186,153-158
ANALYTICALBIOCHEMISTRY
(1990)
Attenuated Total Reflectance Fourier Transform Infrared Analysis of an Acyl-Enzyme Intermediate of a-Chymotrypsin Sally A. Swedberg,”
Joseph J. Pesek,? and Anthony
L. Fink*,’
*The Department of Chemistry, The University of California, Santa Cruz, California and tThe Department of Chemistry, San Jose State University, San Jose, California
Received
September
8,1989
Attenuated total reflectance Fourier transform infrared (FTIR) spectroscopy was used to obtain signal enhancement of the spectrum of the trans-cinnamoyla-chymotrypsin acyl-enzyme intermediate. Dilute solutions (as low as 2.5 mg/ml) of enzyme or stabilized acyl-enzyme intermediate were used to form thin films on a germanium crystal surface. The secondary structure of the enzyme thin film was shown to be consistent with the native secondary structure using deconvoluted FTIR data. A novel subtraction technique was used to eliminate interfering spectra of water vapor and protein in critical regions of analysis for esters. This permitted the difference spectra of the one new ester carbony1 bond to be discerned from the 300 or so amide bonds in the protein. The results suggest that the acylenzyme exists in two different conformations. This study demonstrates that ir structural information of enzyme-substrate or enzyme-inhibitor complexes can be obtained with dilute protein solutions. 0 mgo Academic Press.
95064, 95192
Inc.
Several examples exist in the literature which demonstrate the use of FTIR2 spectroscopy to obtain the structural evidence required in mechanistic studies of enzyme action (1). The FTIR instrumentation currently available offers significant advantages over older instruments in providing signal averaging and subtraction routines as part of their powerful computational capabilities. These capabilities are necessary when looking for the weak signals produced by a single residue-ligand interaction with other sources of informational noise present, such as water vapor or protein. 1 To whom correspondence should be addressed. 2 Abbreviations used: FTIR, Fourier transform infrared; deuterotriglycine sulfate: ATR, attenuated total reflectance. 0003.2697/90 $3.00 Copyright 0 1990 by Academic Press, All rights of reproduction in any form
DTGS,
While literature examples exist which emphasize the potential of FTIR spectroscopy as a tool in elucidating enzyme mechanism studies, there are some drawbacks to the conventional transmission experiment which make it lessthan universal as an experimental approach to these studies. In particular, to obtain sufficient S/N, these studies have the requirement of: (i) high protein concentrations. Most studies reported have used concentrations in excess of 50 mg/ml and are generally in the range of 100-300 mg/ml (2-4). For many enzymes, even if sufficient quantities of purified enzyme can be obtained, limited solubility and/or subsequent aggregation may prohibit the use of such high protein concentrations in the transmission experiment. (ii) relatively long path length, necessitating the use of D,O. In addition to high protein concentrations, path lengths in excess of 25 pm are also used; typically, 70pm and loo-pm path lengths have been reported (2,3,5). Subtraction of the water band at 1640 cm-’ is not a problem for FTIR instruments when path lengths < 25 pm are used. For path lengths > 25 pm, water becomes completely absorbing. This requires the use of DzO which results in additional sample preparation and perturbation of the position of absorbance bands. In this study, ATR-FTIR was used to investigate the acyl-enzyme intermediate, trans-cinnamoyl-a-chymotrypsin. The intermediate was formed at pH 4.0 where it is relatively stable (tl12 = 400 min (6)). A small aliquot was deposited on the surface of a Ge crystal and gently dried to a thin film to permit signal amplification. Deconvoluted spectra from transmission experiments in H,O were compared to thin film spectra in order to assess whether significant changes in protein secondary structure had occurred due to formation of the thin film. A novel spectral subtraction algorithm, least-squares subtraction, was used to remove the spectral interfer153
Inc. reserved.
154
SWEDBERG.
PESEK,
ence of residual water vapor and protein (7). This subtraction technique proved to be much more suitable than a conventional signal subtraction routine.
