Chem. Res. Toxicol. 1995, 8, 552-559
552
Direct Characterization of Protein Adducts of the Lipid Peroxidation Product 4-Hydroxy-2-nonenalUsing Electrospray Mass Spectrometry Bernd A. Bruenner,"??A. Daniel Jones,? and J. Bruce Germad Facility for Advanced Instrumentation and Department of Food Science and Technology, University of California, Davis, California 95616 Received November 4, 1994@ Oxidative stress and exposures to xenobiotic substances generate reactive substances including the cytotoxic aldehyde 4-hydroxy-2-nonenal. This aldehyde exhibits a variety of biological effects and has been reported as a marker of lipid peroxidation. The toxicity and atherogenicity of 4-hydroxy-2-nonenal have been attributed t o the formation of covalent protein adducts. I n the current study, two model proteins, P-lactoglobulin B and human hemoglobin, were exposed to 4-hydroxy-2-nonena1, and the protein adducts were characterized using electrospray ionization mass spectrometry. Our findings provided clear and direct evidence that '99% of protein modification occurred via Michael addition, and only trace amounts of Schiff base adducts were formed. Confirmation of this result was obtained via quantitative conversion of the modified proteins to oxime and pentafluorobenzyl oxime derivatives as demonstrated by electrospray ionization mass spectrometry, spectrophotometric protein carbonyl assays, and gas chromatography/mass spectrometry determination of 4-hydroxy-2nonenal released upon treatment with hydroxylamine. These results further demonstrate the availability of the protein-bound aldehyde for subsequent reaction or a s a site of molecular recognition. The preponderance of Michael addition products over Schiff base adducts also suggests that most methods for determining 4-hydroxy-2-nonenal in biological tissues or fluids are based on erroneous assumptions that hydrazines or hydroxylamines release 4-hydroxy-2nonenal from proteins.
Introduction Free radical formation has been implicated as a contributing factor in diverse degenerative conditions including toxicity of xenobiotic chemicals, atherosclerosis, and aging (11, and regulated events including apoptosis and destruction of pathogens by phagocytes (2). Free radical oxidations of polyunsaturated fatty acids, generally in lipoproteins and membranes, lead to formation of aldehydes and many other products (3). There is great interest in developing noninvasive methods to investigate biological effects of free radicals in vivo (41, but reliable biomarkers of in vivo oxidative stress have remained elusive because relationships between biomarker levels and biological effects are not well established. It has been proposed that multiple measures of oxidative stress are needed to determine the clinical significance of biological oxidation ( 5 ) . Aldehydic products of lipid peroxidation are more stable than their precursor lipid radicals or peroxides, and this stability allows them to diffuse from the initial site of lipid oxidation. For this reason, aldehydes are considered toxic second messengers of oxidative stress and candidate biomarkers. Reactive aldehydes may be metabolized, may undergo further reactions with cellular constituents, or may be found in tissues before they have had a chance to react (6). Most work on the subsequent * Author to whom correspondence should be addressed, at the Facility for Advanced Instrumentation (F.A.I.),University of California, Davis, Davis, CA 95616.Phone: (916)752-0284;FAX: (916)752-4412; E-mail address: BABruennefiUCDavis.edu. ' Facility for Advanced Instrumentation, University of California, Davis. * Department of Food Science and Technology, University of California, Davis. Abstract published in Advance ACS Abstracts, April 15, 1995. @
reactions of oxidatively generated aldehydes has been conducted assuming that aldehydes are likely to react with nitrogen nucleophiles to form Schiff bases (imines). One of the most cytotoxic aldehydic products of lipid peroxidation is (E)-4-hydroxy-2-nonenal(HNE)' which is an unusually reactive a$-unsaturated aldehyde (7). Nucleophilic sulfhydryl groups undergo rapid 1,4-addition (Michael addition) to the electrophilic double bond of HNE even without benefit of enzymatic catalysis (8). It has been hypothesized that proteins become modified in vivo by HNE under conditions of oxidative stress, and these HNE-modified proteins represent an important adduct class with unique reactivities and toxicities. Indeed, HNE adducts of low density lipoprotein (LDL) have been detected in human atherosclerotic lesions using immunological techniques (9). Because HNE has a short lifetime in the presence of sulfhydryl-containing molecules such as glutathione, the persistence of aldehyde-protein adducts makes them more suitable than the free aldehydes as biomarkers of lipid oxidation in vivo. Levels of adducts of aldehydes with proteins which exhibit long half-lives, such as hemoglobin, may serve as measures of chronic oxidative stress (10). Mass spectrometry has become essential to the identification of protein modifications, including xenobiotic adducts (11)and post-translational modifications. The recent development (12) of electrospray ionization (ESI) allows for determination of molecular weights of proteins with sufficient resolution and accuracy that protein adducts can be easily characterized. Another singular Abbreviations: HNE, (E)-4-hydroxy-2-nonenal; LDL, low density lipoprotein; ESI, electrospray ionization; Hb, hemoglobin; P-LgB, B-lactoglobulin B; HEPES, N-(2-hydroxyethyl)piperazine-N'-2-ethane sulfonic acid.
