Protein Stability Monitored by Fourier Transform Infrared Spectrometry

Degradation of proteins during storage or processing is a major concern in both manufacturing and research. A detailed structural analysis is often not required; a simple, fast, “good/bad” determination is needed. Fourier transform infrared spectrometry can make this discrimination, provided proper care in data collection is taken. This paper outlines the challenges, limitations, and potential of FTIR, and gives an example involving the temperature-induced changes in bovine serum albumin.

Protein characterization

Proteins are characterized by four levels of structure: 1) primary—the sequence of amino acids in the peptide chain, 2) secondary—organized regions within the peptide chain, 3) tertiary— folding of the organized and unorganized regions, and 4) quaternary—spatial combinations of multiple peptide chains. The amide bonds of the amino acids in a protein are all similar, but the molecular environment of each amino acid residue depends upon the local structure, in particular, the secondary structure. Amino acids in proteins participate in hydrogen bonding relationships with spatially close neighbors, forming secondary structures such as α-helices, β-sheets, β-turns, and random coils.

The bonds in molecules can be stimulated to vibrate by exposure to infrared radiation. A full description of why the secondary structure affects the IR spectrum (in infrared and Raman spectroscopy) is available elsewhere.¹ Simply stated, the molecular environment (such as hydrogen bonding) acts like the tuning knobs on a guitar, changing the frequency slightly. Many publications deal with correlating the frequencies of amino acid vibrations to the presence of various secondary structural elements.² There are limitations to this method,³ but the fact that structure affects the spectrum has been clearly demonstrated.

The primary vibration of interest in proteins is the amide I stretch (around 1650 cm^–1), which results from the concerted vibration of the O=C–N amide group. The oxygen and nitrogen both participate in the hydrogen bonding; thus the influence of secondary structure on the vibration can be pronounced. Unfortunately, both liquid and vapor phase water also have vibrations in this region, and the removal of these two interferences is the biggest challenge to the infrared analysis of proteins.

There are several mechanisms for avoiding the water peaks. The vapor can be removed either through purging the spectrometer with nitrogen or dry air, or through digital means. Physical removal by purging is preferred, since any digital processing always moves the analysis one step further away from the raw data. The liquid peak can be removed by using D₂O as a solvent (the deuterium vibration is shifted away from the amide I band), but this is both expensive and inconvenient.

Figure 1 - Infrared spectra of BSA in solution, the buffer, the difference spectrum, and a water vapor overlay. The first three are on a common scale, showing the tiny protein signals relative to the original spectra. The water vapor (not to scale) shows the extreme overlap of the vapor peaks, often misinterpreted as noise.

Figure 2 - Spectrum of Tris buffer as a function of temperature. Note the peak shifts and narrows at high temperature due to the breakdown of the hydrogen bonds.

These issues are highlighted in Figures 1 and 2. Figure 1 shows the spectrum of water vapor and the spectrum of a protein in an aqueous buffer. Also shown are the buffer spectrum and the difference spectrum (protein minus buffer), all on a common scale (offset for clarity). The protein signal is small—about 0.02 absorbance units versus the 1.5 Au of the buffer peak. The overlap of the water vapor is also apparent. Figure 2 demonstrates the effect of variable temperature on the (liquid) water peak. Protein stability studies often focus on the influence of temperature, but subtraction of a buffer spectrum taken at room temperature from a protein spectrum taken at higher or lower temperatures will result in unusable data because the water peak shifts and changes shape.

Experimental

Bovine serum albumin (BSA) (Sigma Aldrich, St. Louis, MO) was diluted in a Tris buffer (.05 M Tris, 0.05 M NaCl, 0.0005 M ethylenediaminetetraacetic acid [EDTA], pH 7.2) to yield a 4-mg/mL solution. A Nicolet™ 6700 spectrometer (Thermo Electron Corp., Madison, WI) equipped with midinfrared optics (KBr beamsplitter, deuterated triglycine sulfate [DTGS] detector) and its Proteus™ protein analysis kit temperature control accessory (Thermo Electron Corp.) were purged with dry air overnight. The sample was contained in a 6-μm pathlength cell with CaF₂ windows. The @Temp™ program (Simplex Scientific, Middleton, WI) provided with the Proteus kit was used to collect spectral data from 5 °C to 80 °C in 5 °C steps. A 5-min equilibration time followed each temperature step. The spectra resulted from 256 scans collected at 4 cm^–1 resolution. Data were processed and analyzed using the OMNIC™ spectroscopy software suite (Thermo Electron Corp.), including the Peak Resolve™ feature available in OMNIC. Information supplied in the guidebook⁴ provided with the Proteus protein analysis kit was used to define the experiment and analysis. The results were exported to Microsoft^® Excel™ (Microsoft, Redmond, WA) for final analysis and presentation.

