- Top of page
- Results and Discussion
Quantitative studies in molecular and structural biology generally require accurate and precise determination of protein concentrations, preferably via a method that is both quick and straightforward to perform. The measurement of ultraviolet absorbance at 280 nm has proven especially useful, since the molar absorptivity (extinction coefficient) at 280 nm can be predicted directly from a protein sequence. This method, however, is only applicable to proteins that contain tryptophan or tyrosine residues. Absorbance at 205 nm, among other wavelengths, has been used as an alternative, although generally using absorptivity values that have to be uniquely calibrated for each protein, or otherwise only roughly estimated. Here, we propose and validate a method for predicting the molar absorptivity of a protein or peptide at 205 nm directly from its amino acid sequence, allowing one to accurately determine the concentrations of proteins that do not contain tyrosine or tryptophan residues. This method is simple to implement, requires no calibration, and should be suitable for a wide range of proteins and peptides.
- Top of page
- Results and Discussion
Accurate determination of protein concentration is essential for quantitative biochemical, biophysical, molecular, and structural biology studies. A wide range of spectrophotometric methods are available for doing so, each with its own advantages, disadvantages, and specific requirements. The absorbance of ultraviolet (UV) radiation by intrinsic chromophores is one commonly used method; particularly useful is absorbance at 280 nm (A280), which offers high specificity, as it arises strictly from tryptophan and tyrosine residues (and to a small extent from disulfide bonds if present). Thus, the molar absorptivity (extinction coefficient) for a protein at 280 nm (ε280) can be accurately estimated directly from its amino acid sequence.[2-4] Absorbance at wavelengths other than 280 nm is also used less commonly, generally either in a non-sequence-specific manner or by calibrating absorbance data on a protein-by-protein basis. Calibration can be time-consuming and technically difficult, especially if it requires directly weighing lyophilized protein or peptide (which is confounded by the presence of any residual salt or water remaining in the dried protein). Other methods (e.g., Bradford, Lowry assays) that do not rely on intrinsic chromophores also require calibration and have the added disadvantage of requiring additional reagents and time. Thus, the ability to predict the molar absorptivity (e.g., at 280 nm) is advantageous in terms of both accuracy and efficiency.
If a protein contains no tyrosine or tryptophan residues, however, A280 cannot be used to determine the concentration of that protein in solution. One alternative is A205, which arises primarily from the peptide bond.[5, 6] Although the maximum absorbance of a protein actually occurs closer to 190 nm, A205 has been favored in part due to the technical limitations of measuring at lower wavelengths.[6, 7] Still, whereas most common buffers and solutions in biological research are essentially transparent at 280 nm, many solutes will exhibit some absorbance at 205 nm (although this effect is even more pronounced at lower wavelengths).[1, 5, 8] Thus, one must consider, and carefully control for, any A205 stemming from the buffer alone, as described below. On the other hand, the much higher sensitivity of A205 relative to A280 offers an additional advantage that helps counteract this. The ratio of A205 to A280 is ∼30 on average, although it varies widely from protein to protein (and is virtually infinite for a protein lacking tryptophan and tyrosine residues).
Absorbance at 205 nm arises primarily from the peptide backbone. Thus, one can roughly estimate the concentration of a protein solution (in terms of mg·mL−1) without any knowledge of the protein sequence. A commonly used absorptivity value is ε205 = 31 mL·mg−1·cm−1.[1, 5, 8] However, side chain absorbance at 205 nm is still significant; in particular, the aromatic side chains (tryptophan, phenylalanine, tyrosine, and histidine) all have a greater A205 than a single peptide bond. Considering this, Scopes proposed using the A280/A205 ratio for a given protein to estimate a more accurate ε205 (in effect taking into account tryptophan and tyrosine side chains). However, this does not include the contributions of phenylalanine, histidine, methionine, arginine, or cysteine/cystine, all of which absorb significantly at 205 nm. Thus, by taking into account the contributions of individual side chains to the overall A205, one can accurately and specifically (i.e., in a sequence-specific manner) determine the concentration of virtually any protein in solution, as we demonstrate in this report.
Results and Discussion
- Top of page
- Results and Discussion
We measured the absorbance of six proteins and two peptides at both 205 nm and 280 nm to compare the values at these two wavelengths. Calmodulin (CaM) is a 148-residue calcium-binding protein, which has a predicted molar absorptivity at 280 nm of ε280 = 2980 M−1·cm−1 from two tyrosine residues and no tryptophan residues. Measurements were performed on both apo CaM and CaM saturated with four calcium ions (4Ca2+-CaM). Calmodulin-dependent kinase 1 (CaMK1), is a serine/threonine kinase that is activated by calcium-loaded CaM. One CaMK1 construct used in this study corresponds to the kinase domain (residues 1–296) and has a predicted ε280 of 42,860 M−1·cm−1 from four tryptophans and 14 tyrosines. The other CaMK1 construct (residues 299–320) is a short peptide corresponding to the CaM-binding domain of CaMK1, with a predicted ε280 of 5500 M−1·cm−1 from one tryptophan. The other peptide used in this study is the 26-residue M13 peptide derived from skeletal muscle myosin light-chain kinase (skMLCK), which also has a predicted ε280 of 5500 M−1·cm−1, based on one tryptophan. The remaining four proteins measured were the B1 domain of Streptococcal protein G (GB1), the talin2 F3 domain, maltose-binding protein (MBP), and Enzyme I. These range in size from 66 to 575 residues in length, and have predicted ε280 values ranging from 9970 M−1·cm−1 to 66,350 M−1·cm−1, stemming from multiple tryptophan and tyrosine residues.
