Near-Infrared and Mid-Infrared Fourier Transform ... - SAGE Journals

Near-Infrared and Mid-Infrared Fourier Transform Vibrational Circular Dichroism of Proteins in Aqueous Solution SHENGLI MA,* TERESA B. FREEDMAN, RINA K. DUKOR, and LAURENCE A. NAFIE Department of Chemistry, Syracuse University, Syracuse, New York 13244 (S.M., T.B.F., L.A.N.); and BioTools Inc, 17546 Bee Line Hwy, Jupiter, Florida 33458 (R.K.D., L.A.N.)

Vibrational circular dichroism (VCD) of a series of proteins in H2O solution with differing secondary structure are reported for the first time in the near-infrared (NIR) region as well as the NH-stretching region. The Fourier transform (FT) near-infrared (NIR) measurements were carried out between 6000 to 4000 cm1. FT-VCD measurements were simultaneously carried out for the mid-infrared (mid-IR) region from 2000 to 800 cm1 for direct comparison to VCD in the NIR region. The NIR VCD spectra of proteins show distinct spectral features for different protein structural motifs, indicating a valuable new method to study protein conformations. The principal VCD transitions in the NIR region are two combination bands, the amide A-II and B-II bands, of the amide A and B fundamentals with the amide II fundamental, and the second overtone of the amide II, referred to as the amide 36II band. VCD in the amide A and B band region consisting primarily of NH stretching motions were successfully obtained in H2O for the first time for an insulin fibril sample. Similar to the enhanced VCD signal observed in amide I and II regions, the amide A and B VCD of insulin fibril shows strong intensity enhancements, providing an additional valuable probe of protein fibril growth and development in solution. The relative sensitivities of the midIR, N–H stretching, and NIR regions are discussed. Index Headings: Vibrational circular dichroism; VCD; Fourier transform; FT; Near-infrared spectroscopy; NIR spectroscopy; Protein secondary structure.

INTRODUCTION Protein structures play a crucial role in biological activity, covering virtually all biological processes and functions from the single cell to intact living organisms.1 Thus, understanding protein structure on the molecular level is a central problem in biological chemistry. Owing to the complexity of their structures and the somewhat limited availability of effective analytical techniques, only a small fraction (,10%) of the more than 44 700 proteins with known amino-acid sequences (primary structures) in the Protein Data Bank (PDB) have been determined with complete three-dimensional structures.2–4 Among currently available techniques, X-ray crystallography and nuclear magnetic resonance (NMR) spectroscopy are the most widely used to study the structure of proteins; however, both have limitations. X-ray can provide atomic-level information,5 but a single crystal of sufficient size is required to carry out X-ray structure analysis, and growing crystals of proteins is as much an art as a science, and obtaining crystals for X-ray analysis for some proteins is difficult, if not impossible, to achieve. Determination of protein structure by nuclear magnetic resonance (NMR) is limited to proteins of modest size. When possible, a complete determination of protein structure requires multidimensional NMR techniques Received 21 October 2009; accepted 25 March 2010. * Current address: Boehringer Ingelheim Pharmaceuticals, Inc., Ridgefield, CT 06877. Author to whom correspondence should be sent. E-mail: lnaﬁ[email protected].

Volume 64, Number 6, 2010

with high magnetic field strength and in some cases isotope labeling of specific nuclei.6 These requirements make X-ray and NMR expensive, difficult, and sometimes not currently possible for the determination of protein conformations. Vibrational spectroscopy, principally infrared (IR) and Raman spectroscopy, is another widely used approach to the study of protein structure.7 Compared to X-ray crystallography and NMR spectroscopy, vibrational spectroscopy permits a variety of sampling methods, including solutions and films, which provide structural information on the molecular level in environments close to natural biological media, although not at the same level of atomic resolution as X-ray or NMR. With spectral parameters, such as band shape, intensity, and band position, detailed information about protein structure, including backbone, side-chain, and bound-ligand structures, can be elucidated. These advantages make vibrational spectroscopy a cost-efficient and valuable tool for the investigation of protein structure and dynamics. Due to the homochirality of natural proteins, a form of vibrational spectroscopy sensitive to molecular chirality, vibrational circular dichroism (VCD), as well as its companion technique, Raman optical activity, is especially useful for studying peptides and proteins.8 Since the 1980s, VCD has been successfully applied to the analysis of protein structure determination.9,10 Compared to its ultraviolet (UV) visible form, electronic circular dichroism (ECD), VCD shows many well-resolved spectral bands that carry information about specific types of vibrational modes. In addition, owing to the longer wavelength of light in the infrared region, various sampling methods, such as high-concentration solutions, films, and solid particles, can be used in VCD with only low levels of scattering interferences, which can obscure ECD in the UVvisible region. With recent advances in instrumentation,11 VCD has been extended to the near-infrared (NIR) region where overtone and combination bands can be studied. Recent publications of the NIR VCD of terpenes,12 small chiral drug molecules, amino acids,13 and quantitative percent enantiomeric excess (%EE) analyses14 demonstrate that NIR VCD is a particularly useful new spectroscopic tool for the study of chiral molecular structures and their reactions. In addition, NIR VCD retains the basic analytical advantages of NIR spectroscopy, such as ease of sampling methods with longer path lengths and the potential use of fiber optics for industrial process and quality-control analyses. In this paper, we report the first measurements of VCD in proteins in aqueous solution beyond the mid-infrared (mid-IR) region into the NH-stretching and NIR regions. VCD in both the mid-IR and NIR regions were carried out for the same samples at approximately the same time to directly compare the relative sensitivities of these two spectral regions for protein secondary structure.15 The mid-IR VCD results are consistent