MATERIALS
AND
METHODS
Reagents. The a-chymotrypsin and trans-cinnamoyl imidazole were purchased from Sigma. All other reagents used were of the highest grade available. Preparation of the sample for analysis. Protein concentrations from 1 to 50 mg/ml of a-chymotrypsin were investigated. For samples in the 1 to 10 mg/ml range the enzyme was dissolved in deionized water and the pH adjusted to pH 4.0 with dilute solutions of NaOH or HCl. To prevent protein aggregation at high protein concentrations (>25 mg/ml), the enzyme was dissolved in 100 mM phosphate buffer, pH 7.0, prior to adjustment to pH 4.0. Solutions of trans-cinnamoyl imidazole in the 30 to 150 mM range were prepared in dry acetonitrile such that 5 to 10 ~1 could be used to deliver the appropriate quantity to 1 ml of enzyme solution (6). The stoichiometry was slightly greater than 1 M equivalent of enzyme. The substrate solutions were made up immediately prior to use. It was found that 10 to 50 nmol of protein deposited over 3-4 cm2 on the Ge crystal gave a sufficiently strong signal for these experiments. Protein concentrations of 2.5 to 50 mg/ml, using volumes of 100 to 25 ~1, respectively, were compared. The reaction was initiated by the addition of the appropriate volume of substrate stock solution to 1 ml of enzyme solution at 23°C. A thin film was created by smearing the sample over the crystal surface and then removing the solvent by vacuum desiccation. Drying was also achieved by flowing a stream of dry gas over the surface. While this latter method was faster than vacuum desiccation (3 min vs 20 min), it resulted in nonuniform deposition, creating significant film-tofilm variation. FTIR analysis. Spectra were taken using a PerkinElmer Model 1800 FTIR equipped with a DTGS detector. The instrument was interfaced to the Perkin-Elmer Model 7000 computer workstation. To obtain the best S/N, the single beam ratio mode with a mirror speed of 0.05 cm/s was used. The measurements were collected at a resolution of 4 cm-l, using a triangular apodization function. The chamber was purged for one-half hour with dry Nz before data collection was initiated. Four spectra were collected per sample with a total analysis time of under 4 min. A conventional trapezoidal Ge crystal was used to collect the ATR spectra. Deconvolution was performed by software provided with the Model 7000 computing workstation. The leastsquares subtraction algorithm was that of Lacey (7). It was programmed into the Model 7000 computing system using the OBEY language.
AND
FINK
RESULTS Informational noise in the carbonyl region. As previously noted, protein and water vapor signals were the two major contributions to spectral interference. The spectrum in Fig. 1 is an example of the raw data obtained from the thin film experiment. The areas shown include the carbonyl (1720-1708 cm-‘) and C-O ester stretch (1310-1100 cm-‘) regions for &unsaturated esters (8). The protein amide I band at 1638 cm-’ with an intensity of >l.O AU is about 150X greater in absorbance than that of the observed acyl-enzyme carbonyl. The carbonyl stretch region falls on the steep-rising portion of the amide I band. In addition water vapor signals in the carbony1 region were particularly troublesome on two accounts. First, they were in close proximity to the signals of interest. Additionally, the small residual water vapor signals left even after lengthy purging were on the same order of magnitude in intensity as the acyl-enzyme carbony1 signal. Comparison of secondary structure by deconvoluted FTIR. As well as possible effects on structure due to the removal of bulk solvent, it is known that proteinsurface interactions may result in conformational changes (9,lO). To check for changes in protein secondary structure in the solution compared to thin film experiments, deconvoluted amide I/II and amide III data sets were evaluated. In Fig. 2, published transmission data for a-chymotrypsin (11) is compared to the thin film data from this study. Analysis of regions of interest for a&unsaturated esters. Figure 3 shows spectra which are the result of least-squares subtraction, Figures 3A and 3C represent the subtraction of an enzyme thin film from an enzymesubstrate thin film over the particular regions of interest. The experiment