The design of an appropriate test suite for software testing is a challenging task. It requires a suitable tradeoff between effectiveness, e.g., a sufficient amount of test cases to satisfy the test goals of a given coverage criterion, and efficiency, e.g., a redundancy-reduced selection of test cases. Recent test suite optimization approaches, therefore, usually require an explicit enumeration of existing test cases to select from. The test suite design for covering entire software product lines is even more problematic as the dependency between test cases, test goals and product configurations has to be taken into account. Due to the exponential number of configurations w.r.t. the number of features, an explicit enumeration of all products for optimizing a product-line test suite is impractible. To tackle this problem, we propose an incremental test suite optimization approach for product-line testing that does not require an explicit representation of the set of configurations under test, but rather uses a symbolic representation in terms of feature constraints. The approach is illustrated by means of a running example.