Low-Rank Tensor Function Representation for Multi-Dimensional Data Recovery.

Since higher-order tensors are naturally suitable for representing multi-dimensional data in real-world, e.g., color images and videos, low-rank tensor representation has become one of the emerging areas in machine learning and computer vision. However, classical low-rank tensor representations can solely represent multi-dimensional discrete data on meshgrid, which hinders their potential applicability in many scenarios beyond meshgrid. To break this barrier, we propose a low-rank tensor function representation (LRTFR) parameterized by multilayer perceptrons (MLPs), which can continuously represent data beyond meshgrid with powerful representation abilities. Specifically, the suggested tensor function, which maps an arbitrary coordinate to the corresponding value, can continuously represent data in an infinite real space. Parallel to discrete tensors, we develop two fundamental concepts for tensor functions, i.e., the tensor function rank and low-rank tensor function factorization, and utilize MLPs to paramterize factor functions of the tensor function factorization. We theoretically justify that both low-rank and smooth regularizations are harmoniously unified in LRTFR, which leads to high effectiveness and efficiency for data continuous representation. Extensive multi-dimensional data recovery applications arising from image processing (image inpainting and denoising), machine learning (hyperparameter optimization), and computer graphics (point cloud upsampling) substantiate the superiority and versatility of our method as compared with state-of-the-art methods. Especially, the experiments beyond the original meshgrid resolution (hyperparameter optimization) or even beyond meshgrid (point cloud upsampling) validate the favorable performances of our method for continuous representation.

Correction: Zhu et al. A Feedback Loop Formed by ATG7/Autophagy, FOXO3a/miR-145 and PD-L1 Regulates Stem-like Properties and Invasion in Human Bladder Cancer. Cancers 2019, 11, 349.

In the original article [...].

CoNoT: Coupled Nonlinear Transform-Based Low-Rank Tensor Representation for Multidimensional Image Completion.

Recently, the transform-based tensor nuclear norm (TNN) methods have shown promising performance and drawn increasing attention in tensor completion (TC) problems. The main idea of these methods is to exploit the low-rank structure of frontal slices of the tensor under the transform. However, the transforms in TNN methods usually treat all modes equally and do not consider the different traits of different modes (i.e., spatial and spectral/temporal modes). To address this problem, we suggest a new low-rank tensor representation based on the coupled nonlinear transform (called CoNoT) for a better low-rank approximation. Concretely, spatial and spectral/temporal transforms in the CoNoT, respectively, exploit the different traits of different modes and are coupled together to boost the implicit low-rank structure. Here, we use the convolutional neural network (CNN) as the CoNoT, which can be learned solely from an observed multidimensional image in an unsupervised manner. Based on this low-rank tensor representation, we build a new multidimensional image completion model. Moreover, we also propose an enhanced version (called Ms-CoNoT) to further exploit the spatial multiscale nature of real-world data. Extensive experiments on real-world data substantiate the superiority of the proposed models against many state-of-the-art methods both qualitatively and quantitatively.

Self-Supervised Nonlinear Transform-Based Tensor Nuclear Norm for Multi-Dimensional Image Recovery.

Recently, transform-based tensor nuclear norm (TNN) minimization methods have received increasing attention for recovering third-order tensors in multi-dimensional imaging problems. The main idea of these methods is to perform the linear transform along the third mode of third-order tensors and then minimize the nuclear norm of frontal slices of the transformed tensor. The main aim of this paper is to propose a nonlinear multilayer neural network to learn a nonlinear transform by solely using the observed tensor in a self-supervised manner. The proposed network makes use of the low-rank representation of the transformed tensor and data-fitting between the observed tensor and the reconstructed tensor to learn the nonlinear transform. Extensive experimental results on different data and different tasks including tensor completion, background subtraction, robust tensor completion, and snapshot compressive imaging demonstrate the superior performance of the proposed method over state-of-the-art methods.

Isorhapontigenin (ISO) inhibits stem cell-like properties and invasion of bladder cancer cell by attenuating CD44 expression.

Cancer stem cells (CSC) are highly associated with poor prognosis in cancer patients. Our previous studies report that isorhapontigenin (ISO) down-regulates SOX2-mediated cyclin D1 induction and stem-like cell properties in glioma stem-like cells. The present study revealed that ISO could inhibit stem cell-like phenotypes and invasivity of human bladder cancer (BC) by specific attenuation of expression of CD44 but not SOX-2, at both the protein transcription and degradation levels. On one hand, ISO inhibited cd44 mRNA expression through decreases in Sp1 direct binding to its promoter region-binding site, resulting in attenuation of its transcription. On the other hand, ISO also down-regulated USP28 expression, which in turn reduced CD44 protein stability. Further studies showed that ISO treatment induced miR-4295, which specific bound to 3'-UTR activity of usp28 mRNA and inhibited its translation and expression, while miR-4295 induction was mediated by increased Dicer protein to enhance miR-4295 maturation upon ISO treatment. Our results provide the first evidence that ISO has a profound inhibitory effect on human BC stem cell-like phenotypes and invasivity through the mechanisms distinct from those previously noted in glioma stem-like cells.

