The challenging problem of energy efficiency in security of the Internet of Things (IoT) is tackled in this article. The authors consider the upcoming generation of mobile networks, 5G, as a communication architecture for the IoT. The concept of adaptive security is adopted, which is based on adjusting the security level as per the changing context. It has the potential of reducing energy consumption by adapting security rather than always considering the worst case, which is energy consuming. The consideration of 5G introduces new dynamics that can be exploited to perform more adaptation. The proposed solution introduces an intelligence in the application of security, from the establishment phase to the use phase (end-to-end). The security level related to the used cryptographic algorithm/key is adapted for each node during the establishment phase, so to match with the duration of the provided services. A new strategy is formulated that considers both IoT and 5G characteristics. In addition, a solution based on the framework of the coalitional game is proposed in order to associate the deployed objects with the optimized security levels. Moreover, the application of security is also adapted during the use phase according to the threat level. Trust management is used to evaluate the threat level among the network nodes, while existing works focus on performing the adaptation during the use phase. The proposed approach achieves more adaptation through the consideration of both IoT and 5G dynamics. The analysis and performance evaluations are conducted to show the effectiveness of the proposed end-to-end approach.