Mass spectrometry in aging research

Authors

  • Christian Schöneich

    Corresponding author
    1. Department of Pharmaceutical Chemistry, University of Kansas, 2095 Constant Avenue, Lawrence, Kansas 66047
    • Department of Pharmaceutical Chemistry, University of Kansas, 2095 Constant Avenue, Lawrence, KS 66047.
    Search for more papers by this author

Abstract

This review covers the application of mass spectrometric techniques to aging research. Modern proteomic strategies will be discussed as well as the targeted analysis of specific proteins for the correlation of post-translational modifications with protein function. Selected examples will show both the power and also current limitations of the respective techniques. Experimental results and strategies are discussed in view of current theories of the aging process. © 2004 Wiley Periodicals, Inc., Mass Spec Rev 24:701–718, 2005

I. INTRODUCTION

The last years have witnessed an enormous growth in the application of mass spectrometry techniques to solve biological and biomedical problems. Successful collaborations between analytical and biomedical researchers develop best if both fields understand the power and limitations of the respective techniques of their collaborators. The fact that an antibody is able to detect minute amounts of a specific protein does not necessarily mean that a full analysis of all possible post-translational modifications by multidimensional mass spectrometry is possible with complete sequence coverage. On the other hand, often only limited amounts of biological material can be collected at a high price so that analytical scientists need to make every effort to preserve sample and maximize the amount of information, which can be obtained. The present article is written with the intention to review current applications of mass spectrometry to gerontology, to introduce the analytical scientist to some of the questions important for molecular aging research, and to demonstrate the capabilities of modern analytical techniques to biomedical scientists.

II. BIOLOGICAL AGING

Biological aging is a function of several closely interrelated parameters such as metabolic rate, caloric intake, genetics, lifestyle, and environmental factors (Jazwinski, 1996; Smith & Pereira-Smith, 1996; Sohal & Weindruch, 1996; Finch & Tanzi, 1997; Lamberts, Van den Beld, & Van der Lely, 1997; Morrison & Hof, 1997). Multiple theories for the aging process have been developed (Masoro, 1993) including the “rate of living theory,” the “somatic mutation theory,” the “error catastrophy theory,” the “cross-linkage theory,” and the “free radical theory of aging.” Many of these theories are based on some common phenomena. For example, the rate of living theory inversely correlates metabolic rate with lifespan. Higher metabolic rates are usually associated with higher levels of reactive oxygen species (ROS), which become available for the chemical modification of biomolecules. Such modifications will result in some post-translational cross-linking of proteins and DNA, supporting the “cross-linkage theory” of aging. Moreover, the “free radical theory” of aging is built on a comparable argument: an age-dependent increase in the oxidative modifications of biomolecules caused by an age-dependent increase of the steady-state levels of ROS. In recent years, much support has been accumulated for the involvement of oxidative stress in aging (Finkel & Holbrook, 2000; Hekimi & Guarente, 2003). The overexpression of both antioxidant enzymes Cu,Zn superoxide dismutase (SOD1) and catalase in Drosophila melanogaster has resulted in a ca. 34% increase of the lifespan (Orr & Sohal, 1994), and the overexpression of SOD1 only in the motorneurons of Drosophila melanogaster has resulted in an up to 40% increase of the lifespan (Parkes et al., 1998). An extension of the lifespan by caloric restriction has been noted for multiple, unrelated organisms such as yeast, worms, flies, and mice (Longo & Finch, 2003). Moreover, through the central involvement of glucose levels and insulin/insulin-like growth factor-I (IGF-1)-dependent signaling pathways, a mechanistic link between caloric restriction and the expression of antioxidant enzymes as well as stress proteins (e.g., heat shock proteins) has been identified. Mutations that decrease the activity of the insulin/IGF-1-dependent signaling cascade, as well as lower plasma levels of IGF-1, increase longevity and the expression of SOD. Usually, caloric restriction lowers glucose and insulin/IGF-1-dependent signaling, linking caloric restriction to increased expression of SOD.

III. BIOLOGICAL AGING AND POST-TRANSLATIONAL PROTEIN MODIFICATION

Post-translational protein modifications expand the genetic code in a sense that they allow for multiple variants of a single gene product through specific, potentially reversible, covalent alterations of amino acid side chains. Approximately 200 distinct post-translational protein modifications are known (Khidekel & Hsieh-Wilson, 2004) but we are far from a complete understanding of all their specific functional and biological consequences, and as of yet there has not been a full characterization of all possible post-translational modifications in aged tissue. Several important mechanisms of non-enzymatic post-translational modifications include deamidation, oxidation, racemization, β-elimination, and covalent aggregation (e.g., the formation of disulfide, dityrosine, or lysylalanine).

It is now well established that biological aging correlates with the accumulation of oxidized biomolecules in tissue such as oxidized proteins (Oliver et al., 1987; Sohal, Sohal, & Orr, 1995; Berlett & Stadtman, 1997), lipids (Yu et al., 1996), DNA bases (Ames & Shigenaga, 1993; Sohal, Sohal, & Orr, 1995), advanced glycation endproducts (AGEs) (Wolff, 1993; Wells-Knecht, Brinkmann, & Baynes, 1995a,b; Chen et al., 1996; Fu et al., 1996; Xu & Sayre, 1998a), and lipofuscin (Chen et al., 1996; Ivy et al., 1996), supporting the “Free Radical Theory of Aging” (Beckman & Ames, 1998). The importance of protein oxidation for biological aging is reflected in the age-dependent upregulation of stress response genes that process damaged or misfolded proteins (e.g., heat shock proteins), demonstrated for skeletal muscle (Lee et al., 1999).

