Recently, mobile agents use in wireless sensor networks (WSNs) for data aggregation has acquired considerable attention. In client-server, sensor nodes (SNs) collect data and transmitted back to the base station or sink and thus consuming sensor nodes energy and surcharge the network bandwidth. Software entities called mobile agents (MAs) migrate among SNs in parallel for data collection, and then send this data back to the sink. It has been proven that itinerary planning for MAs in this paradigm is an NP-hard problem. Therefore, many solutions have been proposed to solve this problem. This paper, presents a novel itinerary planning algorithm for MAs based on cluster heads (CHs). In our proposed approach, instead of planning itinerary among all SNs, we plan itinerary just among CHs. First, we group SNs in clusters based on density of SNs then select some SNs as CHs. Second, we plan itineraries for MAs just among CHs based on minimum spanning tree algorithm (MST). At the end, we dispatch an optimal number of MAs for data collection and gathering from CHs. Simulation result shows that our MA Protocol based Energy aware data aggregation (MAPE) for WSNs performs better than previous proposed approaches.