Protein Assay Data Analysis
With most protein assays, sample protein concentrations are determined by comparing their assay responses to that of a dilution-series of standards whose concentrations are known. Protein samples and standards are processed in the same manner by mixing them with assay reagent and using a spectrophotometer to measure the absorbances. The responses of the standards are used to plot or calculate a standard curve. Absorbance values of unknown samples are then interpolated onto the plot or formula for the standard curve to determine their concentrations.
This comparative method for determining the concentration of an "unknown" is conceptually simple and straightforward. However, its implementation in an assay protocol is complicated by pipetting and dilution steps, evaluation of replicates, blank-corrections and other factors. These steps frequently cause confusion with regard to the calculations that are necessary to obtain a final determination.
Principles of standard curve assays
Sample assay responses are directly comparable to each other if they are processed in exactly the same manner. Variation in amount of protein is the only possible cause for differences in final absorbance (color intensity) if all three of the follow conditions are met:
- Samples are dissolved in the same buffer
- The same lot and stock of assay reagent is used for all samples
- All samples are mixed and incubated at the same time and temperature
- No pipetting errors are introduced
Of course, because of differences in the chemistry of protein assay methods, different proteins will generate different absorbance values even at the same concentration. This is called "protein-protein variation" or "protein uniformity" and is discussed more fully in other protein methods articles.
The unit of measure used to express the standards is by definition the same unit of measure associated with the calculated value for the unknown sample (i.e., final results for unknown samples will be expressed in the same unit of measure as was used for the standards). For example, if the standards are expressed as micrograms per milliliter (µg/mL), then the value for the unknown sample, which is determined by comparison to the standard curve, is also expressed as micrograms per milliliter.
Contrary to what many people assume, it is neither necessary nor even helpful to know the actual amount (e.g., micrograms) of protein applied to each well or cuvette of the assay. Consider a simple example in which the Coomassie Plus Protein Assay Kit (Part No. 23236) is used to assay two protein samples: a test sample whose concentration is not known, and a standard whose concentration is 1mg/mL (= 1000 µg/mL). (Ordinarily, an entire set of standards is necessary to establish a response curve, but this is a simplified example.)
In the microplate protocol (see figure), one adds 10µL of sample (test or standard) and 300µL of assay reagent per well. Because 10µL of the standard sample is added to a well, there is 0.010mL x 1,000 µg/mL = 10µg of protein in the well. If the assay results in the test sample having the same final absorbance as the standard sample, then the conclusion is that the test sample contains the same amount of protein as the standard sample. Because there was 10µg of standard per well, one could report the determined concentration of test sample as 10µg/well. However, the amount of protein per well is almost certainly not the value of interest; instead, one usually wants to know the protein concentration of the original test sample. Because the original standard was 1000µg/mL, the test sample that produced the same absorbance in the assay also must be 1000µg/mL.
Furthermore, it is neither necessary nor helpful to know the protein concentration as it exists when diluted in assay reagent. In the above example, because the 10 µg standard was diluted to 310µL after adding 300µL of assay reagent, the final concentration in the well is 10µg/310µL = 0.0323µg/µL = 32.3µg/mL. Therefore, one could report the determined concentration of test sample as 32.3µg/mL. However, the protein concentration when diluted by assay reagent is almost certainly not the value of interest; instead, one wants to know the protein concentration of the original test sample. Because the original standard was 1000µg/mL, the test sample that produced the same assay absorbance also must be 1000µg/mL.
One situation in which the dilution factor is important to consider is when the original sample has been pre-diluted relative to the standard sample. Continuing with the same example, suppose that the original protein sample is actually known to be approximately 5mg/mL. This is too concentrated to be assayed by the Coomassie Plus Protein Assay Kit, whose assay range in the standard microplate protocol is 100-1500µg/mL. However, one could dilute it 5-fold in buffer (i.e., 1 part sample plus 4 parts buffer) and then use that diluted sample as the test sample in the protein assay. If the test sample produces the same absorbance as the 1000µg/mL standard sample, then one can conclude that the test (5-fold diluted) sample is 1000µg/mL, and therefore the original (undiluted) sample is 5 x 1000µg/mL = 5000µg/mL = 5mg/mL.