An important age-dependent hydrolytic modification is the deamidation of protein asparagine to aspartic and isoaspartic acid, and racemization reactions interconverting L- and D-aspartic acid as well as L- and D-isoaspartic acid (Lindner & Helliger, 2001; Clarke, 2003). Deamidation can have profound biological consequences such as the initiation of amyloid formation (Nilsson, Driscoll, & Raleigh, 2002; Shimizu et al., 2002) or the loss of the anti-apoptotic activity of Bcl-xL (Deverman et al., 2002, 2003). For characterization, deamidated peptides can be obtained through proteolytic digestion of the proteins of interest, and the deamidated peptides separated from their original peptides by high performance liquid chromatography and capillary electrophoresis prior to MS analysis (Lindner et al., 1998). Fourier-transform ion cyclotron resonance (FT-ICR) MS can be used to analyze mixtures of deamidated and parent peptides as this technique allows baseline resolution of the monoisotopic peak of the deamidated peptide from the first isotope peak of the parent peptide, i.e., a mass difference of 0.0193 Da (Schmid et al., 2001). Some protein deamidation is subject to repair by the enzyme protein L-isoaspartate (D-aspartate) O-methyltransferase (PCMT1). Polymorphism results in different human PCMT1 isoforms with either isoleucine or valine at position 119; these differ in specific activity and substrate affinity (De Vry & Clarke, 1999; Clarke, 2003). It appears that healthy aging may correlate with an increased frequency of a heterozygous genotype, simultaneously expressing both isoforms (Clarke, 2003). Mutation of the Escherichia coli protein PCMT1 did not affect bacterial growth but significantly enhanced survival during conditions of stationary state or heat stress (Li & Clarke, 1992). A 50% lower expression of the methyl transferase was observed in epileptic hippocampus compared to controls, correlating with the accumulation of damaged tubulin (Lanthier et al., 2002).

Aging can also result in changes of yield, specificity, and lifetime of regulatory modifications such as phosphorylation, methylation, and/or acetylation (Xu & Narayanan, 1998; MacCoss et al., 2002; Sarg et al., 2002). Much research effort has been placed on the characterization of the age-dependent heterogeneity of soluble and insoluble crystallins in the lens, and the mechanisms underlying the conversion of soluble into insoluble protein. By means of MS characterization, several truncated, S-glutathiolated, intrachaindisulfide-containing, deamidated, phosphorylated, methylated, and acylated isoforms were characterized, potentially important for cataract development (Lampi et al., 1998, 2002; Ma et al., 1998; Cherian-Shaw et al., 1999; Feng, Smith, & Smith, 2000; Zhang et al., 2001a; Lapko et al., 2002a; Ueda, Duncan, & David, 2002; Ueda et al., 2002; Schaefer et al., 2003). Advanced glycation end products were characterized in the water-insoluble fraction of lens proteins (Biemel, Friedl, & Lederer, 2002).

For a functional evaluation, both the location and the quantity of these post-translational modifications within a given protein need to be analyzed. In a single protein, a detailed functional correlation of activity with a specific modification can be compromised by the fact that multiple amino acid residues of the same type and of different types may be modified to various extents. Moreover, a single amino acid may be converted to different products depending on environmental conditions, protein conformation, the concentration of the modifying agent (e.g., an oxidant), and the time of exposure to the modifying agent. Functional characterization would ideally require that individual isoforms be isolated, each carrying only one specific modification at one specific site. Often the functional integrity of proteins requires the formation of protein complexes. If not only one but several constituents of these complexes suffer age-dependent modifications, an exact correlation of a specific modification with activity can be a very complex task. In this respect, it must also be noted that in protein complexes a larger functional decline may be the result of smaller levels of inactivation of each of the constituents. If in a heterotrimeric complex each post-translationally modified protein constituent shows an average activity of 75% compared to non-modified wild-type, the maximum possible overall activity of the complex may be mathematically approximated to (0.75)3 = 0.42, or 42%, compared to an intact complex built with only non-modified protein constituents.

An important limitation in the characterization of modified proteins from aged tissue is the fact that only a subset of the proteome is available. Proteins are constantly synthesized and degraded in most tissues. The accumulation of an oxidized protein depends on the rate of formation of oxidizing species, the relative rate constants by which these oxidizing species react with either the protein of interest or competing substrates (e.g., antioxidants), and the rates by which the oxidized protein is either repaired or degraded. These processes are summarized in Figure 1.

Figure 1.

Summary of pathways leading to the accumulation of oxidized proteins. The formulas ΣΣPiXj and ΣΣν(PiXj) represent the sum of all proteins containing various modifications and the total rate of degradation of all proteins with the different modifications, respectively. Abbreviations: MsrA/B, methionine sulfoxide reductase A/B; PDI, protein disulfide isomerase; RNS, reactive nitrogen species; ROS, reactive oxygen species.

IV. MASS SPECTROMETRIC ANALYSIS OF POST-TRANSLATIONAL MODIFICATIONS

The strategies for protein analysis in aging tissue may be divided into two main approaches. On one hand, profiling of the entire proteome or sub-proteomes may be pursued by proteomic techniques. In most cases, these techniques will permit the differential comparison of proteomes from different sources, the identification of specific proteins present or absent in certain tissues, and the indication of some common post-translational modifications. However, generally the proteomic approach will not allow for the quantitative analysis of all potential post-translational modifications of a given protein of interest as well as its functional characterization. Such studies require the enrichment or targeted purification of the specific protein in as large as possible quantities from a tissue of interest to achieve maximal sequence coverage (ideally 100%) for MSn analysis. This review deals separately with the proteomic analysis of aged tissue and with selected examples of targeted isolation of specific proteins from aged tissue for mass spectrometric analysis.

