Hydrogen Peroxide-induced Structural Alterations of RNase A*

Proteins exposed to oxidative stress are degraded via proteolytic pathways. In the present study, we undertook a series of in vitro experiments to establish a correlation between the structural changes induced by mild oxidation of the model protein RNase A and the proteolytic rate found upon exposure of the modified protein toward the isolated 20 S proteasome. Fourier transform infrared spectroscopy was used as a structure-sensitive probe. We report here strong experimental evidence for oxidation-induced conformational rearrangements of the model protein RNase A and, at the same time, for covalent modifications of amino acid side chains. Oxidation-related conformational changes, induced by H2O2exposure of the protein may be monitored in the amide I region, which is sensitive to changes in protein secondary structure. A comparison of the time- and H2O2concentration-dependent changes in the amide I region demonstrates a high degree of similarity to spectral alterations typical for temperature-induced unfolding of RNase A. In addition, spectral parameters of amino acid side chain marker bands (Tyr, Asp) revealed evidence for covalent modifications. Proteasome digestion measurements on oxidized RNase A revealed a specific time and H2O2 concentration dependence; at low initial concentration of the oxidant, the RNase A turnover rate increases with incubation time and concentration. Based on these experimental findings, a correlation between structural alterations detected upon RNase A oxidation and proteolytic rates of RNase A is established, and possible mechanisms of the proteasome recognition process of oxidatively damaged proteins are discussed.

In studies using mammalian cells, cell lysates or extracts, strong evidence was presented for the key role of a single proteolytic complex, the 20 S proteasome, in the selective recognition and degradation of oxidatively damaged proteins (5,6,14,19). The basis of this recognition still remains unknown. However, some experimental evidence suggests the selective recognition of hydrophobic moieties at the protein surface (14, 16, 20 -22) or the importance of the methionine oxidation product, methionine sulfoxide (14,21). The groups of Davies (16,20) and Stadtman and Levine (14,21) demonstrated several times the enhanced proteolytic susceptibility of oxidized proteins with increased surface hydrophobicity using various methods such as separation on hydrophobic interaction chromatography (16,20) or the 8-anilino-1-naphtalenesulfonic acid-dependent fluorescence intensity (21). Levine et al. (14) demonstrated a correlation between methionine sulfoxide formation and proteolytic degradation. It was concluded that the oxidative modification of amino acid side chains disrupts, at least locally, the tertiary protein structure, which is, in turn, accompanied by exposure of hydrophobic moieties to the surface of the protein. This increase in surface hydrophobicity seems to be the recognition signal for the 20 S proteasome for binding and degradation of the substrate protein (14, 16, 18, 20 -22). However, until now it could not be shown directly that protein oxidation is accompanied by the disruption of secondary and tertiary protein structure.
The present investigation addresses the question of whether protein oxidation is followed by amino acid side chain modifications or changes in secondary and tertiary protein structure and whether these effects correlate with increased proteolytic susceptibility. For our studies, we selected the small single domain cytoplasmic protein RNase A as a model, since this protein is exceptionally well characterized by a variety of structure-sensitive techniques such as x-ray crystallography (23), NMR spectroscopy (24), UV-visible and CD spectroscopy (25), differential scanning calometric (for a review, see Pace et al. (26)), and Fourier transform infrared (FT-IR) 1 spectroscopy (27)(28)(29)(30). In particular, FT-IR spectroscopy has proved to be a sensitive tool for following conformational changes in proteins (Ref. 31; for a review, see Jackson & Mantsch (32)). Peptide backbone and side chain infrared "marker" bands can be employed as conformation-sensitive monitors to derive structural parameters during refolding of RNase A in solution (33,34). Recently, time-resolved infrared spectroscopic techniques have been employed to follow refolding processes of RNase A in the time range of 50 ms to 15 minutes (29,35,36).
In the present paper, we report structural alterations, detected by FT-IR spectroscopy, of the model protein RNase A upon exposure toward hydrogen peroxide and upon thermal unfolding. The changes of the protein structure were related to the results obtained by degradation measurements of the oxidized RNase A samples by isolated 20 S proteasomes.

EXPERIMENTAL PROCEDURES
Materials-Highly purified lyophilized RNase A (from bovine pancreas) was purchased from Sigma Chemie GmbH (Deisenhofen, Germany).
Proteasome Preparation-Proteasome was isolated from erythrocytes of outdated human blood conserves according to Hough et al. (37). Erythrocytes were lysed in 1 mM dithiothreitol. After the removal of membranes and nonlysed cells by centrifugation, the proteasome was isolated by DEAE chromatography, sucrose density gradient ultracentrifugation, and separation on a Mono Q column of a fast protein liquid chromatography system.
RNase A Treatment with Hydrogen Peroxide-All RNase A treatments were performed under standardized conditions to ensure reproducibility of the FT-IR spectroscopic and proteasome degradation experiments. Stock solutions containing RNase A were prepared in a 10 mM cacodylat/D 2 O-buffer at pH* 7.1 to yield an enzyme concentration of 20 mg/ml. The protein solutions were heated for 15 min at 60°C to achieve complete H/D exchange of the amide protons (28 Measurement of Proteolytic Susceptibility-Prior to the measurements of the proteolytic susceptibility, all RNase A solutions were diluted to 1 mg/ml. The degradation of native and oxidized RNase A by the 20 S proteasome was measured by incubation of 40 g of substrate protein with 0.6 g of proteasome in a proteolysis buffer containing 50 mM HEPES (pH 7.8), 20 mM KCl, 5 mM magnesium acetate, and 1 mM dithiothreitol for 2 h at 37°C. The reaction was stopped by the addition of an equal volume of trichloroacetic acid (20%). After centrifugation (15 min, 12,000 ϫ g) the supernatants containing primary amines were neutralized with 1 M HEPES (pH 7.8). Fluorescamine (0.3 mg/ml in acetone) was added under vigorous shaking, and the fluorescence was determined at 390-nm excitation/470-nm emission. Leucine was used as a standard. Proteolysis rates were calculated as the difference between the measured value and the blank value (sum of incubated substrate protein without protease and of incubated protease only).
