Protein kinases are clinically relevant, attractive drug targets for cancer. One major problem with kinase inhibitors is broad promiscuity, causing off-target actions and side effects. In silico prediction of targets of a compound would immensely facilitate and accelerate drug development. Using a virtual "inverse" screening approach, where single compounds are docked into protein structures from a database, we identify among known targets of indirubin derivatives phosphoinositide-dependent kinase 1 (PDK1) as a target of one derivative (6BIO) in particular. This prediction is functionally supported by an in vitro kinase assay, inhibition of intracellular phosphorylation of PDK1-substrates, and inhibition of endothelial cell migration, which highly depends on PDK1. Virtual inverse screening combined with biological tests, thus, is proposed as a valuable tool for the drug discovery process and re-examination of already established kinase inhibitors.