The inhibitory effect of compound ChlA-F on human bladder cancer cell invasion can be attributed to its blockage of SOX2 protein.

Sex-determining region Y-box 2 (SOX2), a well-known stemness biomarker, is highly expressed in a variety of cancers, including human highly invasive bladder cancer (BC). However, the role of SOX2 may vary in different kinds of malignancy. In the present study, we discovered that ChlA-F, a novel conformation derivative of isolate Cheliensisin A (Chel A), remarkably inhibits the invasive ability of human invasive BC cells through downregulation of SOX2 protein expression. We found that ChlA-F treatment dramatically decreases SOX2 protein expression in human high-grade invasive BC cells. Ectopic expression of SOX2 reversed ChlA-F inhibition of cell invasion ability in human bladder cancer cells, suggesting that SOX2 is a major target of ChlA-F during its inhibition of human BC invasion. Mechanistic studies revealed that ChlA-F downregulates SOX2 at both the protein degradation and protein translation levels. Further studies revealed that ChlA-F treatment induces HuR protein expression and that the increased HuR interacts with USP8 mRNA, resulting in elevation of USP8 mRNA stability and protein expression. Elevated USP8 subsequently acts as an E3 ligase to promote SOX2 ubiquitination and protein degradation. We also found that ChlA-F treatment substantially increases c-Jun phosphorylation at Ser63 and Ser73, initiating miR-200c transcription. The increased miR-200c directly binds to the 3'-UTR of SOX2 mRNA to suppress SOX2 protein translation. These results present novel mechanistic insight into understanding SOX2 inhibition upon ChlA-F treatment and provide important information for further exploration of ChlA-F as a new therapeutic compound for the treatment of highly invasive/metastatic human BC patients.

Transcriptionally elevation of miR-494 by new ChlA-F compound via a HuR/JunB axis inhibits human bladder cancer cell invasion.

Muscle invasive bladder cancer (MIBC) is characterized by a poor overall survival rate in patients. Therefore, innovation and evaluation of idea anti-cancer compounds is of importance for reducing the mortality of MIBCs. The chemotherapeutic activity of ChlA-F, a novel C8 fluoride derivative of cheliensisin A with potent anti-neoplastic properties, was barely investigated. We reported here that ChlA-F treatment significantly induced miR-494 expression and suppressed cell invasion in human MIBC cells. Our results indicated that miR-494 was downregulated in M1 metastatic BC patients in comparison to non-metastatic (M0) BC patients, and such downregulation was also well correlated with over survival rate for MIBC patients. Mechanistically, ChlA-F-induced upregulation of miR-494 was due to a HuR-mediated increase in JunB mRNA stabilization and protein expression, which led to an increase in miR-494 transcription via directly binding to the miR-494 promoter region, while the upregulated miR-494 was able to bind the 3'-UTR region of c-Myc mRNA, resulting in decreased c-Myc mRNA stability and protein expression and further reducing the transcription of c-Myc-regulated MMP-2 and ultimately inhibiting BC invasion. Our results provide the first evidence showing that miR-494 downregulation was closely associated with BC metastatic status and overall BC survival, and ChlA-F was able to reverse the level of miR-494 with a profound inhibition of human BC invasion in human invasive BC cells. Our studies also reveal that ChlA-F is a promising therapeutic compound for BCs and miR-494 could also serve as a promising therapeutic target for the treatment of MIBC patients.

A Feedback Loop Formed by ATG7/Autophagy, FOXO3a/miR-145 and PD-L1 Regulates Stem-Like Properties and Invasion in Human Bladder Cancer.

Programmed cell death protein 1 (PD-1) and its ligand PD-L1 blockade have been identified to target immune checkpoints to treat human cancers with durable clinical benefit. Several studies reveal that the response to PD-1-PD-L1 blockade might correlate with PD-L1 expression levels in tumor cells. However, the mechanistic pathways that regulate PD-L1 protein expression are not understood. Here, we reported that PD-L1 protein is regulated by ATG7-autophagy with an ATG7-initiated positive feedback loop in bladder cancer (BC). Mechanistic studies revealed that ATG7 overexpression elevates PD-L1 protein level mainly through promoting autophagy-mediated degradation of FOXO3a, thereby inhibiting its initiated miR-145 transcription. The lower expression of miR-145 increases pd-l1 mRNA stability due to the reduction of its direct binding to 3'-UTR of pd-l1 mRNA, in turn leading to increasing in pd-l1 mRNA stability and expression, and finally enhancing stem-like property and invasion of BC cells. Notably, overexpression of PD-L1 in ATG7 knockdown cells can reverse the defect of autophagy activation, FOXO3A degradation, and miR-145 transcription attenuation. Collectively, our results revealed a positive feedback loop to promoting PD-L1 expression in human BC cells. Our study uncovers a novel molecular mechanism for regulating pd-l1 mRNA stability and expression via ATG7/autophagy/FOXO3A/miR-145 axis and reveals the potential for using combination treatment with autophagy inhibitors and PD-1/PD-L1 immune checkpoint blockade to enhance therapeutic efficacy for human BCs.