was repeated several times and consistent results were obtained in which two peaks were observed in the difference spectra for the carbonyl region, at 1723 and 1710 cm-i, and two in the C-O ester stretch region, one broadband with X,,, at 1277-1264 cm-’ and one at 1206 cm-‘. To provide a suitable control, the spectrum of an enzyme thin film was subtracted from that of a different enzyme thin film. Examples of these controls are shown in Figs. 3B and 3D. Additional controls. In addition to the enzyme thin film control, the following compounds were analyzed separately as thin films, peaks in the regions of interest are given in parentheses (s, strong; m, medium; w, weak; sh, shoulder): (i) trans-cinnamoyl imidazole [i709(s), 1238(s)] (ii) trans-cinnamic acid [1683(s), 1286(s), 1205(sh)] (iii) imidazole [1264(m)]. By comparing the spectra pounds with those shown
1288(w), 1232(m),
in both regions for these comin Figs. 3A and 3C, it was de-
FOURIER 1.0
TRANSFORM
INFRARED
ANALYSIS
538
AU
0.8
0.6
0.4 1300-l
150cm
4'
0.2
0.0 1850
1600
1400
1150
Cm-l
FIG. 1. Spectrum of an enzyme-substrate subtraction. Boxed areas indicate the regions sociated with a&unsaturated esters.
thin film before spectral of interest for bands as-
termined that the spectra of the enzyme-substrate thin film was unique. Finally, after obtaining a typical response for a sample at pH 4.0, the pH of the sample was raised to pH 7.0 and immediately reanalyzed. In that sample, in the regions of the difference spectrum inspected for the presence of a&unsaturated ester, no bands appeared. DISCUSSION
Key questions to be answered in this work were (i) whether the conformation of the protein was the same in solution and in thin films, (ii) whether dilute protein solutions could be used to form the thin film, (iii) to determine the minimum amount of protein necessary to obtain a satisfactory S/N, and (iv) to resolve previous controversy regarding the nature of the ester linkage in the acyl-enzyme. Comparison of secondary structure. The significant features in deconvoluted FTIR spectra for different
OF
AN
ACYL-ENZYME
INTERMEDIATE
155
structural classes of proteins have been reported (11,12). The fingerprint for a p + low LYstructure in deconvoluted amide I and III regions (13) can be found in both the thin film and the solution data of cu-chymotrypsin and is consistent with the crystallographic structure. The band at 1638 cm-’ in the amide I region (Figs. 2A and 2C) is characteristic of a @structure (11). Since the amount of a-helix in chymotrypsin is low, its contributions in the amide I position are negligible, and hence comparison of the data to typical P-sheet proteins (Fig. 2A) can be done. The thin film spectrum shows a characteristic pattern for @+ low a structure. In the amide III region, the major features assigned to /3 + low LYstructures are apparent in both the thin film and the solution data. For this class of structures, the characteristic 1235-cm-’ band for the P-sheet structure is split into two components at about 1254 and 1236 cm-i (11). This pattern is apparent in the thin film preparation. Thus comparison of the protein amide signals indicate that the secondary structure, at least, is very similar in both the solution and the thin film states of chymotrypsin. Consequently it is reasonable to assume that formation of the film does not cause any artifact in the signal from the acyl-enzyme, compared to that in solution. Recent studies (see below) indicate that many proteins have similar structures in the solution and thin film state. a-Chymotrypsin acyl intermediate. To determine the validity of the peaks at 1723 and 1710 cm-l the S/N for the difference spectrum was calculated. For the worstcase signal (1723 cm-‘) and worst-case noise (peak-topeak residual informational noise, 1850-1750 cm-‘) the S/N was 2.5 for several data sets. The S/N for the peak at 1710 cm-‘, using the worst-case noise, was 4.0. Thus these peaks exceed the requirement for the analytical definition of minimum detectable signal. As discussed in Bellamy (8) the carbonyl stretching regions for a$-unsaturated esters (1724-1719 cm-‘), amides (1690-1650 cm-‘), and cinnamic acid (
ANALYTICALBIOCHEMISTRY
(1990)
Attenuated Total Reflectance Fourier Transform Infrared Analysis of an Acyl-Enzyme Intermediate of a-Chymotrypsin Sally A. Swedberg,”
