We consider one non-real-time (NRT) and N realtime (RT) cellular users interested in downloading delay-tolerant and delay-sensitive packets from the base station (BS), respectively. Each RT packet is a multicast packet and RT user i is satisfied if more than qi% of the data is received before the deadline. At each slot, one of the RT users is allowed to retransmit the data it has received from the BS to its RT neighbors. This reduces the load on the BS but causes interference to the NRT user. The problem is formulated as a joint scheduling and rate-allocation problem. We present a throughput-optimal algorithm that maximizes the NRT rate in this system under interference. Our simulations show that, in high density networks, increasing the number of RT users N in the system improves the NRT user's throughput due to the multi-user diversity effect. We also show that, in low density networks, the throughput is maximized at an optimal value of N.