V. PROTEOMIC ANALYSIS OF AGING TISSUE

Proteomic analysis encompasses the qualitative, quantitative, and functional characterization of the entire protein profile of a given cell, tissue, and/or organism. Such analysis includes the profiling of native proteins as well as of isoforms, splice variants, mutants, and post-translationally modified species, and the definition of protein–protein interactions. While the last years have witnessed an immense improvement in the techniques available for the analysis of the proteome, largely because of advances in MS analysis (Aebersold & Goodlett, 2001), it has also been realized that none of the individual methods will be able to cover all aspects of proteomic analysis. With a myriad of different proteomic techniques available, each specific research problem may be solved by a different approach, ranging from the traditional one-(1D) or two-dimensional (2D) gel electrophoretic separation to more novel strategies such as surface-enhanced laser desorption/ionization (SELDI) (Merchant & Weinberger, 2000), or the mass spectrometric profiling of peptides directly from tissue slices (Chaurand, Stoeckli, & Caprioli, 1999; Chaurand & Caprioli, 2002). The current state and future directions of proteomic research has been summarized by several authors (Pandey & Mann, 2000; Aebersold & Goodlett, 2001; Griffin & Aebersold, 2001) and the application to aging research has been reviewed (Gromov et al., 2002; Dierick et al., 2002a). Chang et al. have characterized the reproducibility and sensitivity of proteomic analysis of mitochondrial proteins from aged mouse skeletal muscle, suggesting that a sample size of 10 animals will be sufficient to detect a 100% difference in 97% of the 505 mitochondrial proteins resolved in their 2D electrophoresis separation (Chang et al., 2003).

VI. PROTEOMIC ANALYSIS OF POST-TRANSLATIONAL MODIFICATIONS

Considering the progressive age-dependent functional changes of most cell types and tissues, an accurate profile of age-dependent post-translational protein modifications is mandatory.

A. Phosphorylation

Much research effort has been placed on the characterization of the “phosphoproteome,” i.e., the nature and levels of transient protein phosphorylation (Conrads, Issaq, & Veenstra, 2002; Mann et al., 2002). Age-dependent changes in protein phosphorylation have been noted, for example, for crystallins (Ueda et al., 2002) and various sarcoplasmic reticulum proteins (Xu & Narayanan, 1998). Key to the sensitive analysis of phosphopeptides and/or phosphoproteins is a separation of phosphorylated peptides or proteins from excess non-phosphorylated species. Phosphoproteins can be enriched by immuno-affinity methods utilizing antibodies recognizing phosphoserine, phosphothreonine, or phosphotyrosine, respectively (Grønborg et al., 2002; Pandey et al., 2002). The immuno-affinity purified protein species are then characterized by MS and tandem MS analysis after proteolytic digestion. Generally, in the negative electrospray ionization (ESI) mode, the presence of either phosphotyrosine, phosphoserine, or phosphothreonine is indicated by a characteristic fragment of m/z 79 (PO3) while in the positive ESI mode phosphothreonine and phosphoserine show neutral losses of HPO3 (80 Da) or H3PO4 (98 Da) (Schlosser et al., 2001). Steen et al. (2002) have shown that the phosphotyrosine-specific immonium ion with m/z 216.043 can be used as a “reporter ion” for precursor tyrosine-phosphorylated peptides in complex mixtures. Moreover, a similar strategy can be used for phosphoserine and phosphotyrosine-containing peptides when they are chemically modified into dimethylamine-containing sulfenic acids (Steen & Mann, 2002a). The resolution of such immonium ions from other small peptide fragments is possible because the reporter ions contain a higher incidence of mass-deficient atoms (O, P, S), creating an inherent mass-deficient mass tag (MaDMaT) (Steen & Mann, 2002b). The chemical modification of phosphoserine and phosphothreonine residues is based on their lability to undergo β-elimination at alkaline pH. The resulting reactive dehydroalanine residue can be derivatized via Michael addition of appropriate nucleophiles. This concept has been utilized to biotinylate original phosphoserine and phosphothreonine residues for affinity purification (Oda, Nagasu, & Chait, 2001), selective immobilization on solid phases (Zhou, Watts, & Aebersold, 2001), and incorporation of a cleavage-site for a lysine-specific protease through reaction of the dehydroalanine moiety with cysteamine (Knight et al., 2003). However, it should be noted that β-elimination is not restricted to phospho-amino acids. For example, O-linked β-N-acetylglucosamine (O-GlcNAQc)-modified serine or threonine suffer β-elimination, on which basis Wells et al. developed a procedure termed mild β-elimination followed by Michael addition with dithiothreitol (BEMAD) for the selective mapping of O-GlcNAc sites (Wells et al., 2002). In an alternative approach, phosphopeptides can be selectively enriched using immobilized metal affinity chromatography (IMAC) on supports containing trivalent metal ions such as Fe3+ or Ga3+ (Zarling et al., 2000; Riggs, Sioma, & Regnier, 2001; Chen et al., 2002; Ficarro et al., 2002). However, optimal resolution of phosphorylated from non-phosphorylated peptides requires chemical esterification of peptide carboxylate groups (Ficarro et al., 2002).