0003-7028/10/6406-0615$2.00/0 Ó 2010 Society for Applied Spectroscopy

APPLIED SPECTROSCOPY

615

with previously published spectra8,9 and are shown to have high signal-to-noise ratio (SNR) and excellent VCD baselines. The NIR VCD spectra reveal, for the first time, the sensitivity of this region to protein secondary structure. It is shown that three major NIR VCD bands, originating from two combination bands of the amide II and amide A and B modes, named amide A-II and amide B-II, and the second overtone of the amide II mode, called the amide 33II band, are characteristic of the secondary structure of proteins. We attribute the absence of combination bands between the amide I and the amide A and B transitions as indicative of the absence of significant NH motion in the amide I transition compared to the amide II transition. Furthermore, the NIR region is particularly sensitive to modes rich in hydrogen motion, which likewise favors combination modes involving amide II. We further extend the Fourier transform (FT) VCD measurements to the amide A and B region for aqueous solutions where NH stretching modes are observed. Although no VCD spectra have been observed for native proteins under normal sampling conditions, due to the extremely large H2O background absorption, we were able to obtain enhanced VCD in the amide A and B regions from insulin fibrils in aqueous solutions. Compared to the enhanced VCD reported in the amide I and II regions,16 similar levels of enhancement of VCD in the amide A and B region were observed, opening this region as a supplemental region for studying protein fibril growth and development.

EXPERIMENTAL Materials. Horse skeletal muscle myoglobin, bovine serum albumin (BSA), horse heart cytochrome c, hen egg-white lysozyme, ovalbumin, a-chymotrypsin from bovine pancreas, and concanavalin A from jack beans, as well as insulin for the fibril measurements, were purchased from Sigma-Aldrich and used without further purifications. Solutions were prepared by dissolving proteins in a 50 mM phosphate buffer at pH 7 at a concentration of 200 mg/mL. Insulin fibrils were prepared using the reported protocol16 as follows. Insulin was first dissolved in water at pH 2 (adjusted with HCl), followed by heating for two hours at 65 8C to initiate the growth of fibrils and then measured after approximately one day when growth had stopped or was very slow on the VCD measurement timescale. Instrumentation. Three dual-source17 FT-VCD ChiralIR spectrometers (BioTools, Inc., Jupiter, FL), equipped with liquid nitrogen cooled HgCdTe (MCT), thermo-electrically cooled MCT, and an InGaAs detector, respectively, were used to carry out mid-IR (800–2000 cm1), NH-stretching (2000– 4000 cm1), or NIR VCD (3800–6200 cm1) measurements, respectively. Single-photoelastic modulation (PEM) spectrometers were re-configured with dual polarization modulation (DualPEM)18 to improve and stabilize the VCD baseline near zero VCD intensity and thereby significantly reduce artifacts. Vibrational Circular Dichroism Measurements. A single drop of protein solution in a BioCell (BioTools), approximately, 7 lL, was used for mid-IR VCD and amide A and B VCD measurements, and 0.1 mL solution was used for NIR VCD measurements. Protein concentrations for the mid-IR and NIR regions were 200 mg/mL in 50 mM phosphate buffer. The midIR path lengths were 6 lm and those in the NIR regions were 1 mm. All mid-IR VCD spectra were measured for 12 hours with 8 cm1 resolution. NIR VCD measurements were carried out

616


FIG. 1. VCD and IR spectra of myoglobin in the region 1000–2500 cm1 measured at a concentration of 200 mg/mL in pH7 phosphate buffer at 8 cm1 resolution for 12 hours. VCD and IR spectra were solvent (buffer) subtracted. The noise level was offset for clarity.

concurrently on the NIR VCD spectrometer for the same duration with 32 cm1 resolution. For the amide A and B region, the spectral collection time was 1 hour, and a special sampling method was used to carry out the measurements. Insulin fibrils were prepared from 100 mg/mL of insulin solution adjusted to pH 2, heated at 65 8C for two hours, and allowed to develop for 24 hours. Approximately 7 lL of insulin fibril solution was placed between two flat BaF2 windows without a spacer. A path length of ;3–4 lm was obtained with a total absorbance of ;0.9 at ;3300 cm1.

