Technically, if you are only interested in a single target of interest i.e. Does the protein abundance of my gene of interest change following some variable (e.g. treatment with some drug, or between different cell types), then you should look at that specific target protein.

Then, depending on what method of protein loading normalization you want to use - you would then either blot for a housekeeper protein (Tubulin, GAPDH, Actin etc), or look at total protein by some kind of protein stain e.g. ponceau-S across all samples. 

It depends entirely on what question you are trying to answer and what your experimental design is. Hopefully you will already have some idea what targets you expect to see changes in, these are clearly the ones to start with.

I hope that helps!