Machine learning-based monosaccharide profiling for tissue-specific classification of Wolfiporia extensa samples.

Machine learning (ML) has been used for many clinical decision-making processes and diagnostic procedures in bioinformatics applications. We examined eight algorithms, including linear discriminant analysis (LDA), logistic regression (LR), k-nearest neighbor (KNN), random forest (RF), gradient boosting machine (GBM), support vector machine (SVM), Naïve Bayes classifier (NB), and artificial neural network (ANN) models, to evaluate their classification and prediction capabilities for four tissue types in Wolfiporia extensa using their monosaccharide composition profiles. All 8 ML-based models were assessed as exemplary models with AUC exceeding 0.8. Five models, namely LDA, KNN, RF, GBM, and ANN, performed excellently in the four-tissue-type classification (AUC > 0.9). Additionally, all eight models were evaluated as good predictive models with AUC value > 0.8 in the three-tissue-type classification. Notably, all 8 ML-based methods outperformed the single linear discriminant analysis (LDA) plotting method. For large sample sizes, the ML-based methods perform better than traditional regression techniques and could potentially increase the accuracy in identifying tissue samples of W. extensa.

Multiomics analyses reveal DARS1-AS1/YBX1-controlled posttranscriptional circuits promoting glioblastoma tumorigenesis/radioresistance.

The glioblastoma (GBM) stem cell-like cells (GSCs) are critical for tumorigenesis/therapeutic resistance of GBM. Mounting evidence supports tumor-promoting function of long noncoding RNAs (lncRNAs), but their role in GSCs remains poorly understood. By combining CRISPRi screen with orthogonal multiomics approaches, we identified a lncRNA DARS1-AS1-controlled posttranscriptional circuitry that promoted the malignant properties of GBM cells/GSCs. Depleting DARS1-AS1 inhibited the proliferation of GBM cells/GSCs and self-renewal of GSCs, prolonging survival in orthotopic GBM models. DARS1-AS1 depletion also impaired the homologous recombination (HR)-mediated double-strand break (DSB) repair and enhanced the radiosensitivity of GBM cells/GSCs. Mechanistically, DARS1-AS1 interacted with YBX1 to promote target mRNA binding and stabilization, forming a mixed transcriptional/posttranscriptional feed-forward loop to up-regulate expression of the key regulators of G1-S transition, including E2F1 and CCND1. DARS1-AS1/YBX1 also stabilized the mRNA of FOXM1, a master transcription factor regulating GSC self-renewal and DSB repair. Our findings suggest DARS1-AS1/YBX1 axis as a potential therapeutic target for sensitizing GBM to radiation/HR deficiency-targeted therapy.

Critical role of lncEPAT in coupling dysregulated EGFR pathway and histone H2A deubiquitination during glioblastoma tumorigenesis.

Histone 2A (H2A) monoubiquitination is a fundamental epigenetics mechanism of gene expression, which plays a critical role in regulating cell fate. However, it is unknown if H2A ubiquitination is involved in EGFR-driven tumorigenesis. In the current study, we have characterized a previously unidentified oncogenic lncRNA (lncEPAT) that mediates the integration of the dysregulated EGFR pathway with H2A deubiquitination in tumorigenesis. LncEPAT was induced by the EGFR pathway, and high-level lncEPAT expression positively correlated with the glioma grade and predicted poor survival of glioma patients. Mass spectrometry analyses revealed that lncEPAT specifically interacted with deubiquitinase USP16. LncEPAT inhibited USP16's recruitment to chromatin, thereby blocking USP16-mediated H2A deubiquitination and repressing target gene expression, including CDKN1A and CLUSTERIN. Depletion of lncEPAT promoted USP16-induced cell cycle arrest and cellular senescence, and then repressed GBM cell tumorigenesis. Thus, the EGFR-lncEPAT-ubH2A coupling represents a previously unidentified mechanism for epigenetic gene regulation and senescence resistance during GBM tumorigenesis.

tRNA Function and Dysregulation in Cancer.

