Mobile cloud computing enables the offloading of computationally heavy applications, such as for gaming, object recognition or video processing, from mobile users (MUs) to a cloud server connected to wireless access points. The optimization of the operation of a mobile cloud computing system amounts to the problem of minimizing the energy required for offloading across all MUs under latency constraints at the application layer. In a scenario with multiple MUs transmitting over a shared wireless medium across multiple cells, this problem requires the management of interference for both the uplink, through which MUs offload the data needed for computation in the cloud, and for the downlink, through which the outcome of the cloud computation are fed back to the MUs, as well as the allocation of backhaul resources for communication between wireless edge and cloud and of computing resources at the cloud. In this paper, this problem is formulated for general multi-antenna, or MIMO, channels, and tackled by means of successive convex approximation methods. The numerical results illustrate the advantages of a joint allocation of computing and communication resources.