Mass Spectroscopy of Oxidized RNase A-To exclude the possibility of hydrogen peroxide-induced RNase A backbone fragmentation, RNase A samples (10 mg/ml) were incubated for 48 h with H 2 O 2 (0, 1, 2, and 4 mol of H 2 O 2 /mg of protein) and subsequently analyzed by matrixassisted laser desorption ionization-time of flight mass spectroscopy. A comparative analysis of the mass spectra indicated that no new peaks of the oxidized RNase A occurred between 300 and 60,000 Da. Thus, it could be experimentally confirmed that mild oxidation of RNase A does not cause protein backbone fragmentation. However, small shifts of RNase A peaks were observed. We interpreted these experimental findings as a result of covalent side chain modification of RNase A.
FT-IR Spectroscopic Measurements-Infrared spectra were recorded using a Bruker IFS28B FT-IR spectrometer equipped with a computercontrolled sample changer and a deuterated triglycin sulfate detector. To eliminate spectral contributions due to atmospheric water vapor, the instrument was continuously purged with dry air. After mixing, the solutions of RNase A and H 2 O 2 were quickly transferred to an IR sample cell consisting of a pair of CaF 2 windows separated by a 50-m spacer. FT-IR measurements were carried out at a temperature of 30°C. For each spectrum, 1024 interferograms were coadded and Fourier-transformed employing a Happ-Genzel apodization function and a zero filling factor of 4. Nominal resolution was 4 cm Ϫ1 . For Fourier self-deconvolution, the software routines implemented in the manufac-turer software package OPUS (Bruker) were used. Second derivative spectra were evaluated applying a Savitzky-Golay algorithm with seven smoothing points. To visualize and to interpret small spectroscopic changes, difference spectra D t were calculated in the following way, where D t is the difference spectrum at a given time t, while A last and A t are absorbance spectra recorded at the end of a measurement series or at time t. Therefore, negative bands of the difference spectra indicate decreasing amounts of the respective structures, and vice versa. FT-IR Temperature Gradient Measurements-Temperature profiles were carried out on a Bruker IFS66 FT-IR spectrometer, which was equipped with a deuterated triglycin sulfate detector. Again, a Happ-Genzel apodization function, a zero filling factor of 4, and a nominal resolution of 4 cm Ϫ1 was used. For temperature profile measurements, infrared spectra were collected continuously in a thermostated IR cuvette with an optical path length of 50 m. For this type of measurement, RNase A was dissolved in 10 mM cacodylat/D 2 O buffer at pH* 7.1 to a concentration of 20 mg/ml. The protein was unfolded and refolded in two consecutive cycles by applying a linear temperature gradient of 0.5 K/min between 20 and 80°C and back to 20°C. Complete H/D exchange of the amide protons was obtained after the first heating cycle. For data evaluation, only spectra of the second heating run (unfolding) were utilized. Infrared spectra were corrected for spectral contributions of buffer and water vapor as described previously (33). Difference spectra were calculated in the same way as outlined above.
Calculation of the Accessible Surface Area-The accessible surface areas (ASA) were calculated using the program Naccess version 2.1 (38), which is an implementation of the Lee and Richards method (39). The program calculates the atomic accessible surface defined by rolling a probe of given size around a van der Waals surface. A slice thickness of 0.05 Å and a probe size of 1.4 Å for H 2 O and 2.1 Å for H 2 O 2 was used. The latter value was estimated on the basis of the bond length and the atom radii of O 2 and F 2 . We utilized an output file containing summed atomic ASA over each residue. For these calculations, the coordinates from the Protein Data Bank ID 1rbx (RNase A without ligand) by J. L. H. Dunbar and G. K. Farber (40) were used.

RESULTS
To obtain new insights into the recognition process of oxidized proteins by the proteasome, we combined two different techniques: structural changes of the model protein RNase A upon oxidation by hydrogen peroxide were followed by FT-IR spectroscopy, while the changes of susceptibility to proteasome degradation were tested by biochemical methods. For this purpose, stock solutions of RNase A and H 2 O 2 were prepared and used for both kinds of experiments. Although the use of D 2 O buffers or fully H/D-exchanged proteins is not essential for the proteasome degradation characterizations, we tried to maintain the experimental protocols as much as possible. Therefore, the first steps of proteasome digestion were performed in D 2 O buffers after a complete H/D exchange of the protein.
Proteolytic Susceptibility of RNase A-The degradation of various proteins by the 20 S proteasome was already measured by several groups for a number of different proteins (5,14,15). Since most of these experiments were performed under standardized conditions, we used the same experimental protocol to study the proteolytic susceptibility of RNase A toward the 20 S proteasome. The results of the investigations are demonstrated in Fig. 1. When RNase A is treated with hydrogen peroxide concentrations up to 4 mol/mg of protein (approximately 55 molecules of H 2 O 2 per molecule of RNase A), an increase in proteolytic susceptibility of the resulting oxidized proteins is detected. At higher concentrations of hydrogen peroxide, a decrease of the proteolytic susceptibility was measured. As previously postulated and shown earlier (5,13,14,41,42), this decline is due to the irreversible formation of protein aggregates. These protein aggregates are poor substrates for the protease or may even be able to inhibit the 20 S proteasome (15,18,41,42). One of the objectives of the present work was to investigate the structural rearrangements of the protein induced by oxidation, which causes the recognition and the deg-radation of the protein by the 20 S proteasome. Consequently, hydrogen peroxide concentrations below 4 mol/mg of protein were applied in all experiments. Since FT-IR spectroscopy requires relatively high protein concentrations (ϳ10 mg/ml), the hydrogen peroxide concentration was varied from 0 (control) to 40 mM. The hydrogen peroxide/protein concentration ratio was therefore identical to that utilized in the proteasome degradation experiments (0 -4 mol of H 2 O 2 /mg of protein).
