No abstract

Despite aggressive surgery, radiotherapy, and chemotherapy, treatment of malignant glioma remains formidable. Although the concept of cancer stem cells reveals a new framework of cancer therapeutic strategies against malignant glioma, it remains unclear how glioma stem cells could be eradicated. Here, we demonstrate that autocrine TGF-beta signaling plays an essential role in retention of stemness of glioma-initiating cells (GICs) and describe the underlying mechanism for it. TGF-beta induced [corrected] expression of Sox2, a stemness gene, and this induction was mediated by Sox4, a direct TGF-beta target gene. Inhibitors of TGF-beta signaling drastically deprived tumorigenicity of GICs by promoting their differentiation, and these effects were attenuated in GICs transduced with Sox2 or Sox4. Furthermore, GICs pretreated with TGF-beta signaling inhibitor exhibited less lethal potency in intracranial transplantation assay. These results identify an essential pathway for GICs, the TGF-beta-Sox4-Sox2 pathway, whose disruption would be a therapeutic strategy against gliomas.

Temporal action localization is an important yet challenging problem. Given a long, untrimmed video consisting of multiple action instances and complex background contents, we need not only to recognize their action categories, but also to localize the start time and end time of each instance. Many state-of-the-art systems use segmentlevel classifiers to select and rank proposal segments of predetermined boundaries. However, a desirable model should move beyond segment-level and make dense predictions at a fine granularity in time to determine precise temporal boundaries. To this end, we design a novel Convolutional-De-Convolutional (CDC) network that places CDC filters on top of 3D ConvNets, which have been shown to be effective for abstracting action semantics but reduce the temporal length of the input data. The proposed CDC filter performs the required temporal upsampling and spatial downsampling operations simultaneously to predict actions at the frame-level granularity. It is unique in jointly modeling action semantics in space-time and fine-grained temporal dynamics. We train the CDC network in an end-toend manner efficiently. Our model not only achieves superior performance in detecting actions in every frame, but also significantly boosts the precision of localizing temporal boundaries. Finally, the CDC network demonstrates a very high efficiency with the ability to process 500 frames per second on a single GPU server. Source code and trained models are available online at https://bitbucket. org/columbiadvmm/cdc.

No abstract

Inhibitory Smads (I-Smads) have conserved carboxy-terminal MH2 domains but highly divergent amino-terminal regions when compared with receptor-regulated Smads (R-Smads) and common-partner Smads (co-Smads). Smad6 preferentially inhibits Smad signaling initiated by the bone morphogenetic protein (BMP) type I receptors ALK-3 and ALK-6, whereas Smad7 inhibits both transforming growth factor β (TGF-β)- and BMP-induced Smad signaling. I-Smads also regulate some non-Smad signaling pathways. Here, we discuss the vertebrate I-Smads, their roles as inhibitors of Smad activation and regulators of receptor stability, as scaffolds for non-Smad signaling, and their possible roles in the nucleus. We also discuss the posttranslational modification of I-Smads, including phosphorylation, ubiquitylation, acetylation, and methylation.