I think the first important point is to clarify if one of the housekeepers is really expressed in the tissue you want to analyse. You could go on websites like the Human Protein Atlas or UniProt to check this.

On the other hand it is important to test how variable the marker is in your set of samples. So you can do a kind of trial to test several loading controls, assess the variation of the expression levels from sample to sample, and compare it with a full protein stain (e.g. SYPRO Ruby). You would definitely not go for something which has a higher variability than the protein of interest you want to normalize. So it's - to a certain extent - a trial and error approach.

What has to be considered as well is that the way you are extracting your proteins is maybe influencing the protein content. When doing e.g. a nuclear fractionation of a tissue and you are assaying this kind of samples on a WB, then you should not choose a protein which is not present in this fraction.

You can of course easily check other publications of studies analysing these tissues and get an idea what they used. In one publication (see link) they used e.g. ß-actin as a loading control, though it's varying a bit in their WBs.

A website (see second link) was suggesting a-actin as a loading control for the aortic smooth muscle tissue.

http://circ.ahajournals.org/content/119/18/2498.figures-only

http://www.labome.com/method/Loading-Controls-for-Western-Blots.html

 Actually, I tested GAPDH, β-actin, β-tubulin in the samples, but I don't know which one is suitable. Different loading control was used by previously published articles in the aortic tissue, including  GAPDH, β-actin, β-tubulin. Importantly, there is no golden standard to reference, although the protein concentration was detected by BCA kit. when I choose different loading control to calibration loading protein, the expression pattern of target protein is different. 

As I mentioned before, you could analyze the variability of the tested loading controls between the samples. The one which is the most stable through all conditions should be actually the best choice.

Have you considered using a full-protein stain like SYPRO Ruby for membranes as an alternative approach for normalization of the protein of interested?

@Jonasz Jeremiasz Weber

Thank you, SYPRO Ruby maybe a good choice! I will try soon!

Hi, I have concern about choosing the best loading control. I use three types of loading control which are 18rs, b-actin, and GAPDH. GAPDH give the most suitable findings for ppar-a and cbep-a genes expression. Though, most others paper use b-actin as loading control. Can we depend on our result not the best suit loading control that were used by others?

Thank you

@ Christian P. Moritz

Thanks! Very useful suggestions!