RESULTS AND DISCUSSION Vibrational Circular Dichroism and Infrared Spectroscopy of Proteins in the Mid-Infrared Region. In Fig. 1, we present the FT-IR and FT-VCD spectra of myoglobin in the spectral region of 1000–2500 cm1. Clear VCD bands compared with noise level are observed from the amide III (1200–1350 cm1) to the amide I (1600–1750 cm1) regions. Although no VCD bands exist in the region beyond 1800 cm1 (1800–2500 cm1), this region is used to ensure accurate subtraction of water in the protein absorbance (IR) spectra in aqueous solution.19 In particular, the small broad IR band at ;2125 cm1 shown in Fig. 2 from a combination band of water is subtracted from the protein solution spectrum to produce a flat (zero) baseline in the region of 1800 to 2500 cm1 (Fig. 1). Given the appropriate solvent-subtracted IR spectra, together with proper calibration of FT-VCD instrumentation,17 the VCD intensities obtained are consistent with previous measurements carried out on a dispersive spectrom-

FIG. 2. Overlay of IR spectra of water and protein aqueous solution (myoglobin). The experimental conditions are the same as in Fig. 1.

eter.9 Thus, the mid-IR spectra shown here not only support structural analysis, but also can be used to benchmark VCD intensities for future measurements of proteins in H2O solutions. Fourier transform VCD measurement possesses all the general benefits of FT-IR spectroscopy compared to the corresponding dispersive methodology and thereby yields simultaneous broadband measurement of the amide I, II, and III regions (Fig. 1). Among these three bands, amide I and II bands are the most used for structural determination due to

their higher intensities and increased sensitivity to secondary structure compared to the amide III band. Therefore, the experimental settings (e.g., sample cell path length and concentrations) and ensuing analysis of VCD spectra in the mid-IR are focused on the amide I and amide II regions and include the spectral range 1800–1400 cm1. The VCD and IR spectra of proteins with primarily a-helical structures, myoglobin and albumin (BSA), are shown in Fig. 3. IR spectra (lower trace) show two major bands, amide I and amide II bands located at 1654 cm1 and 1546 cm1, respectively. These two band locations and shapes exhibit the basic IR signature profiles used to identify a-helical structures in protein IR absorbance spectra. The VCD spectra (upper trace) of both myoglobin and albumin show a positive couplet ( þ) in the amide I region with a zero crossing point near 1650 cm1 and an intense negative VCD band in the amide II region with a maximum at 1515 cm1. These VCD features have been well established as characteristic of the a-helix secondary structure.9,10 In particular, it has been shown that the amide II band is smaller than or equal in intensity to the amide I couplet intensities. To facilitate the analysis of the structural contents of all seven proteins, the detailed information for all seven proteins in this study is summarized in Table I. Although these data are based on crystal-state data, previous quantitative analyses of protein structures based on VCD are in good agreement with crystal data with minor variations.20 Figure 4 shows the VCD and IR spectra of cytochrome c and lysozyme. These proteins have a-helical structures with some b-turns and a mixture of b-sheet and other secondary elements. Compared to myoglobin and BSA, cytochrome c and lysozyme

FIG. 3. VCD and IR spectra of myoglobin and bovine serum albumin measured under conditions as in Fig. 1.


617

TABLE I. Protein secondary structure fractions determined by X-ray from the PDB. Protein Myoglobin Bovine serum albumin Cytochrome c Lysozyme Ovalbumin a-Chymotrypsin Concanavalin A

Category

a-Helix

b-Sheet

b-Turn

Remainder

High helix High helix Helix þ others Helix þ others Helix þ sheet Low helix þ sheet High sheet

0.76 0.61 0.41 0.39 0.31 0.12 0.04

0.00 0.00 0.00 0.08 0.31 0.31 0.45

0.12 0.14 0.23 0.21 0.16 0.20 0.23

0.12 0.25 0.36 0.32 0.22 0.37 0.28

have a relatively smaller, positive VCD couplet in the amide I region. Interestingly, both proteins also show a small positive VCD shoulder at 1689 cm1, which indicates that some degree of b-turn structure exists in both proteins. In addition, cytochrome c shows a small positive feature at 1616 cm1 and lysozyme shows a small negative band at 1635 cm1. At the same time, the amide II VCD band is broader and less intense. These results confirm that there is major a-helical content with minor b-turn content and mixtures of other structures in these two proteins, all of which is consistent with X-ray crystal data. These are the most difficult proteins to measure in the amide I region due to cancellation of positive and negative bands. Vibrational circular dichroism and IR spectra of ovalbumin and a-chymotrypsin, as mixed a-helix/b-sheet proteins, are presented in Fig. 5. Compared to VCD spectra originating from proteins with high a-helix content, some new amide I VCD features are observed for both proteins. Ovalbumin shows a ( þ ) pattern of VCD features with negative bands at 1662 cm1 and 1623 cm1 and a positive band at 1647 cm1. Similar

to ovalbumin, the a-chymotrypsin VCD spectrum exhibits these three amide I bands with an additional positive VCD band at 1689 cm1. This gives a-chymotrypsin an overall (þ þ ) VCD pattern. The band at 1689 cm1 is usually assigned to b-turns; the next two VCD bands (negative at 1662 cm1 and positive at 1647 cm1) indicate the presence of some a-helix structure; and the intense negative VCD band at 1627 cm1 is the signature of b-sheet structure. In addition, both proteins show VCD amide II bands with a negative couplet with a positive peak at ;1558 cm1 and a negative peak at ;1515 cm1. These VCD features confirm a large amount of b-sheet structure in both proteins. Concanavalin A is a protein with dominant b-sheet structure. The amide I band shows an intense negative band located at 1627 cm1, shown in Fig. 6, which is the typical signature of bsheet structure. In the amide II region, an intense negative couplet is observed with a positive peak at 1558 cm1 and a negative peak at 1515 cm1. Vibrational circular dichroism and IR spectra of all seven of these proteins are compared in Fig. 7. As a benefit of the FT-

FIG. 4. VCD and IR spectra of cytochrome c and lysozyme measured under conditions as in Fig. 1.

