The development and widespread adoption of connected and autonomous vehicle (CAV) technology are expected to generate a broader range of vehicle usage scenarios while offering new opportunities to mitigate imbalanced travel demand. It can be explored and implemented by traffic managers, based on the existing road network, so that flexible lane management strategies can be adopted to significantly enhance the utilization of roadway resources and the overall operational efficiency of the transportation network. Therefore, a reversible lane layout optimization problem for CAV-enabled networks was proposed, and a corresponding mathematical programming model was developed. CAVs were first classified into two categories based on service target: individual CAVs (ICAVs) for private users and public CAVs (PCAVs) for shared mobility services. A mixed multi-user equilibrium model was established to characterize their distinct route choice behaviors. Second, a system-optimal reversible lane layout model was formulated from the perspective of traffic managers by embedding the equilibrium constraints. The model was solved using an improved genetic algorithm. Finally, the proposed framework was validated on the Sioux-Falls network, and sensitivity analyses were conducted with respect to the PCAV penetration rate and various origin-destination (OD) demand levels. The results indicate that reversible lanes can substantially improve network performance in CAV environments, achieving a 66.87% reduction in total travel time under the saturated OD demand conditions.