Emerging virtual reality ( <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">VR</i> ) applications require high data rate transmission and low end-to-end latency, which has become one of the main challenges for future wireless networks. Unmanned aerial vehicle ( <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">UAV</i> ) mounted base stations and computing facilities can be used to provide better wireless connectivity and computing services to edge VR users to meet their computing needs and reduce the end-to-end latency. We propose a novel UAV assisted mobile edge computing ( <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">MEC</i> ) network to enable high-quality mobile 360-degree video VR applications by leveraging UAVs to provide the required communication and computing needs. Then, we formulate the joint <underline xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">UAV</u> placement, <underline xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">M</u> EC and radio resource allocation, and 360-degree <underline xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">v</u> ideo content layer assignment ( <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">UAV-MV</i> ) problem, which aims to select the allocation of computing and communications resources and the location of the UAVs such that the delivered quality of experience ( <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">QoE</i> ) is maximized across the mobile VR users, given various system constraints. We show that the problem is NP-hard, and decompose it into three lower-complexity subproblems that we solve sequentially. We design an approximation algorithm with performance guarantees that solves the UAV-MV problem based on the solutions to the three subproblems. Our simulation results show that the average QoE enabled by the proposed algorithm is 15% and 90% greater relative to two competitive reference methods.