Several journals including Journal of Biological Chemistry discourage using housekeeping proteins (beta-actin, GAPDH, histone, and etc) as a loading control for western blots, unless you can demonstrate that expression of these proteins are constant under your experimental conditions. Instead, total protein stains such as amido black, Coomassie, and Ponceau are recommended. An editorial article by JBC ( doi: 10.1074/jbc.E115.000002 ) explains this along with several papers describing methods for nomalizing western blots using total protein stains. As an editorial board member of JBC, I sometimes need to request authors to revise western blot data based on this recommendation.

Loading control in WB has been used to provide more or less correct evaluation of a given protein in different samples to be tested. The idea is based on choosing as a loading control the protein of similar expression level. In this regard, b-actin or other housekeeping proteins can be used. Amido black or coomassie blue used for coloration can be applied to detect protein loading controls.

So, you should know or suggest whether b-actin or another loading control protein is equally expressed in the cells of your interest. The choice depends also on the compartiment location (membrane, cytoplasme or nucleus) of a protein of your interest.

However, I think that using as a loading control the expression level of a total protein is more correct and precise. Look at Li-Cor how to use fluorescent total protein for reliable quantification and estimations.

Normalisation is essentially the same process in both instances: you are normalising your signal of interest to some universal standard to correct for uneven loading. In the first instance, you're using crude total protein, in the second you're using a marker protein as a representative surrogate for total protein.

All that differs is the precise conditions, and these conditions determine which is the most appropriate.

Total protein (amido/ponceau/coomassie) is cheap, easy, and can give you a general impression of overall protein load, though strictly speaking any protein assay should allow you to do this beforehand. You should generally always be loading equal amounts of protein per lane (though see caveats, below).

This method is also crude, generally low sensitivity, and (in the case of any analysis done post-hoc) hampered by signal from the large quantities of whatever generic non-specific protein you used to block your membrane.

Most importantly, it's bad for comparing samples from different tissues, or from samples where the cell populations might be variable.

Use of an internal marker protein is much more sensitive, can usually be done alongside your actual western for your signal of interest (thus is more directly comparable), and allows you to correct for your cell type or tissue of interest.

For example: imagine you are interested in a muscle protein, and interested in whether it is up- or down-regulated in a fibrotic condition. Use of a total protein marker like coomassie or amido might give you misleading results, because fibrotic muscle samples will contain lots more connective tissue and consequently less muscle tissue. Connective tissue will never express this muscle protein, but will nevertheless contribute to your 'total protein' readout, so for equal apparent protein loads, your gene of interest will appear to be down in fibrotic samples even if it is not actually down-regulated.

If you use an internal marker for muscle specifically (vinculin, alpha actinin etc), you can correct for this, and you will be able to determine whether your GOI is up or down-regulated relative to the total amount of muscle protein (specifically) that is present, regardless of the amount of total protein.

Generally speaking (in my view), an internal marker is a better method, and total protein approaches should be favoured only when better options are not available. On the other hand, done badly (i.e. using a protein that might vary between test and control samples) the internal marker approach can lead to wildly inaccurate results. Total protein methods are harder to screw up, simply because they're more crude.

There is no "one size fits all" answer, and you should tailor your approach to your specific situation.

Dear John first of all, a deep appreciation for such an informative explanation, second I am working on hHsp90 variants (phosphomimetic and non-phosphorylatable) expressed in yeast and I am addressing hHsp90 expression level in each variant and in the same blot I have detected actin as well, and now I would like to quantify the expression level in an accurate level, therefore, I got confused between the two methods of normalization. Then I think based on your clarification I should follow the second method right??

Dear Vehary Sakanyan , thanks a lot for your helpful answer, I am detecting the expression level of hHsp90 phosphomimetic and non phosphorylatable variants, I have detected actin in the same blot as well, should I use the second strategy of normalization??

If you have a good reason to assume actin expression is unaffected by Hsp90 expression, then yes.

Dear Rasha, I believe that the normalization can be done based on your β-actin expression data. Just to mention that you must be sure that there is no cell detachment (like anoikis) in some samples that appear to be processed (can be assumed from your excel date). If this happens, you must take this into account. You can find out more in our article:

Article Targeting Degradation of EGFR through the Allosteric Site Le...

Best wishes

Dear Vehary Sakanyan deep appreciation and thanks a lot