Deep generative models are rapidly advancing structure-based drug design, offering substantial promise for generating ligand molecules that bind to specific protein targets. However, most current approaches assume a rigid protein binding pocket, neglecting the dynamic nature of protein structure and ligand-induced conformational changes, limiting their applicability in practical drug discovery. Here, we propose Apo2Mol, a diffusion-based generative framework for structure-based 3D molecule design that explicitly accounts for conformational flexibility in protein binding pockets. To support this, we curate a new dataset of over 24,000 experimentally resolved Apo-Holo protein structure pairs from the Protein Data Bank, enabling the modeling of protein conformational changes associated with ligand binding. Apo2Mol employs a hierarchical graph-based diffusion model that simultaneously generates 3D ligand molecules and their corresponding induced Holo pocket conformation from an input Apo state. Extensive experiments demonstrate that Apo2Mol can produce chemically valid ligands with state-of-the-art binding affinities and realistic pocket conformation changes. To our knowledge, Apo2Mol is the first open-sourced model trained on experimental Apo-Holo structure pairs to explicitly model coupled ligand-pocket dynamics, representing a crucial advance and offering a valuable resource for future research in structure-based drug design.