730
Chem. Res. Toxicol. 1998, 11, 730-740
Reaction of N-Acetylglycyllysine Methyl Ester with 2-Alkenals: An Alternative Model for Covalent Modification of Proteins Andrew Baker, Luka´sˇ Zˇ ´ıdek, Don Wiesler, Josef Chmelı´k,† Marty Pagel, and Milos V. Novotny* Department of Chemistry, Indiana University, Bloomington, Indiana 47408 Received September 11, 1997
Among the various reactions of lipid peroxidation products with proteins, 2-alkenals have been shown to react extensively with the -amino group of lysine residues [Zˇ ´ıdek et al. (1997) Chem. Res. Toxicol. 10, 702-710]. To obtain additional information about the kinetic and mechanistic aspects of this modification, a model peptide (N-acetylglycyllysine O-methyl ester) was reacted with 2-hexenal. The reaction products were characterized through a combination of NMR and MS techniques. The structural elucidation efforts have shown the formation of pyridinium salts through the reaction of two or more alkenals with one amino group. Kinetic data were obtained using a continuous infusion of the reaction mixture into an electrospray ionization mass spectrometer. A mechanism is proposed that offers an alternative model for the formation of stable protein cross-links. The reaction progresses through a Schiff base intermediate to form a dihydropyridine species which can be alternatively reduced to form various 3,4- or 2,5-substituted pyridinium species or react with another Schiff base to form a trialkyl-substituted pyridinium structure. The stoichiometry of this structure (aldehyde/amine) is 3:2, in contrast to the widely accepted 1:2. Therefore, it represents another possible crosslinking mechanism for bifunctional products of lipid peroxidation.
Introduction The secondary products of lipid peroxidation include alkanes, alkenes, ketones, alkanals, alkenals, and hydroxyalkenals (1, 2). Several products of lipid peroxidation have now been implicated in a large number of biological processes, including cytotoxic and atherogenic events. The early emphasis on the easily measurable malonaldehyde, among the most reactive secondary products of peroxidation, was soon augmented by an increasing interest in 4-hydroxy-trans-2-nonenal (HNE1) since its first identification as a cytotoxic agent (3). In addition, the related R,β-unsaturated aldehydes have been studied extensively. Various aspects of chemistry and biochemistry of the main lipid peroxidation products have been reviewed by Esterbauer et al. (4). The reactivity of carbonyl compounds originating from lipid peroxidation appears widely expressed in their modification of the biologically important proteins. HNE has been found to modify substantially the behavior of low-density lipoproteins (5-8) and alter enzymatic activities (9, 10). The effects of 2-alkenals on the glycolytic enzymes were also noted (11). Numerous other studies * Author for correspondence. Phone: (812) 855-4532. Fax: (812) 8558300. E-mail:
[email protected]. † Current address: Institute of Analytical Chemistry, Academy of Sciences of the Czech Republic, Veverˇ´ı 97, 61142 Brno. 1 Abbreviations: HNE, 4-hydroxy-2-nonenal; AcGKOMe, N-acetylglycyllysine O-methyl ester; ESI, electrospray ionization; FAB, fast atom bombardment; MALDI-TOF, matrix-assisted laser desorption/ ionization time-of-flight; CID, collision-induced dissociation; COSY, 1H correlation spectroscopy; NOE, nuclear Overhauser effect; HETCOR, 1H-13C heteronuclear correlation spectroscopy; PEG, poly(ethylene glycol); Py, pyridinium; SDS-PAGE, sodium dodecyl sulfate-polyacrylamide gel electrophoresis.
(for a review, see refs 12-14) implicate the importance of protein-aldehyde interactions. To unravel the seemingly complex chemistry of the reactions between aldehydes and proteins, it has been appropriate to use model systems: various primary amines; amino acids; peptides and relatively uncomplicated proteins. Although findings from such studies may or may not always be pertinent to more complicated biological systems, the relative reactivity of various protein residues needs to be established before considering proteins’ advanced structural aspects. The most widely studied HNE has been shown to act as an electrophile, adding to cysteine (15), methionine, histidine (16-18), and lysine (19) moieties. While the formation of thiol-derived Michael adducts with the protein containing sulfhydryl groups has been clarified (15) many years ago, there is less literature agreement on identity of the products formed in the reactions between HNE (or other R,β-unsaturated aldehydes) and various amines, in particular, the -amino group of lysine residues (5). At least some of the differences in structural assignments are attributable to differences in the used methodologies. The two primary reactions between aldehydes and lysine residues are the Michael addition and formation of the Schiff base. Stadtman and co-workers (16) consider the Michael addition to be a dominant pathway for modification of both histidine and lysine residues. Using two model proteins (human hemoglobin and β-lactoglobulin B), Bruenner et al. (20, 21) found that there is very little, if any, Schiff base formation. Kantiainen (22) determined that unsaturated aldehydes could react with the Nterminal valine in human hemoglobin to form either a Schiff base or a Michael adduct.
