User cooperation is an emerging transmission framework where users act as relays of each other to provide extra diversity paths for better overall performance. However, most work in the literature focused only on the physical-layer properties and operations of a fixed pair of source and destination at a particular time instant. In contrast, we consider the downlink throughput optimization for an adaptive cellular network with both user cooperation and scheduling. Firstly, we evaluate several commonly used cooperation strategies under different fairness constraints. We show that the decode-and- forward protocol cannot deliver any capacity gain over its non-cooperative counterpart when users are scheduled for maximizing the overall system throughput Secondly, the placement of relaying timeslots, which directly affects the cooperation performance, is investigated over existing frame structures so that user cooperation can be enabled easily in conventional networks. It is shown that multiple relays can actually be supported in different timeslots concurrently with other transmissions through scheduling and the corresponding optimal relays selection and power allocation are derived.