In this paper, we propose a novel user cooperation framework aimed at optimizing power efficiency in MIMO-OFDM uplink, where cooperation is directed towards minimizing the transmit signal PAPR at all cooperating users. We consider different degrees of cooperation: 1) joint data transmission with PAPR reduction, 2) only joint data transmission and 3) the reference non-cooperative case based on a zero forcing receiver at the base station. In the cooperative case, we also account for the power required to exchange data between users. The main idea of our scheme is to exploit the unused space-frequency resources induced by the user cooperation to assign dummy symbols on the empty subcarriers and optimize these symbols to affect the outbound waveform. This is achieved as a result for a convex optimization problem which aims to simultaneously minimize the transmit PAPR and dummy symbol power allocation. We further introduce a rate maximization scheme, which maximizes the data power allocation for a given transmit peak power constraint by executing a one-dimensional search. The results show that our proposed PAPR minimizing bit and power allocation scheme can achieve significantly higher throughput at short and medium UE-UE distances.