Cancer invasion is driven by complex interactions between tumor cells and tumor microenvironment (TME), which promote cancer cell plasticity and remodel the TME to support invasive behavior. In bladder cancer, invasion into the muscularis propria reduces the five-year survival rate to below 30%. Despite this clinical significance, the molecular mechanisms underlying bladder cancer invasion remain poorly understood. Current studies predominantly focus on the comparison between non-invasive and invasive tumor tissue, leaving dynamic TME variation largely unexplored. Leveraging the capabilities of spatial information, we integrated Stereo-seq spatial transcriptomics with single-nucleus RNA sequencing (snRNA-seq) on whole-layers bladder cancer specimens. These integrated datasets delineated a spatially resolved whole-layers landscape of bladder cancer at single cell resolution, elucidating the localization and function of principle cell types during the invasive process. We discovered a bladder cancer invasive leading structure characterized by EPCAM and KRT17 co-expressing, whose plasticity is induced and maintained through interactions with POSTN+ cancer associated fibroblasts (CAFs) and APOE+ macrophages. Additionally, we uncovered the specific spatial distribution of different CAF subtypes during bladder cancer progression, highlighting their roles in shaping distinct TME. Notably, we revealed a progressive increase in immunosuppressive states from superficial to muscle-invading bladder tumors. Our findings underscore the orchestrated dynamics of bladder cancer progression driven by intricate tumor-stroma interactions within the TME. These insights provide a framework for understanding invasive behavior in other muscle-invasive cancers, guiding future research into shared mechanisms of tumor progression and microenvironmental remodeling.