Search

Detail description of microarray analysis and comparison with cancer stem cell dataset, and TMA of prostate cancer patients

Supplemental Figure 1. Trop2 ICD abundance is maintained by Src kinase. Supplemental Figure 2. Canertinib, a tyrosine kinase inhibitor targeting EGFR reduces cell proliferation but does not affect Trop2 ICD abundance. Supplemental Figure 3. Schematic representation of the mechanism by which Src induction of cyclin D1 enhances cleavage of the membrane associated Trop2 complex to thereby enhance Trop2 ICD. Supplemental Figure 4. Human prostate cancer Tissue Microarray (TMA) IHC staining.

Supplemental Figure 1. Trop2 ICD abundance is maintained by Src kinase. Supplemental Figure 2. Canertinib, a tyrosine kinase inhibitor targeting EGFR reduces cell proliferation but does not affect Trop2 ICD abundance. Supplemental Figure 3. Schematic representation of the mechanism by which Src induction of cyclin D1 enhances cleavage of the membrane associated Trop2 complex to thereby enhance Trop2 ICD. Supplemental Figure 4. Human prostate cancer Tissue Microarray (TMA) IHC staining.

Proteomic analysis of castration-resistant prostate cancer demonstrated the enrichment of Src tyrosine kinase activity in approximately 90% of patients. Src is known to induce cyclin D1, and a cyclin D1–regulated gene expression module predicts poor outcome in human prostate cancer. The tumor-associated calcium signal transducer 2 (TACSTD2/Trop2/M1S1) is enriched in the prostate, promoting prostate stem cell self-renewal upon proteolytic activation via a γ-secretase cleavage complex (PS1, PS2) and TACE (ADAM17), which releases the Trop2 intracellular domain (Trop2 ICD). Herein, v-Src transformation of primary murine prostate epithelial cells increased the proportion of prostate cancer stem cells as characterized by gene expression, epitope characteristics, and prostatosphere formation. Cyclin D1 was induced by v-Src, and Src kinase induction of Trop2 ICD nuclear accumulation required cyclin D1. Cyclin D1 induced abundance of the Trop2 proteolytic cleavage activation components (PS2, TACE) and restrained expression of the inhibitory component of the Trop2 proteolytic complex (Numb). Patients with prostate cancer with increased nuclear Trop2 ICD and cyclin D1, and reduced Numb, had reduced recurrence-free survival probability (HR = 4.35). Cyclin D1, therefore, serves as a transducer of v-Src–mediated induction of Trop2 ICD by enhancing abundance of the Trop2 proteolytic activation complex. Cancer Res; 76(22); 6723–34. ©2016 AACR.

PDF file - 111K, Cyclin D1 dependent DHT regulated gene expression.

Supplemental Figure 1. Quantitation and statistical evaluation of the microarray data for Fig. 1A and Fig. 2H, I. Supplemental Figure 2. Somatic excision of Dach1 in prostate epithelial cells. Supplemental Figure 3. Transgenic deletion of Dach1 in the prostate epithelial cells enhances cellular proliferation and reduces cellular apoptosis. Supplemental Figure 4. Prostate specific expression of Dach1 enhances cyclin D1 and cyclin A1. Supplemental Figure 5. Quantitation and statistical evaluation of the microarray data for Fig. 3F and Fig. 4A, B. Supplemental Figure 6. ChIP analysis of cyclin A2 and cyclin E1 promoter regions. Supplemental Figure 7. Endogenous Dach1 is a key restraint of prostate cytokine production. Supplemental Figure 8. Time lapse microscopy of Dach1-/- vs. Dach1+/+ anterior prostate epithelial cells. Supplemental Figure 9. Transgenic deletion of Dach1 in prostate epithelial cells increases IL-6 and IL-8 expression.Supplemental Figure 10. Time lapse microscopy of Dach1+/+ or Dach1-/- ventral prostate epithelial cells. Supplemental Figure 11. The Expression of DACH1 in PC3 cells. Supplemental Figure 12. The relationship between DACH1 and IL-6, IL-8 in metastatic prostate cancer.

PDF file - 132KB, Legends for Supplementary Table 1 and Figures 1-7, and Supplementary Methods

Supplementary Methods from The Canonical NF-κB Pathway Governs Mammary Tumorigenesis in Transgenic Mice and Tumor Stem Cell Expansion

PDF file - 220KB, Analysis of gene copy number of isogenic prostate cancer cell lines.

Supplementary Tables 1-2 from The Canonical NF-κB Pathway Governs Mammary Tumorigenesis in Transgenic Mice and Tumor Stem Cell Expansion

PDF file - 131K, Endogenous cyclin D1 maintains prostate cancer stem cell population.