Transfer RNA (tRNA) is a central component of protein synthesis and plays important roles in epigenetic regulation of gene expression in tumors. tRNAs are also involved in many cell processes including cell proliferation, cell signaling pathways and stress response, implicating a role in tumorigenesis and cancer progression. The complex role of tRNA in cell regulation implies that an understanding of tRNA function and dysregulation can be used to develop treatments for many cancers including breast cancer, colon cancer, and glioblastoma. Moreover, tRNA modifications including methylation are necessary for tRNA folding, stability, and function. In response to certain stress conditions, tRNAs can be cleaved in half to form tiRNAs, or even shorter tRNA fragments (tRF). tRNA structure and modifications, tiRNA induction of stress granule formation, and tRF regulation of gene expression through the repression of translation can all impact a cell's fate. This review focuses on how these functions of tRNAs, tiRNA, and tRFs can lead to tumor development and progression. Further studies focusing on the specific pathways of tRNA regulation could help identify tRNA biomarkers and therapeutic targets, which might prevent and treat cancers.

The Roles and Regulation of m6A Modification in Glioblastoma Stem Cells and Tumorigenesis.

Glioblastoma is the most common and most lethal primary malignant brain tumor. N6-methyladenosine (m6A) is a widespread and abundant internal messenger RNA (mRNA) modification found in eukaryotes. Accumulated evidence demonstrates that m6A modification is aberrantly activated in human cancers and is critical for tumorigenesis and metastasis. m6A modification is also strongly involved in key signaling pathways and is associated with prognosis in glioblastoma. Here, we briefly outline the functions of m6A and its regulatory proteins, including m6A writers, erasers, and readers of the fate of RNA. We also summarize the latest breakthroughs in this field, describe the underlying molecular mechanisms that contribute to the tumorigenesis and progression, and highlight the inhibitors targeting the factors in m6A modification in glioblastoma. Further studies focusing on the specific pathways of m6A modification could help identify biomarkers and therapeutic targets that might prevent and treat glioblastoma.

Robust Aggregation for Federated Learning by Minimum γ-Divergence Estimation.

Federated learning is a framework for multiple devices or institutions, called local clients, to collaboratively train a global model without sharing their data. For federated learning with a central server, an aggregation algorithm integrates model information sent from local clients to update the parameters for a global model. Sample mean is the simplest and most commonly used aggregation method. However, it is not robust for data with outliers or under the Byzantine problem, where Byzantine clients send malicious messages to interfere with the learning process. Some robust aggregation methods were introduced in literature including marginal median, geometric median and trimmed-mean. In this article, we propose an alternative robust aggregation method, named γ-mean, which is the minimum divergence estimation based on a robust density power divergence. This γ-mean aggregation mitigates the influence of Byzantine clients by assigning fewer weights. This weighting scheme is data-driven and controlled by the γ value. Robustness from the viewpoint of the influence function is discussed and some numerical results are presented.

Alteration of power law scaling of spontaneous brain activity in schizophrenia.

Nonlinear dynamical analysis has been used to quantify the complexity of brain signal at temporal scales. Power law scaling is a well-validated method in physics that has been used to describe the dynamics of a system in the frequency domain, ranging from noisy oscillation to complex fluctuations. In this research, we investigated the power-law characteristics in a large-scale resting-state fMRI data of schizophrenia and healthy participants derived from Taiwan Aging and Mental Illness cohort. We extracted the power spectral density (PSD) of resting signal by Fourier transform. Power law scaling of PSD was estimated by determining the slope of the regression line fitting to the logarithm of PSD. t-Test was used to assess the statistical difference in power law scaling between schizophrenia and healthy participants. The significant differences in power law scaling were found in six brain regions. Schizophrenia patients have significantly more positive power law scaling (i.e., more homogenous frequency components) at four brain regions: left precuneus, left medial dorsal nucleus, right inferior frontal gyrus, and right middle temporal gyrus and less positive power law scaling (i.e., more dominant at lower frequency range) in bilateral putamen compared with healthy participants. Moreover, significant correlations of power law scaling with the severity of psychosis were found. These findings suggest that schizophrenia has abnormal brain signal complexity linked to psychotic symptoms. The power law scaling represents the dynamical properties of resting-state fMRI signal may serve as a novel functional brain imaging marker for evaluating patients with mental illness.