FT-IR Spectroscopic Characterization of Oxidized RNase A-FT-IR spectroscopic studies of proteins in aqueous solutions are frequently performed in D 2 O buffers for the following reasons. The amide I band (1620 -1690 cm Ϫ1 ), arising predominantly from the CϭO oscillators of the secondary amide function of protein backbone, is superimposed on the deformation band at 1643 cm Ϫ1 of the solvent H 2 O. Thus, in many FT-IR spectroscopic studies on protein structure, H 2 O is substituted by 2 H 2 O (D 2 O), which exhibits its deformation band at 1210 cm Ϫ1 . Thus, the use of D 2 O instead of H 2 O as the solvent makes it possible to analyze spectral features in the secondary structure-sensitive amide I region without interference of bulk water.
The interpretation and quantification of structural changes of RNase A during oxidation by hydrogen peroxide was carried out, comparing these data with infrared spectra for thermally unfolded protein. It is known that RNase A can be reversibly unfolded thermally, giving rise to a characteristic spectral unfolding pattern, particularly in the amide I region (29,33,35,43). In the present study, FT-IR spectra were collected between 20 and 80°C, applying two consecutive heating and cooling cycles with a linear temperature gradient of 0.5 K/min. The FT-IR spectrum of native RNase A is shown in Fig. 2. This spectrum was recorded at a temperature of 30°C after one heating and one cooling cycle (i.e. after complete thermal unfolding and refolding), which accelerates the exchange of the amide protons by deuterons (in the following, this procedure is called H/D exchange). The dotted line in Fig. 2A displays the absorbance spectrum corrected for the spectral contributions of the buffer, while the solid line represents the corresponding Fourier self-deconvolution spectrum ( Fig. 2A) to demonstrate the fine structure of the amide I band. A second derivative spectrum (Fig. 2B) was also calculated from the absorbance spectrum (positive bands of an absorbance spectrum appear in second derivative spectra as negative bands). From the literature, it is known that the energy of the CϭO oscillators of the protein backbone depends on the coupling to adjacent CϭO oscillators and the strength of the hydrogen bonds. Furthermore, the strength of these bonds and the symmetry of hydrogen bond patterns is characteristic for distinct secondary structure elements of proteins. It is therefore possible to distinguish various secondary structures from the experimentally observed band components of the amide IЈ band. For RNase A, the band assignment was carried out according to literature data (27,29,43,44). In good agreement with these studies, IR marker bands for antiparallel ␤-pleated sheets were found at 1631 and 1680 cm Ϫ1 , for ␣-helix at 1651 cm Ϫ1 , and for unordered turn structures at 1665 cm Ϫ1 (cf. Fig. 2). Also, infrared absorption bands of the amino acid side chains such as the "tyrosine band" (aromatic tyrosine ring vibration at 1515 cm Ϫ1 ) or absorptions of aspartate and glutamate residues at 1584 and 1566 cm Ϫ1 , respectively, were observed in agreement with previous studies (45,46).
Spectroscopic parameters of the amino acid side chain absorption bands can be utilized to derive structural information on the specific microenvironment of these functional groups. For example, the frequency of the tyrosine band at 1515 cm Ϫ1 was used to monitor specifically the formation of tertiary contacts upon refolding (29). Fig. 3A displays a series of Fourier self-deconvolution infrared spectra (corrected for buffer) obtained by a linear temperature gradient measurement. As previously described for RNase A (29,43), the appearance of a broad and featureless amide I band suggests the lack of stable secondary structure elements at temperatures above 70°C.
A frequently used approach to illustrate temperature-induced spectral changes is given in Fig. 3B, which shows a series of FT-IR difference spectra. These difference spectra have been calculated according to D x ϭ A 80°C Ϫ A x , where A 80°C is the absorbance spectrum at 80°C and A x is an absorbance spectrum at the temperature x. Two negative bands at 1631 and 1680 cm Ϫ1 reflect the disappearance of antiparallel ␤-pleated sheet structures upon temperature-induced unfolding. The broad band between 1651 and 1680 cm Ϫ1 (Fig. 3A) indicates the thermally induced formation of unordered structures. These spectroscopic changes induced by thermal unfolding of RNase The dotted curve displays an FT-IR spectrum of RNase A (buffer subtracted), while the solid curves were obtained applying "resolution enhancement techniques" to the original data (curve A, Fourier selfdeconvolution; curve B, second derivative spectrum). The amide I region of RNase A spectra (1690 -1620 cm Ϫ1 ) is dominated by band components at 1631 and 1680 cm Ϫ1 , both assigned to antiparallel ␤-pleated sheets. Other amide I band components at 1665 cm Ϫ1 (unordered structure) and 1651 cm Ϫ1 (assigned to ␣-helical structures) are indicated. Absorptions of amino acid side chains at 1515 cm Ϫ1 (tyrosine ring vibration) and near 1584 and 1566 cm Ϫ1 (-COO Ϫ vibrational modes of the aspartate and glutamate residues), although less intense than the amide I band, can be used to obtain additional structural information. AU, absorbance units.
A will be compared quantitatively with the spectroscopic effects observed during the oxidation of RNase A by hydrogen peroxide. For quantification of the fraction of unfolded protein, the absorbance/temperature dependence of the most prominent amide IЈ contour at 1631 cm Ϫ1 (low frequency ␤-band) was analyzed (see Fig. 4). Obviously, three main phases of this absorbance/temperature plot can be observed: a linear low temperature region (below 53°C), a sigmoidal transition region (53-72°C), and a second linear post-transitional high temperature region above 72°C. The midpoint (inflection point) of the protein melting curve T m , at which 50% of RNase A is supposedly in the unfolded state, was determined by curve fitting to be 64.5°C, in accordance with literature data (Backmann et al. Two linear fits of the high and low temperature region and the extrapolation of both curves to the temperature of the protein degradation experiments of 30°C allowed us to estimate the value of absorbance difference ⌬A 30°C . This value indicates the absorbance difference between the completely folded and the completely unfolded states of the protein at a temperature of 30°C. Finally, to account for the protein concentration dependence, the ⌬A 30°C value was divided by the absorbance of the completely folded RNase A species A f,30°C (see Fig. 4). The relation (⌬A/A f ) 30°C was set to 100% and was found to be particularly useful in quantifying the IR-spectroscopic protein denaturation features induced by H 2 O 2 . The procedure described can be applied similarly to obtain quantitative information from other infrared marker bands (e.g. from the tyrosine band at 1515 cm Ϫ1 ).