B. Oxidative Modifications

1. Protein-Associated Carbonyls

Consistent with the “free radical theory” of aging, most tissues show an age-dependent increase of the steady-state concentrations of oxidized proteins. Nyström and co-workers have used stationary phase bacterial cells as a model for aging somatic cells of higher eukaryotic organisms to elucidate several pathways leading to the accumulation of oxidized proteins (Nyström, 2002). Oxidized proteins were detected based on increased levels of protein-associated carbonyls following derivatization with 2,4-dinitrophenylhydrazine, 2D gel electrophoresis, and Western blotting with anti-2,4-dinitrophenyl (DNP) antibodies (Dukan & Nyström, 1999; Aguilaniu et al., 2001; Ballesteros et al., 2001). The specificity of the anti-DNP antibody has been characterized through specific reaction with DNP conjugated to proteins (Eshhar, Ofarim, & Waks, 1980; Shacter et al., 1994), and a lower limit of detection of protein-associated carbonyls was estimated to 1 pmol (Shacter et al., 1994). The term “protein-associated carbonyls” classifies carbonyl-containing covalent protein modifications, which either result from the direct chemical conversion of an amino acid residue into a carbonyl product or from the covalent cross-link of an amino acid residue with an α,β-unsaturatd carbonyl (e.g., 4-hydroxy-2-nonenal, 4-HNE) via Michael addition (Berlett & Stadtman, 1997). Experiments with growth-arrested Escherichia coli demonstrated that protein carbonylation is fairly selective, targeting specific proteins responsible for peptide chain elongation (EF-Tu), protein folding, and reconstruction (DnaK), DNA architecture and gene expression (H-NS, basic isoform), central carbon catabolism (Icd), Mdh, AceF, SucC, Pyk, PtsI), amino acid biosynthesis and nitrogen assimilation (GlnA, GltD), and general stress protection (UspA). The additional carbonylation of GroEL, EF-G, and the acidic isoform of H-NS was only detected in SOD- and/or catalase-deficient strains. Mechanistically, not the rate of respiration per se but the degree of coupling in the mitochondrial respiratory apparatus, switching from state 3 to state 4, appears to be responsible for the increased accumulation of oxidized proteins, somehow arguing against the “rate-of-living” theory of biological aging (Aguilaniu et al., 2001). Importantly, during cell division mother cells retain a large fraction of the oxidized proteins, somehow protecting their daughter cells from the consequences of oxidative insult (Aguilaniu et al., 2003). For example, the levels of oxidized proteins in first-generation daughter cells were six-fold lower compared to the mother cells while the rates of protein degradation in both cells were similar. This trend of mother cells to retain oxidized proteins decreases with replicative age (Aguilaniu et al., 2003). While such result would suggest that old tissues would have a reduced ability to resist oxidative stress, recent results with Arabidopsis thaliana show that plants have developed a mechanism, by which old leaves can eliminate oxidized proteins prior to bolting and flowering of the plant, i.e., prior to reproduction (Johannson, Olsson, & Nyström, 2004).

Overall, the levels of carbonylated proteins appear to depend on the synthesis of aberrant, i.e., misfolded, proteins while turnover rates of oxidized proteins remained stable (Dukan et al., 2000). This finding suggests that the presentation of reactive substrates (i.e., misfolded proteins) may be one rate-limiting factor in the age-dependent accumulation of oxidized proteins. An intact metal-binding site is an important factor determining the efficiency of site-specific metal-catalyzed oxidation, one prominent process leading to the formation of protein-associated carbonyls (Requena et al., 2001). However, abberant folding may increase the levels of non-specific metal binding and promote a less selective oxidation of a non-native protein. Moreover, misfolded proteins may provide better access to nucleophilic amino acid side chains for reactive carbonyl compounds (e.g., 4-HNE), originating from lipid peroxidation (Refsgaard, Tsai, & Stadtman, 2000).

The Western blot analysis of protein-associated carbonyl compounds has been utilized to monitor quite selective age-dependent protein oxidation in Drosophila melanogaster (Das et al., 2001) and in human Alzheimer's disease brain (Castegna et al., 2002a,b). Moreover, bacterial cells exposed to hydrogen peroxide or superoxide-generating conditions show fairly selective patterns of protein carbonylation (Tamarit, Cabiscol, & Ros, 1998; Cabiscol et al., 2000). It is now important that the identified proteins are carefully examined with sensitive mass spectrometry techniques to (i) confirm the carbonyl content by a complementary technique, and (ii) locate the modified amino acids. A potential source of error in the immunochemical identification of carbonyl-containing proteins could be the contamination of a non-modified protein in a 2D gel spot by a trace amount of a second, carbonylated, protein. Modern chemiluminescence detection methods for Western blots are very sensitive and the contamination of a major fraction of a native protein with a minor fraction of a carbonylated protein may lead to the erroneous assignment of carbonylation to the major protein fraction. The unambiguous localization of protein-associated carbonyl(s) to a specific protein would require the tandem MS sequencing of the modified peptide(s). Fenaille, Tabet, and Guy, (2002) described this strategy for the identification of 4-HNE-modified peptides following immuno-affinity purification with immobilized anti-HNE- and anti-2,4-dinitrophenylhydrazone antibodies. The antibodies were immobilized with cyanogen bromide (CNBr)-activated sepharose 4B, and aliquots of these immunosorbents loaded into disposable solid-phase extraction cartridges. Peptide derivatives resulting from the Michael addition of HNE were well retained by the immobilized anti-HNE antibody whereas Schiff base-containing products of HNE were detected in the flow through. After derivatization of the HNE-modified (as well as malondialdehyde-modified) peptides with 2,4-dinitrophenylhydrazine, the 2,4-dinitrophenylhydrazone-containing products were selectively retained by the immobilized anti-DNP antibody.

