Résumé : Social media-transmitted online information, which is associated with emotional expressions, shapes our thoughts and actions. In this study, we incorporate social network theories and analyses and use a computational approach to investigate how emotional expressions, particularly \textit{negative discrete emotional expressions} (i.e., anxiety, sadness, anger, and disgust), lead to differential diffusion of online content in social media networks. We rigorously quantify diffusion cascades' structural properties (i.e., size, depth, maximum breadth, and structural virality) and analyze the individual characteristics (i.e., age, gender, and network degree) and social ties (i.e., strong and weak) involved in the cascading process. In our sample, more than six million unique individuals transmitted 387,486 randomly selected articles in a massive-scale online social network, WeChat. We detect the expression of discrete emotions embedded in these articles, using a newly generated domain-specific and up-to-date emotion lexicon. We apply a partial-linear instrumental variable approach with a double machine learning framework to causally identify the impact of the negative discrete emotions on online content diffusion. We find that articles with more expressions of anxiety spread to a larger number of individuals and diffuse more deeply, broadly, and virally. Expressions of anger and sadness, however, reduce cascades' size and maximum breadth. We further show that the articles with different degrees of negative emotional expressions tend to spread differently based on individual characteristics and social ties. Our results shed light on content marketing and regulation, utilizing negative emotional expressions.