I am testing the drug effect on expression of a particular protein in mouse brain. We need to have three test groups- one without drug treatment and others with two different concentrations of drug.
It is hard to say without more information. In general 10 animals is a good suggestion, but you may want more if your behavioral effects is likely to be mild, or if your molecular analysis is likely to be highly variable. For example, samples of 8-10 often work for our own behavioral data, but the gene expression data from brain tissue samples can be highly variable, so we use sample sizes of 14 to address that.