2. Nitrosation and 3-Nitrotyrosine Formation

The nitric oxide (NO)-dependent protein modification plays an important role in the regulation of many cellular processes (Greenacre & Ischiropoulos, 2001; Turku & Murad, 2002; Ischiropoulos & Beckman, 2003). Proteomic strategies have been devised to specifically monitor S-nitrosocysteine- and 3-nitrotyrosine-containing proteins (for 3-nitrotyrosine, see structure 1). For the identification of S-nitrosated proteins, Jaffrey et al. (2001) selectively reduced protein-S-nitrosocysteine to free thiol with ascorbate following the derivatization of all other free thiol groups with methyl methanethiosulfonate. Subsequently, the free thiols were derivatized with a cleavable biotin-containing reagent, affinity purified, and released for SDS–PAGE separation and mass spectrometric identification.

Thumbnail image of

Aulak and co-workers used 2D gel electrophoresis for the resolution and Western blot detection of 3-nitrotyrosine-containing proteins, which were subsequently identified by MS and tandem MS analysis (Aulak et al., 2001; Miyagi et al., 2002). Similar methodology was applied by Kanski et al. to monitor the age-dependent accumulation of 3-nitrotyrosine in rat skeletal muscle (Kanski, Alterman, & Schöneich, 2003) and heart (Kanski et al., 2004) and by Turku et al. (2003) for the identification of 3-nitrotyrosine-containing proteins in the mitochondria of diabetic mice. Elfering et al. (2003) characterized the relative distribution of nitrated proteins in rat liver mitochondria and quantified half-lives of nitrated proteins. An important discovery was the presence of an LPS-inducible enzymatic activity, which reduces 3-nitrotyrosine immunoreactivity for specific proteins such as histone H1.2 (Irie et al., 2003). The presence of such enzymatic activity, in addition to protein turnover and some chemical selectivity of the nitrating species for specific tyrosine residues in selected proteins can rationalize the fact that, so far, proteomic studies have shown that 3-nitrotyrosine accumulation does not occur randomly but on a defined subset of proteins. The positive identification of a nitrated protein actually requires the enrichment and characterization of 3-nitrotyrosine-containing peptides by MSn analysis. Such data are available for the cardiac mitochondrial electron transfer flavoprotein isolated from aged rats (Kanski et al., 2004). Such experiments may benefit from specific enrichment of 3-nitrotyrosine-containing peptides. One strategy involves selective reduction of 3-nitrotyrosine to 3-aminotyrosine followed by covalent coupling to a cleavable, disulfide-containing, biotin affinity tag (Nikov et al., 2003). Quantitative analysis of 3-nitrotyrosine-containing peptides can be achieved by ESI-MS analysis using the native reference peptide (NRP) method, i.e., relative to the abundance of unmodified peptides of a given protein of interest (Willard et al., 2003). The ESI-MSn analysis of 3-nitrotyrosine-containing peptides yields unambiguous results, where the introduction of the nitro group increases the molecular weight of the original peptide by +45 atomic mass units (amu). However, during matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF) MS analysis 3-nitrotyrosine suffers photochemical cleavage reactions leading to the formation of 3-nitroso- and 3-hydroxylaminotyrosine (Sarver et al., 2001).

3. Other Oxidative Modifications

MacCoss et al. (2002) applied a “shotgun” approach to identify modified proteins from a cataractous human lens by MSn analysis. Here, complete unresolved protein mixtures were digested before separation of the resulting peptides by multidimensional liquid chromatography, identifying up to 11 different crystallins together with specific oxidation sites (Met, Tyr, Trp), acetylation sites (Lys), and methylation sites (Arg, Lys). This strategy is termed “multidimensional protein identification technology” (MudPIT) (Washburn, Wolters, & Yates, 2001; Wolters, Washburn, & Yates, 2001). The specific tandem MS analysis of human γS crystallin identified surface-exposed Cys residues as additional age-dependent methylation sites, potentially inhibiting the formation of disulfide-linked aggregates (Lapko, Smith, & Smith, 2002b). More proteomic evidence for Cys oxidation under conditions of oxidative stress comes from the tandem MS characterization of cysteic acid-containing peroxiredoxins following the exposure of Jurkat T-cell lymphoma cells to an organic peroxide or glucose oxidase (Rabilloud et al., 2002). To specifically characterize the sensitivity of mitochondrial protein thiols towards oxidative stress, Lin et al. (2002) utilized a lipophilic cation, (4-iodobutyl)triphenylphosphonium (IBTP), which enriches inside the mitochondria, and an antibody directed against the triphenylphosphonium moiety. Especially the oxidative modification of intracellular proteins occurs in the presence of large concentrations of the endogenous antioxidant glutathione (GSH). Therefore, S-glutathiolation is a common endogenous process, and proteomic methods for the detection of S-glutathiolated proteins have been devised either using biotinylated GSH ester for the selective immuno-affinity purification of glutathiolated proteins (Sullivan et al., 2000), or [35S]-labeled GSH for the specific detection of [35S] incorporation into proteins resolved on 2D gels (Fratelli et al., 2002). Obviously, both methods require the use of cell cultures and cannot be performed in vivo, where, instead S-glutathiolation would need to be characterized by direct tandem MS analysis of S-glutathiolated peptide sequences. A specific subset of proteins, which are susceptible to S-glutathiolation by S-nitrosoglutathione was identified utilizing S-nitrosoglutathione-sepharose (Klatt et al., 2000).

