On studies with a high number of treatment groups and/or animals, it takes several gels to run all our samples. We always run duplicate gels, and we have noted that even between gels with the best pipetting, and using the SAME prepared antibody solution, we can get significant differences for the same sample between blots when we finally image results. (These differences are present whether we use chemiluminescence or IR on a Licor system, or chemiluminescence to film). This is not specific to an antibody or protein type. I have read of approaches such as cutting your gels and transferring all your samples to a single membrane to reduce variability in transfer and antibody binding, but this is not practical in all cases, esp. since we run mini-gels. The best concept I have heard of is to take a standard (nontreatment) loading sample, such as a whole brain (WB) lysate from a stock WT animal, and run the same protein concentration loaded for the study samples in one lane on every gel, and then normalize to that. I'd like some feedback/verification that this is an acceptable approach:
1) For loading control, normalize all lanes, including your WB lane, to Total Protein (we have moved away from using housekeeping proteins).
Formula - take highest total protein (TP) lane, divide all TP lane signals (or IDVs if you are using imageJ) by this signal to get a TP Ratio. Divide all sample signals by the TP Ratio to get Normalized Signal value.
(this all assumes that background has been removed via standard processes in the application you are using. We are currently using Li-Cor Image Studio)
2) For inter-membrane (IM) normalization, Take the highest (normalized) signal for the WB control lane, and divide all other WB control lane signal values by this to obtain IM ratio.
3) Divide all Normalized Signal Values from samples by IM ratio.
Does this sound like a reasonable/acceptable approach? Are there other methodologies that can be used? If a standard inter-membrane lane is not present are there any statistical tools that can be used to account for/remove the inter-membrane variability (which can mask any true signal changes)
Your input is greatly appreciated!
Hi Natalie, the simple truth is that Western-/Immunoblots are not made for quantification. Whatever you do or calculate, it will be semi-quantitative in the end. To quantify, use a good ELISA and many "n"s.
Hi Natalie. I don't have much experience in this field, but from what I read and heard the optimal stratagy is to load equal ammounts of total protein onto each lane and normalize the results using actin antibodies (or similar reference protein) which level doesn't change under various conditions in your experiment.
Best wishes.
Dear Natalie,
in my opinion it's a good idea to use the same untreated sample on every blot. But in the end, as Markus Klotz already stated, a Western Blot is never an accurate method for quantification since there are so many steps and possibilities to lose protein/signal.
Hi Natalie, I am using the LI-COR Odyssey System for detection and total protein staining (Revert from LI-COR) just like you. Based from what I know about Immunoblotting and its quantification, your approach is acceptable given your limitations regarding equipment etc. (I face the same problem using mini-gels).
I perform inter-membrane normalization exactly as you described above. Prior to data analysis, I always check (visually) whether there are any systematic differences between blots, both regarding control lanes and sample lanes. I actually only perform IM-normalization when I detect such systemic differences which is not always the case.
Honestly speaking, although IM normalization sounds reasonable, I am skeptical as to its merits. Variability in signal intensities do not only stem from inter-membrane variability, but also within-membrane (left-to-right, top-to-bottom) variability. Normalization to a control lane which was affected by such variability is confounding data interpretation. What do you think?
Best
Thomas
Natalie, you can normalize by averaging your total protein across blots. I did this for loading controls. We can chat about it sometime if you want.
You could use a standard that binds your secondary antibody on each blot to have an ancor point to normalize the antibody perfomance. Have a look into this paper and if you have questions, contact me
https://www.researchgate.net/publication/305394544_Normalized_Quantitative_Western_Blotting_Based_on_Standardized_Fluorescent_Labeling
Chapter Normalized Quantitative Western Blotting Based on Standardiz...
Thanks everyone for the suggestions! Kristen, the paper looks very interesting - definitely will give it a read and get back to you. Thomas, I agree that intermembrane normalization using a loading control lane is subject to issues with transfer, etc., which is why I put the question out there.
I'm not trying to get absolute quantification, but just some idea if the differences between treatment groups are significant. Unfortunately, when your membrane number is the only statistically significant factor, you know you have a problem with intermembrane variability.
Nafisa, I'll come see you about the total protein thing. :-)
Hi Natalie! I am in the same boat now. I see that the question is two years old. Did you publish any paper using this analysis strategy? It would be great to get some guidance from it. Thank you
- Badges
- Science method