Crocetin, a high-value apocarotenoid in saffron, is widely applied to the fields of food and medicine. However, the existing method of obtaining crocetin through large-scale cultivation is far from meeting the market demand. Microbial synthesis of crocetin is a potential alternative to traditional resources, and it is found that carotenoid cleavage dioxygenase (CCD)is the critical enzyme to synthesize crocetin. So, in this study, we used "hybrid-tunnel" engineering to obtain variants of Crocus sativus-derived CsCCD2, essential for zeaxanthin conversion into crocetin, with a broader substrate specificity and higher catalytic efficiency. Variants including S323A, with a lower charge bias and a larger tunnel size than the wild-type, showed a 5-fold higher crocetin titer in yeast-based fermentations. S323A could also convert the ?-carotene substrate to crocetin dialdehyde and exhibited a 12.83-fold greater catalytic efficiency (kcat/Km)toward zeaxanthin than the wild-type in vitro. This strategy enabled the production of 107 mg/L crocetin in 5 L fed-batch fermentation, higher than that previously reported. Our findings demonstrate that engineering access tunnels to expand the substrate profile by in silico protein design represents a viable strategy to refine the catalytic properties of enzymes across a range of applications.