The spectroscopic features of the temperature-induced reversible unfolding and the partial denaturation of RNase A The absorbance values were obtained from buffer-subtracted FT-IR spectra of a heating run from 20 to 80°C applying a constant heating rate of 0.5 K/min. Both the dependence in the low (Ͻ50°C) and the high temperature region (Ͼ72°C) were fitted with a linear function and extrapolated to a temperature of 30°C. Assuming that RNase A is completely unfolded at T Ͼ 72°C and completely folded at 30°C, one can equate the absorbance difference ⌬A 30°C (equal to 0.073 absorbance units (AU)) between both states to 100% loss of secondary structure of the protein. The ratio (⌬A/A f ) 30°C of the temperature unfolding experiment is concentration-independent and was therefore used for quantification of H 2 O 2 -induced structural alterations of RNase A. Quantitative estimations of other structuresensitive IR bands (e.g. the tyrosine ring vibration band at 1515 cm Ϫ1 ) were carried out in the same way. where RNase A is expected to exist in a completely unfolded state. The second derivative spectra a-d were normalized utilizing the tyrosine ring vibration band at 1515 cm Ϫ1 as an internal standard. A comparison of spectra b (temperature induced unfolding) and c (oxidation) indicates a high degree of conformity of the IR spectroscopic changes, particularly in the secondary structure-sensitive amide IЈ region.
tions after 48 h of incubation of RNase A with H 2 O 2 at a concentration of 4 mol/mg of protein. A comparison of spectra b and c clearly demonstrates that thermal unfolding and hydrogen peroxide degradation lead to similar FT-IR spectroscopic changes, characterized by the partial disappearance of several narrow amide IЈ band components. These are the high and low frequency ␤-bands at 1680 cm Ϫ1 and 1631 cm Ϫ1 and the peak at 1651 cm Ϫ1 , which was assigned to ␣-helical structures. However, it should be pointed out that spectra b and c were recorded at different temperatures (30 and 65°C), which may help explain why they are not identical. Spectrum d of  Fig. 6, difference spectra were calculated according to D t ϭ A last Ϫ A t , where A t is the absorbance spectrum at a given time t, and A last is the last absorbance spectrum of the FT-IR spectroscopic time series. Again, negative bands indicate disappearance of respective structure elements upon hydrogen peroxide-induced protein oxidation, and vice versa. A comparison of the IR spectral differences observed for the temperature-induced unfolding of RNase A (Fig. 3) and the hydrogen peroxide-mediated RNase A degradation (Fig. 6) suggests remarkable similarities in the structure change between both processes. The most notable spectral changes are observed in the amide IЈ region, particularly at 1631 and 1680 cm Ϫ1 (low and high frequency antiparallel ␤-pleated sheet structures). Another interesting feature that coincides with variations in the amide IЈ region is a subtle but highly reproducible frequency shift of the tyrosine band at 1515.20 cm Ϫ1 of about ⌬ 30°C Ϸ 0.5 cm Ϫ1 to higher wave numbers. This behavior is interpreted in the literature as a consequence of changes in the microenvironment of the aromatic tyrosine residues (36,47). Upon unfolding of RNase A, the aromatic tyrosine residues become more solvent-exposed, resulting in the appearance of a new band at 1515.70 cm Ϫ1 . This band cannot be resolved spectroscopically from the original tyrosine band at 1515.20 cm Ϫ1 but is observed via a "band shift" to higher frequency. The "frequency shift" of the tyrosine band near 1515 cm Ϫ1 is accompanied by changes of the absorbance values of this band. Upon temperature-induced unfolding of RNase A between 53 and 72°C, the absorbance value of the tyrosine band increases. This temperature-dependent spectral effect is very small (cf. Fig. 3) compared with the changes observed in the amide I region. Interestingly, the difference spectra shown in Fig. 6, A-C, display a clear decrease of absorbances of the tyrosine band with increasing hydrogen peroxide concentration or time, respectively. Furthermore, for the aspartate absorption band at 1584 cm Ϫ1 , the decrease of absorbance (at 4 mol of hydrogen peroxide/mg of protein, 48-h incubation time) was about twice as high as the corresponding effect observed during complete unfolding of RNase A. Obviously, some specific marker bands due to distinct amino acid residues may indicate covalent modifications of amino side chains by hydrogen peroxide. In contrast, changes in absorbance value of the amide I band contour indicate global conformation changes.
