With various consumer electronics deployed in Internet of Vehicles (IoV), human-centric consumer in-vehicle applications (e.g., driver assistance, path planning, and healthcare system) can supply high-quality driving experience and enhance travel safety within a short time. In addition, Unmanned Aerial Vehicles (UAV) are expected to be critical to assist terrestrial vehicular networks in delivering delay-sensitive contents of services. However, due to the mutual coupling of trajectory planning of UAVs, serving the same task requests repeatedly in the same area results in wasted resources. Hence, it is challenging to supply high-quality services while ensuring energy-efficient content caching. To solve this dilemma, a content Caching scheme with Trajectory design through differential evolution and Deep Reinforcement learning (CTDR) is introduced. Specifically, a content caching scheme based on differential evolution (DE) is first proposed. Next, a trajectory design optimization based on multi-agent proximal policy optimization (MAPPO) is designed to minimize system energy consumption. Eventually, the superiority of CTDR is demonstrated through various simulated experiments.