RNA modification has emerged as an important epigenetic mechanism that controls abnormal metabolism and growth in acute myeloid leukaemia (AML). However, the roles of RNA N4-acetylcytidine (ac4C) modification in AML remain elusive. Here, we report that ac4C and its catalytic enzyme NAT10 drive leukaemogenesis and sustain self-renewal of leukaemic stem cells/leukaemia-initiating cells through reprogramming serine metabolism. Mechanistically, NAT10 facilitates exogenous serine uptake and de novo biosynthesis through ac4C-mediated translation enhancement of the serine transporter SLC1A4 and the transcription regulators HOXA9 and MENIN that activate transcription of serine synthesis pathway genes. We further characterize fludarabine as an inhibitor of NAT10 and demonstrate that pharmacological inhibition of NAT10 targets serine metabolic vulnerability, triggering substantial anti-leukaemia effects both in vitro and in vivo. Collectively, our study demonstrates the functional importance of ac4C and NAT10 in metabolism control and leukaemogenesis, providing insights into the potential of targeting NAT10 for AML therapy.
Advancements in wearable sensor technologies and the digitization of medical records have contributed to the unprecedented ubiquity of biomedical time series data. Data-driven models have tremendous potential to assist clinical diagnosis and improve patient care by improving long-term monitoring capabilities, facilitating early disease detection and intervention, as well as promoting personalized healthcare delivery. However, accessing extensively labeled datasets to train data-hungry deep learning models encounters many barriers, such as long-tail distribution of rare diseases, cost of annotation, privacy and security concerns, data-sharing regulations, and ethical considerations. An emerging approach to overcome the scarcity of labeled data is to augment AI methods with human-like capabilities to leverage past experiences to learn new tasks with limited examples, called few-shot learning. This survey provides a comprehensive review and comparison of few-shot learning methods for biomedical time series applications. The clinical benefits and limitations of such methods are discussed in relation to traditional data-driven approaches. This paper aims to provide insights into the current landscape of few-shot learning for biomedical time series and its implications for future research and applications.
Interest in the concept of a virtual human model that can encompass human physiology and anatomy on a biophysical (mechanistic) basis, and can assist with the clinical diagnosis and treatment of disease, appears to be growing rapidly around the globe. When such models are personalised and coupled with continual diagnostic measurements, they are called 'digital twins'. We argue here that the most useful form of virtual human model will be one that is constrained by the laws of physics, contains a comprehensive knowledge graph of all human physiology and anatomy, is multiscale in the sense of linking systems physiology down to protein function, and can to some extent be personalized and linked directly with clinical records. We discuss current progress from the IUPS Physiome Project and the requirements for a framework to achieve such a model.
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：