The difference spectra, given in Figs. 3B and 6B, also exhibit significant variations in the spectral region from 1620 to 1580 cm Ϫ1 . While the difference spectra obtained upon reversible thermal unfolding of RNase A display no significant changes in this spectral region (Fig. 3B), at least one positive band at 1594 cm Ϫ1 is found for the RNase A oxidation measurement series (Fig. 6B). This band is assigned in the literature to an asymmetric carboxylate stretching vibration ((-COO Ϫ ) as ). The generation of additional carboxylate groups during protein oxidation may be interpreted as a result of disruption of covalent bonds of amino acid side chains. To some extent, however, these spectral features may overlap with a so-called ␤-aggregation band near 1615 cm Ϫ1 . The formation of protein aggregates is widely observed for many proteins (48 -51). A more detailed examination of difference spectra in Fig. 6B reveals the simultaneous appearance of an additional shoulder at 1714 cm Ϫ1 in the positive band contour around 1695 cm Ϫ1 , which is not present in the difference spectra of the unfolding experiments (Fig. 3B). Carbonyl stretching bands can be expected around this wave number. Some authors reported the generation of carbonyl groups during the oxidation of amino acid side chains producing carbonyl derivatives (1,52). Fig. 6C shows the difference spectra of Fig. 6B below 1500 cm Ϫ1 . These difference spectra display a small positive peak at 1047 cm Ϫ1 ((R 2 )-SϭO stretching vibration), which is indicated by an arrow. The amplitude of this difference peak is compa- rable with the intensity changes of the tyrosine band at 1515 cm Ϫ1 (also marked by an arrow). The difference pattern in the spectral region 1180 -1240 cm Ϫ1 results from the very strong D 2 O deformation signal at 1210 cm Ϫ1 . Fig. 7 illustrates the time-dependent structural changes of RNase A induced by H 2 O 2 treatment. These plots were derived using the absorbances of the low frequency ␤-band at 1631 cm Ϫ1 (Fig. 7A) and the absorbance values of the tyrosine band at 1515 cm Ϫ1 (Fig. 7B). To estimate quantitatively the changes of distinct structural elements as a function of H 2 O 2 treatment, the following model was used. From the temperature gradient measurements, the value of the parameter (⌬A/A f ) 30°C is known. In the example of Fig. 4, this ratio indicates the maximal absorbance change of the low frequency antiparallel ␤-pleated sheet band at 1631 cm Ϫ1 , which is a concentrationindependent ratio describing complete unfolding of the protein.
According to Equation 2, this ratio can be used to estimate the percentage of hydrogen peroxide induced structural changes of RNase A.
Loss of structure elements ͑t͒ ϭ The denominator of this equation was calculated from spectral parameters of the temperature profile measurements, and the numerator was calculated from parameters of the hydrogen peroxide experiments. P 0 and P t are any structure-sensitive spectroscopic parameter like the absorbance at a given frequency, band frequency, or the half-width at t ϭ 0 or at a given time t, respectively. We used Equation 2 to compare semiquantitatively the structural changes of RNase A oxidation with the structural changes detected during temperature-induced unfolding. This approach allowed us to analyze structural changes in RNase A using the infrared marker bands from the amino acid side chains or secondary structure-sensitive components of the amide I band. Fig. 7 shows the time and concentration dependence of the absorbances of two distinct spectral marker bands: the low frequency antiparallel ␤-pleated sheet band at 1631 cm Ϫ1 , and the tyrosine band at 1515 cm Ϫ1 . Generally, the higher the concentration of hydrogen peroxide and the longer the incuba-tion time, the larger were the structural alterations detected by FT-IR spectroscopy. The plots of Fig. 7, A and B, and the quantitative results of Table I derived from these plots suggest that substantial differences exists between various secondary structure-sensitive marker bands (amide I band, absorbances at 1680, 1651, and 1631 cm Ϫ1 ) and IR bands of defined amino acid residues (absorbance of the tyrosine and aspartate band at 1515 and 1584 cm Ϫ1 ). While the direction and the time-dependent changes in the amide I region indicate an "unfolding-like" behavior of the protein as a result of H 2 O 2 treatment, the information derived from the amino acid side chain bands cannot be explained simply on the basis of protein unfolding (see ordinate values of Fig. 7B). These interesting findings will be discussed below.
Oxidation of RNase A by Hydrogen Peroxide (120-h Incubation)-Armed with the knowledge that the oxidative damage of RNase A may be accompanied by conformational and covalent modifications, we undertook a series of experiments to detect subtle spectral changes, especially of the amino acid side chain absorptions. Those measurements required a very high spectral signal/noise ratio and instrument stability. Therefore, the sampling time for each individual spectrum of these experiments was increased to 20 min (2200 scans), and the time of data collection was increased to 120 h. The experiments shown in Fig. 7 (48-h duration) were carried out at a temperature of 30°C. The protein concentration and the composition of the buffer solutions were not changed. To analyze the smallest spectroscopic effects detectable and to minimize unavoidable base line shifts (a common problem of long time measurements), the spectroscopic parameters were calculated exclusively from second derivative spectra (see Fig. 8). High signal/ noise ratio (see Fig. 8, B or C), stability of the experimental setup over 5 days (temperature, water vapor content, etc.), and state-of-the-art data evaluation were essential prerequisites to be able to obtain information from these types of experiments.
The interpretation of band parameters derived from second derivative spectra is not as straightforward as for data obtained from the original absorbance spectra. The calculation of d(A) 2 /d 2 () values from a simulated Lorentz band yields a negative band where the position of the minimum of the second derivative band coincides with the frequency value of the maximum of the original Lorentzian, while the ordinate value at the minimum depends on the half-width and the absorbance value of the Lorentz band. Therefore, one can utilize second derivative intensities for comparative purposes in the same way as absorbances from original spectra, provided that the half-widths of the bands are constant and do not depend on temperature or other experimental parameters. For the determination of band frequency, second derivatives are even more precise than the original absorbance spectra.
In Fig. 8, the low frequency ␤-pleated sheet band absorbance at 1631 cm Ϫ1 (Fig. 8, A and D) and the absorbance and the frequency values of the tyrosine band at 1515 cm Ϫ1 (Fig. 8, B, C, E, and F) were evaluated from second derivative spectra. While the curves in the left column of Fig. 8 (A-C)  series was carried out using Equation 2, where the spectral parameters P i were obtained from second derivative spectra.