An interesting approach was used by van der Vliess et al. (2002) to monitor proteins sensitive to oxidative bityrosine formation. They incubated normal human dermal fibroblasts with a membrane-permeable fluorescein-labeled tyrosine analog, tyramine (acetylTyrFluo). Following the exposure of these cells to hydrogen peroxide, acetylTyrFluo-tyrosine cross-linked proteins were visualized after 2D gel electrophoresis and Western blotting with an anti-fluorescein antibody, showing predominantly cross-linking of proteins from the endoplasmic reticulum.

One oxidatively modified protein repeatedly identified in several of these studies is enolase (Tamarit, Cabiscol, & Ros, 1998; Cabiscol et al., 2000; Fratelli et al., 2002; Castegna et al., 2002b; Kanski, Alterman, & Schöneich, 2003). The high sensitivity of this protein towards oxidative modification may be caused by its expression levels and/or specific conformational parameters.

4. Protein Glycation

The accumulation of advanced glycation endproducts is a hallmark of age-dependent protein modification. AGEs represent a class of chemical structures generated via breakdown or cross-linking of initial sugar-protein adducts. Poggioli et al. could demonstrate that age-dependent glycation does not occur randomly but appears to selectively target a few proteins (Poggioli, Bakala, & Friguet, 2002) though no mass spectrometric studies to characterize these proteins were presented. In contrast, Crabb et al. (2002) performed both HPLC-tandem MS and Western blot experiments to identify proteins, including AGE-modified proteins, present in debris-like material, referred to as “drusen,” accumulating below the retinal pigment epithelium on Bruch's membrane during age-related macular degeneration (AMD).

VII. ALTERNATIVES TO 2D GEL ELECTROPHORESIS IN PROTEOMICS

Despite many limitations, 2D gel electrophoresis continues to be a standard technique for proteomic analysis in many laboratories and continuous improvements are made with regard to reproducibility, solubility (especially of membrane proteins), alternative protein stains, and the differential display of proteins. For example, Zhou et al. (2002a) have used differential in-gel electrophoresis (DIGE) to screen for cancer-specific protein markers. Two separate pools of proteins from normal and pathologic cells/tissue, respectively, are labeled with two different fluorescent dyes, 1-(5-carboxypentyl)-1′-propylindocarbocyanine halide (Cy3) N-hydroxy-succinimidyl ester and 1-(5-carboxypentyl)-1′-methylindodi-carbocyanine halide (Cy5) N-hydroxy-succinimidyl ester, which can be individually excited (540 and 620 nm, respectively) for exclusive monitoring of the labeled proteins from one specific pool (emission at 590 and 680 nm, respectively). Specific software allows for the three-dimensional representation of the individual protein spots on the gel. Dierick et al. used 2D gel electrophoresis to study stress-induced premature senescence, induced by either an organic peroxide or ethanol (Dierick et al., 2002b,c) and Benvenuti et al. identified protein candidates for replicative senescence in rat embryo fibroblast cells (Benvenuti et al., 2002a,b).

Alternative separation methods for protein and peptide fractionation are multidimensional (micro) liquid chromatography methods and/or capillary electrophoresis on-line coupled to MSn, recently reviewed by Shen and Smith (2002). In these multidimensional separations orthogonal separation modes of any choice can be coupled, provided that the eluate composition of an earlier separation mode is compatible with the following separation mode. Increasing the resolution of peptides by multidimensional separation, eventually following prefractionation (e.g., through affinity purification) increases the dynamic range of the analytical method allowing for the monitoring of even low abundance proteins. This is necessitated by the MS detector, where, for example, large amounts of high abundance peptides can suppress the ionization of lower abundant peptides. Searching for potential cancer markers, Lubman and co-workers have used liquid isoelectric focussing as a first dimension with reversed-phase chromatography as a second dimension for the protein profiling of an ovarian carcinoma-derived cell line (Kachman et al., 2002; Wang et al., 2002) while Tomlinson et al. have coupled strong cation exchange chromatography (SCX) with reversed-phase chromatography to screen a human gastric carcinoma cell line (Tomlinson et al., 2002). Gygi et al. have combined three dimensions, SCX, biotin affinity chromatography, and reversed-phase chromatography, to profile low-abundance proteins from Saccharomyces cervisiae after derivatizing protein Cys residues with a biotin-containing isotope-coded affinity tag (ICAT) (Gygi et al., 2002). The multidimensional separation of large complex peptide mixtures prior to ESI-MS/MS analysis affords fairly long analysis times. To achieve higher throughput, Lee et al. (2002) described a multiplexed microcapillary liquid chromatography, which performs the parallel separation of four different samples, and, therefore, reduces the total analysis time four-fold. Such applications are possible with new generation MALDI tandem mass spectrometers such as MALDI QqTOF (Shevchenko et al., 2000, 2001; Baldwin et al., 2001) and MALDI TOF-TOF (Medzihradszky et al., 2000). Nowadays, protein identification is facilitated by a number of databases, summarized in Table 1, for matching mass spectrometric data to existing protein sequences.