Results and discussion

Elevated temperatures cause the breakdown of hydrogen bonds. The protein secondary structure is held together by hydrogen bonds; thus their loss can result in subtle or catastrophic changes to the protein. This, in turn, leads to a loss of protein function.

Bovine serum albumin is a transport protein for fats and other water-insoluble materials in the blood. No structure for BSA is presently available, but the protein apparently has between 54% and 68% α-helix,⁵ with the rest turns and random coil (no β-sheet). The analogous human origin protein (HSA) exhibits a heart-shaped 3- D structure with at least five “slots” for transporting fatty acids.

Figure 3 - BSA spectrum at low and high temperatures. The large change in the spectrum due to degradation of the protein is clear even without further analysis.

The spectrum of BSA at low and high temperatures is shown in Figure 3. Even without further analysis, changes are already evident. There are now a number of paths open to extract the desired information. The analysis chosen will determine whether the user requires simple qualitative information (has the protein changed?) or more detailed information about the secondary structure.

Figure 4 - Curve fit results for the high-temperature data. The peak due to α-helix makes up less than 35% of the total area (as opposed to over 60% initially), indicating a change to the protein secondary structure.

First, spectral analysis techniques such as curve fitting can be used to extract information about the specific secondary structures, using correlation tables like that in Ref. 4. This process is shown applied to the BSA spectrum in Figure 4. The initial fit parameters and allowed variation are chosen using the behavior of well-characterized proteins as reference points. The areas of the various fitted peaks are proportional to the “concentrations” of the helix, sheet, etc., and therefore percentages can be calculated and tabulated.

Alternatively, the spectra may be compared to databases of proteins with known characteristics to assign the percentages of each secondary structural element. A database correlating protein information (percent α-helix, β-sheet, etc.) with infrared spectral information is built in a chemometrics package such as TQ Analyst (Thermo Electron Corp.). As with any chemometrics method, the quality of the results depends on the quality of the input data. A wide range of secondary structures is required, and hence the maximum variability can be modeled. Various algorithms (such as partial least squares) are then applied to build a calibration, and the method is validated.

Figure 5 - Screen capture from TQ Analyst showing a “calibration” for BSA degradation. This was done using internal validation; the 0.92 correlation shows the method is reliable.

A simpler method can be applied to the study of protein degradation. Here, a database involving only the protein of interest is built, and some target parameter, such as reactor yield or catalytic efficacy, is measured for the standards. Now, comparison of a spectrum to the database directly yields a prediction of performance. Figure 5 shows a similar application of chemometrics to the BSA data. Here, the percentages of α-helix were calculated from curve fitting. This was then entered into TQ Analyst, and a cross-validation (1-point out) was performed. The excellent correlation shows that the data are internally consistent and may be useful in predicting behavior.

Correlation tables and chemometrics profiles for proteins must always be used with care. Some proteins, such as the α-helix in poly-L-lysine, behave quite differently from most other proteins. Jackson and Mantsch³ discuss the misuse of infrared data and establish some boundaries for the analysis. In short, changes are easy to assess, but absolute conformations are generally not.

Conclusion

FTIR has great potential for rapidly and simply providing quality control information about proteins in aqueous solutions. By following a standard set of procedures, the data can yield insights into whether processing has altered a protein, without requiring crystallizing or concentrating the protein. The method of data analysis will depend upon what type of information is desired, and may be tailored to a specific protein. The Proteus protein analysis kit makes the method accessible for users unfamiliar with FTIR.

References

1. Bradley, M. Thermo Electron Corp. Application Note AN 50733, “Curve Fitting in Raman and IR Spectroscopy: Basic Theory of Line Shapes and Applications”; 2004.
2. Oberg, K.A.; Ruysschaert, J.-M.; Goormagtigh, E. Eur. J. Biochem.2004, 271, 2937–48.
3. Jackson, M.; Mantsch, H.H. Crit.Rev. Biochem. Molec. Biol. 1995, 30(2), 95–120.
4. Bradley, M.S.; Nishikida, K. Infrared Measurements of Proteins: Theory and Applications. Booklet included in Proteus protein analysis kit (not available separately), 2005.
5. Grdadolnik, J.; Marechal, Y. Appl. Spectrosc. 2005, 59(11), 1347–56.

Dr. Bradley is Product Specialist, Thermo Electron Corp., 5225 Verona Rd., Madison, WI 53534, U.S.A.; tel.: 608-276-5620; fax: 608-276-6328; e-mail: [email protected].