Curve I in Fig. 8A indicates a significant decrease of the absorbances of the low frequency ␤-band intensity at 1631 cm Ϫ1 in accordance with the measurement series of Figs. 5-7. In general, the magnitude and the direction of the time-dependent changes detected in the amide I region coincide with those observed during thermal unfolding of RNase A. In contrast, the spectral changes of the tyrosine band at 1515 cm Ϫ1 (induced by hydrogen peroxide or by temperature variation) have an opposite direction. As indicated by the FT-IR difference spectra of Fig. 6, A-C, the intensity of the tyrosine band is decreasing as a function of hydrogen peroxide concentration and incubation time, while it is increasing with temperatureinduced unfolding of the protein. Consequently, the ordinate values of Fig. 8B are negative (see also Fig. 7B). Specifically, the absorbance changes in the amide I region indicate a decrease of about 70% of secondary structure elements after 120 h of incubation time at 1 mol of H 2 O 2 /mg of protein, while the corresponding tyrosine band absorbances changes are comparable in the magnitude but opposed in direction. Interestingly, the absorbance time-dependence of the "aspartate band" at 1584 cm Ϫ1 exhibits much larger changes in the oxidation experiment compared with the effects of the unfolding measurements (cf. Table I). But in contrast to the behavior of the tyrosine band, the intensity of the aspartate band at 1584 cm Ϫ1 is diminishing upon unfolding of RNase A as well as upon oxidation. Thus, relative absorbance changes calculated by Equation 2 were found to be positive for the aspartate band at 1584 cm Ϫ1 and negative for the tyrosine band at 1515 cm Ϫ1 (Table 1).
In analogy to the analysis of the band intensity of the ␤-structure band at 1631 cm Ϫ1 , the frequency analysis of the tyrosine ring vibration band at 1515 cm Ϫ1 indicates similarity between temperature (Fig. 8F) and hydrogen peroxide (Fig.  8C)-induced structural alterations of RNase A. The temperature-induced unfolding is accompanied by a tyrosine "peak shift" to higher wave numbers by about ⌬ 30°C Ϸ 0.4 cm Ϫ1 . A similar effect was observed for the 120-h H 2 O 2 incubation experiment, where a tyrosine peak shift of about 0.35 cm Ϫ1 indicates a global unfolding of about 70% of the RNase A molecules. Interestingly, this value is close to that determined from the high frequency ␤-pleated sheet band at 1680 cm Ϫ1 (not shown) and from the low frequency ␤-pleated sheet band at 1631 cm Ϫ1 (cf. Fig. 8A). Presumably, the intensity of the IR marker bands such as the high and the low frequency ␤-pleated sheet band at 1680 and 1631 cm Ϫ1 and the frequency of the tyrosine band monitor global folding/unfolding events of the protein, while other spectral markers such as the intensities of the tyrosine and the aspartate bands reflect more local consequences of protein oxidation (chemical modifications).
Time Dependence of the Proteolytic Susceptibility of RNase A-As demonstrated by Fig. 1, the reaction with hydrogen peroxide up to 4 mol of H 2 O 2 /mg of RNase A leads to an increase of proteolytic susceptibility of RNase A. Since it is All data are given in percent relative to spectral alterations of RNase A during complete thermal unfolding (see "Results"). While infrared bands for secondary structure elements (e.g. "␤-bands") or the tyrosine peak shift (tertiary contacts) monitor the global unfolding process, other spectral markers like the absorbance of Tyr or Asp bands probably reflect local events, presumably covalent modifications of the amino acid side chains induced by hydrogen peroxide.  8. A comparison of spectral changes induced by temperature increase of RNase A solutions (unfolding) and hydrogen peroxide-induced structural alterations of RNase. All spectroscopic parameters of this figure were calculated from second derivative spectra. Left row, time dependence of band parameters obtained on a hydrogen peroxide/RNase A incubation experiment (I) and the respective control experiment (II). The relation of intensity values at a given time t and t ϭ 0 (A and B) or frequency values (C) were calculated and plotted as a function of time. Curves in A were obtained, using the absorbance information at 1631 cm Ϫ1 (low frequency ␤-pleated sheet band), while B and C display the time dependence of the absorbance or frequency parameters of the tyrosine band at 1515 cm Ϫ1 , respectively. Right, temperature profile measurements of RNase A. D and E display the normalized second derivative/temperature dependence at 1631 and 1515 cm Ϫ1 , respectively, while F shows the frequency/temperature dependence of the tyrosine band. A comparison of the curves in the left and right rows indicates that hydrogen peroxide-induced structural alterations of RNase A are to some extent similar to spectral changes during the unfolding of the protein. However, the structural information derived particularly from infrared marker bands of the amino acid side chains makes it evident that additional "local" events (e.g. oxidation of aromatic residues) may take place (see "Results"). known that protein damage is not only a concentration-dependent but also a time-dependent process, we also investigated the time dependence of the proteolytic susceptibility of RNase A. In this study, only the incubation time of RNase A with hydrogen peroxide was varied, while the reaction time of the 20 S proteasome with the oxidized RNase A was kept constant (2 h). In this way, we measured the proteolytic susceptibility of the substrate and not a maximal amount of protein oxidation. As shown in Fig. 9, a time-dependent increase in the proteolytic susceptibility of the RNase A up to 48 h was detected. At all time points tested, a clear dependence of proteolytic rates on the H 2 O 2 concentration was found.
Correlation of Proteolytic Susceptibility and FT-IR Spectroscopic Changes-One aim of this study was to get new insights into the molecular basis of the recognition process of mildly oxidized RNase A by the 20 S proteasome. Assuming that distinct structure-sensitive IR spectroscopic parameters of oxidized RNase A are related to the proteolysis rates, we correlated hydrogen peroxide-induced changes of specific infrared bands with the proteolytic susceptibility of RNase A toward the 20 S proteasome. Fig. 10 clearly shows that hydrogen peroxide-induced structural reorganization of the protein coincides with increased proteolysis rates of RNase A. For all infrared bands analyzed in this study, i.e. for "global" as well as for "local" parameters, a clear linear dependence between structural changes and the proteolytic susceptibility was detected. Consequently, this approach did not allow the identification of a specific recognition site causing proteasome binding and subsequent protein degradation. However, an improved experimental setup to increase the sensitivity and the time resolution of the experiment may be helpful to address this problem more adequately. DISCUSSION Covalent modifications of proteins by oxidative agents are thought to play a key role in various physiological and pathological conditions such as inflammation, ischemia reperfusion, or aging (for reviews see Oliver et al. (54) and Stadtman (55)). It has been shown that oxidative modification of proteins can be mediated by a number of different systems including oxidases, ozone, hydrogen peroxide, hypochloride, superoxide, ␥-irradiation, and metal-catalyzed oxidation, to mention a few. As a result of protein oxidation, an accumulation of enzymes with partially altered structure and function is observed. These proteins exhibit changes in thermostability (41) and, when mildly oxidized, show an increased susceptibility to degradation by proteasome (1,2,5,6,15,18), which is known to be a part of a complex antioxidant repair and removal system. This main intracellular proteolytic system for the degradation of oxidatively damaged proteins exists in at least two forms, the ATP-and ubiquitin-dependent 26 S form and an independent 20 S form. The 20 S proteasome used in this study is a 700-kDa soluble proteinase complex that is found in the cytosol and the nucleus of mammalian cells (56). Proteolysis of oxidatively damaged proteins by the 20 S form of the proteasome seems to be the major pathway (5,6,18). As observed by a number of studies, proteolysis by the 20 S proteasome is most efficient after treatment of the protein with moderate oxidative stress, whereas greater oxidative damage actually leads to decreased proteolytic susceptibility (15,17,57). It has been shown that the formation of protein aggregates, by whatever mechanism, may contribute to the decrease of proteolytic rate (41).