For greatest accuracy, a wide range of half-log dilutions was measured, though in everyday practice such extensive measurement would not be necessary or even warranted. All dilutions were performed in water. The stocks of Enzyme I and the two short peptides were already in water, but the stocks of the other proteins used in this study were in other buffers, some with significant absorbance at 205 nm. In these cases, dilutions were also made of the buffer alone, so that its absorbance could be subtracted from that of the protein solution. Thus, although the buffers used for CaMK11–296 and 4Ca2+-CaM, for example, had undiluted A205 values of 23.74 and 27.45, respectively, we were still able to accurately and precisely measure the A205 of CaMK11–296 (uncorrected = 678.2, corrected = 654.5) and CaM (uncorrected = 665.6, corrected = 638.2) using water dilutions in the range of 1:1000 to 1:10,000 (Table 1). The buffer used for apo CaM also had a relatively high absorbance of A205 = 31.21, but the other buffers had much lower absorbances (0.59–4.37). Regardless, buffer absorbance corrections were applied in all of these cases. However, no such corrections were needed for skMLCK, CaMK1299–320, or Enzyme I, as these stock solutions were already in water. Although one could instead ensure that solutes free of absorbance at 205 nm are used, our results show that this is unnecessary. In practice (as done here) one would ideally measure A205 values much greater for the protein than for the buffer solution alone, but at the very least, the difference between the two measurements must be significantly larger than the experimental error.
Table 1. Analysis of Data Collected for This Study
|Protein||MW (g·mol−1)a||ε280 (M−1·cm−1)b|| ||A280c||A205||Conc. (µM)d||εbb (M−1·cm−1)e|
From the data collected at 205 nm and 280 nm, and under the assumption that the molar absorptivities calculated for 280 nm are correct, we estimated molar absorptivities at 205 nm for each of the polypeptides in this study, ranging from ε205 = 96,705 M−1·cm−1 (37.21 mL·mg−1·cm−1) for CaMK1299–320 to ε205 = 1,918,988 M−1·cm−1 (30.24 mL·mg−1·cm−1) for Enzyme I (Table 2). Values of ε205 could be calibrated for any protein or peptide containing tryptophan or tyrosine in a similar manner, or using some other method for determining the concentration (e.g., directly weighing lyophilized powder) for those that do not. However, in this study we demonstrate that one can calculate the ε205 directly from the amino acid sequence, making such additional measurements or standards completely unnecessary.
Table 2. Comparison of Different Methods of Calculating Molar Absorptivity at 205 nm (ɛ205)
|A280a||31 mL·mg−1·cm−1 b||Scopes methodc||New methodd|
|% Errorg|| ||−16.68||−9.09||1.66|
|% Errorg|| ||−18.96||−14.20||−1.28|
|% Errorg|| ||−15.13||−13.71||−8.17|
|% Errorg|| ||−10.40||−7.25||−2.27|
|apo CaM||ε205 (M−1·cm−1)e||514,838||517,889||462,669||524,190|
|% Errorg|| ||0.59||−10.13||1.82|
|% Errorg|| ||5.08||−6.02||6.35|
|% Errorg|| ||−6.69||−4.80||3.96|
|% Errorg|| ||−16.43||−12.97||−3.27|
|Enzyme I||ε205 (M−1·cm−1)e||1,918,988||1,967,462||1,810,473||1,973,840|
|% Errorg|| ||2.53||−5.65||2.86|
A great advantage of using A205 for determining protein concentration is that absorbance at 205 nm arises primarily from the peptide backbone, although side chains, and in particular aromatic ones, can contribute significantly. To take this into account, we referenced molar absorptivities at 205 nm given by Goldfarb et al. and Saidel et al. These reported molar absorptivities for individual amino acids ranged from 54 M−1·cm−1 (glycine) to 20,400 M−1·cm−1, though only those with ε205 > 200 M−1·cm−1 were included in this study (Table 3). The literature, however, gives a wide range of values for the ε205 of the peptide bond, and Goldfarb et al. suggested a range of 2500–2800 M−1·cm−1 per peptide bond. This likely represents true variation stemming from local structure, rather than experimental error.[7, 13]
Table 3. Molar Absorptivity Values at 205 nm (ɛ205) Used in This Study for Protein Side Chains and the Backbone Peptide Bond
|Side chain/feature||ε205 (M−1·cm−1)|
|Backbone peptide bondc||2780 ± 168|