S0893-228x(97)00167-7 CCC: $15.00 © 1998 American Chemical Society Published on Web 06/11/1998
2-Alkenal Modification of Lysine Amino Groups
Napetschnig et al. determined that two 4-hydroxy-2pentenal molecules react with the amino group of glycine to form a pyridinium salt (23). Alaiz and Barragan have found that in the presence of phosphate buffer, 2-alkenals react with the -amino group of lysine to form pyridinium-carboxylic acid betaines (24). The products described in these studies involve the loss of acetaldehyde and water (23) or the loss of water and a butyl group from one of the alkenals (24). A recent work of this laboratory (25), using equine cytochrome c as a model protein without disulfide bonds or sulfhydryl groups, has led to the conclusion that there are some Schiff base modifications and some Michael additions, but mostly yet another class of modifications in which one amino group reacts with two alkenal moieties (with a loss of two water molecules). Through the use of peptide mapping and mass spectrometry, we found these modifications to be distributed along the polypeptide chain in a random fashion. Additional studies by laser desorption mass spectrometry on smaller peptides (26) verified these results. To elucidate further the structures of adducts with 2:1 stoichiometry, we have now chosen a simple peptide, N-acetylglycyllysine methyl ester, to react with trans-2hexenal (in a 3-fold molar excess). Using a continuous infusion of the reaction mixture into an electrospray mass spectrometer, real-time measurements were obtained that subsequently enabled us to get a “kinetic picture” of this modification. The modification produced a number of reaction products, all featuring pyridinium moieties, whose structures were unambiguously determined through a combination of tandem mass spectrometry and multidimensional nuclear magnetic resonance techniques. These structural and kinetic data suggest a mechanism where one lysine -amino group can react with two alkenals to form a dihydropyridine ring. This species can react further to form pyridinium moieties or stable pyridinium cross-links.
Experimental Section Chemical Reactions and Separations. The model peptide, N-acetylglycyllysine methyl ester as the acetate salt (AcGKOMe), was purchased from Bachem (King of Prussia, PA), while trans-2-hexenal originated with Sigma (St. Louis, MO). All other chemicals were from Aldrich (Milwaukee, WI). Distilled and deionized water was used for all experiments. The model peptide (0.0156 M) was incubated in the dark with a 3-fold molar excess of trans-2-hexenal in a 1:1 mixture of acetonitrile and water (pH 8.1) at 37 °C under air for a 24-h period. After reaction, the solvent was removed through lyophilization. The residue was redissolved in distilled water, filtered, and analyzed using high-performance liquid chromatography (HPLC). To assess the stability of our reaction products (and to ensure that these products were, in fact, not intermediate species) longer incubations were carried out as well. The analytical-scale separations were performed on a 250mm × 4.6-mm (i.d.) C18 (5-µm particle size) Vydac (The Separations Group, Hesperia, CA) column, using a model 200 PerkinElmer HPLC unit equipped with a model 235C diode array detector (Perkin-Elmer, Norwalk, CT). Preparative-scale separations were carried out with a 300-mm × 21.4-mm (i.d.) C18 Dynamax Microsorb column (Rainin, Woburn, MA), using a Perkin-Elmer model 3B liquid chromatograph with a Linear model UV/vis 20S detector operating at 230 nm. Aqueous trifluoroacetic acid (0.1%) and acetonitrile with 0.1% trifluoroacetic acid were used as the mobile phases in both chromato-