Radiomic Features at CT Can Distinguish Pancreatic Cancer from Noncancerous Pancreas.

Purpose To identify distinguishing CT radiomic features of pancreatic ductal adenocarcinoma (PDAC) and to investigate whether radiomic analysis with machine learning can distinguish between patients who have PDAC and those who do not. Materials and Methods This retrospective study included contrast material-enhanced CT images in 436 patients with PDAC and 479 healthy controls from 2012 to 2018 from Taiwan that were randomly divided for training and testing. Another 100 patients with PDAC (enriched for small PDACs) and 100 controls from Taiwan were identified for testing (from 2004 to 2011). An additional 182 patients with PDAC and 82 healthy controls from the United States were randomly divided for training and testing. Images were processed into patches. An XGBoost (https://xgboost.ai/) model was trained to classify patches as cancerous or noncancerous. Patients were classified as either having or not having PDAC on the basis of the proportion of patches classified as cancerous. For both patch-based and patient-based classification, the models were characterized as either a local model (trained on Taiwanese data only) or a generalized model (trained on both Taiwanese and U.S. data). Sensitivity, specificity, and accuracy were calculated for patch- and patient-based analysis for the models. Results The median tumor size was 2.8 cm (interquartile range, 2.0-4.0 cm) in the 536 Taiwanese patients with PDAC (mean age, 65 years ± 12 [standard deviation]; 289 men). Compared with normal pancreas, PDACs had lower values for radiomic features reflecting intensity and higher values for radiomic features reflecting heterogeneity. The performance metrics for the developed generalized model when tested on the Taiwanese and U.S. test data sets, respectively, were as follows: sensitivity, 94.7% (177 of 187) and 80.6% (29 of 36); specificity, 95.4% (187 of 196) and 100% (16 of 16); accuracy, 95.0% (364 of 383) and 86.5% (45 of 52); and area under the curve, 0.98 and 0.91. Conclusion Radiomic analysis with machine learning enabled accurate detection of PDAC at CT and could identify patients with PDAC. Keywords: CT, Computer Aided Diagnosis (CAD), Pancreas, Computer Applications-Detection/Diagnosis Supplemental material is available for this article. © RSNA, 2021.

EGFR/SRC/ERK-stabilized YTHDF2 promotes cholesterol dysregulation and invasive growth of glioblastoma.

Glioblastoma (GBM) is the most common type of adult malignant brain tumor, but its molecular mechanisms are not well understood. In addition, the knowledge of the disease-associated expression and function of YTHDF2 remains very limited. Here, we show that YTHDF2 overexpression clinically correlates with poor glioma patient prognosis. EGFR that is constitutively activated in the majority of GBM causes YTHDF2 overexpression through the EGFR/SRC/ERK pathway. EGFR/SRC/ERK signaling phosphorylates YTHDF2 serine39 and threonine381, thereby stabilizes YTHDF2 protein. YTHDF2 is required for GBM cell proliferation, invasion, and tumorigenesis. YTHDF2 facilitates m6A-dependent mRNA decay of LXRA and HIVEP2, which impacts the glioma patient survival. YTHDF2 promotes tumorigenesis of GBM cells, largely through the downregulation of LXRα and HIVEP2. Furthermore, YTHDF2 inhibits LXRα-dependent cholesterol homeostasis in GBM cells. Together, our findings extend the landscape of EGFR downstream circuit, uncover the function of YTHDF2 in GBM tumorigenesis, and highlight an essential role of RNA m6A methylation in cholesterol homeostasis.

YTHDF3 Induces the Translation of m6A-Enriched Gene Transcripts to Promote Breast Cancer Brain Metastasis.

Brain metastasis is a major cause of cancer mortality, but its molecular mechanisms are severely understudied. In addition, little is known regarding the role of m6A reader YTHDF3 in human diseases. Here, we show that YTHDF3 overexpression clinically correlates with brain metastases in breast cancer patients. YTHDF3 promotes cancer cell interactions with brain endothelial cells and astrocytes, blood-brain barrier extravasation, angiogenesis, and outgrow. Mechanistically, YTHDF3 enhances the translation of m6A-enriched transcripts for ST6GALNAC5, GJA1, and EGFR, all associated with brain metastasis. Furthermore, overexpression of YTHDF3 in brain metastases is attributed to increased gene copy number and the autoregulation of YTHDF3 cap-independent translation by binding to m6A residues within its own 5' UTR. Our work uncovers an essential role of YTHDF3 in controlling the interaction between cancer cells and brain microenvironment, thereby inducing brain metastatic competence.