0893-228x/95/2708-0552$09.00/00 1995 American Chemical Society
Chem. Res. Toxicol., Vol. 8, No. 4, 1995 553
Electrospray Mass Spectra of HNE-Protein Adducts advantage of the mass resolution is the potential to distinguish mechanisms of chemical modification occurring with proteins (23). ESI mass spectrometry generates spectra in which proteins are ionized with a distribution of charge states, with each charge state giving rise to a peak in the spectrum a t a distinct masdcharge ( m/ z ) value. Numerical algorithms can then transform the multiply charged mass spectral data to obtain a n accurate measure of the molecular weight of the intact protein. This investigation examined two classes of proteins modified in vitro by reaction with HNE. Hemoglobin, which is present a t millimolar concentrations in red blood cells, has a half-life of several weeks and thus would be expected to accumulate oxidative damage over time. /?-Lactoglobulin B (P-LgB) is a small protein that is a member of the lipocalin family of hydrophobic ligand binding proteins (14). Since both its sequence and threedimensional structure are known (15),this serves as a useful model for those proteins which due to ligand binding properties may constitute inordinately sensitive targets of modification by HNE. Using ESI mass spectrometry, we have investigated the extent to which binding of HNE to these proteins occurs by Michael addition and also by Schiff base formation.
Materials and Methods Caution: The range of toxicological effects of HNE are not yet known. I t is, however, cytotoxic a n d mutagenic a n d should be handled to minimize exposure. Preparation of HNE. HNE was prepared by synthesis following the Esterbauer procedure (161, and the deuterated analog [2,3-2H]HNEwas prepared as previously described ( 17). The aldehydes were purified by column chromatography using silica gel and hexane/ethyl acetate (7525) as solvent. Purity of the synthesized HNE was established by G C M S to be a t least 97%. G C M S was performed on a Trio-2 G C M S system (VG Masslab, Altrincham, U.K.) in the full scan mode using both electron impact and chemical ionization on a 15 m DB-5 column (J&W Scientific, Folsom, CAI. NMR spectra of the aldehydes were obtained on a General Electric Omega 300 MHz instrument using CDC13 as solvent. Proton chemical shifts (ppm from tetramethylsilane) for [2,3-2H]HNE [O=CHaCD=CDCHb(OH)CH2C(CH2d)3CH3e]were observed at 0.85 (t, 3H, He), 1.27 (m, 6H, Hd), 1.58 (9, 2H, Hc), 3.12 (s, l H , OH), 4.38 (t, l H , Hb), and 9.51 (s, l H , Ha). No NMR signal corresponding to vinylic hydrogen was detected, indicating isotopic purity of the deuterated compound. For HNE [O=CHaCHb=CHcCHd(OH)CH2e(CH2q3CH3gl,hydrogen absorption was observed at 0.89 (t,3H, Hg), 1.28 (m, 6H, Hf), 1.58 (9, 2H, He), 2.70 (s, l H , OH), 4.38 (m, l H , Hd), 6.26 (dd, l H , Hb), 6.81 (dd, l H , He), and 9.51 (d, l H , Ha). Preparation of Protein Adducts. Proteins incubated with HNE in vitro included human hemoglobin A (Sigma Chemical Co., St. Louis, MO) and purified p-LgB (18). Several milligrams of each protein were dissolved in 10 mM N-(2-hydroxyethyl)piperazine-"-2-ethanesulfonic acid (HEPES) buffer (pH 7.41, and the aldehyde was added to the proteins as a n 8 mM solution in the same buffer. HNE concentration in buffer was calculated from measurements of the absorbance at 223 nm against a buffer blank, using a molar absorptivity of 13 750 M-* cm-l(16). The final concentration of HNE resulted in a 46-fold molar excess of the aldehyde to hemoglobin (m = 64 450 Da) and a 58-fold molar excess to P-LgB (m = 18 277 Da). Corresponding protein controls without HNE were also prepared in HEPES buffer. The protein-HNE mixtures were allowed to react at 37 "C for 3 h. The samples were then extensively dialyzed against 10 mM ammonium bicarbonate buffer (pH 7.4) to remove unreacted HNE (seven changes, 1:400 viv) using 1000 molecular weight cutoff dialysis tubing. Heme was removed by
precipitation in acidic acetone (19). The proteins were subsequently lyophilized and stored at -80 "C. Reaction of HNE with N-Acetyllysine. N"-Acetyl-L-lysine (1.5 mg) was dissolved in 100 mL of deionized water to yield a solution of 80 pmoUpL. The solution was adjusted to pH 7.5 by addition of a small amount of 1M NaOH, and a 1.5 mL aliquot (0.12 pmol) was pipetted into a 2 mL Eppendorf tube. To this was added 95 p L of 12.7 mM aqueous HNE solution (1.2 pmol), a 10-fold molar excess of aldehyde. The reaction mixture was incubated for 3 h at 37 "C and then stored at -25 "C until further analysis. Samples were prepared for analysis by ESU MS by diluting the reaction mixture with a n equal volume of acetonitrile. Aliquots were removed and analyzed using electrospray ionization mass spectrometry in negative ion mode.
