The paper proposes a framework for building ontology-aware learning object (LO) content. Previously ontologies were exclusively employed for enriching LOs' metadata. Although such an approach is useful, as it improves retrieval of relevant LOs from LO repositories, it does not enable one to reuse components of a LO, nor to incorporate an explicit specification of domain semantics into the LO content. We propose the use of domain ontologies to annotate LO content as well as content structure ontologies to enable direct access to LOs' components. That way, the same LO can be used in different ways and by different users, that is, it can be repurposed. In order to show the benefits of our proposal we discuss its application in adaptive learning systems. We also explore Semantic Web technologies and tools that are needed to support the presented approach.