Epigenetic regulation of cancer stem cell and tumorigenesis.

As a unique subpopulation of cancer cells, cancer stem cells (CSCs) acquire the resistance to conventional therapies and appear to be the prime cause of cancer recurrence. Like their normal counterparts, CSCs can renew themselves and generate differentiated progenies. Cancer stem cells are distinguished among heterogenous cancer cells by molecular markers and their capacity of efficiently forming new tumors composed of diverse and heterogenous cancer cells. Tumor heterogeneity can be inter- or intra-tumor, molecularly resulting from the accumulation of genetic and non-genetic alterations. Non-genetic alterations are mainly changes on epigenetic modifications of DNA and histone, and chromatin remodeling. As tumor-initiating cells and contributing to the tumor heterogeneity in the brain, glioblastoma stem cells (GSCs) attract extensive research interests. Epigenetic modifications confer on tumor cells including CSCs reversible and inheritable genomic changes and affect gene expression without alteration in DNA sequence. Here, we will review recent advances in histone demethylation, DNA methylation, RNA methylation and ubiquitination in glioblastomas and their impacts on tumorigenesis with a focus on CSCs.

Wnt-Induced Stabilization of KDM4C Is Required for Wnt/ß-Catenin Target Gene Expression and Glioblastoma Tumorigenesis.

Wnt/ß-catenin signaling activates the transcription of target genes to regulate stem cells and cancer development. However, the contribution of epigenetic regulation to this process is unknown. Here, we report that Wnt activation stabilizes the epigenetic regulator KDM4C that promotes tumorigenesis and survival of human glioblastoma cells by epigenetically activating the transcription of Wnt target genes. KDM4C protein expression was upregulated in human glioblastomas, and its expression directly correlated with Wnt activity and Wnt target gene expression. KDM4C was essential for Wnt-induced gene expression and tumorigenesis of glioblastoma cells. In the absence of Wnt3a, protein kinase R phosphorylated KDM4C at Ser918, inducing KDM4C ubiquitination and degradation. Wnt3a stabilized KDM4C through inhibition of GSK3-dependent protein kinase R activity. Stabilized KDM4C accumulated in the nucleus and bound to and demethylated TCF4-associated histone H3K9 by interacting with ß-catenin, promoting HP1γ removal and transcriptional activation. These findings reveal that Wnt-KDM4C-ß-catenin signaling represents a novel mechanism for the transcription of Wnt target genes and regulation of tumorigenesis, with important clinical implications. SIGNIFICANCE: These findings identify the Wnt-KDM4C-ß-catenin signaling axis as a critical mechanism for glioma tumorigenesis that may serve as a new therapeutic target in glioblastoma.

Methylation of RNA N6-methyladenosine in modulation of cytokine responses and tumorigenesis.

Among myriads of distinct chemical modification in RNAs, the dynamic, reversible and fine-tuned methylation of N6-methyladenosine (m6A) is the most prevalent modification in eukaryotic mRNAs. This RNA mark is generated by proteins that act as m6A writers and can be reversed by proteins that act as m6A erasers. The RNA m6A modification is also mediated by another group of proteins capable of recognizing m6A that act as m6A readers. The m6A modification exerts direct control over the RNA metabolism including mRNA processing, mRNA exporting, translation initiation, mRNA stability and the biogenesis of long-non-coding RNA (LncRNA), thereby can influence various aspects of cell function. Evidently, m6A is intimately associated with cancer development and progression such as self-renewal capacity of cancer stem cells, proliferation, apoptosis and therapeutic resistance, and immune response. In this review, we will discuss the regulation and function of m6A, the various functions ascribed to these proteins and the emerging concepts that impact our knowledge of these proteins and their roles in the epitranscriptome. Conceivably, m6A may play pivotal roles in cytokine and immune response and carcinogenesis.

