The paper is devoted to the selection of leader nodes in evolving directed random networks for information spreading. We consider non-homogeneous networks consisting of several connected subgraphs with different distributions of nodes characteristics. This is a plausible situation for real complex networks. The evolution of the network in time starting from a seed set of nodes is provided by the preferential attachment tool. The seed set is used to send messages to the entire developing graph. Our objectives are to compare the spreading information by the graph evolution with a well-known algorithm.