Reaction of HNE-Hemoglobin Adducts with Hydroxy. lamine Reagents. To 1mg of the hemoglobin-HNE reaction product were added 200 pL of a 50 mM solution of hydroxylamine hydrochloride (Fisher Scientific, Fair Lawn N J ) in 0.1 M ammonium acetate (pH 7.01, and 150 p L of acetonitrile. Similarly, to a separate sample was added 200 pL of a 50 mM solution of 0-(2,3,4,5,6-pentafluorobenzyl)hydroxylaminehydrochloride (Aldrich Chemical Co., Milwaukee, WI) in the same buffer. After the mixtures were allowed to react at room temperature for 30 min, the samples were extracted three times with 200 p L of hexane to remove aldehydes displaced a s oxime derivatives and then dialyzed overnight using a Pierce Microdialyzer System 100 and a 1000 molecular weight cutoff membrane.
Analysis of HNE Schiff Base Adducts as the OximeBis(tert-butyldimethylsilyl)Derivative. The derivatization scheme used is a modification of a previously published method that was designed to measure free HNE and HNE-Schiff base adducts (20). A 1mg quantity of lyophilized hemoglobin-HNE reaction product was added to 300 pL of 0.1 M ammonium acetate (pH 7) containing 50 mM hydroxylamine hydrochloride and 100 ng of [2,3-2HlHNE as a n internal standard. The mixture was vortexed and allowed to react a t room temperature for 15 min, and extracted three times with 300 pL of dichloromethane. The solvent extracts were combined and dried under a stream of nitrogen. Subsequently, 30 ,uL of dimethylformamide and 30 pL of N-methyl-N-(tert-butyldimethylsily1)trifluoroacetamide (Pierce, Rockford, IL) were added and incubated a t 60 "C for 1 h. Quantitative determination of HNE released from hemoglobin was performed by analyzing the oxime-bis(tert-butyldimethylsilyl) derivatives using a Trio-2 G C M S and a 15 m DB-5 column. Oven temperature was programmed from 80 to 270 "C at a rate of 10 W m i n , and the injector temperature was 270 "C. Ions of m / z 342 and 344 were monitored by selected ion monitoring using 70 eV electron impact ionization. These are fragment ions corresponding to loss of the tert-butyl group [M - 571+ from the HNE and [2,32HlHNE oxime derivatives, respectively. Calibration curves were generated based upon integrated peak areas for the more abundant, later eluting isomer of the cis/trans oxime isomer pair and were used to quantify the amount of HNE released from protein. Reduction of/.?-LgBwith Dithiothreitol. A portion of the P-LgB exposed to HNE was dissolved i n water to give a concentration of 0.05 mM and mixed with a 100-fold molar excess of dithiothreitol in distilled water to reduce the two intramolecular disulfide bridges. The mixture was allowed to react a t room temperature overnight and then was extracted three times with an equal volume of hexane. Protein Carbonyl Assay. Protein-bound carbonyls were determined by reaction with (2,4-dinitrophenyl)hydrazineand ultraviolet spectrophotometry using a Model PC-2101 spectrophotometer (Shimadzu Scientific Instruments, Columbia, MD) following a previously published method (211. The absorbance maximum was found to occur at 370 nm, and carbonyl contents were calculated using a molar absorption coefficient of 22 000 M-I cm-'. Amino Acid Analysis. Protein amino acid composition was determined by quantitative amino acid analysis on a Beckman
Bruenner et al.
554 Chem. Res. Toxicol., Vol. 8, No. 4, 1995 6300 amino acid analyzer a t the UC Davis Protein Structure Laboratory. Samples underwent a standard 24 h HCI hydrolysis prior to separation of the amino acids by ion-exchange chromatography using a sodium citrate buffer system. To quantitate cysteine and methionine, which are destroyed during the acid hydrolysis, a n identical sample was oxidized with performic acid prior to the hydrolysis, converting the cysteine to cysteic acid and methionine to methionine sulfone. Because Michael additions and Schiff base formation can be reversible in strong acid, HNE-protein adducts were reduced with NaBH4 before the acid hydrolysis to prevent release of these adducts during this procedure. Mass Spectrometry of Proteins and Adducts. Lyophil% lized proteins were dissolved in 50/50 methanollwater formic acid to give a final concentration of approximately 20 pmollpL for analysis by electrospray ionization mass spectrometry in positive ion mode. Intact proteins were immediately analyzed on a VG/Fisons Quattro-BQ triple quadrupole mass spectrometer (VG BioTech, Altrincham, U.K.) using 50150 methanollwater 1%formic acid as the mobile phase delivered at a flow rate of 5 pL/min using a pLC-500 pump (1x0, Lincoln, NE). Loop injections of 10 pL were made. Capillary voltage was f 3 . 5 kV and source temperature was 65 "C. Electrospray spectra were generated using a nozzle-skimmer bias of 50 V which imparts sufficient energy to eliminate noncovalent associations between proteins and ligands. Electrospray mass spectra of N-acetyllysine adducts were generated in negative ion mode using 50/50 acetonitrile/water with no formic acid as the mobile phase.