Supplemental Figures S1-S12. Supplement 1. High resolution images of normal prostate and v-Src-PEC tumor Supplement 2. AR, CK5 and CK8 abundances in cultured v-Src-PECs compared with primary cultured normal PECs detected by western-blot. Supplement 3. High resolution images of v-Src-PEC bone metastasis tumor and adjacent bone marrow Supplement 4. Kegg-pathway analysis of up-regulated genes in v-Src-PEC tumor vs normal prostate Supplement 5. High resolution gene expression profile comparing normal FVB murine prostate epithelial cells Supplement 6. Validation of CCR5 expression by qtPCR in v-Src-PEC tumors derived by subcutaneous injection compared with normal prostate Supplement 7. CCR5 expression in public available database Supplement 8. FACS analysis of CCR5 positive population in v-Src-PECs Supplement 9. High resolution image of immunohistochemical staining for CCR5 in v-Src-PEC metastasis in bone which showed in figure 5 Supplement 10. High resolution image of (A-D) histological images Supplement 11. CK5 (A) and CK8 (C) immunohistochemical staining in the prostate of castrated mice vs normal mice. Supplement 12. CCR5 IHC staining in bone metastasis v-Src-PEC tumor.

Supplemental Figure Legend from The Endogenous Cell-Fate Factor Dachshund Restrains Prostate Epithelial Cell Migration via Repression of Cytokine Secretion via a CXCL Signaling Module

Legend for Supplementary Figures S1-S12.

Endogenous cyclin D1 promotes LNCaP cell proliferation and DNA synthesis (S1); The epithelial population in Cyclin D1+/+ PECs is predominantly composed of luminal CK8+ cells (S2).

Supplementary Figure Legends 1-5 from The Canonical NF-κB Pathway Governs Mammary Tumorigenesis in Transgenic Mice and Tumor Stem Cell Expansion

PDF file - 675KB, Supplementary Figure 1: Histology of tumors induced by transformed prostate cancer cell lines in mice. Supplementary Figure 2: vWF and KLK3 staining of tumors induced by transformed cell lines. Supplementary Figure 3: Gene Ontology enrichment analysis. Supplementary Figure 4: c-Myc prostate cancer cell line gene expression. Supplementary Figure 5: Kaplan Meier curves for high and low expression populations for the c-Myc overexpression signature. Supplementary Figure 6: Androgen responsiveness of transformed PEC lines. Supplementary Figure 7: CK5 or CK8 IFC staining in cell lines and tumor lung metastasis.

PDF file - 124K

Supplementary Figures 1-5 from The Canonical NF-κB Pathway Governs Mammary Tumorigenesis in Transgenic Mice and Tumor Stem Cell Expansion

PDF file - 208K, DHT regulated gene pathways in cyclin D1+/+ and cyclin D1-/- mouse prostates.

PDF file - 15KB, Endogenous cyclin D1 mediates both E2 and EDC-induced yH2AX in MCF-7 cells.

XLSX file - 118K, Functional annotation clustering of cyclin D1 induced genes.

Prostate cancer is the second leading form of cancer-related death in men. In a subset of prostate cancer patients, increased chemokine signaling IL8 and IL6 correlates with castrate-resistant prostate cancer (CRPC). IL8 and IL6 are produced by prostate epithelial cells and promote prostate cancer cell invasion; however, the mechanisms restraining prostate epithelial cell cytokine secretion are poorly understood. Herein, the cell-fate determinant factor DACH1 inhibited CRPC tumor growth in mice. Using Dach1fl/fl/Probasin-Cre bitransgenic mice, we show IL8 and IL6 secretion was altered by approximately 1,000-fold by endogenous Dach1. Endogenous Dach1 is shown to serve as a key endogenous restraint to prostate epithelial cell growth and restrains migration via CXCL signaling. DACH1 inhibited expression, transcription, and secretion of the CXCL genes (IL8 and IL6) by binding to their promoter regulatory regions in chromatin. DACH1 is thus a newly defined determinant of benign and malignant prostate epithelium cellular growth, migration, and cytokine abundance in vivo. Cancer Res; 75(10); 1992–2004. ©2015 AACR.

XLSX file - 398K, Lists of genes differentially regulated by DHT in cyclin D1+/+ mice, DHT regulated genes in cyclin D1-/- mice and genes differentially regulated by cyclin D1.

PDF file - 83K, Cyclin D1 is required for prostate development in vivo.

PDF file - 107K, Cyclin D1 expression signature is induced in prostatospheres and induction of Wnt.