Aberrant Activation of ß-Catenin Signaling Drives Glioma Tumorigenesis via USP1-Mediated Stabilization of EZH2.

Aberrant activation of ß-catenin signaling is a critical driver for tumorigenesis, but the mechanism underlying this activation is not completely understood. In this study, we demonstrate a critical role of ß-catenin signaling in stabilization of enhancer of zeste homolog 2 (EZH2) and control of EZH2-mediated gene repression in oncogenesis. ß-Catenin/TCF4 activated the transcription of the deubiquitinase USP1, which then interacted with and deubiquitinated EZH2 directly. USP1-mediated stabilization of EZH2 promoted its recruitment to the promoters of CDKN1B, RUNX3, and HOXA5, resulting in enhanced enrichment of histone H3K27me3 and repression of target gene expression. In human glioma specimens, expression levels of nuclear ß-catenin, USP1, and EZH2 correlated with one another. Depletion of ß-catenin/USP1/EZH2 repressed glioma cell proliferation in vitro and tumor formation in vivo. Our findings indicate that a ß-catenin-USP1-EZH2 axis orchestrates the interplay between dysregulated ß-catenin signaling and EZH2-mediated gene epigenetic silencing during glioma tumorigenesis. SIGNIFICANCE: These findings identify the ß-catenin-USP1-EZH2 signaling axis as a critical mechanism for glioma tumorigenesis that may serve as a new therapeutic target in glioblastoma.

Sufficient dimension reduction via random-partitions for the large-p-small-n problem.

Sufficient dimension reduction (SDR) continues to be an active field of research. When estimating the central subspace (CS), inverse regression based SDR methods involve solving a generalized eigenvalue problem, which can be problematic under the large-p-small-n situation. In recent years, new techniques have emerged in numerical linear algebra, called randomized algorithms or random sketching, for high-dimensional and large scale problems. To overcome the large-p-small-n SDR problem, we combine the idea of statistical inference with random sketching to propose a new SDR method, called integrated random-partition SDR (iRP-SDR). Our method consists of the following three steps: (i) Randomly partition the covariates into subsets to construct an envelope subspace with low dimension. (ii) Obtain a sketch of the CS by applying a conventional SDR method within the constructed envelope subspace. (iii) Repeat the above two steps many times and integrate these multiple sketches to form the final estimate of the CS. After describing the details of these steps, the asymptotic properties of iRP-SDR are established. Unlike existing methods, iRP-SDR does not involve the determination of the structural dimension until the last stage, which makes it more adaptive to a high-dimensional setting. The advantageous performance of iRP-SDR is demonstrated via simulation studies and a practical example analyzing EEG data.

The Role of Fibrinogen-Like Protein 2 on Immunosuppression and Malignant Progression in Glioma.

BACKGROUND: Virtually all low-grade gliomas (LGGs) will progress to high-grade gliomas (HGGs), including glioblastoma, the most common malignant primary brain tumor in adults. A key regulator of immunosuppression, fibrinogen-like protein 2 (FGL2), may play an important role in the malignant transformation of LGG to HGG. We sought to determine the mechanism of FGL2 on tumor progression and to show that inhibiting FGL2 expression had a therapeutic effect. METHODS: We analyzed human gliomas that had progressed from low- to high-grade for FGL2 expression. We modeled FGL2 overexpression in an immunocompetent genetically engineered mouse model to determine its effect on tumor progression. Tumors and their associated microenvironments were analyzed for their immune cell infiltration. Mice were treated with an FGL2 antibody to determine a therapeutic effect. Statistical tests were two-sided. RESULTS: We identified increased expression of FGL2 in surgically resected tumors that progressed from low to high grade (n = 10). The Cancer Genome Atlas data showed that LGG cases with overexpression of FGL2 (n = 195) had statistically significantly shorter survival (median = 62.9 months) compared with cases with low expression (n = 325, median = 94.4 months, P < .001). In a murine glioma model, HGGs induced with FGL2 exhibited a mesenchymal phenotype and increased CD4+ forkhead box P3 (FoxP3)+ Treg cells, implicating immunosuppression as a mechanism for tumor progression. Macrophages in these tumors were skewed toward the immunosuppressive M2 phenotype. Depletion of Treg cells with anti-FGL2 statistically significantly prolonged survival in mice compared with controls (n = 11 per group, median survival = 90 days vs 62 days, P = .004), shifted the phenotype from mesenchymal HGG to proneural LGG, and decreased M2 macrophage skewing. CONCLUSIONS: FGL2 facilitates glioma progression from low to high grade. Suppressing FGL2 expression holds therapeutic promise for halting malignant transformation in glioma.