618


FIG. 5. VCD and IR spectra of ovalbumin and a-chymotrypsin measured under conditions as in Fig. 1.

FIG. 6. VCD and IR spectra of concanavalin A measured under conditions as in Fig. 1.

FIG. 7. Comparison of VCD and IR spectra of seven proteins described previously. The IR and VCD spectra of each protein are assigned a letter.


619

TABLE II. Mid-IR VCD band assignment of proteins. Amide I (cm1)

Proteins Myoglobin Bovine serum albumin Cytochrome c Lysozyme Ovalbumin a-Chymotrypsin Concanavalin A

1658(), 1662(), 1689(þ), 1689(þ), 1662(þ), 1689(þ), 1627()

1643(þ) 1640(þ) 1658(), 1660(), 1647(), 1662(),

1643(þ), 1616(þ) 1648(þ) 1623() 1647(þ), 1627()

VCD instrumentation used here, specifically dual-source and dual-PEM optical components, VCD spectra of proteins in H2O solution have been obtained with a high signal-to-noise ratio (SNR) and zero-baseline offset without any data treatment except solvent spectral subtraction. From a visual inspection of the frequencies and shapes of VCD bands, we can clearly distinguish spectral transitions between proteins with different secondary structural motifs. Qualitative band assignments for the amide I and II bands are listed in Table II as a general guide to secondary content from IR and VCD spectroscopy. Because the amide I and II regions are collected simultaneously and presented in a stacked plot in the same figure, the discriminating features for these two regions can now be easily observed. Of particular note is the gradual loss of intensity in the amide I region with loss of a-helix content and the growth of a new high-frequency positive component in the amide II band leading to a distinct (þ ) couplet for high b-sheet content. Although VCD and IR spectra of these proteins were reported previously,21–24 it is valuable to view FT-VCD spectra of these proteins in a unified format across a wide spectral range. To demonstrate the quality of the FT-VCD spectra measured, a direct comparison with previously published data25 is presented in Fig. 8. Two proteins with different structures, a-chymotrypsin (b-sheet) and cytochrome c (ahelix), are selected as examples, shown in Figs. 8a and 8b, respectively. The previously published data were collected 15 years ago on a dispersive VCD spectrometer for ;10 hours (10 scans) with ;10 cm1 resolution,9,25 which are similar in time of collection but with lower resolution compared to the corresponding FT-VCD spectra (;12 hours, 8 cm1). Direct comparison shows that the FT-VCD spectra have improved SNR, although it should be noted that noise features differ between FT and dispersive spectra and no noise spectra were measured for the dispersive spectra. Nevertheless, with the noise level available as reference, the FT-VCD show that some previously detected small bands in the dispersive VCD spectra, such as a few small positive bands between 1665 and 1640 cm1 (Fig. 8a, upper trace) for a-chymotrypsin and some small features between 1620 and 1540 cm1 (Fig. 8b, upper trace) for cytochrome c, are likely due to noise. With the FT-VCD spectra, previously unresolved spectral features, such as the small positive couplet (negative peak at 1662 cm1 and positive band at 1647 cm1 ) for achymotrypsin and two positive low frequency amide I bands (1643 cm1 and 1616 cm1) for cytochrome c, were clearly detected. Such small features of necessity carry additional structural information. Moreover, it appears that quantitative analysis of protein structures will become more accurate if based on FT-VCD spectra. In addition, and very important for protein conformation elucidation, FT-VCD shows the growth

620


Amide II (cm1)

Major secondary structure

1515() 1515() 1515() 1512() 1562(þ), 1512() 1558(þ), 1518() 1558(þ), 1515()

a-Helix a-Helix a-Helix and b-turns a-Helix and b-turns a-Helix, b-sheets, and b-turns b-Sheet and b-turns b-Sheet and b-turns

of a positive VCD band in the amide II region at 1558 cm1. This band, combined with the previously identified negative amide II VCD band at 1515 cm1, is an excellent marker for conformation changes, and its growth corresponds to increasing b-sheet content. This band has never previously been associated with b-sheet content in native proteins but has been identified as one of five marker bands associated with extremely intense VCD in protein fibrils.16 It needs to be mentioned that some advances in VCD instrumentation, such as the implementation of a narrow-band MCT detector and digital signal processing, have been recently reported.26,27 Improved results were obtained on the dispersive VCD and are comparable to FT-VCD (ChiralIR) in terms of SNR on some biological samples, including poly-lysine and BSA in D2O solution.27 However, the FT-VCD results shown in those comparisons were not measured on the most recent FT instrumentation (e.g., dual-source17). Therefore, a direct comparison of the two instruments (FT versus dispersive) with the most recent advancements is currently not available. Such a comparison needs to be carried out on the same sample under the same conditions and is beyond the focus of this paper. One additional advantage of FT-VCD versus dispersive VCD should be mentioned in the context of this discussion. FT-VCD can be used easily to follow kinetic changes in a sample that change slowly with respect to the block collection time. FT-VCD spectra exhibit no time bias across the spectrum over the course of the time-block collection. For example, recently the kinetics of protein fibril development have been measured with mid-IR FT-VCD16 using collection times from 90 minutes to 30 minutes in unpublished data with high signalto-noise ratios. Dispersive VCD collection has an intrinsic time

