A gene expression module (module for short) is a set of genes with shared expression behavior under certain experimental conditions. Discovering of modules enables us to uncover the function of uncharacterized genes or genetic networks. In recent years, several biclustering methods have been suggested to discover modules from gene expression data matrices, where a bicluster is defined as a subset of genes that exhibit a highly correlated expression pattern over a subset of conditions. Biclustering however involves combinatorial optimization in selecting the rows and columns composing modules. Hence most existing algorithms are based on heuristic or stochastic approaches and produce possibly sub-optimal solutions. In this paper, we propose a novel biclustering method, BiModule, based on a closed itemset enumeration algorithm. By exhaustive enumeration of such biclusters, it is possible to select only biclusters satisfying certain criteria such as a user-specified bicluster size, an enrichment of functional annotation terms, etc. We performed comparative experiments to existing salient biclustering methods to test the validity of biclusters extracted by BiModule using synthetic data and real expression data. We show that BiModule provides high performance compared to the other methods in extracting artificially-embedded modules as well as modules strongly related to GO annotations, protein-protein interactions and metabolic pathways.