However, the molecular basis of the recognition process of mildly oxidized proteins by the proteasome still remains questionable. Experimental evidence for a possible role of increased surface hydrophobicity (16, 20 -22), the formation of dityrosines (57), the conversion of methionine to methionine sulfoxide (21), and the increase of reactive carbonyl content (21) were found. Nevertheless, an in vitro study showing a direct relationship between structural changes of a model protein and proteolytic susceptibility to the 20 S proteasome is still lacking. To get new insights into the recognition process of mildly oxidized proteins by the 20 S proteasome, we combined a structure-sensitive spectroscopic technique, FT-IR spectroscopy, with measurements of the proteolytic susceptibility of RNase A toward the 20 S proteasome. We report here experimental evidence for oxidation-induced conformational rearrangements of the secondary structure of the model protein RNase A and, at the same time, for covalent modifications of amino acid side chains such as tyrosine and aspartate residues. These modifications could be correlated to the proteasome-induced proteolysis rate of oxidatively damaged RNase A. Oxidatively damaged RNase A can be degraded by the 20 S proteasome. This degradation exhibits a typical biphasic dependence; at comparably low concentrations of H 2 O 2 , the proteolytic rate of the mildly oxidized protein is increased if the oxidant concentration is increased. At concentrations above 4 mol of hydrogen peroxide/mg of protein, an inverse relation is observed (cf. Fig. 1). These findings are in good agreement to literature data for other model systems (6,15,58), which were a starting point for the present study.
The remarkably high degree of similarity between the infrared difference patterns, particularly in the amide I region, for oxidation-and temperature-induced unfolding of RNase A prompted us to propose a correlation between hydrogen peroxide-induced protein modifications and protein unfolding. In fact, the similarity of spectral characteristics observed for both processes (cf. spectra b and c in Fig. 5) support the hypothesis that oxidative damage of proteins may result in global unfolding rearrangements of the polypeptide. This is also corroborated by the thermal and denaturant-induced unfolding of RNase A (29). The proposed correlation is supported by two facts. First, a quantitative comparison of RNase A oxidation using several spectroscopic parameters of the amide I region (such as high and low frequency ␤-band or the band for ␣-helical structure) indicate a remarkably high degree of internal consistence (Table I). Second, even when relatively high concentrations of the oxidants were applied for time periods up to 120 h, the changes of all secondary structure elements of the amide I region were always smaller then the changes observed during the temperature-induced two-state transition of global unfolding. Thus, the results of the protein oxidation experiments suggest that the hydrogen peroxide-induced structural rearrangements of RNase A are processes similar to global unfolding, involving all parts of the protein. Protein backbone fragmentation may be a potential candidate producing unfolded fragments of RNase A. However, the results of mass spectrometry demonstrate that oxidative unfolding of RNase A does not perturb the protein sequence.
The detailed analysis of some amino acid side chain absorption bands indicates distinct discrepancies that cannot be explained on the basis of the simple model of protein unfolding. While the absorbance values of the tyrosine band at 1515 cm Ϫ1 are increasing during temperature-induced unfolding of RNase A, the oxidation experiments exhibit a decrease of this band. Furthermore, for the absorption band of the aspartate residues at 1584 cm Ϫ1 we found a decrease in intensity twice as large compared with the corresponding changes observed during complete unfolding. From the literature, it is known that side chains of proteins are primary targets of oxidation (e.g. Ref. 57). Therefore, it is possible that some of the tyrosine (and aspartate) side chains of RNase A were oxidized by hydrogen peroxide. Tyrosine is one amino acid residue that can readily undergo oxidation, forming a number of reaction products such as dityrosines (57). RNase A contains six tyrosine residues and, as the calculations of the ASA show, only one of these side chains is initially not accessible by the solvent (Tyr 97 ). The remaining residues revealed relative ASA values between 11 and 65% (Tyr 25 and Tyr 65 ). Therefore, an oxidative modification of at least some tyrosine residues seems likely.
Unlike the absorbances of the tyrosine band, the time dependence of the tyrosine band frequency confirmed quantitatively the corresponding structural alterations found in the amide I region (see Table 1). It is expected that any covalent modification of the tyrosine ring will change its spectral characteristics considerably. Thus, oxidation of the tyrosine residues will necessarily be followed by a decrease of the tyrosine band intensity at 1515 cm Ϫ1 (and accordingly by the appear-ance of a number of new bands), whereas the frequency of the tyrosine band should remain unchanged. Therefore, we interpret frequency shifts of the tyrosine band during RNase A oxidation as a consequence of changes in the microenvironment, in this case due to the unfolding of the protein.
