Despite its increasing popularity the widespread adoption of product line engineering is still hampered by a lack of flexible and extensible approaches that can be tailored to deal with diverse organizational specifics such as architectural styles, languages, or modeling notations. Many existing product line approaches focus on process aspects and provide general-purpose modeling approaches. In this paper we present a flexible and extensible variability modeling approach that can be adapted to domain-specific needs. The approach is supported by the meta-tool DecisionKing. The tool treats variability as a prime modeling concept and supports the domain-specific definition of dependencies between model elements. We demonstrate the feasibility of our approach with two case studies in the areas of industrial automation and service-oriented systems.