Chem. Res. Toxicol., Vol. 11, No. 7, 1998 731 graphic systems. The major separated fractions were lyophilized before being redissolved in an appropriate solvent for the electrospray ionization (ESI), fast atom bombardment (FAB), or matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF) mass spectrometry (MS), as well as the nuclear magnetic resonance (NMR) spectrometric studies. In a comparative study where trans-2-pentenal was used as a reactant, similar procedures were used to prepare and analyze the samples. NMR Spectroscopy. The conventional 1H one-pulse experiment,1H correlation spectroscopy (COSY) experiment, and 1H nuclear Overhauser effect (NOE) experiments were carried out using a 500-MHz Varian UnityINOVA spectrometer (Palo Alto, CA) equipped with a Nalorak (Martinez, CA) Z-Spec IDTXG probe. The 13C and two-dimensional 1H-13C heteronuclear correlation (HETCOR) experiments were performed using a 400MHz Varian Unity spectrometer. All reported chemical shifts were measured relative to TMS. To determine the chemical shifts of exchangeable protons, the NMR measurements were carried out in 90% H2O/10% D2O. Proton COSY data were also acquired in this solvent system to determine the connectivity of these exchangeable protons. Energy-minimized models of the peptides were generated using PC-Model (Serena Software, Bloomington, IN). Mass Spectrometry. For acquisition of the accurate mass data on the isolated HPLC fractions, FAB-MS was carried out on a Kratos MS-80 mass spectrometer, using 4 keV of xenon. Peak matching was accomplished using the sodiated adducts of poly(ethylene glycol) (PEG 600) as calibrant masses. The ESIMS studies were performed using a model 4600 Finnigan singlequadrupole mass spectrometer (San Jose, CA) equipped with an Analytica of Branford electrospray source (Branford, CT) and a Teknivent data system (St. Louis, MO). The ESI source was operated in the positive ion mode using a needle voltage of 4.5 kV. A stream of heated (190 °C) UHP grade nitrogen was used to aid droplet desolvation. For the LC/MS studies, an Upchurch Scientific microsplitter valve (Upchurch Scientific, Oak Harbor, WA) was used to split the effluent from a 250-mm × 4.6-mm (i.d.), 5-µm (particle size) octadecylsilica column; 5 µL/min was transferred to the ion source using a short piece of 50-µm i.d. fused silica tubing. Sheath liquid comprised of 75% propionic acid/25% 2-propanol was delivered to the ion source at 3 µL/min using a syringe pump (27). The rest of the eluent was passed to an Isco model µLC-10 UV detector (Lincoln, NE) operating at 215 nm. The measurements involving MALDI-TOF MS were carried out with a Voyager RP-DE time-of-flight mass spectrometer (PerSeptive Biosystems, Framingham, MA) operating in the positive ion mode. A 337-nm pulsed nitrogen laser was used to irradiate the samples mixed with R-cyanohydroxycinnamic acid as the matrix. Argon was the collision gas for the collisioninduced dissociation (CID) MS/MS studies. The kinetic studies were performed on the ESI mass spectrometer, using solutions of the peptide and trans-2-hexenal, in 50% aqueous acetonitrile (pH 8.1), introduced through the barrel of a disposable plastic syringe. The concentration of peptide and alkenal used in these studies is shown in Table 1. Warm water from an Endocal RTE-100 (Neslab, Newington, NH) recirculator was passed through a piece of 1/16-in. stainless steel tubing wrapped several times around the syringe barrel to maintain the incubation mixture at 37 °C. A short piece of 50-µm, i.d., fused silica tubing was used as a transfer line between the syringe and the electrospray source. The solution was pumped at 1.1 µL/min into the ion source using a KD Scientific model 100 infusion pump (Boston, MA). Another infusion pump operating at 1.1 µL/min was used to supply acetonitrile as a sheath liquid. A mixed-aldehyde species was synthesized by incubating the peptide (0.0156 M) in 50:50 acetonitrile/water with a stoichiometric quantity of 2-hexenal (0.0156 M) at 37 °C for 15 min. A 10-fold excess of 2-pentenal was added. After 5 min of reaction, the mixture was injected into the HPLC/MS system. The major
732 Chem. Res. Toxicol., Vol. 11, No. 7, 1998
Baker et al.
Table 1. Concentration of Aldehyde and Peptide in Kinetic Determinations peptide concn (mM)
aldehyde concn (mM)
ratio
15.6 15.6 15.6 15.6 15.6 15.6 15.6 15.6 15.6 15.6 15.6 7.80 15.6 31.2 46.8 78.0
0.780 3.90 7.80 15.6 31.2 39.0 46.8 78.0 156 234 312 15.6 15.6 15.6 15.6 15.6
1:0.05 1:0.25 1:0.5 1:1 1:2 1:2.5 1:3 1:5 1:10 1:15 1:20 0.5:1 1:1 2:1 3:1 5:1
product containing one pentenal and one hexenal was isolated and used for subsequent NMR and mass spectrometric studies.