FIG. 8. Comparisons of previously published dispersive VCD spectra9,25 with FT-VCD spectra presented here. (a) Comparison of a-chymotrypsin spectra (bottom trace) with that from Ref. 25 (top trace); (b) comparison of cytochrome c spectra (bottom trace) with that from Ref. 25 (top trace). Noise level from FTVCD measurement is shown in the middle trace for both panels.

FIG. 9. NIR absorbance of a 200 mg/mL BSA solution (solid line) and ph 7 phosphate buffer solution (dashed line) measured at 32 cm1 resolution and a 1 mm path length.

bias involving the time for a given scan, and this technique can only be used to follow kinetic changes that essentially do not change during the period of time of typically several minutes or longer, during which the scan is executed. Significantly faster scanning of dispersive VCD spectra can only be conducted using a suitably shorter lock-in time constant, on the order of milliseconds, in which case individual scans become very noisy and spectral collection becomes more difficult, if possible at all. By contrast FT-VCD spectra collected in blocks of

arbitrary time length, with a trade-off between signal quality and time resolution, are always time-equivalent across the entire spectrum, whereas dispersive VCD can never be measured in a time-equivalent way due to the length of time needed to scan the spectrum. For this reason, FT spectral measurement for kinetics across a spectral region is always superior to the corresponding dispersive spectral measurement. Vibrational Circular Dichroism of Proteins in the NearInfrared Region. In the NIR region for aqueous solutions between 4000 and 6000 cm1, there are two strong bands of H2O that dominate the absorbance spectra. Figure 9 shows the overlay of the NIR absorbance of protein (BSA) and the H2O buffer solution. The off-scale band centered near 5150 cm1 is associated with the combination mode of the fundamental OHstretching and OH-deformation vibrations of water. The other off-scale band below 4000 cm1 arises directly from the OHstretching fundamental vibration of water. As a result, the observation of NIR protein VCD bands is confined to the regions between 4300 cm1 and 4900 cm1 and above 5600 cm1. To eliminate appropriately by subtraction the absorbance of pure water from that of the protein solution, a scale factor of 0.86 was applied to the absorbance spectrum of pure water at the same path length to compensate for the presence of protein in the solution spectrum.28 In Fig. 10, we present the NIR VCD spectra of primarily ahelical proteins, myoglobin and BSA. Three combination and overtone IR bands, located at 4875 cm1, 4600 cm1, and 4366 cm1, correspond to combination modes of amide A (amide NH stretching) and amide II, amide B (hydrogen-bonded amide NH stretching) and amide II, and the second overtone (n ¼ 3) of the amide II mode, respectively. These results are in agreement

FIG. 10. NIR VCD and absorbance of (left) myoglobin and (right) BSA measured with conditions as in Fig. 9.


621

FIG. 11. NIR VCD and absorbance spectra of (left) cytochrome c and (right) lysozyme measured with conditions as in Fig. 9.

with previous studies of the NIR absorption spectra of proteins.29,30 In the VCD spectra, the two combination bands show clearly negative intensity at 4860 cm1 (A þ II) and 4600 cm1 (B þ II ), respectively. Due to the fact that both myoglobin and BSA exhibit intense negative amide II VCD bands, it is reasonable to observe negative VCD features in the combination bands containing the amide II vibrational mode. The IR band at 4366 cm1 (33II) has a very weak VCD band centered at ;4366 cm1 (for BSA, ;4351 cm1) that rides on the negative VCD background of what we call the higher frequency 4600 cm1 band. To facilitate discussion of these NIR protein bands we have named them amide A-II (4860 cm1), amide B-II (4600 cm1), and amide 33II (4366 cm1). Figure 11 shows NIR VCD spectra of cytochrome c and lysozyme. Similar to myoglobin and BSA, cytochrome and lysozyme give rise to negative amide A-II and B-II VCD bands. On the other hand, these two negative VCD bands are smaller and broader, relative to their parent NIR absorbance bands, due to mixtures of secondary structures and lower ahelical content in both proteins. This result is consistent with VCD in the mid-IR region, which shows a corresponding sensitivity to protein secondary structure. The NIR VCD and absorbance spectra of a-chymotrypsin and ovalbumin are shown in Fig. 12. Compared to the previous proteins with higher a-helical structure, a-chymotrypsin and ovalbumin show obvious new features in both the NIR absorbance (4875 and 4536 cm1) and VCD spectra (4875, 4505, and 4443 cm1). The absorbance spectra of both the amide B-II and amide 33II bands of a-chymotrypsin and ovalbumin are broader. In particular, the shapes of the amide

