Optimizing reduction of Western blotting analytical variations: use of replicate test samples, multiple normalization methods, and sample loading positions

Abstract

AbstractWestern blot (WB) analysis is widely used, but obtaining consistent results can be problematic, particularly if all the treatments cannot be placed on one gel. Sources of variation were investigated by analyzing a set of samples which should produce statistically identical results, an approach commonly applied to analytical instruments. Test samples were lysates from RAW 264.7 murine macrophages treated with LPS to activate MAPK and NF-kB signaling targets. The resulting WB were probed for p-ERK, ERK, IkBβ and non-target protein levels. The first test set (Set 1) was a pooled cell lysate and the second (Set 2) four independently prepared lysates.Analytical replicate data were generated by using multiple WB. Different normalization methods and sample groupings were applied to the density values and the resulting coefficients of variation (CV) and ratios of maximal to minimal values (Max/Min) were compared. Ideally, as the samples were identical (Set 1), or nearly so (Set 2), CVs would be 0 and the Max/Min 1. Results showed the raw density data variance was composed of variability between gel lanes and between gels; but only between gels after applying normalization calculations. CV and Max/Min were 30-70% and 1.5 for raw data, respectively. Common normalization methods, total lane protein, % Control, and p-ERK/ERK ratios, did not have the lowest CVs or Max/Min values. Some of these methods were counterproductive, greatly increasing CV and Max/Min. The best normalization methods resulted in CV and Max/Min value, as low as 5-10% and 1.1, respectively, but more typically 20-25% and 1.2. Application of these methods should allow reliable interpretation of complex experiments that require samples to be placed on multiple gels to provide results, even if fold changes in the target proteins are relatively modest.