How does PCA differ from ordinary feature selection?

A. PCA changes observations into class labels B. PCA constructs new combined features, whereas feature selection retains a subset of original features C. PCA increases the original dimension D. PCA always uses target labels, while selection never can