+
+
Results and Discussion Proteins were incubated with large excesses of HNE using conditions not normally encountered in vivo in order to probe the chemistry of addition of HNE a t less reactive amino acid residues, such as lysine. Lysine residues have been widely postulated to participate in the formation of Schiff base linkages with HNE (22,231, whereas cysteine and histidine residues are softer nucleophiles not expected to react via 1,2-addition to the carbonyl group. Reduction of disulfide bonds was not performed before incubation of protein with HNE. Electrospray mass spectra of proteins incubated with excess HNE revealed extensive protein modification. Multiple peaks a t each charge state in the mass spectrum of hemoglobin exposed to HNE attest to the presence of a mixture resulting from the covalent attachment of several HNE molecules to protein (Figure 1A). Interpretation of such complex results became simpler when the raw electrospray mass spectrum was mathematically transformed to a true mass scale (Figure 1B). Some unmodified a-chain was observed ( m = 15 126 Da), as well as peaks corresponding to the addition of from one to four molecules of HNE per a-chain. Interestingly, the P-chain of hemoglobin was observed only in the mass spectrum of the unexposed control ( m = 15 867 Da), and at much lower intensity than the a-chain. No modified or unmodified P-chain was detected in the hemoglobin sample exposed to 4-hydroxy-2-nonenal. The B-chain is believed to have formed intermolecular cross-links, developing a higher molecular weight polymer which may have been rendered insoluble during removal of the heme (see discussion below). Sequential peaks in the transformed spectrum of HNE-hemoglobin (Figure 1B) all differed in molecular mass by 156 Da (within experimental uncertainty of f 5 Da), the molecular mass of HNE. This increase in molecular weight provides strong evidence that reaction between HNE and all protein nucleophiles occurred via Michael addition, since Schiff base formation involves
loss of water and the corresponding mass shift would be 18 Da less, or 138. No evidence of peaks differing by 138 Da (characteristic of Schiff base formation) was found in transformed spectra. P-LgB ( m = 18 277 Da) also exhibited a large degree of modification. For this protein, electrospray spectra demonstrated the formation of HNE-protein adducts containing from three to eight HNE molecules per molecule of protein (Figure 2). Again, the peaks in the mass spectrum differed in molecular weight by 156 Da, indicating that formation of Schiff base linkages to protein was negligible. In contrast to the a-chain of hemoglobin, no unmodified P-LgB was observed. It is not yet clear whether the increased number of adducts relative to the hemoglobin sample reflects the higher molar concentration of HNE used or represents a greater reactivity of nucleophilic sites on the P-LgB molecule. As Schiff base formation can be a reversible process, the possibility exists that these adducts were lost via hydrolysis during dialysis of the proteins. To further probe whether Schiff base adducts formed, proteins (PLgB and hemoglobin) incubated with HNE were analyzed directly by electrospray analysis without prior dialysis. The spectra obtained by this approach demonstrated substantial heterogeneity as judged by the complex electrospray spectra and were further complicated by the presence of salts that cause formation of salt adduct peaks in the spectra. Application of conventional mathematical processes used to transform the multiply charged electrospray data of the dialyzed proteins to a true mass scale did not give interpretable spectra owing to this heterogeneity. However, application of a maximum entropy algorithm (MaxEnt), a probabilistic technique that enhances resolution and signal-to-noiseratios, yielded a true mass scale spectrum containing numerous peaks. For the nondialyzed HNE-hemoglobin incubation mixture, peaks of major relative intensity corresponding to Michael addition adducts were observed in the MaxEnt transform, but a t most only a few minor peaks could be assigned to Schiff base adducts. Thus the lack of observable Schiff base adducts appears to be due to the preferential formation of HNE Michael adducts and does not appear to be due to their breakdown during dialysis. To more clearly explore the importance of Schiff base formation in the reaction of HNE with lysine side chains, a simpler model system was employed. N-Acetyllysine was incubated with a 10-fold excess of HNE in deionized water and analyzed using negative ion mode ESI mass spectrometry without further treatment. Negative ion mass spectra are less likely to be complicated by the formation of alkali metal adducts, and compounds containing carboxyl groups readily form deprotonated molecular anions [M - HI-. There is the additional benefit that dehydration of [M - HI- in the electrospray source is less likely than dehydration of the protonated molecule which is a common process in positive ion spectra of low molecular weight HNE adducts. In this case, the mass spectrum was dominated by a peak a t m l z 343, corresponding to the deprotonated molecular anion of the N-acetyllysine-HNE Michael addition product (Figure 3). The presence of some unreacted N-acetyllysine is indicated by the [M - HI- peak a t m l z 187. Ionization of unreacted HNE was found to be insignificant owing to its lack of ionizable functional groups. The intensity of the [M - HI- peak corresponding to the presence of Schiff base adducts ( m l z 325) was less than 2% of the
Chem. Res. Toxicol., Vol. 8, No. 4, 1995 555
Electrospray Mass Spectra of HNE-Protein Adducts
"1
v
I
1
-
v
01',l
14800
15000
15200
15400
15600
15800
1
-"i mass
Figure 1. (A) Electrospray mass spectrum of human hemoglobin modified with HNE. Intact hemoglobin was incubated in vitro with HNE; heme and unreacted HNE were subsequently removed. Numerous peaks occur a t each charge state, indicating a mixture of HNE-protein adducts. Multiple charging reduces the mass to charge ratio (Dale) of the protein to the observed values. Spectrum was background subtracted and smoothed; charge states are indicated by brackets. (B) Deconvoluted electrospray mass spectrum of hemoglobin modified with HNE. Multiply charged mass spectrum of panel A has been transformed to a true mass scale. Only the a-globin chain was observed, modified with up to four HNE adducts. Peaks are separated by 156 Da, the mass of HNE, indicating that reaction has occurred by Michael addition to the double bond of the aldehyde.
