AJ, You can use offset mirror tokens to achieve this. Pick your initial tokens for DC1N1 and DC1N2 as if they were the only nodes in your cluster. Now increment each by 1 and use them as the tokens for DC2N1 and DC2N2. This will give you a complete keyspace within each data center with even distribution between nodes.
If you want a more detailed description, there is a recipe for this titled "Calculating Ideal Initial Tokens for use with Network Topology Strategy and Random Partitioner" in the last part of the sample chapter of "Cassandra High Performance" book http://www.packtpub.com/cassandra-apache-high-performance-cookbook/book -Eric