Joseph J. Pesek,? and Anthony
L. Fink*,’
*The Department of Chemistry, The University of California, Santa Cruz, California and tThe Department of Chemistry, San Jose State University, San Jose, California
Received
September
8,1989
Attenuated total reflectance Fourier transform infrared (FTIR) spectroscopy was used to obtain signal enhancement of the spectrum of the trans-cinnamoyla-chymotrypsin acyl-enzyme intermediate. Dilute solutions (as low as 2.5 mg/ml) of enzyme or stabilized acyl-enzyme intermediate were used to form thin films on a germanium crystal surface. The secondary structure of the enzyme thin film was shown to be consistent with the native secondary structure using deconvoluted FTIR data. A novel subtraction technique was used to eliminate interfering spectra of water vapor and protein in critical regions of analysis for esters. This permitted the difference spectra of the one new ester carbony1 bond to be discerned from the 300 or so amide bonds in the protein. The results suggest that the acylenzyme exists in two different conformations. This study demonstrates that ir structural information of enzyme-substrate or enzyme-inhibitor complexes can be obtained with dilute protein solutions. 0 mgo Academic Press.
95064, 95192
Inc.
Several examples exist in the literature which demonstrate the use of FTIR2 spectroscopy to obtain the structural evidence required in mechanistic studies of enzyme action (1). The FTIR instrumentation currently available offers significant advantages over older instruments in providing signal averaging and subtraction routines as part of their powerful computational capabilities. These capabilities are necessary when looking for the weak signals produced by a single residue-ligand interaction with other sources of informational noise present, such as water vapor or protein. 1 To whom correspondence should be addressed. 2 Abbreviations used: FTIR, Fourier transform infrared; deuterotriglycine sulfate: ATR, attenuated total reflectance. 0003.2697/90 $3.00 Copyright 0 1990 by Academic Press, All rights of reproduction in any form
DTGS,
While literature examples exist which emphasize the potential of FTIR spectroscopy as a tool in elucidating enzyme mechanism studies, there are some drawbacks to the conventional transmission experiment which make it lessthan universal as an experimental approach to these studies. In particular, to obtain sufficient S/N, these studies have the requirement of: (i) high protein concentrations. Most studies reported have used concentrations in excess of 50 mg/ml and are generally in the range of 100-300 mg/ml (2-4). For many enzymes, even if sufficient quantities of purified enzyme can be obtained, limited solubility and/or subsequent aggregation may prohibit the use of such high protein concentrations in the transmission experiment. (ii) relatively long path length, necessitating the use of D,O. In addition to high protein concentrations, path lengths in excess of 25 pm are also used; typically, 70pm and loo-pm path lengths have been reported (2,3,5). Subtraction of the water band at 1640 cm-’ is not a problem for FTIR instruments when path lengths < 25 pm are used. For path lengths > 25 pm, water becomes completely absorbing. This requires the use of DzO which results in additional sample preparation and perturbation of the position of absorbance bands. In this study, ATR-FTIR was used to investigate the acyl-enzyme intermediate, trans-cinnamoyl-a-chymotrypsin. The intermediate was formed at pH 4.0 where it is relatively stable (tl12 = 400 min (6)). A small aliquot was deposited on the surface of a Ge crystal and gently dried to a thin film to permit signal amplification. Deconvoluted spectra from transmission experiments in H,O were compared to thin film spectra in order to assess whether significant changes in protein secondary structure had occurred due to formation of the thin film. A novel spectral subtraction algorithm, least-squares subtraction, was used to remove the spectral interfer153
Inc. reserved.
154
SWEDBERG.
PESEK,
ence of residual water vapor and protein (7). This subtraction technique proved to be much more suitable than a conventional signal subtraction routine.
