NCBI is a good starting point. BUT BE CAREFUL. The annotation of genes in NCBI is rather poor. I would start by removing all environmental sequences (if this is applicable to you). I would then look at the top BLAST hits and try and find a sequence for which the gene has been experimentally confirmed. From experience with my work, gene annotation in NCBI or from genomes does not accurately portray a genes function. Only 40% (roughly) of E. coli genes have been experimentally confirmed. You can use IMG JGI to look at your sequence as well. This is a great database and only uses genomes from isolates. You can get gene neighbourhood maps etc to help better identify your gene.