SUMMARY Unresectable glioblastoma (GBM) cells in the invading tumor edge can act as seeds for recurrence. The molecular and phenotypic properties of these cells remain elusive. Here, we report that the invading edge and tumor core have two distinct types of glioma stem-like cells (GSCs) that resemble proneural (PN) and mesenchymal (MES) subtypes, respectively. Upon exposure to ionizing radiation (IR), GSCs, initially enriched for a CD133 + PN signature, transition to a CD109 + MES subtype in a C/EBP-β-dependent manner. Our gene expression analysis of paired cohorts of patients with primary and recurrent GBMs identified a CD133-to-CD109 shift in tumors with an MES recurrence. Patient-derived CD133 − /CD109 + cells are highly enriched with clonogenic, tumor-initiating, and radiation-resistant properties, and silencing CD109 significantly inhibits these phenotypes. We also report a conserved regulation of YAP/TAZ pathways by CD109 that could be a therapeutic target in GBM.

SummaryATG4B stimulates autophagy by promoting autophagosome formation through reversible modification of ATG8. We identify ATG4B as a substrate of mammalian sterile20-like kinase (STK) 26/MST4. MST4 phosphorylates ATG4B at serine residue 383, which stimulates ATG4B activity and increases autophagic flux. Inhibition of MST4 or ATG4B activities using genetic approaches or an inhibitor of ATG4B suppresses autophagy and the tumorigenicity of glioblastoma (GBM) cells. Furthermore, radiation induces MST4 expression, ATG4B phosphorylation, and autophagy. Inhibiting ATG4B in combination with radiotherapy in treating mice with intracranial GBM xenograft markedly slows tumor growth and provides a significant survival benefit. Our work describes an MST4-ATG4B signaling axis that influences GBM autophagy and malignancy, and whose therapeutic targeting enhances the anti-tumor effects of radiotherapy.

We introduce PubMedQA, a novel biomedical question answering (QA) dataset collected from PubMed abstracts. The task of Pub-MedQA is to answer research questions with yes/no/maybe (e.g.: Do preoperative statins reduce atrial fibrillation after coronary artery bypass grafting?) using the corresponding abstracts. PubMedQA has 1k expert-annotated, 61.2k unlabeled and 211.3k artificially generated QA instances. Each PubMedQA instance is composed of (1) a question which is either an existing research article title or derived from one, (2) a context which is the corresponding abstract without its conclusion, (3) a long answer, which is the conclusion of the abstract and, presumably, answers the research question, and (4) a yes/no/maybe answer which summarizes the conclusion. Pub-MedQA is the first QA dataset where reasoning over biomedical research texts, especially their quantitative contents, is required to answer the questions. Our best performing model, multi-phase fine-tuning of BioBERT with long answer bag-of-word statistics as additional supervision, achieves 68.1% accuracy, compared to single human performance of 78.0% accuracy and majority-baseline of 55.2% accuracy, leaving much room for improvement. PubMedQA is publicly available at https://pubmedqa.github.io.

Head and neck squamous cell carcinoma (HNSCC) is characterized by complex relations between stromal, epithelial, and immune cells within the tumor microenvironment (TME). To enable the development of more efficacious therapies, we aim to study the heterogeneity, signatures of unique cell populations, and cell-cell interactions of non-immune and immune cell populations in 6 human papillomavirus (HPV)+ and 12 HPV– HNSCC patient tumor and matched peripheral blood specimens using single-cell RNA sequencing. Using this dataset of 134,606 cells, we show cell type-specific signatures associated with inflammation and HPV status, describe the negative prognostic value of fibroblasts with elastic differentiation specifically in the HPV+ TME, predict therapeutically targetable checkpoint receptor-ligand interactions, and show that tumor-associated macrophages are dominant contributors of PD-L1 and other immune checkpoint ligands in the TME. We present a comprehensive single-cell view of cell-intrinsic mechanisms and cell-cell communication shaping the HNSCC microenvironment.

Precision oncology involves identifying drugs that will effectively treat a tumor and then prescribing an optimal clinical treatment regimen. However, most first-line chemotherapy drugs do not have biomarkers to guide their application. For molecularly targeted drugs, using the genomic status of a drug target as a therapeutic indicator has limitations. In this study, machine learning methods (e.g., deep learning) were used to identify informative features from genome-scale omics data and to train classifiers for predicting the effectiveness of drugs in cancer cell lines. The methodology introduced here can accurately predict the efficacy of drugs, regardless of whether they are molecularly targeted or nonspecific chemotherapy drugs. This approach, on a per-drug basis, can identify sensitive cancer cells with an average sensitivity of 0.82 and specificity of 0.82; on a per-cell line basis, it can identify effective drugs with an average sensitivity of 0.80 and specificity of 0.82. This report describes a data-driven precision medicine approach that is not only generalizable but also optimizes therapeutic efficacy. The framework detailed herein, when successfully translated to clinical environments, could significantly broaden the scope of precision oncology beyond targeted therapies, benefiting an expanded proportion of cancer patients. .