MATERIALS
AND
METHODS
Reagents. The a-chymotrypsin and trans-cinnamoyl imidazole were purchased from Sigma. All other reagents used were of the highest grade available. Preparation of the sample for analysis. Protein concentrations from 1 to 50 mg/ml of a-chymotrypsin were investigated. For samples in the 1 to 10 mg/ml range the enzyme was dissolved in deionized water and the pH adjusted to pH 4.0 with dilute solutions of NaOH or HCl. To prevent protein aggregation at high protein concentrations (>25 mg/ml), the enzyme was dissolved in 100 mM phosphate buffer, pH 7.0, prior to adjustment to pH 4.0. Solutions of trans-cinnamoyl imidazole in the 30 to 150 mM range were prepared in dry acetonitrile such that 5 to 10 ~1 could be used to deliver the appropriate quantity to 1 ml of enzyme solution (6). The stoichiometry was slightly greater than 1 M equivalent of enzyme. The substrate solutions were made up immediately prior to use. It was found that 10 to 50 nmol of protein deposited over 3-4 cm2 on the Ge crystal gave a sufficiently strong signal for these experiments. Protein concentrations of 2.5 to 50 mg/ml, using volumes of 100 to 25 ~1, respectively, were compared. The reaction was initiated by the addition of the appropriate volume of substrate stock solution to 1 ml of enzyme solution at 23°C. A thin film was created by smearing the sample over the crystal surface and then removing the solvent by vacuum desiccation. Drying was also achieved by flowing a stream of dry gas over the surface. While this latter method was faster than vacuum desiccation (3 min vs 20 min), it resulted in nonuniform deposition, creating significant film-tofilm variation. FTIR analysis. Spectra were taken using a PerkinElmer Model 1800 FTIR equipped with a DTGS detector. The instrument was interfaced to the Perkin-Elmer Model 7000 computer workstation. To obtain the best S/N, the single beam ratio mode with a mirror speed of 0.05 cm/s was used. The measurements were collected at a resolution of 4 cm-l, using a triangular apodization function. The chamber was purged for one-half hour with dry Nz before data collection was initiated. Four spectra were collected per sample with a total analysis time of under 4 min. A conventional trapezoidal Ge crystal was used to collect the ATR spectra. Deconvolution was performed by software provided with the Model 7000 computing workstation. The leastsquares subtraction algorithm was that of Lacey (7). It was programmed into the Model 7000 computing system using the OBEY language.
AND
FINK
RESULTS Informational noise in the carbonyl region. As previously noted, protein and water vapor signals were the two major contributions to spectral interference. The spectrum in Fig. 1 is an example of the raw data obtained from the thin film experiment. The areas shown include the carbonyl (1720-1708 cm-‘) and C-O ester stretch (1310-1100 cm-‘) regions for &unsaturated esters (8). The protein amide I band at 1638 cm-’ with an intensity of >l.O AU is about 150X greater in absorbance than that of the observed acyl-enzyme carbonyl. The carbonyl stretch region falls on the steep-rising portion of the amide I band. In addition water vapor signals in the carbony1 region were particularly troublesome on two accounts. First, they were in close proximity to the signals of interest. Additionally, the small residual water vapor signals left even after lengthy purging were on the same order of magnitude in intensity as the acyl-enzyme carbony1 signal. Comparison of secondary structure by deconvoluted FTIR. As well as possible effects on structure due to the removal of bulk solvent, it is known that proteinsurface interactions may result in conformational changes (9,lO). To check for changes in protein secondary structure in the solution compared to thin film experiments, deconvoluted amide I/II and amide III data sets were evaluated. In Fig. 2, published transmission data for a-chymotrypsin (11) is compared to the thin film data from this study. Analysis of regions of interest for a&unsaturated esters. Figure 3 shows spectra which are the result of least-squares subtraction, Figures 3A and 3C represent the subtraction of an enzyme thin film from an enzymesubstrate thin film