For GSEA you don't specify a set of differentially expressed genes. Instead, you specify sets of genes and then decide if those genes appear in aggregate to be differentially expressed in your samples, where 'differentially expressed in aggregate' can take at least two forms; competitive or self-contained. You can also test a whole battery of gene sets (like the Broad c1-c6 gene sets) or just a few.
If you are already using limma, then you can do a competitive GSEA for a battery of gene sets using romer() or for a few sets using camera(), and the analogous self-contained GSEA using mroast() and roast(). Or basic GSEA using geneSetTest(). See 10.GeneSetTests for more information, as well as http://bioinformatics.oxfordjournals.org/content/23/8/980.short for more information about the various types of GSEA.
The GeneSetEnrichment biocViews shows relevant packages, although James's curated list is probably more helpful:
I think you will have to map to human Entrez or HGNC gene symbols, unless you know of rat gene sets that you want to test.