Therapy resistance and poor outcome in prostate cancer is associated with increased expression of cyclin D1. Androgens promote DNA double-strand break repair to reduce DNA damage, and cyclin D1 was also shown to enhance DNA damage repair (DDR). In this study, we investigated the significance of cyclin D1 in androgen-induced DDR using established prostate cancer cells and prostate tissues from cyclin D1 knockout mice. We demonstrate that endogenous cyclin D1 further diminished the dihydrotestosterone (DHT)-dependent reduction of γH2AX foci in vitro. We also show that cyclin D1 was required for the androgen-dependent DNA damage response both in vitro and in vivo. Furthermore, cyclin D1 was required for androgen-enhanced DDR and radioresistance of prostate cancer cells. Moreover, microarray analysis of primary prostate epithelial cells from cyclin D1-deficient and wild-type mice demonstrated that most of the DHT-dependent gene expression changes are also cyclin D1 dependent. Collectively, our findings suggest that the hormone-mediated recruitment of cyclin D1 to sites of DDR may facilitate the resistance of prostate cancer cells to DNA damage therapies and highlight the need to explore other therapeutic approaches in prostate cancer to prevent or overcome drug resistance. Cancer Res; 76(2); 329–38. ©2015 AACR.

PDF file - 33KB

Improved clinical management of prostate cancer has been impeded by an inadequate understanding of molecular genetic elements governing tumor progression. Gene signatures have provided improved prognostic indicators of human prostate cancer. The TGF-β/BMP-SMAD4 signaling pathway, which induces epithelial–mesenchymal transition (EMT), is known to constrain prostate cancer progression induced by Pten deletion. Herein, cyclin D1 inactivation reduced cellular proliferation in the murine prostate in vivo and in isogenic oncogene–transformed prostate cancer cell lines. The in vivo cyclin D1–mediated molecular signature predicted poor outcome of recurrence-free survival for patients with prostate cancer (K-means HR, 3.75, P = 0.02) and demonstrated that endogenous cyclin D1 restrains TGF-β, Snail, Twist, and Goosecoid signaling. Endogenous cyclin D1 enhanced Wnt and ES cell gene expression and expanded a prostate stem cell population. In chromatin immunoprecipitation sequencing, cyclin D1 occupied genes governing stem cell expansion and induced their transcription. The coordination of EMT restraining and stem cell expanding gene expression by cyclin D1 in the prostate may contribute to its strong prognostic value for poor outcome in biochemical-free recurrence in human prostate cancer. Cancer Res; 74(2); 508–19. ©2013 AACR.

Effectively recognizing different sceneries with complex backgrounds and varied lighting conditions plays an important role in modern AI systems. Competitive performance has recently been achieved by the deep scene categorization models. However, these models implicitly hypothesize that the image-level labels are 100% correct, which is too restrictive. Practically, the image-level labels for massive-scale scenery sets are usually calculated by external predictors such as ImageNet-CN. These labels can easily become contaminated because no predictors are completely accurate. This article proposes a new deep architecture that calculates scene categories by hierarchically deriving stable templates, which are discovered using a generative model. Specifically, we first construct a semantic space by incorporating image-level labels using subspace embedding. Afterward, it is noticeable that in the semantic space, the superpixel distributions from identically labeled images remain unchanged, regardless of the image-level label noises. On the basis of this observation, a probabilistic generative model learns the stable templates for each scene category. To deeply represent each scenery category, a novel aggregation network is developed to statistically concatenate the CNN features learned from scene annotations predicted by HSA. Finally, the learned deep representations are integrated into an image kernel, which is subsequently incorporated into a multiclass SVM for distinguishing scene categories. Thorough experiments have shown the performance of our method. As a byproduct, an empirical study of 33 SIFT-flow categories shows that the learned stable templates remain almost unchanged under a nearly 36% image label contamination rate.

Accurately categorizing million-scale Internet users (e.g., Flickr or Google Picasa) into multiple communities based on their genre tastes is an indispensable techniquein machine learning and multimedia. It can facilitate a series of applications, such as fashion recommendation and 3D non-realistic photo rendering. Conventional methods cannot handle this task appropriately because of the inherent contaminated image labels, which are produced by auxiliary image label predictors. In this article, we propose a noise-tolerant deep architecture which optimally encodes stable templates, 1 1 Stable template is our proposed new concept. It denotes the distribution of co-occurring semantic categories toward an image set. Both theoretical and empirical analysises have demonstrated that such distribution (i.e. stable templates) remains almost unchanged in the presence of contaminated image tags. discovered from a collection of images with contaminated semantic labels. Specifically, we first construct a semantic space by encoding image labels using manifold embedding. Afterward, we observe that in the semantic space, the distribution of superpixels from images with the same label remains stable, regardless of the noises from image labels. According to this observation, a probabilistic generative model (Hidden Stable Analysis) is proposed to learn the stable templates toward each image label. To globally represent the composition of a user's images, a deep aggregation network is developed which statistically concatenates the CNN features learned from all its generated stable templates. Subsequently, an affinity graph is built, in which the genre difference among users is determined by their deep features. Finally, we employ a dense subgraph discovery technique which effectively mines the communities toward various genre tastes. Experiments on a million-scale image set ( $>$ 1.4 million) compiled from Flickr have demonstrated the effectiveness of our method. Additionally, empirical study on the 33 SIFT-flow categories have shown that the detected stable templates maintain almost unchanged under nearly 32% contaminated image labels.