622


B-II bands have changed significantly. For the VCD spectra, although the amide A-II and amide B-II bands still show negative peaks, the intensities have become much smaller and broader. Furthermore, a broad and small negative VCD couplet appears for 4505 cm1 (positive) and 4443 cm1 (negative, not marked in the ovalbumin VCD spectrum). At the same time, a small shoulder can be seen at 4536 cm1 on the amide B-II band in the NIR absorbance spectrum for ovalbumin (also present for a-chymotrypsin but weaker and not marked). Since a-chymotrypsin and ovalbumin belong to the class of proteins with significant b-sheet content, it can be concluded that the negative VCD couplet and small IR shoulders are characteristic of b-sheet content. These results are confirmed in Fig. 13 by the IR and VCD spectra of concanavalin A, which has a dominant b-sheet content. Compared to proteins with less b-sheet content, concanavalin A shows broader amide B-II and amide 33II IR bands. In addition, the negative couplet of the amide 33II VCD band (þ couplet at 4520 cm1 and 4443 cm1, respectively) becomes larger and sharper compared to chymotrypsin. Moreover, negative VCD intensity in the amide A-II band disappears and eventually becomes a small positive feature and the amide B-II band shows a small positive peak as well, located at 4613 cm1. Comparison of NIR VCD and absorbance of all seven proteins in Fig. 14 clearly shows that the two negative amide A-II and amide B-II VCD bands are the signatures of a-helical structure. From these results it is expected that the relative intensity and band shape will determine the relative amounts of a-helical structure in these proteins. Proteins with increasing

FIG. 12. NIR absorbance and VCD spectra of (left) a-chymotrypsin and (right) ovalbumin measured with conditions as in Fig. 9.

mixtures of b-structures show different VCD features. In particular, the negative amide A-II band becomes smaller and eventually turns positive with more b-sheet structure as in achymotrypsin to concanavalin A. Furthermore, a negative VCD couplet becomes evident and grows larger in the region from 4520 cm1 to 4443 cm1, with shoulders appearing on amide B-II and amide 33II IR bands. Compared to NIR absorbance, it is clear that NIR VCD is more sensitive to protein secondary structure. We summarize the combination and overtone band assignments in Table III, as well as the key signatures of the NIR VCD bands for typical secondary structures. Comparison of Mid-Infrared Vibrational Circular Dichroism with Near-Infrared Vibrational Circular Dichroism. Since both mid-IR VCD and NIR VCD spectra show distinctive features for protein secondary structures, it is worthwhile to compare the two techniques in terms of sensitivity and relative advantages. In the mid-IR region, the amide I and II bands are the two major marker bands for structural determination. In particular, the amide II band shows one or two VCD bands, while the amide I band shows one to four VCD bands for different proteins. As a result, from three to six major VCD bands, in combination and overlap, can be used for the analysis of protein structure in the mid-IR region. In the NIR region between 4300 and 4900 cm1, there are three major marker bands from two combination modes and a second overtone mode. In particular, the amide A-II band shows one negative or positive VCD band, while the amide B-II and 33II bands, separately or with some overlap, show two to four VCD bands for different proteins. Thus, three to five VCD bands are available in the NIR region for structural analysis. Since VCD bands in both regions carry structural information and change

FIG. 13. NIR VCD and absorbance spectra of concanavalin A measured with conditions as in Fig. 9.


623

FIG. 14. Comparison of NIR absorbance and VCD spectra of different proteins. (A) Myoglobin; (B) BSA; (C) cytochrome c; (D) lysozyme; (E) ovalbumin; (F) a-chymotrypsin; and (G) concanavalin A. The IR spectra in the lower panel are plotted in the same order as the corresponding VCD spectra. Five vertical lines (dashed lines), located at 4860 cm1, 4600 cm1, 4520 cm1, 4443 cm1, and 4366 cm1, were plotted for reference.

significantly with changes in protein structure, it can be concluded that mid-IR VCD and NIR VCD possess similar sensitivities to protein structure, although a more complete evaluation of the relative effectiveness of each region is best carried out using chemometric methods such as principal component analysis. In other areas of comparison, mid-IR VCD and NIR VCD have their own relative advantages for specific applications originating from the same relative advantages of mid-IR and NIR absorbance spectroscopy. For example, the ability to use long path length cells in the NIR region gives a relative ease of sampling and a wider variety of sampling techniques (e.g., fiber optics). On the other hand, NIR VCD requires larger amounts of protein sample, approximately 20 mg of protein in 100 lL