intensity of the deprotonated Michael adducts and was judged to be insignificant. Though Schiff base formation may indeed occur, the results of this investigation suggest that Schiff base adducts of HNE with lysine residues are, a t most, short-lived intermediates. Therefore, formation of Schiff base adducts with lysine residues is unlikely to be a significant reaction in the modification of proteins by HNE. To confirm that the adducts in the dialyzed P-LgB were not merely the result of noncovalent hydrophobic associations, the two intramolecular disulfide bridges of P-LgB were reduced with dithiothreitol to alter the tertiary structure of the protein. Extraction of the reduced
protein with hexane would then have removed any noncovalently bound aldehyde, but the electrospray spectrum of the HNE-/3-LgB did not undergo measurable change afier reduction and extraction. Furthermore, electrospray ionization mass spectra were conducted using conditions expected to dissociate noncovalent protein-ligand interactions. Further evidence for the covalent nature of these adducts was obtained by using the well-established protein carbonyl assay (211. The carbonyl content of the p-LgB modified with HNE was 293 & 180 nmol/mg of protein, or 5.4 & 3.3 mol of carbonyllmol of protein (average of two replications). As one HNE-Michael
556 Chem. Res. Toxicol., Vol. 8, No. 4, 1995
Bruenner et al.
YO-
i .hb 0-
\,j
+---
,
,
,
,
,
,
,
,
mass
,
343 [M-HI- N-acetyllysine-HNE
Michael Adduct
100-
,
, 187 [M-HI- N-acetyllysine
0
180
200
220
240
260
280
300
320
340
360
380
400
420
440
Dale
460
Figure 3. Negative ion electrospray mass spectrum of N-acetyllysine-HNE reaction mixture. The peak at m l z 343 corresponds to the deprotonated molecular anion of the N-acetyllysine-HNE Michael addition product. No significant peak for the corresponding Schiff base adduct ( ml z 325) is observed. The peak at m l z 187 is due to unreacted N-acetyllysine.
adduct would produce 1 equiv of protein carbonyl, this value is in good agreement with the average number of HNE adducts observed in the electrospray mass spectrum (the distribution shows a peak of 5-6 HNE molecules/ protein) and further substantiates the Michael addition mechanism, as Schiff base adducts would be released and not detected by the protein carbonyl method. Modified hemoglobin had a n average carbonyl value of 190 f 40 nmol/mg, or 12.2 2.5 mol of carbonyUmo1 of €Ib ( m = 64 450 Da), somewhat higher than the distribution of adducts evident in the electrospray mass spectrum. In the case of hemoglobin, the higher degree of modification as determined by the protein carbonyl assay relative to the ESI results is attributed to adducts on the p chain, which was not observed by ESI.
As both model proteins have too few free sulfhydryl residues to account for the number of aldehyde molecules attached to protein, HNE must react with amino acids other than cysteine. Lysine and histidine also possess nucleophilic side chains, and addition of HNE to these amino acids may account for the high degree of modification as demonstrated in earlier studies that employed chemical degradation methods (24-26). Individual amino acid residues modified by reaction with HNE were indirectly determined by comparing amino acid analysis of the hydrolyzed native protein with analysis of the hydrolyzed HNE-protein adduct (Table 1). In the case of p-LgB modified with HNE, the one free thiol residue and on average between two and three of the 15 lysine residues were modified. In addition, both
Electrospray Mass Spectra of HNE-Protein Adducts
Chem. Res. Toxicol., Vol. 8, No. 4, 1995 557
Table 1. Amino Acid Analysis of Unmodified and HNE-ModifiedProteinsa
Scheme 1. Reaction Schemes for HNE Addition to Protein Nucleophiles"
mol of amino acidmol of protein amino acid
p-LgB control
CYS His LYS
4.4 2.5 15
P-LgB
+ HNE
3.9 0.6 12.6
Hb control 2.7 19.7 22
Hb
OH
Acne
+ HNE
/
1.7 8.9 19.3
f
a Amino acids listed showed large decreases after exposure to HNE as determined by conventional quantitative amino acid analysis.
