Online social networks (OSNs) are changing the way information spreads throughout the Internet. A deep understanding of information spreading in OSNs leads to both social and commercial benefits. In this paper, dynamics of information spreading (e.g., how fast and widely the information spreads against time) in OSNs are characterized, and a general and accurate model based on Interactive Markov Chains (IMCs) and mean-field theory is established. This model shows tight relations between network topology and information spreading in OSNs, e.g., the information spreading ability is positively related to the heterogeneity of degree distributions whereas negatively related to the degree-degree correlations in general. Further, the model is extended to feature the time-varying user behavior and the ever-changing information popularity. By leveraging the mean-field theory, the model is able to characterize the complicated information spreading process (e.g., the dynamic patterns of information spreading) with six parameters. Extensive evaluations based on Renren's data set illustrate the accuracy of the model, e.g., it can characterize dynamic patterns of video sharing in Renren precisely and predict future spreading dynamics successfully.