancer treatment (chemotherapy, radiotherapy), current biomaterials are also summarized for novel cancer therapy based on hyperthermia, photothermal, photodynamic effects, and clinical imaging-guided surgery. In conclusion, biomedical materials for imaging-guided therapy are becoming one of the mainstream treatments for cancer in the future. It is hoped that this review might provide new impetus to understand nanotechnology and nanomaterials employed for imaging-guided cancer therapy.