Results and Discussion Structural Aspects. In our previous work on cytochrome c (25), the lysine residues throughout the entire polypeptide structure were shown to be major reactive sites for an attack by 2-alkenals. While only small amounts of both the Schiff base and Michael adducts were spectroscopically detected in different peptides after digestion with proteases, a stable modification with a higher mass was regularly observed, suggesting involvement of two aldehyde molecules at a single lysine site. The model peptide, N-acetylglycyllysine methyl ester (AcGKOMe), was chosen in this study to mimic the situation of an isolated lysine site. A representative chromatogram of the products from the reaction between 2-hexenal and AcGKOMe is shown in Figure 1, indicating an unexpected mixture complexity. On the basis of NMR experiments, high-resolution FABMS measurements, and tandem mass spectrometry (MS/ MS) data, seven major reaction products were identified. When the mixtures were incubated for much longer times (up to 1 month) with or without the phosphate buffer at either pH 7.0 or 8.1, similar, if not identical, chromatograms were obtained. The corresponding structures are shown in Figure 2, while the pertinent spectrometric data are provided as Supporting Information. A further clarification of the determination of structures I-VII is provided below. ESI- and MALDI-MS experiments performed on the products of reactions using alkenals of different chain length (2-pentenal and 2-hexenal) indicated a one peptide:two alkenal stoichiometry for structures I-IV (MW 390 and 392 for the pentenal reaction vs 418 and 420 for the hexenal-derived products). High-resolution FAB-MS further indicated that structure II was formed from one peptide and two aldehyde molecules, with the loss of two water molecules; structures I, III, and IV were formed through a reaction of one peptide with two aldehydes, with the loss of two water molecules and two hydrogen atoms; structures V and VI were formed from the reaction of two peptides plus three aldehydes, with the loss of three water molecules; and structure VII was the result of the reaction of three aldehydes with one peptide, with the loss of three water molecules. The results of high-resolution FAB-MS studies yielded exact mass values and empirical formulas for the major
Figure 1. Reversed-phase liquid chromatogram of the reaction products from a 12-h reaction of acetylglycyllysine methyl ester with trans-2-hexenal. Peaks marked with Roman numerals were isolated and studied through NMR and high-resolution mass spectrometry. The large peaks at 30 and 32 min are due to trans2-hexenal.
peptide-containing reaction products that were not consistent with the set of structures described by Alaiz and Barragan (24). These structures are 2,5-substituted pyridinium salts where one 2-alkenal loses a butyl group during formation of the pyridinium core. The aldehyde moiety in this group is alternatively oxidized to a carboxylic acid or reacts with acetaldehyde to form a 2,5substituted pyridinium salt, which is two carbons longer, and then oxidized to the corresponding carboxylic acid. On the basis of the proton NMR experiments, the downfield signals in the aromatic region for three hydrogens and the associated coupling suggested that these structures were variously substituted pyridinium salts. The fact that the addition products from one peptide and two aldehydes (2+1 species) had one singlet and two doublet hydrogens, while the addition product from either two peptides and three aldehydes (3+2 species) or three aldehydes and one peptide (3+1 species) had only two singlet hydrogens, indicates that there are three positions open on the ring in the 2+1 species, but only two free ring sites on the 3+2 and 3+1 species. More importantly, the two free sites on the ring in the 3+2 and 3+1 species are not adjacent, while two of the aromatic hydrogens in each 2+1 species are adjacent. The one-dimensional 1H and 13C NMR experiments indicated that there are two methyl groups in each of the 2+1 species and three methyl groups in each of the 3+2 and 3+1 species. This demonstrates that there is no scission of carbon-carbon bonds in the formation of the pyridinium core. Coupling constants for the vinyl protons were used to determine the cis/trans isomerism. 13 C NMR spectra from species I, III, and IV indicate that there are seven downfield resonances, in addition to the three carbonyl resonances present in the unmodified peptide. Two of these are accounted for by the double bond. The other five signals can be attributed to the
2-Alkenal Modification of Lysine Amino Groups
Chem. Res. Toxicol., Vol. 11, No. 7, 1998 733
Figure 2. Structures of the compounds isolated from the reaction of acetylglycyllysine methyl ester with trans-2-hexenal.