Previous literature demonstrated (25) that under nonphysiological high concentrations of the oxidant, proteins may aggregate to large and densely packed complexes that are not accessible to proteases. Aggregates can be either formed by covalent cross-linking (dityrosines; Ref. 57) or by intermolecular hydrophobic interactions (intramolecular ␤-sheets; Ref. 20). According to these well established facts, the decrease of proteasome degradation activity at high H 2 O 2 /protein concentration ratios (cf. Fig. 1) was correlated to the formation of aggregates of this protein. Previous IR studies have shown (48,50,51) that densely packed intermolecular ␤-aggregates give rise to highly characteristic infrared bands. Particularly, a so-called "␤-aggregation band" at 1615 cm Ϫ1 is known to be typical for the presence of these aggregates. The FT-IR measurements of the present study, however, were carried out at comparably low H 2 O 2 /protein concentration ratios, i.e. when protein aggregation is supposed to play only an inferior role. It cannot be ruled out completely that the difference spectra shown in Fig. 6B show small signs of an aggregation band. However, the presence or absence of a minor ␤-aggregation band seems to be irrelevant for the interpretation of the data of the present study.
Oxidative modification of amino acid side chains can severely reduce the conformational stability of proteins. Stadtman and colleagues (52) used an example of two-model proteins to demonstrate that during exposure of amino acids to ozone. Primarily methionine and aromatic amino acid residues were oxidized in the order Met Ͼ Trp Ͼ Tyr Ϸ His Ͼ Phe (52). Hence, methionine residues appear to be the primary targets of oxidation. It was furthermore proposed that methionine residues constitute an effective antioxidant on the surface of proteins (14). Since the oxidation product of methionine, methionine sulfoxide, can be recycled by a catalytic system, methionine residues may act as oxidant scavengers. Interestingly, for many proteins it was demonstrated that protein conformation is little affected if solvent-accessible methionines on the protein surface are oxidized (14, 59 -61). These studies support the hypothesis that surface exposed methionines generally preserve the biological function of the protein.
The S-C stretching vibration of methionines (R-CH 2 -S-CH 3 ) can be expected in the spectral region of 730 -570 cm Ϫ1 (62). The experimental setup used in this study (CaF 2 -windows) did not allow us to monitor IR-spectroscopic changes in this low frequency region. Nevertheless, the absorption band arising from the SϭO stretching vibration of methionine sulfoxide ((R 2 )-SϭO) is known to occur between 1060 and 1015 cm Ϫ1 (62). In fact, at 1047 cm Ϫ1 a small and narrow positive band (cf. Fig.  6C) was observed for the RNase A oxidation experiments, while the control measurements did not indicate any additional peak at these wave numbers. In RNase A, the four methionine residues are poorly accessible by the oxidant. For the side chain of Met 29 , an ASA value of 18% was calculated (Met 13 , 7%; Met 30 , 0%; Met 79 , 9%). However, we demonstrated that large scale disorganization of the protein conformation occurs upon H 2 O 2 treatment, which implies that Met residues could have been oxidized that were initially shielded.
The other aromatic amino acid side chains of RNase A such as Phe (there is no Trp in RNase A) are also poorly accessible. For three Phe side chains, ASA values between 0 and 8% were calculated. On the other hand, the three His residues of RNase A, which are also known to undergo oxidative modification, exhibit ASA values of Ͼ35%. Therefore, we think that in native RNase A, the primary targets of oxidation are tyrosine and histidine residues. This could be shown convincingly by FT-IR difference spectroscopy at least for tyrosine residues.
A summarizing description of possible conformational and covalent modifications of proteins during oxidative damage is given in Fig. 11. Areas A-D of Fig. 11 schematically illustrate how the concentration of the different RNase A fractions may vary if the concentration of the oxidant is increased. Initially, at comparably low oxidant concentration, some of the solventaccessible amino acid residues such as histidine, methionine, or aromatic amino acid residues are modified (Fig. 11B). The FT-IR spectroscopic results of this study clearly indicate that tyrosine and aspartate amino acid residues are involved. Possibly, the more side chains are oxidized by hydrogen peroxide, the more the conformational stability of the protein is reduced and the higher the probability of local or even global unfolding (Fig. 11C). Unfolding causes the exposure of initially buried hydrophobic groups to the solvent and accordingly to the oxidant. Thus, residues that were initially not exposed to the solvent were made accessible to oxidation by hydrogen peroxide and other secondary oxidants. As a result, the protein surface hydrophobicity, potentially one of the recognition signals for proteasome binding, is increased. However, with increasing concentrations of the unfolded protein, the probability of intermolecular formation of protein aggregates that are not accessible to proteasome digestion will increase as well (Fig. 11D). Therefore, at nonphysiologically high concentrations of the oxidants, the proteasome turnover is not further enhanced but will be reduced instead. This situation is reflected in Fig. 11, where a difference curve between the unfolded fraction of RNase A (C) and aggregated forms of the protein (D) is displayed. Interestingly, the shape of this difference curve matches the main features of the proteasome turnover plot of Fig. 1. However, it cannot be completely ruled out that products of the oxidative modification of amino acid side chains, such as methionine sulfoxide, play a significant role in the proteasome recognition process as well. Yet we believe that hydrophobicity is the key signal for the 20 S proteasome because of the initially low accessibility of the methionine residues. However, additional experimental efforts are necessary to answer these questions. FT-IR difference spectroscopy is one of the potentially techniques for detecting oxidant-induced changes in protein structure due to its very high sensitivity and specificity. Further increase of the specificity of the method may be achieved by the use of isotopic labeled compounds e.g. of methionines marked by 13 C and 15 N and/or the substitution of specific amino side residues. FIG. 11. Schematic illustration of the H 2 O 2 concentration-dependent effects on RNase A structure. At low hydrogen peroxide concentrations, solvent-exposed amino acid side chains of RNase A are potentially targets of the oxidants (curve B). In our model, with an increasing number of modified side chains, the stability of the model protein is more and more reduced, leading initially to local and finally to global unfolding (curve C). The higher the concentration of unfolded RNase A species, the higher the probability of protein aggregation (curve D). Possibly, unfolded RNase A species are recognized by the 20 S proteasome due their exposure of hydrophobic surfaces (curves C and D), whereas aggregates are not accessible. Please note that all processes are concentration-and time-dependent.