3. The dye does not bind to free arginine or lysine, or to peptides smaller than about 3000 Da (4,11). Many peptide hormones and other important bioactive peptides fall into the latter category, and the Bradford assay is not suitable for quantifying the amounts of such compounds.
4. The assay technique described here is subject to variation in sensitivity between individual proteins (see Table 2) . Several modifications have been suggested that reduce this variability (5–7,12). Generally, these rely on increasing either the dye content or the pH of the solution. In one variation, adjusting the pH by adding NaOH to the reagent improves the sensitivity of the assay and greatly reduces the variation observed with different proteins (7). (This is presumably caused by an increase the proportion of free dye in the blue form, the ionic species that reacts with protein.) However, the optimum pH is critically dependent on the source and concentration of the dye (see Note 5). Moreover, the modified assay is far more sensitive to interference from detergents in the sample.
Particular care should be taken when measuring the protein content of membrane fractions. The conventional assay consistently underestimates the amount of protein in membrane-rich samples. Pretreatment of the samples with membrane-disrupting agents such as NaOH or detergents may reduce this problem, but the results should be treated with caution (13). A useful alternative is to precipitate protein from the sample using calcium phosphate and remove contaminating lipids (and other interfering substances, see Note 1) by washing with 80% ethanol (9,10).
5. The amount of soluble dye in Coomassie Blue G250 varies considerably between sources, and suppliers’ figures for dye purity are not a reliable estimate of the Coomassie Blue G250 content (14). Generally, Serva Blue G is regarded to have the greatest dye content