Table 1. Databases for protein identification (adapted from Aebersold and Goodlett (2001))Thumbnail image of

Surface-enhanced laser desorption/ionization (SELDI) represents a concept with great promise for sample size reduction and enrichment of specific classes of analytes (Merchant & Weinberger, 2000). Protein-chip arrays with specifically modified surfaces allow the affinity purification of analytes prior to on-chip digestion and MALDI-TOF MS analysis. Various different chemical and biochemical surfaces have been designed containing immobilized hydrophobic, ionic, or mixed-mode phases or antibodies, IMAC supports, polynucleotides, specific enzymes, or receptors. This concept has been applied to clinical problems such as the identification of biomarkers for certain types of cancer (Merchant & Weinberger, 2000; Ardekani, Liotta, & Petricoin, 2002; Li et al., 2002).

If the proteomic analysis does not require the characterization of post-translational modifications or de novo sequencing, protein microarrays with immobilized antibodies offer a suitable alternative for the rapid screening of the expression levels of known proteins (for which antibodies are available) (Schweitzer et al., 2002).

VIII. QUANTITATIVE PROTEOMICS

An important aspect of proteomics is the direct quantitative comparison of protein profiles from different sources such as, e.g., young and old or healthy and a pathologic tissue, or cells in different states of the cell cycle. Differential maps of peptides and proteins can be obtained when samples from two different sources can be analyzed in a single analytical run, avoiding run-to-run variations of the analytical method. For such experiments, the peptide and protein samples of the different sources need to be chemically derivatized for distinction by fluorescence- or MS-detection without significant alteration of the relative mobilities of the peptides and proteins during separation. Above, we have described such a strategy, DIGE, for the differential display of proteins after 2D gel electrophoresis following labeling of the proteins with fluorescent dyes (Zhou et al., 2002a). For the same purpose, Aebersold and co-workers have developed the ICAT methodology (Gygi et al., 1999; Han et al., 2001): proteins/peptides are covalently derivatized with a bifunctional reagent (structure 2) containing an electrophile on one end, usually to react with Cys residues, and biotin on the other end, for affinity purification of the labeled proteins/peptides. Both functional groups are connected with an organic linker containing either normal (H) or heavy (D) hydrogen isotopes. When two pools of proteins are labeled with the normal or heavy isotope-containing ICAT reagent, their relative abundance can be determined by quantitative MS analysis of the ratio of normal to heavy ICAT-derivatized peptides of specific proteins. During 2D gel electrophoresis, normal and heavy isotope ICAT-derivatized proteins comigrate, ensuring an accurate relative quantification of the proteins from different sources (Smolka, Zhou, & Aebersold, 2002). However, slight differences in the elution profiles of normal and heavy isotope ICAT-labeled peptides have been noted during reverse-phase chromatography (Zhang et al., 2001b). Therefore, an accurate relative quantification of ICAT-labeled peptides requires MS analysis at the respective peak maxima of the individual chromatographic peaks of normal and heavy isotope ICAT-labeled peptides. Complications have also been noted with the recovery of ICAT-labeled peptides during biotin–avidin affinity purification (Gygi et al., 2002), and with tandem MS sequencing of ICAT-labeled peptides. Therefore, recent improvements of ICAT reagents include the design of a solid-phase isotope tagging method, avoiding biotin, and incorporation of a photocleavable linker (structure 3); after solid-phase purification of ICAT-labeled peptides and photolysis, the remaining peptides essentially contain an additional normal or heavy isotope-containing Leu residue (Zhou et al., 2002b). Instead, Qiu et al. (2002) designed an acid-labile isotope-coded extractant (ALICE; structure 4).

Thumbnail image of Thumbnail image of Thumbnail image of

Using the ICAT methodology, complex mixtures of peptides from digested proteins can be reduced to one or a few peptides per protein (essentially as many peptides as the protein contains reduced or reducible Cys residues). Such a reduction in sample size significantly enhances the dynamic range of the proteomic analysis. For example, using the codon bias value as a measure for protein expression, Gygi et al. (2000) showed that traditional 2D gel electrophoresis was not able to indicate low abundant proteins with codon bias values of <0.1 whereas many of these proteins were detected with a multidimensional separation strategy including ICAT-labeling and affinity purification (Gygi et al., 2002). We note that irreversibly oxidized Cys residues (to cysteic acid) will not be amenable to ICAT labeling. Therefore, especially investigations of biological aging and oxidative stress would need to take into account that the differential display of ICAT-labeled Cys containing peptides may be compromised by Cys oxidation and not truly reflect protein expression. An alternative strategy for sample size reduction without ICAT labeling was devised by Weinberger et al., termed tagless extraction-retentate chromatography (Weinberger, Viner, & Ho, 2002). This method is based on the reversible, covalent alkylation of methionine with bromoacetyl groups attached to beads. However, this method does not allow for differential display in a single analytical run, because the reductive dissociation of methionine-containing beads regenerates native peptides without any incorporation of isotope labels, unless isotopic labels are incorporated at an earlier stage, for example, during proteolytic digestion in either H218O or H216O (Liu & Regnier, 2002).

In addition to the differential display of peptides and proteins, attempts for the absolute quantification of proteins in biological samples by MS analysis have been made (Bucknall, Fung, & Duncan, 2002; Chelius & Bondarenko, 2002; Wu et al., 2002). Such “targeted proteomics” studies usually focus on one or a few proteins, for which authentic standards are available.