A
of the two histidine residues in the protein were lost. Similar findings were obtained from analysis of the HNE-hemoglobin samples, which consist of a- and P-chains. Again, the one free cysteine residue and three of 22 lysines were lost. A large proportion of histidine residues were also modified; 10 out of a total of 19 (aplus P-chains) were not recovered after exposure to HNE. Since only a maximum of four HNE adducts was observed per a-chain by ESI, nearly all of the available histidine residues on the @-chain must have become modified. Decreases in relative amounts of other amino acids were determined to be minimal. Indeed, this high degree of modification would account for the complete disappearance of the modified P-chain in electrospray mass spectra. After lyophilization of the modified protein, visible amounts of the protein remained insoluble when the protein was dissolved in the mobile phase solvent. Addition of a large number of hydrophobic residues such as HNE would be expected to change the structure and decrease the aqueous solubility of a protein. The decrease in protein solubility was also observed in the protein carbonyl assay. Insoluble material is removed by centrifugation prior to photometric carbonyl determination, and such material losses of heavily modified protein would reconcile the discrepancy between the higher degree of modification suggested by amino acid analysis relative to HNE modification as determined by protein carbonyl content. This solubility problem may thus underestimate the degree of modification, and care must be taken in the quantitation of heavily modified proteins. These observations suggest an important consequence of Michael addition of HNE to proteins. The aldehyde moiety remains free (or in equilibrium with a hemiacetal form) and capable of undergoing further reactions, even when the addition is to a lysine side chain. Further condensation would depend on the accessibility of a second nucleophilic group capable of addition to the carbonyl group and is expected to result in intra- or intermolecular cross-linking. The participation of protein carbonyl groups in chemical reactions can be followed using mass spectrometry, provided the mass resolving power and mass measurement accuracy are capable of distinguishing reactants from products. Condensation of hydroxylamine with the free aldehyde group of HNE results in a net increase of 15 Da in molecular mass; therefore, each HNE-oxime adduct adds 171 Da to the molecular mass of the protein (Scheme 1). Evidence of the progress of the oximation of HNE -hemoglobin reacted with hydroxylamine is shown in the electrospray spectrum presented in Figure 4. The distribution of the number of HNE groups attached to Hba is virtually unchanged after treatment with hydroxylamine, and the oximation reaction gave quantitative conversion of the free aldehyde groups.
t
H2N-
HNE MW = 156
C
NH H
A& /
O
CH =N
Schiff Base Adduct (t 138 Da)
t
H2N-OH
Michael Addition Adduct (t 156 Da) t
H2N -OH
OH
Displaced HNE oxime
& Protein bound HNE oxime (t 171 Da) a (a) Michael addition to cysteine, lysine, and histidine side chains. Free carbonyl group undergoes subsequent reaction with hydroxylamine to form oxime derivatives that remain bound to the protein. (b) Schiff base formation followed by displacement of HNE from the protein. The tert-butyldimethylsilyl ether derivatives of the displaced HNE oximes were analyzed by GC/MS. The data indicate Schiff base formation to be a minor reaction pathway.
Adjacent peaks in the mass spectrum now differed by 171 Da, revealing the predicted increase in mass. Similar results were observed when the aldehyde groups of the globin bound HNE were reacted with (pentafluorobenzy1)hydroxylamine. In this case the expected value of 351 Da separated the adduct peaks (spectrum not shown). Additional evidence supporting the near-exclusive formation of Michael adducts on the modified hemoglobin came from the stable isotope dilution GCMS analysis of HNE displaced from Schiff base linkages by hydroxylamine. It has long been assumed that, in the presence of proteins, HNE would bind both by Michael addition and by Schiff base formation, and several accounts have assumed that HNE predominantly forms Schiff base adducts with lysine residues (22, 23). Most analytical methods for HNE determination in biological tissues and fluids were developed to measure both free and Schiff base-bound HNE. Addition of amine reagents such as hydroxylamine and (pentafluorobenzy1)hydroxylamine has been touted to displace Schiff base-bound HNE (20, 27). These findings are not disputed; however, we recovered only a minor fraction (1.5 nmollmg) of HNE from Schiff base linkages a s the oxime derivative. This compares to 190 nmol/mg of Michael addition adduct as determined by the protein carbonyl assay. In other words, less than 1%of the HNE is bound through Schiff base linkages or noncovalently associated with the protein even when an excess of HNE was used to ensure addition of HNE to the less reactive lysine side chains. This low yield of Schiff base-bound HNE supports the conclusions derived from electrospray mass spectra and
558 Chem. Res. Toxicol., Vol. 8, No. 4, 1995
Bruenner et al.