The application of machine learning algorithms in the field of power grid improves the service level of power enterprises and promotes the development of power grid. NVIDIA Volta and Turing GPUs powered by Tensor Cores can accelerate training and learning performance for these algorithms. With Tensor Cores enabled, FP32 and FP16 mixed precision matrix multiplication dramatically accelerates the throughput and reduces AI training times. In order to explore the cause of this phenomenon, we choose a convolutional neural network (CNN), which is widely used in computer vision, as an example and show the performance characteristics with tensor core on general matrix multiplications and convolution calculations as benchmark. Building a CNN based on cuDNN and TensorFlow, we analyze the performance of CNN from various aspects and optimize performance of it by changing the shape of convolution kernel and using texture memory, etc. The experimental results prove the effectiveness of our methods.

With the development of the State Grid, the power lines, equipment and transmission scale are expanding. In order to ensure the stability and safety of electricity, it is necessary to patrol and inspect the power towers and other equipment. With the help of deep learning, neural networks can be used to learn the features in patrol image. In this paper, feature learning model named CNN Transformer Detect Anomalies (CTran_DA) is proposed to detect anomalies in patrol images. CTran_DA uses CNN to learn local features in the image, and uses Transformer to learn global features. This paper innovatively combines the advantages of CNN and Transformer to learn the local details as well as the global feature associations in images. By comparing experiments on out self-constructed dataset, the model outperforms state-of-the-art baselines. Moreover, the Floating Point Operations (FLOPs) and parameters of the model in this paper are smaller than other algorithms. In general, CTran_DA is an efficient and lightweight model to detect anomalies in images.

Power system dispatch benefits from accurate wind power predictions. To increase the prediction precision for wind power, this paper proposes a combined model for predicting short-term wind power based on the autoregressive moving average-gated recurrent unit (ARMA-GRU). Firstly, we build the ARMA model and GRU model respectively to predict wind power. Then we optimize the combined model’s weights by quantum particle swarm algorithm (QPSO). Finally, we build an error correction model for the prediction errors to acquire the final results for the wind power predictions. Our experimental results prove the model’s reliability and the model’s high predictability is verified by comparing different prediction models.

In recent years, network security has been frequently threatened by complex network attacks. Intrusion detection system(IDS) is an active network security protection, which can effectively detect the network system when it is attacked. However, most intrusion detection system only consider the contribution of each feature to the result. In this paper, we proposes an intrusion detection system based on Global-feature Contribution Net-work(GCN). The Global-feature Contribution Network consists of two parts: Factorization Machine(FM) part and Global-Local network part. Factorization machine carry out second-order combination of features to consider the correlation between different features in attack samples and reduce the impact of sparse sample features on detection results. Global-Local network continuously integrates the global features of samples into local features, so that the network can learn the local features of samples under the grasp of global information. We evaluated our model on KDDCup99 dataset to show the effectiveness of our method. Compared with traditional methods, the model improves the accuracy and recall rate, and makes an accurate judgment on the perception of network security situation.

It is important to detect adversarial samples in the physical world that are far away from the training data distribution. Some adversarial samples can make a machine learning model generate a highly overconfident distribution in the testing stage. Thus, we proposed a mechanism for detecting adversarial samples based on semisupervised generative adversarial networks (GANs) with an encoder-decoder structure; this mechanism can be applied to any pretrained neural network without changing the network’s structure. The semisupervised GANs also give us insight into the behavior of adversarial samples and their flow through the layers of a deep neural network. In the supervised scenario, the latent feature (or the discriminator’s output score information) of the semi-supervised GAN and the target network's logit information are used as the input of logistic regression classifier to detect the adversarial samples. In the unsupervised scenario, first, we proposed a one-class classier based on the semisupervised Gaussian mixture conditional generative adversarial network (GM-CGAN) to fit the joint feature information of the normal data, and then, we used a discriminator network to detect normal data and adversarial samples. In both supervised scenarios and unsupervised scenarios, experimental results show that our method outperforms latest methods.