Stain-Free total-protein normalization enhances the reproducibility of Western blot data

Abstract

We compared the accuracy of three common methods of total protein normalization. The Stain-Free method was accurate across different types/brands of western blotting membrane and for various protein loads, unlike Ponceau S and Amido Black. Normalizing to the housekeeping proteins Actin and β-Tubulin could match the accuracy of the Stain-Free method. However, compared to Actin or β-Tubulin, normalizing to the Stain-Free signal reduced variability that led to enhanced reproducibility and a reduction in the number of samples needed to obtain statistically significant results by >50%. Stain-Free normalization can enhance the reproducibility and hence the confidence in Western Blot data.

Keywords: Amido black; Ponceau S; Rigour and reproducibility; Stain-free; Total protein normalization; Western blot.

Fig. 1.. There are large differences in accuracy between commonly used techniques of total protein normalization that are influenced by the type and brand of Western blotting membrane and the amount of protein loaded.

(A) BEAS-2B cell lysate was loaded onto 10% Tris-Glyeine Stain-Free gels and transferred onto either PVDF or 0.2 μM/0.45 μM nitrocellulose membranes from different manufacturers. A Stain-Free image was captured, the membrane stained with Ponceau S, imaged, destained, and then stained with Amido Black and imaged again. (B) The signal in each lane was assessed using Image Lab v6.0. Data are represented as fold change from the 5 μg lane i.e., the normalization factor relative to the 5 μg load, except for Ponceau S and Amido Black on Brand 1 0.45 μM NC and Amido Black on Brand 1 PVDF where the 10 μg signal was used because the 5 μg signal was very low and variable. Accuracy was defined as the degree to which the observed normalization factor matched the expected normalization factor (dashed line). Data is from 3 technical replicates for each stain on each type of membrane. NC = nitrocellulose; PVDF: polyvinylidene difluoride.

Fig. 2.. Normalization to Actin and β-Tubulin can match the accuracy of Stain-Free total protein normalization but will less precision.

(A) Increasing amounts of BEAS-2B cell lysate were loaded onto 10% Tris-Glycine Stain-Free gels and transferred onto PVDF membranes. The Stain-Free image was captured, the membranes cut horizontally at 50KD and 40KD, and the respective pieces immunoblotted for β-Tubulin, Actin, and GAPDH. (B) Data are represented as fold change relative to the 10 μg load i.e., the normalization factor relative to the 10 μg load. Accuracy was defined as the degree to which the observed normalization factor matched the expected normalization factor (dashed line). Precision was assessed based of the size of the standard deviation.

Fig. 3.. Normalization to the Stain-Free signal improves reproducibility of Western blot data relative to normalization to Actin or β-Tubulin.

(A–C) BEAS-2B cells were transfected with dsiRNA control vs. dsiRNA targeting an E3 ligase that degrades FoxP1 protein. Lysates (20 μg) were electrophoresed using 10% Stain-Free gels and transferred onto PVDF membranes. A Stain-Free image was captured, the membranes cut horizontally at 65KD and 50KD, and the respective pieces immunoblotted for FoxP1, β-Tubulin, and Actin. The experiment was repeated 3 times, >2 weeks apart. (D) FoxP1 protein levels with standard deviations when normalized to β-Tubulin, Actin, or the Stain-Free signal. (E) The statistical significance and mean difference in FoxP1 protein levels in dsi control vs. dsi E3 ligase when normalized to β-Tubulin, Actin, or the Stain-Free signal. (F) Levels of Actin or β-Tubulin normalized to Stain-Free signal in dsi control vs. dsi E3 ligase.

Fig. 4.. Normalizing to the Stain-Free signal instead of Actin or β-Tubulin can greatly reduce the number of samples required to obtain statistically significant results.

The estimated total number of samples needed to detect a 15%, 20%, and 30% difference in FoxP1 protein levels by immunoblot using standard deviations assessed in Fig. 3E and assuming a two-sided p < 0.05 and 90% power.