Due to the above uncertainty, we asked whether we could find an averaged value of ε205 for the peptide bond (εbb) that would be generally applicable, keeping in mind that in reality the actual value per peptide bond is probably quite variable. From our data on eight different polypeptides, we found an overall best-fit εbb of 2780 ± 168 M−1·cm−1, which is within the previously reported range. The values calculated for individual peptides ranged from 2567 to 3112 M−1·cm−1 (Table 1). Using the average value of εbb = 2780 M−1·cm−1, we calculated a value of ε205 for each polypeptide in this study by summing all of the backbone and side chain contributions, as described in the methods. These values are presented in the rightmost column of Table 1. Compared to the concentrations determined from A280, the ε205 values based on εbb = 2780 M−1·cm−1 give errors in concentration of at most 8.2% (GB1), though for all but two (GB1 and 4Ca2+-CaM) the error is less than 4% (Table 2) and well within experimental error (Table 1). If, however, one uses a non-sequence-specific estimate for ε205 of 31 mL·mg−1·cm−1, the errors are in general much larger (up to 19.0%). Even using the formula put forth by Scopes, these errors are only slightly reduced to a maximum of 14.2% (Table 2). Only for 4Ca2+-CaM do both of the generic methods give lower errors than the new sequence-specific method, though the differences in error between these methods (6.4% vs. 5.1% vs. 6.0%) are negligible. For apo CaM and Enzyme I, the generic 31 mL·mg−1·cm−1 value gives a slightly lower error than the sequence-specific value, but the difference in errors (1.8% vs. 0.59% for apo CaM, 2.9% vs. 2.5% for Enzyme I) are miniscule and well within the experimental error. In all other cases, the sequence-specific value of ε205 gives the lowest error, often dramatically so.
Based on the data presented here, this method works well for both folded proteins and short unstructured peptides. Although the majority of the A205 in all cases comes from the peptide backbone, the actual contribution for these test cases ranges from 59.4% (CaMK1299–320) to 80.8% (Enzyme I). Thus, one can see why calculating sequence-specific molar absorptivities would be advantageous, since side chains can be responsible for up to 40% of the total A205 in these instances. Additionally, measuring data on both apo CaM and 4Ca2+-CaM offered the opportunity to test whether structural changes within the same protein affect A205. Although these two conditions gave slightly different results (εbb = 2716 ± 324 M−1·cm−1 for apo CaM and 2567 ± 199 M−1·cm−1 for 4Ca2+-CaM), these differences were within the experimental error.
To test the utility of the A205 method for determining protein concentration using sequence-specific calculated molar absorptivities, we performed a concentration-dependent experiment on a protein construct that does not absorb at 280 nm. The N-terminal domain of CaM (CaM1–76) does not contain any 280 nm-absorbing residues (both tyrosines are located in the C-terminal domain). Using the method described here, CaM1–76 has a calculated ε205 of 266,150 M−1·cm−1. We determined the concentrations of CaM1–76 and skMLCK by A205 and then performed a fluorescence anisotropy binding experiment. This titration experiment yielded a CaM1–76:skMLCK stoichiometry of 2.13(±0.17):1 (Fig. 1), well within the experimental error of the expected value of 2:1.14 (Note that whereas the stoichiometry of the full-length CaM:skMLCK interaction is 1:1—with the N and C domains of CaM participating in the interaction11—in isolation, individual domains of CaM bind with a 2:1 stoichiometry, because a second CaM1–76 molecule, in this case, takes the place of the absent C-terminal CaM domain in the complex.)
Figure 1. Testing the stoichiometry of 2Ca2+-CaM1–76 binding to the skMLCK M13 peptide. Fluorescence anisotropy was measured for the tryptophan of skMLCK alone (10 µM) and in the presence of 0–50 µM CaM1–76. Experimental data (average of three measurements) are plotted as filled-in squares, with error bars indicating one standard deviation. The best-fit line is shown in red. The CaM1–76:skMLCK stoichiometry from the fit of the data is n = 2.13 ± 0.17, which is in excellent agreement with the literature value of 2:1.14 The titration was performed with [skMLCK] >> KD to most accurately determine the stoichiometry. The effective KD from the fit is 100 ± 28 nM, several orders of magnitude weaker than the value determined for full-length CaM (∼50 pM).20
Download figure to PowerPoint
In conclusion, we have demonstrated the utility of a method for determining protein concentrations from absorption at 205 nm, using a molar absorptivity calculated specifically from the amino acid sequence. This method allows one to easily and quickly determine protein concentration when there are no tryptophan or tyrosine residues present. It is generally applicable, requiring no additional standards or other calibration, and we have made this tool available on the web (at http://spin.niddk.nih.gov/clore) to assist in calculating ε205 for any protein or peptide. We have also shown that taking the amino acid sequence into account is essential for accurate results, and we believe that this method will be widely applicable.