This paper discusses the requirements for Virtual Learning Environments from a didactical and implementational point of view, and also identifies structure and typical elements. Following modern educational theories one of the most important aspects is to provide interactive elements in order to "grasp" the learning matter. In this context different interaction types are elaborated. Besides this, the paper also shows that narration, just as with (serious) games and digital storytelling, is another crucial principle that helps to provide the learner with some background information and structure for understanding the new concepts and building adequate mental models. This implies, that meaningful virtual objects and actions should be arbitrarily utilizable and combinable, in order to be able to create Virtual Learning Environments which are customized to the learning matter. Finally methods and building blocks for enabling instructors to "write" interactive 3d learning environments are presented.