A peptide encoded by circular form of LINC-PINT suppresses oncogenic transcriptional elongation in glioblastoma.

Circular RNAs (circRNAs) are a large class of transcripts in the mammalian genome. Although the translation of circRNAs was reported, additional coding circRNAs and the functions of their translated products remain elusive. Here, we demonstrate that an endogenous circRNA generated from a long noncoding RNA encodes regulatory peptides. Through ribosome nascent-chain complex-bound RNA sequencing (RNC-seq), we discover several peptides potentially encoded by circRNAs. We identify an 87-amino-acid peptide encoded by the circular form of the long intergenic non-protein-coding RNA p53-induced transcript (LINC-PINT) that suppresses glioblastoma cell proliferation in vitro and in vivo. This peptide directly interacts with polymerase associated factor complex (PAF1c) and inhibits the transcriptional elongation of multiple oncogenes. The expression of this peptide and its corresponding circRNA are decreased in glioblastoma compared with the levels in normal tissues. Our results establish the existence of peptides encoded by circRNAs and demonstrate their potential functions in glioblastoma tumorigenesis.

Profiles of brain metastases: Prioritization of therapeutic targets.

We sought to compare the tumor profiles of brain metastases from common cancers with those of primary tumors and extracranial metastases in order to identify potential targets and prioritize rational treatment strategies. Tumor samples were collected from both the primary and metastatic sites of nonsmall cell lung cancer, breast cancer and melanoma from patients in locations worldwide, and these were submitted to Caris Life Sciences for tumor multiplatform analysis, including gene sequencing (Sanger and next-generation sequencing with a targeted 47-gene panel), protein expression (assayed by immunohistochemistry) and gene amplification (assayed by in situ hybridization). The data analysis considered differential protein expression, gene amplification and mutations among brain metastases, extracranial metastases and primary tumors. The analyzed population included: 16,999 unmatched primary tumor and/or metastasis samples: 8,178 nonsmall cell lung cancers (5,098 primaries; 2,787 systemic metastases; 293 brain metastases), 7,064 breast cancers (3,496 primaries; 3,469 systemic metastases; 99 brain metastases) and 1,757 melanomas (660 primaries; 996 systemic metastases; 101 brain metastases). TOP2A expression was increased in brain metastases from all 3 cancers, and brain metastases overexpressed multiple proteins clustering around functions critical to DNA synthesis and repair and implicated in chemotherapy resistance, including RRM1, TS, ERCC1 and TOPO1. cMET was overexpressed in melanoma brain metastases relative to primary skin specimens. Brain metastasis patients may particularly benefit from therapeutic targeting of enzymes associated with DNA synthesis, replication and/or repair.

Robust mislabel logistic regression without modeling mislabel probabilities.

Logistic regression is among the most widely used statistical methods for linear discriminant analysis. In many applications, we only observe possibly mislabeled responses. Fitting a conventional logistic regression can then lead to biased estimation. One common resolution is to fit a mislabel logistic regression model, which takes into consideration of mislabeled responses. Another common method is to adopt a robust M-estimation by down-weighting suspected instances. In this work, we propose a new robust mislabel logistic regression based on γ-divergence. Our proposal possesses two advantageous features: (1) It does not need to model the mislabel probabilities. (2) The minimum γ-divergence estimation leads to a weighted estimating equation without the need to include any bias correction term, that is, it is automatically bias-corrected. These features make the proposed γ-logistic regression more robust in model fitting and more intuitive for model interpretation through a simple weighting scheme. Our method is also easy to implement, and two types of algorithms are included. Simulation studies and the Pima data application are presented to demonstrate the performance of γ-logistic regression.