ELISA normalization for relative quantification - (Nov/17/2012 )

Hi all,

I am a newbie in elisa and I need some urgent help.

I have an ELISA kit that does not have any standards so I am assuming it is for relative quantification instead of finding absolute values.

I need to find the extent of phosphorylation of psmad in the lysates or some kind of fold expression with respect to positive controls.

I was wondering can I use an unphosphorylated antibody of psmad as positive control, find its O.D. Then can I normalize the O.D levels of my sample lysates with respect to the O.D of the positive control to get fold change? Does this normalized intensity seem right?


I also have the total protein content of my lysates from a BCA Assay. So if the above is wrong, can I normalize the O.D values to the total protein content? For this do I simply divide the O.D values by the total protein value? (I have in ug/ml)

Is this right?

Normalizing the amount of phosphorylated psmad to toal psmad seems reasonable but of course you will want to validate the ELISA assay before using it for experimental samples. This would require having known quantities of both psmad and phosphorylated psmad. Can you buy these as standards to validate the assay? Total protein could probably also be used but you will want to make sure to validate it as well. I've also seen people use cell count as way to normalize ELISAs.

Hi thank you for your reply. I found the total protein value using a BCA assay (not total psmad) cos I read in some papers that you can use BCA assay too. I cannot find standards anywhere for this that is why I thought I can use fold change expression or something like that.

Does comparing the O.D values sound reasonable to you? For eg, dividing the O.D value I got from the ELISA to O.D value of total protein to get a fold change?

Without defining the relationship between your protein concentration and response (which is what a standard curve does) any measurements will simply be relative, and as the response/protein concentration relationship is likely to vary from assay to assay, a two fold change in response in one assay may not represent the same fold response of protein concentration with a 2 fold change in response from another assay (for example).

If you have a positive control which contains something very closely related to the analyte that you are trying to assess, then a standard curve can be created by dilution of this positive control (in an appropriate diluent). If you use total protein as a measure, then this is OK, but you will have to express your results as relative to micrograms of lysate and you don't know how much specific protein this relates to. As the amount of analyte is likely to change from lysate preparation to preparation it may be useful to bulk prepare one lystate and keep enough aliquots as cold as you can to get you through your experiments. You will have to hope that the lysate stays stable over time.

The most important thing to define is the type of data that you require from your assay. Relative protein concentrations within single experiments? Fully quantitative data translatable across multiple experiments and assays? Knowing this will determine the design of the assay (critical reagents and standard curve) and the extent of validation work that is required.

Don't hesitate to re-post with more details for advice on what is needed for your fit-for-purpose assay.

in response to your second post, ratios of ODs does not sound robust or appropriate. Prepare a standard curve somehow which should span the range of response (maximum to minimum OD values or the dynamic range) and back calculate your sample results against this curve. You can do this on graph paper or use curve fitting software to do it. The curve should be prepared in a matrix (diluent) that is fundamentally the same as the diluent for your samples.

Hi Ben, I just need a relative expression. I have 2 groups treated and untreated. Therefore I was hoping hoping I could compare the O.Ds normalized to total protein content and find the fold change between treated and untreated?

drockhere on Tue Nov 20 01:48:04 2012 said:

you can't unless you know that the change in od is linear in the range in which you are working and to which you are calculating. and you have to take the rate of change in od into cosideration.

your best bet would be to evaluate the sample and then normalize.