pyridinium core. There were no additional carbonyl resonances, as would be expected if one of the olefinic side chains had been oxidized. Proton COSY experiments were performed on all of the pyridinium derivatives to determine the alkyl chain connectivity. With the COSY spectrum in Figure 3 (a representative COSY from structure IV), starting from the methyl groups, spin systems for a propyl chain and a 1-butenyl chain were determined. After establishing the connectivities for the alkyl chains, one-dimensional NOE experiments were performed. Table 2 shows the peaks that were irradiated and also the peaks that had any appreciable energy transfer. Results from the semiquantitative NOE enhancements were compared with the distances derived from energy-minimized models of the derivatized peptide species. These data were used to determine placement of the substituents around the pyridinium ring, in addition to confirming the assignment of the cis/trans isomerism. Proton COSY for species I and III and HETCOR spectra for species I, IV, and VI are provided as Supporting Information. A proton COSY spectrum for species VI is shown in Figure 4. The results of the proton COSY, doublequantum-filtered COSY, and one-dimensional NOE experiments indicate that the 3+2 species differed from the 2+1 species only by addition of a second peptide moiety bound through a third hexenal molecule at the 5 position on the pyridinium ring. The connectivities of the third aldehyde and the second peptide of the 3+2 species were ambiguous in these experiments, as the pyridinium ring and nitrogen group have similar shielding characteristics. With the exception of the additional signals from the
Figure 3. Two-dimensional proton COSY of structure IV obtained in deuteriomethanol.
third aldehyde and the second peptide, structures III and V have very similar spectra, as do structures IV and VI. The spectra of both the pentenal- and hexenal-derived species were also very similar. Two-dimensional protoncarbon HETCOR and 13C NMR spectroscopic studies were consistent with these structures.
734 Chem. Res. Toxicol., Vol. 11, No. 7, 1998
Baker et al.
Table 2. NOE Enhancements species I
group irradiateda Py-2H (8.86) CHCH (6.78) Py-CH (6.54) CHCH2 (2.34)
III
Py-CH (6.49) CHCH (6.14) CHCH2 (2.15) Py-6H (8.73) Py-5H (7.94) Lys-CH2 (4.55)
IV
Py-2H (8.94) PyCH2 (2.92)
CHCH2 (2.37) V
Py-2H (8.83) Py-CH (6.52) CHCH (6.17)
VI
Py-6H (8.60) Py-2H (8.82) Py-6H (8.73) Py-CH (6.67)
CHCH (6.54) CH2CH2 (2.37)
energy transfer to
intensityb
CHCH (6.78) Py-CH (6.53) Lys-CH2 (4.55) Py-2H (8.86) Py-4H (8.48) CHCH2 (2.34) Py-2H (8.86) Py-4H (8.48) CHCH2 (2.34) CHCH (6.78) Py-CH (6.54) CHCH (6.14) Py-2H (8.67) Py-CH2 (2.86) Py CH (6.49) CHCH2 (2.15) CHCH (6.14) Py-2H (8.67) Py-5H (7.94) Lys-CH2 (4.55) Py-6H (8.73) PyCH2 (2.86) CH2CH2 (1.72) Py-6H (8.73) Py-2H (8.67) Lys-γCH2 (1.45) Lys-δCH2 (1.92) Py-CH (6.67) CHCH (6.58) Lys-CH2 (4.55) CH2CH2 (1.72) CH2CH3 (1.05) Py-CH (6.67) Py-5H (7.80) CHCH (6.58) CHCH (6.58) Py-CH (6.67) CH2CH3 (1.16) Lys-CH2 (4.55) CHCH (6.17) Py-CH2 (2.8) Py-CH (6.52) CHCH2 (2.16) Lys-CH2 (4.55) Py-CH (6.67) CHCH (6.54) Lys-CH2 (4.55) Lys-CH2 (4.55) Lys-δCH2 (1.68) CH2CH3 (0.90) Py-2H (8.82) Py-CH2 (2.97)‘ CH2CH2 (2.37) CH2CH3 (1.16) Py-2H (8.82) CH2CH2 (2.37) CH2CH3 (1.16) Py-CH (6.67) CHCH (6.54) CH2CH3 (1.16)
strong strong moderate moderate moderate weak moderate moderate moderate moderate moderate strong weak weak strong moderate moderate moderate moderate moderate moderate weak weak moderate moderate weak weak moderate strong moderate moderate weak strong moderate weak moderate moderate weak weak moderate moderate moderate moderate weak weak moderate weak weak weak weak weak weak weak weak moderate moderate weak moderate moderate weak
a
Groups were irradiated with single RF during the delay and pulse portions of the experiment, but not during signal acquisition. b NOEs were determined semiquantitatively. Enhancements were assigned as weak (
