If you are sure that serine is phosphorylated and locates on the surface of your protein, replace it with a leucine shouldn't be a problem. Otherwise, mutate it to alanine is better than a leucine since the larger side chain of leucine might alter the structure of your protein and as a consequence affect the protein function.

I concur with Huanting Liu. Ser to Ala mutation would be better approach and the reviewers will be fine with that when you submit that work for publication. You may also also create Ser to Asp (S to D) mutant which is a phospho-mimicking mutant and might give you a good phenotype (if you get any!).