I
Hba unmodified n
1
I
I I
I
Hb a
I
earlier studies (24-26) that the predominant mechanism of reaction of HNE with these proteins is by Michael addition. As the reaction to form oxime derivatives has been purported to displace HNE bound as Schiff bases to proteins, the observed lack of Schiff base formation calls into question the information gained from such an approach. Since HNE reacts rapidly with proteins via Michael addition, the extent of reaction is expected to depend upon the concentration of reactive protein and other competing nucleophiles as well as rates of HNE metabolism, which are not yet known. Chemical transformations of HNE can thus be expected to be tissuedependent and should also show a dependence on activities of metabolic enzymes, including aldehyde dehydrogenase and glutathione S-transferases. Since Schiff base formation is negligible, most existing methods measure only free HNE. The extent of reaction of HNE with protein is not measured using the common GC/MS and HPLC methods. Consequently, the application of additional methods, such as the immunochemical measurement of HNE-protein adducts (28), is important in determining the actual state of oxidative stress. These observations have several significant biological implications. The degree of covalent addition of HNE to proteins is dependent upon their structure. Proteins having hydrophobic binding regions, as in the case of P-LgB and perhaps membrane-bound proteins, are more predisposed to forming noncovalent associations, which may in turn facilitate the covalent modification of these proteins. If this is so, such proteins would serve as more sensitive biomarkers of oxidative damage. Furthermore, since the modification proceeds by a Michael-type reaction, the binding of HNE to proteins both modifies existing nucleophiles and adds a reactive aldehyde group. In addition to causing immediate dysfunction of the protein by reacting with critical amino acids in active sites or via conformational changes, the aldehyde group of HNE may participate in the formation of inter- and intramolecular cross-links via Schiff base formation with amino residues, resulting in further changes in protein function.
In conclusion, this study has qualitatively demonstrated that HNE adducts form almost exclusively by Michael addition to nucleophilic amino acid residues, even in reactions with lysine. This finding should help direct further development of methods that aim to quantify HNE modifications on specific proteins or a t specific amino acid residues. The ability to resolve these adducts intact illustrates the power of ESI mass spectrometry for directly identifying modifications of proteins and in relating structural determinants of proteins to their susceptibility toward modification. Such direct structural analysis of adducts of pathologically important proteins is needed, particularly in view of the hypotheses that implicate oxidants as causal to disease. In conjunction with methods such as HPLC, which allow separation of the small quantities of HNE-modified proteins expected in vivo, this approach will certainly be crucial in establishing the physiological significance of LDL and other HNE-modified proteins as biomarkers of oxidative damage.
Acknowledgment. This work was supported by the NIEHS Superfund Basic Research Program (Grant 2P42 ES-04699), the NIEHS Center for Environmental Health Science a t UC Davis (Grant 1P30 ES-05707), and the California Dairy Foods Research Center. Amino acid analyses were performed by the staff of the UC Davis Protein Structure Laboratory.
References Ames, B. N., Shigenaga, M. K., and Hagen, T. M. (1993)Oxidants, antioxidants, and the degenertive diseases of aging. Proc. Natl. Acad. Sci. U.S.A. 90, 7915-7922. Babior, B. M. (1984)Oxidants from phagocytes: agents of defense and destruction. Blood 64, 959-966. Hageman, J. J., Bast, A,, and Vermeulen, N. P. E. (1992) Monitoring of oxidative free radical damage in vivo: analytical aspects. Chem.-Bid. Interact. 82, 243-293. Pryor, W. A,, and Godber, S. S. (1991) Noninvasive measures of oxidative stress status in humans. Free Radical Biol. Med. 10, 177-184. Smith, C. V. (1991) Correlations and apparent contradictions in assessment of oxidant stress status in vivo. Free Radical Bid. Med. 10,217-224. Schaur, R. J., Zollner, H., and Esterbauer, H. (1991) Biological effects of aldehydes with particular attention to 4-hydroxynonenal
