Video documents provide a great support for e-learning, and in some cases even constitute the core of the proposed contents. Yet video remains a particular medium that lacks properties that can be found more easily in texts, such as structure or meaningful indexing. This limits the way they can be manipulated by both learners and teachers. One way of overtaking these limitations consists in defining, managing and harnessing video annotations. In this talk we first define what precisely are video annotations, before highlighting various ways in which they can be used to enhance elearning-related processes, be it for appropriation (active learning, sharing), collaboration, course improvement, course retrieval and building, etc. We provide several illustrations from current systems, and then address some technological issues related to models and platforms to manage and use annotations. Lastly, we present the recently launched CominOpenCourseware (COCo) project and the work we intend to do.