- Tech Tip #57: How to use a protein assay standard curve
Unlike the example described above, most assays use an entire set of protein standards whose concentrations span the effective assay range (e.g., 100-1500µg/mL). Rarely, if ever, will the test sample produce an assay response that corresponds exactly to one of the specific standard samples. Therefore, a method is needed to calculate or interpolate between the standard sample points.
Most plate readers and spectrophotometers have associated software that automatically plots a best-fit (linear or curvilinear) regression line through the standard points, interpolates the test samples on that regression line, and reports the calculated value. It is important to choose an appropropriate curve-fitting algorithm because the mathematical formula describing the fitted curve will be used to calculate the concentration of the test sample.
The following figures illustrate how different curve-fitting algorithms affect the accuracy of protein assay calculations. Few, if any, protein assays are perfectly linear over the entire useful assay range. As long as the appropriate curve-fit is used, an assay does not need to be linear to be accurate. If curve-fitting must be done manually, a point-to-point fit will usually be more accurate than a linear fit to the entire range of standard points (see figure). A "point-to-point" fit is a linear fit between each successive pair of points.
- Tech Tip #57: How to use a protein assay standard curve (This Tech Tip describes curve-fitting in greater detail. It includes instructions for using Microsoft Excel Software to calculate a best-fit, 3- or 4-parameter trendline for calculating protein concentrations.)
Several factors affect protein assay accuracy and precision. The only way to evaluate the extent of random error is to include replicates of each standard and test sample. Because all test samples are evaluated by comparison to the standard curve, it is especially important to run the standards in triplicate. The standard deviation (SD) and coefficient of variation (CV) can then be calculated, providing a degree of confidence in the technician's pipetting precision. If replicates are used, curve-fitting is done on the average value (minus obvious outliers).
Published graphs of standard curves (as in the figures on this page) usually show the line going through the origin (0,0). Although visually appealing, this is irrelevant to the calculations. Most protein assay working reagents have absorptivity at the detection wavelength (i.e., they have positive absorbance even when there is no protein present, see figures below).
It is common practice to subtract the absorbance of the zero assay standard(s) from the all other sample absorbance values. However, if replicate zero-assay standards will be used to calculate error statistics, then another independent value may be required for blank-correction. If the standards were prepared in a buffer to match that of the test samples, and this buffer contains components that may interfere with the assay chemistry, it is informative to blank the absorbances with a "water reference" (i.e., a zero-protein, water sample). Differences between the water reference and zero standard sample are then indicative of buffer effects.
The standard curve slope is directly related to assay accuracy and sensitivity. All else being equal, the steepest part of the curve is the most reliable. For most protein assays, the standard curve is steepest (i.e., has the greatest positive slope) in the bottom half of the assay range. In fact, the upper limit of an assay range is determined by the point at which the slope approaches zero; the line there is so flat that even a tiny difference in measured absorbance translates to a large difference in calculated concentration.
However, the very lowest part of the assay range, while having the steepest slope, is not always optimal because random errors and interfering substances have greater relative effects in samples containing very small amounts of protein. Therefore, the clearest results are usually obtained with test samples that are pre-diluted so that they correspond to the lower-middle portion of the assay range. Many technicians test samples at two or three dilutions to ensure that at least one of them "lands" in this part of the assay range.
The measurement wavelengths that are recommended for each protein assay method are optimal because they yield standard curves with maximal slope. This usually, but not always, corresponds to the absorbance maximum. (In certain circumstances, other considerations are also important in choosing the best possible measurement wavelength, such as avoiding interference from sample components that absorb at similar wavelengths).
In fact, for most protein assays, depending on the precision required, acceptable results can be obtained using any measurement wavelengths within a certain range. The following figures illustrate this point (see Tech Tip #25 for details).
- Tech Tip #25: Determine acceptable wavelengths for measuring protein assays (This document describes several protein assay absorbance spectra in greater detail.)
The possible effects of interfering substances were not discussed in this article because the assumption was that all protein samples were treated exactly the same, including the buffers in which the proteins were dissolved. In this situation, any interference caused by components of the buffers is exactly the same for both test and standard samples. Nevertheless, interference by non-protein substances in the samples that block or contribute to the assay color reaction is an important issue for any protein assay system. Refer to related articles and documents for further discussion this topic.
For Research Use Only. Not for use in diagnostic procedures.