solution, to carry out measurements compared to mid-IR VCD, which requires only on the order of 1 mg of protein and 7 lL of solution using the same protein concentration. Thus, mid-IR VCD is more suitable for research applications, where typically small amounts of samples are available. NIR VCD is more applicable to industrial settings, such as process monitoring during the manufacture of biomolecules, where bulk samples are produced in large amounts and ease of sampling is important. However, where sample quantity is not a problem, both mid-IR and NIR VCD appear to be equally valuable, and together more valuable than either method in isolation. Vibrational Circular Dichroism of Amide A and B Bands of Protein Fibrils. The amide A and B bands arising from NH stretching modes also contain important information about protein structures. Due to the exceptionally strong interference from the OH stretching band of water, sampling in this region is so difficult for both IR and VCD measurements of proteins that no spectra from this region appear to have been previously reported for aqueous protein solutions. In addition, compared to amide I and amide II, amide A and amide B bands exhibit weaker intensities that make aqueous spectra in this region even more difficult to detect. On the other hand, VCD in the amide A and amide B region contains sensitive information on the vibrational coupling and hydrogen bonding of amide NH bonds that in turn can provide important experimental data for theoretical calculations, which to date have focused on vibrational modes of the mid-IR region. In a recent paper,16 enhanced VCD signals were described for amide I and II bands of protein fibrils. To further investigate the structures and the unusually large VCD intensities, we extend the VCD study of insulin fibrils to the amide A and B region (3000–3600 cm1). In Fig. 15, we present the VCD and IR spectra of insulin fibrils in both the amide I and II (1500–1700 cm1) and amide A and B (3000– 3600 cm1) regions for the same solution of insulin fibrils. The amide I bands are the same as those published,16 which confirmed that fibrils have predominantly the standard cross bsheet fibril structure. The VCD spectra of the amide A and B bands show strong, interesting features. In particular, amide A shows an intense negative couplet (þ) with a positive peak at 3325 cm1 and negative peak at 3270 cm1, and the amide B band exhibits a small negative VCD peak at ;3074 cm1. Similar to the amide I region, the negative amide A VCD band also exhibits unusually large intensity that is roughly the same magnitude as the largest amide I feature at 1627 cm1. These new VCD results in the NIR region provide further information on the structure of protein fibrils based on their amide NHstretching vibrational modes and should prove invaluable for future analyses and theoretical calculations of protein fibril structure.

TABLE III. Band assignments of NIR VCD of proteins.a Amide A-II (cm1)

Proteins Myoglobin BSA Cytochrome c Lysozyme Ovalbumin a-Chymotrypsin Concanavalin A a

4860s 4860s 4860s 4860s 4875s 4860s 4860m

() () () () () () (þ)

W: weak; M: medium; S: strong.

624


Amide B-II (cm1)

Amide 33II (cm1)

4600s () 4600s () 4597s () 4597s () 4613s (), 4582s () 4597m () 4613w (þ)

4366w () ;4351w () 4366w () 4505w (þ), 4366m (þ) 4505w (þ), 4459m (), 4366m (þ) 4505w (þ), 4443m (), 4366m (þ) 4520s (þ), 4443s (), 4351m (þ)

FIG. 15. VCD and IR spectra of insulin fibril in (left) the amide A and B region and (right) the amide I region. Measurements were carried out concurrently in the two regions for 1 h at 8 cm1 resolution. The path length for the amide A and B region was ;4 lm and the noise level in this region is offset for clarity. The noise level in the amide I region is essentially zero at this level of intensity scale.

CONCLUSION We report, for the first time, the NIR VCD spectra of proteins. We also report the first IR and VCD spectra of aqueous solutions of proteins in the NH-stretching amide A and B regions. The NIR VCD measurements were carried out in aqueous solution at relatively high concentrations, although lower concentrations by up to a factor of ten could be used if required. The NIR VCD spectra of proteins with different compositions of secondary structure exhibit distinct VCD features. The three major combination and overtone bands in the spectral region from 4300 to 4900 cm1 were named amide AII, amide B-II, and amide 33II. These bands were used to classify and facilitate band assignments of different types of secondary structures. Compared to NIR absorbance, the corresponding VCD spectra are shown to be more sensitive to secondary structure and thus constitute a new conformational probe of proteins in aqueous solution. Enhanced signal quality and improved baseline stability of VCD spectra in the mid-IR region are reported in conjunction for the comparison of FT-VCD and dispersive VCD of two proteins for which a direct comparison to previously published results is possible. Based on this comparison, the improved quality of FT-VCD spectra of protein solutions suggests that such spectra should provide more reliable statistical databases and new VCD features that are below the noise level of dispersive VCD spectra. Furthermore, FT-VCD can be used to follow kinetic changes in samples more easily than dispersive

VCD due to the absence of an intrinsic time bias during the period of spectral collection time, FT time-block versus dispersive scan. More recent instrumental improvements reported for the measurement of the dispersive VCD of proteins may make a direct comparison of dispersive VCD and FT-VCD more meaningful if the latest advances of both methods are brought to focus. The stack plots of absorbance and VCD in the NIR and midIR regions for a sequence of seven well-characterized proteins reveals a consistent and near-continuous transition from high a-helix to high b-sheet proteins. The first and last members of this sequence clearly define the limiting spectra for a-helix and high b-sheet, respectively, across two to three hundred wavenumbers of frequency. While the a-helix VCD spectrum is well characterized in the amide I and II regions, a new positive VCD band at 1558 cm1 is observed as part of the amide II signature for an anti-parallel, or possibly parallel, bsheet conformation in a folded protein. A positive VCD band at this frequency is observed in the enhanced mid-IR VCD of protein fibrils, which are also largely composed of b-sheet structure.16 The VCD of insulin fibrils was extended to the amide A and B region. Unusually large VCD intensities, comparable in size to the unusually large VCD features in the amide I and II regions, were observed in the NIR region. These results should lead to important applications of VCD in this region for future studies of protein fibril formation and development. Considering the differing relative advantages and strengths