Electrospray Mass Spectra of HNE-Protein Adducts and malonaldehyde. In Membrane Lipid Oxidation (Vigo-Pelfrey, C., Ed.) Vol. 111, pp 141-163, CRC Press, Boca Raton. (7) Esterbauer, H., Schaur, R. J., and Zollner, H.(1991) Chemistry and biochemistry of 4-hydroxynonenal, malonaldehyde and related aldehydes. Free Radical Biol. Med. 11, 81-128. (8) Esterbauer, H., Zollner, H., and Scholz, N. (1975) Reaction of glutathione with conjugated carbonyls. 2.Naturforsch. 30c, 466473. (9) Yla-Herttuala, S.,Palinski, W., Rosenfeld, M. E., Parthasarathy, S., Carew, T. E., Butler, S., Witztum, J. L., and Steinberg, D. (1989) Evidence for the presence of oxidatively modified low density lipoprotein in atherosclerotic lesions of rabbit and man. J . Clin. Inuest. 84, 1086-1095. (10) Kautiainen, A. (1992) Determination of hemoglobin adducts from aldehydes formed during lipid peroxidation in vitro. Chem.-Biol. Interact. 83, 55-63. (11) Springer, D. L., Bull, R. J., Goheen, S. C., Sylvester, D. M., and Edmonds, C. G. (1993)Electrospray ionization mass spectrometric characterization of acrylamide adducts to hemoglobin. J . Tozicol. Enuiron. Health 40, 161-176. (12) Fenn, J . B., Mann, M., Meng, C. K., Wong, S. F., and Whitehouse, C. M. (1989) Electrospray ionization for mass spectrometry of large biomolecules. Science 246, 64-71. (13) Bruenner, B. A,, Jones, A. D., and German, J. B. (1994)Maximum entropy deconvolution of heterogeneity in protein modification: protein adducts of 4-hydroxy-2-nonenal. Rapid Commun. Mass Spectrom. 8, 509-512. (14) Godovac-Zimmermann, J. (1988) The structural motif of P-lactoglobulin and retinol-binding protein: a basic framework for binding and transport of small hydrophobic molecules? Trends Biochem. Sci. 13, 64-66. (15) Monaco, H. L., Zanotti, G., Spadon, P., Bolognesi, M., Sawyer, L., and Eliopoulos, E. E. (1987) Crystal structure of the trigonal form of bovine /3-lactoglobulin and of its complex with retinol at 2.5 A resolution. J . Mol. Biol. 197, 695-706. (16) Esterbauer, H., and Weger, W. (1967) Uber die wirkungen von aldehyden auf gesunde und maligne zellen, 3. mitt: synthese von homologen 4-hydroxy-2-alkenalen, I1 (Action of aldehydes on normal and malignant cells. 111. Synthesis of homologous 4-hydroxy 2-alkenals. 2). Monatsh. Chem. 98, 1994-2000. (17) Selley, M. L., Bartlett, M. R., McGuiness, J. A,, Hapel, A. J., Ardlie, N. G., and Lacey, M. J. (1989) Determination of the lipid peroxidation product trans-4-hydroxy-2-nonenalin biological
Chem. Res. Toxicol., Vol. 8, No. 4, 1995 559 samples by high-performance liquid chromatography and combined capillary column gas chromatography-negative-ion chemical ionisation mass spectrometry. J. Chromatogr. 488, 329-340. (18) Ebeler, S. E., Phillips, L. G., and Kinsella, J. E. (1990)Purification of P-lactoglobulin: isolation of genetic variants and influence of purification method on secondary structure. Milchwissenschaft 45,694-698. (19) Fanelli, A. R., Antonini, E., and Caputo, A. (1958) Studies on the structure of hemoglobin I. Physicochemical properties of human globin. Biochim. Biophys. Acta 30, 608-615. (20) Kinter, M., Sullivan, S., Roberts, R. J.,and Spitz, D. (1992) Trace quantitation of 4-hydroxy-2-nonenal in biological samples as its oxime-bis-tee-butyldimethylsilyl derivative using 3-hydroxynonanal as an internal standard. J . Chromatogr. 578, 9-16. (21) Levine, R. L., Garland, D., Oliver, C. N., Amici, A., Climent, I., Lenz, A., Ahn, B., Shaltiel, S., and Stadtman, E. R. (1990) Determination of carbonyl content in oxidatively modified proteins. Methods Enzymol. 186, 464-478. (22) Hoff, H.F., ONeil, J., Chisolm, G. M., 111, Cole, T. B., Quehenberger, O., Esterbauer, H., and Jurgens, G. (1989) Modification of low density lipoprotein with 4-hydroxynonenal induces uptake by macrophages. Arteriosclerosis 9, 538-549. (23) Hoff, H. F., and ONeil, J. A. (1991) Oxidation of LDL: role in atherogenesis. Klin. Wochenschr. 69, 1032-1038. (24) Uchida, K., and Stadtman, E. R. (1993) Covalent attachment of 4-hydroxynonenal to glyceraldehyde-3-phosphatedehydrogenase. J . Biol. Chem. 268,6388-6393. (25) Szweda, L. I., Uchida, K., Tsai, L., and Stadtman, E. R. (1993) Inactivation of glucose-6-phosphatedehydrogenase by 4-hydroxy2-nonenal. J . Biol. Chem. 268, 3342-3347. (26) Uchida, K., and Stadtman, E. R. (1992) Modification of histidine residues in proteins by reaction with 4-hydroxynonenal. Proc. Natl. Acad. Sci. U.S.A. 89, 4544-4548. (27) van Kuijk, F. J. G. M., Thomas, D. W., Stephens, R. J.,and Dratz, E. A. (1990) Gas chromatography-mass spectrometry of 4-hydroxynonenal in tissues. Methods Enzymol. 186, 399-406. (28) Uchida, K., Szweda, L. I., Chae, H., and Stadtman, E. R. (1993) Immunochemical detection of 4-hydroxynonenal protein adducts in oxidized hepatocytes. Proc. Natl. Acad. Sci. U.S.A. 90, 8742-8746.
TX940151D