over the particular regions of interest. The experiment was repeated several times and consistent results were obtained in which two peaks were observed in the difference spectra for the carbonyl region, at 1723 and 1710 cm-i, and two in the C-O ester stretch region, one broadband with X,,, at 1277-1264 cm-’ and one at 1206 cm-‘. To provide a suitable control, the spectrum of an enzyme thin film was subtracted from that of a different enzyme thin film. Examples of these controls are shown in Figs. 3B and 3D. Additional controls. In addition to the enzyme thin film control, the following compounds were analyzed separately as thin films, peaks in the regions of interest are given in parentheses (s, strong; m, medium; w, weak; sh, shoulder): (i) trans-cinnamoyl imidazole [i709(s), 1238(s)] (ii) trans-cinnamic acid [1683(s), 1286(s), 1205(sh)] (iii) imidazole [1264(m)]. By comparing the spectra pounds with those shown
1288(w), 1232(m),
in both regions for these comin Figs. 3A and 3C, it was de-
FOURIER 1.0
TRANSFORM
INFRARED
ANALYSIS
538
AU
0.8
0.6
0.4 1300-l
150cm
4'
0.2
0.0 1850
1600
1400
1150
Cm-l
FIG. 1. Spectrum of an enzyme-substrate subtraction. Boxed areas indicate the regions sociated with a&unsaturated esters.
thin film before spectral of interest for bands as-
termined that the spectra of the enzyme-substrate thin film was unique. Finally, after obtaining a typical response for a sample at pH 4.0, the pH of the sample was raised to pH 7.0 and immediately reanalyzed. In that sample, in the regions of the difference spectrum inspected for the presence of a&unsaturated ester, no bands appeared. DISCUSSION
Key questions to be answered in this work were (i) whether the conformation of the protein was the same in solution and in thin films, (ii) whether dilute protein solutions could be used to form the thin film, (iii) to determine the minimum amount of protein necessary to obtain a satisfactory S/N, and (iv) to resolve previous controversy regarding the nature of the ester linkage in the acyl-enzyme. Comparison of secondary structure. The significant features in deconvoluted FTIR spectra for different
OF
AN
ACYL-ENZYME
INTERMEDIATE
155
structural classes of proteins have been reported (11,12). The fingerprint for a p + low LYstructure in deconvoluted amide I and III regions (13) can be found in both the thin film and the solution data of cu-chymotrypsin and is consistent with the crystallographic structure. The band at 1638 cm-’ in the amide I region (Figs. 2A and 2C) is characteristic of a @structure (11). Since the amount of a-helix in chymotrypsin is low, its contributions in the amide I position are negligible, and hence comparison of the data to typical P-sheet proteins (Fig. 2A) can be done. The thin film spectrum shows a characteristic pattern for @+ low a structure. In the amide III region, the major features assigned to /3 + low LYstructures are apparent in both the thin film and the solution data. For this class of structures, the characteristic 1235-cm-’ band for the P-sheet structure is split into two components at about 1254 and 1236 cm-i (11). This pattern is apparent in the thin film preparation. Thus comparison of the protein amide signals indicate that the secondary structure, at least, is very similar in both the solution and the thin film states of chymotrypsin. Consequently it is reasonable to assume that formation of the film does not cause any artifact in the signal from the acyl-enzyme, compared to that in solution. Recent studies (see below) indicate that many proteins have similar structures in the solution and thin film state. a-Chymotrypsin acyl intermediate. To determine the validity of the peaks at 1723 and 1710 cm-l the S/N for the difference spectrum was calculated. For the worstcase signal (1723 cm-‘) and worst-case noise (peak-topeak residual informational noise, 1850-1750 cm-‘) the S/N was 2.5 for several data sets. The S/N for the peak at 1710 cm-‘, using the worst-case noise, was 4.0. Thus these peaks exceed the requirement for the analytical definition of minimum detectable signal. As discussed in Bellamy (8) the carbonyl stretching regions for a$-unsaturated esters (1724-1719 cm-‘), amides (1690-1650 cm-‘), and cinnamic acid (