625

of NIR and mid-IR VCD, it is found that NIR VCD is more suitable to applications in industrial processes. The successful measurement of NIR VCD spectra of proteins opens opportunities for the use of NIR VCD in fundamental studies of protein solution conformation when sample quantities are not a limitation and possibly more importantly in routine process control and quality analysis in the manufacture of biological and biopharmaceutical molecules. Due to similar degrees of sensitivity to secondary structure, it is likely that in the future, quantitative analysis of protein structure in aqueous solution can be extended from the mid-IR to include routine NIR absorbance and VCD. 1. L. Stryer, Biochemistry (W. H. Freeman and Company, New York, 1988), 3rd ed. 2. P. Aloy, M. Pichaud, and R. B. Russell, Curr. Opin. Struct. Biol. 15, 15 (2005). 3. R. Hu¨hne, F.-T. Koch, and J. Su¨hnel, Brief. Funct. Genomic. Proteomic. 6, 220 (2007). 4. J. Roland and L. Dunbrack, Curr. Opin. Struct. Biol. 16, 374 (2006). 5. J. Drenth, Principles of Protein X-Ray Crystallography (Springer-Verlag, Berlin, 2002), 2nd ed. 6. S. J. Opella and F. M. Marassi, Chem. Rev. 104, 3587 (2004). 7. F. S. Parker, Applications of Infrared, Raman, and Resonance Raman Spectroscopy in Biochemistry (Springer-Verlag, Berlin, 1983). 8. T. A. Keiderling, ‘‘Peptide and Protein Conformational studies with Vibrational Circular Dichroism and related Spectroscopies’’, in Circular Dichroism: Principles and Applications, Second Edition, N. Berova, K. Nakanishi, and R. W. Woody, Eds. (Wiley-VCH, New York, 2000), p. 621. 9. V. Baumruk and T. A. Keiderling, J. Am. Chem. Soc. 115, 6939 (1993). 10. T. A. Keiderling, Curr. Opin. Struct. Biol. 6, 682 (2002). 11. L. A. Nafie, R. K. Dukor, J. R. Roy, A. Rilling, X. Cao, and H. Buijs, Appl. Spectrosc. 57, 1245 (2003).

626


12. X. Cao, R. D. Shah, R. K. Dukor, C. Guo, T. B. Freedman, and L. A. Nafie, Appl. Spectrosc. 58, 1057 (2004). 13. S. Ma, T. B. Freedman, X. Cao, and L. A. Nafie, J. Mol. Struct. 799, 226 (2006). 14. C. Guo, R. D. Shah, R. K. Dukor, X. Cao, T. B. Freedman, and L. A. Nafie, Appl. Spectrosc. 59, 1114 (2005). 15. W. Kabsch and C. Sander, Biopolymers 22, 2577 (1983). 16. S. Ma, X. Cao, M. Mak, A. Sadik, C. Walkner, T. B. Freedman, I. K. Lednev, R. K. Dukor, and L. A. Nafie, J. Am. Chem. Soc. 129, 12364 (2007). 17. L. A. Nafie, H. Buijs, A. Rilling, X. Cao, and R. K. Dukor, Appl. Spectrosc. 58, 647 (2004). 18. L. A. Nafie, Appl. Spectrosc. 54, 1634 (2000). 19. D. Francoise, T. Marie, and P. Michel, Appl. Spectrosc. 43, 538 (1989). 20. P. Pancoska, E. Bitto, V. Janota, M. Urbanova, V. P. Gupta, and T. A. Keiderling, Protein Sci. 4, 1384 (1995). 21. V. Baumruk, D. Huo, R. K. Dukor, T. A. Keiderling, D. Lelievre, and A. Brack, Biopolymers 34, 1115 (1994). 22. M. G. Paterlini, T. B. Freedman, and L. A. Nafie, Biopolymers 25, 1751 (1986). 23. P. Xie, Q. Zhou, and M. Diem, J. Am. Chem. Soc. 117, 9502 (1995). 24. C. Zhao and P. L. Polavarapu, J. Am. Chem. Soc. 121, 11259 (1999). 25. V. Baumruk, P. Pancoska, and T. A. Keiderling, J. Mol. Biol. 259, 774 (1996). 26. J. Hilario, D. Drapcho, R. Curbelo, and T. A. Keiderling, Appl. Spectrosc. 55, 1435 (2001). 27. T. A. Keiderling, J. Kubelka, and J. Hilario, ‘‘Vibrational Circular Dichroism of Biopolymers: Summary of Methods and Applications’’, in Vibrational Spectroscopy of Biological and Polymeric Materials, V. G. Gregoriou and M. S. Braiman, Eds. (CRC Press, Boca Raton, FL, 2005), p. 253. 28. K.-i. Izutsu, Y. Fujimaki, A. Kuwabara, Y. Hiyama, C. Yomota, and N. Aoyagi, J. Pharm. Sci. 95, 781 (2006). 29. Y. Wu, B. Czarnik-Matusewicz, K. Murayama, and Y. Ozaki, J. Phys. Chem. B 104, 5840 (2000). 30. B. Yuan, K. Murayama, Y. Wu, R. Tsenkova, X. Dou, S. Era, and Y. Ozaki, Appl. Spectrosc. 57, 1223 (2003).