Search for: All records

Accurately typing entity mentions from text segments is a fundamental task for various natural language processing applications. Many previous approaches rely on massive human-annotated data to perform entity typing. Nevertheless, collecting such data in highly specialized science and engineering domains (e.g., software engineering and security) can be time-consuming and costly, without mentioning the domain gaps between training and inference data if the model needs to be applied to confidential datasets. In this paper, we study the task of seed-guided fine-grained entity typing in science and engineering domains, which takes the name and a few seed entities for each entity type as the only supervision and aims to classify new entity mentions into both seen and unseen types (i.e., those without seed entities). To solve this problem, we propose SEType which first enriches the weak supervision by finding more entities for each seen type from an unlabeled corpus using the contextualized representations of pre-trained language models. It then matches the enriched entities to unlabeled text to get pseudo-labeled samples and trains a textual entailment model that can make inferences for both seen and unseen types. Extensive experiments on two datasets covering four domains demonstrate the effectiveness of SEType in comparison with various baselines. Code and data are available at: https://github.com/yuzhimanhua/SEType. 

Myokines and exosomes, originating from skeletal muscle, are shown to play a significant role in maintaining brain homeostasis. While exercise has been reported to promote muscle secretion, little is known about the effects of neuronal innervation and activity on the yield and molecular composition of biologically active molecules from muscle. As neuromuscular diseases and disabilities associated with denervation impact muscle metabolism, we hypothesize that neuronal innervation and firing may play a pivotal role in regulating secretion activities of skeletal muscles. We examined this hypothesis using an engineered neuromuscular tissue model consisting of skeletal muscles innervated by motor neurons. The innervated muscles displayed elevated expression of mRNAs encoding neurotrophic myokines, such as interleukin-6, brain-derived neurotrophic factor, and FDNC5, as well as the mRNA of peroxisome-proliferator-activated receptor γ coactivator 1α, a key regulator of muscle metabolism. Upon glutamate stimulation, the innervated muscles secreted higher levels of irisin and exosomes containing more diverse neurotrophic microRNAs than neuron-free muscles. Consequently, biological factors secreted by innervated muscles enhanced branching, axonal transport, and, ultimately, spontaneous network activities of primary hippocampal neurons in vitro. Overall, these results reveal the importance of neuronal innervation in modulating muscle-derived factors that promote neuronal function and suggest that the engineered neuromuscular tissue model holds significant promise as a platform for producing neurotrophic molecules. 

Heparan sulfate (HS) glycosaminoglycans are widely expressed on the mammalian cell surfaces and extracellular matrices and play important roles in a variety of cell functions. Studies on the structure–activity relationships of HS have long been hampered by the challenges in obtaining chemically defined HS structures with unique sulfation patterns. Here, we report a new approach to HS glycomimetics based on iterative assembly of clickable disaccharide building blocks that mimic the disaccharide repeating units of native HS. Variably sulfated clickable disaccharides were facilely assembled into a library of mass spec-sequenceable HS-mimetic oligomers with defined sulfation patterns by solution-phase iterative syntheses. Microarray and surface plasmon resonance (SPR) binding assays corroborated molecular dynamics (MD) simulations and confirmed that these HS-mimetic oligomers bind protein fibroblast growth factor 2 (FGF2) in a sulfation-dependent manner consistent with that of the native HS. This work established a general approach to HS glycomimetics that can potentially serve as alternatives to native HS in both fundamental research and disease models.