IX. PROTEOMIC ANALYSIS OF PROTEIN COMPLEXES AND NETWORKS

Proteomic analysis provides a convenient entry to study the composition of protein complexes and protein–protein interactions and networks. The yeast two-hybrid system is a genetic method, where the interaction of two proteins brings into close proximity essential domains of a transcription factor, resulting in the expression of a reporter gene. This method does not require the isolation of proteins, and has been applied to several aspects of aging research (Knudsen et al., 2002). On the other hand, elegant strategies have been devised to isolated protein complexes under non-dissociating conditions, where only one target protein but not the other complexing proteins need to be genetically modified. These methods are described a little more detailed in the following. Rigaut et al. (1999) designed a tandem affinity purification (TAP) tag constructing a fusion cassette, displayed in Figure 2, encoding calmodulin-binding peptide (CBP), a TEV protease cleavage site, and protein A (ProtA) (Rigaut et al., 1999).

Figure 2.

The tandem affinity purification (TAP) methodology.

This TAP tag can be fused to any target protein of interest in cellular systems to co-immunopurify proteins complexing with the target protein. In the first immuno purification, the ProtA-tagged complex is retained by an IgG matrix and washed. Then, TEV protease is added to release the target protein complex, which still contains CBP. The second immuno purification involves capturing the CBP with calmodulin-coated beads from which the target protein complex can be released with EGTA. Recently, the TAP technology has been applied to isolate a manifold of protein complexes from Saccharomyces cervisiae (Gavin et al., 2002) and budding yeast (Shevchenko et al., 2002). A complex protein network was generated based on specific proteins shared by more than one complex. In this way, communication between protein complexes may be characterized, important for signaling, stress response, etc., and it is evident that such type of studies with cell culture models of aging will hold great promise for the analysis of age-dependent biological functional changes. A similar approach, albeit based on a one-step immuno-affinity purification utilizing the Flag epitope, was reported by Ho et al. (2002). Such proteomic studies can be integrated with genomic analysis (DNA microarrays) to correlate mRNA levels with protein expression, as performed by Ideker et al. (2002) for metabolically perturbed networks in Saccharomyces cervisiae.

X. TARGETED ANALYSIS OF MODIFIED PROTEINS FROM AGED TISSUE

The targeted analysis of specific, enriched proteins with maximal sequence coverage will ultimately permit the correlation of covalent modifications with protein function. In such studies, the quantitative analysis of proteins from selected tissues can be complemented by in vitro experiments, which permit the formation of specifically modified proteins through combination of covalent modification and site-specific mutation. Several representative examples are discussed in detail below.

A. Calmodulin

Calmodulin represents a ubiquitous cytosolic “calcium sensor,” which undergoes a profound conformational change upon complexation of 4 calcium ions enabling the protein to interact with/regulate up to 100 different target proteins (Wilson & Brunger, 2000; Chou et al., 2001). Predominantly studies with Fisher 344 rats have shown that this protein suffers an age-dependent oxidation of methionine to methionine sulfoxide resulting in a progressive age-dependent loss of activity (Gao et al., 1998) [we note, however, that negligible yields of methionine oxidation are found with currently available animals (Sharov and Schöneich, unpublished data)]. The sequence of vertebrate calmodulin contains nine methionine residues (see Fig. 3) of which especially oxidation of one of the vicinal residues Met144–Met145 is functionally important for the activation of the plasma membrane Ca-ATPase (Yin, Kuczera, & Squier, 2000; Bartlett et al., 2003).

Figure 3.

Location of methionine residues in the calcium-saturated conformation of calmodulin. Structure taken from the protein databank (pdb accession code 1CLL; Chattopadhyaya et al. (1992)).

In fact, the incubation of calmodulin isolated from aged tissue with methionine sulfoxide reductase A (MsrA) restores ca. 70% of the lost biological activity compared to calmodulin isolated from young adult tissue (Sun et al., 1999). MsrA diastereospecifically reduces protein methionine sulfoxide with S-configuration at the sulfur (Sharov et al., 1999), rationalizing the incomplete restoration of calmodulin activity by MsrA only. Importantly, the original mass spectrometric data by Gao et al. (1998) show little age-dependent calmodulin methionine oxidation at positions Met144–Met145 but up to 2.2 mol methionine sulfoxide/mol calmodulin at the other seven methionine residues. These results, together with the age-dependent loss of calmodulin activation of the plasma membrane Ca-ATPase, suggest that the combined oxidation of some of the other methionine residues could be critical for the activation of the plasma membrane Ca-ATPase. Rate constants measured in vitro for the oxidation of vertebrate calmodulin by two physiologically important oxidants of methionine, hydrogen peroxide, and peroxynitrite (Hühmer et al., 1996; Yin, Kuczera, & Squier, 2000), imply that for unbound calmodulin the lack of methionine sulfoxide at Met144–Met145in vivo cannot be caused by a kinetic preference of these oxidants for alternative methionine targets. However, binding of CaM to target proteins can affect the selectivity of oxidation. For example, the tight association of calmodulin to the inducible nitric oxide synthase largely prevented oxidation of calmodulin by peroxynitrite, and association of calmodulin with melittin resulted in a reduced oxidation sensitivity of especially the C-terminal methionine residues, Met109, Met124, and Met144–Met145 (Hühmer et al., 1997). Importantly, the observed oxidation selectivity in vivo does not only reflect the selectivity of the oxidizing species but also potential selectivities of repair and/or degradation. The proteolytic degradation of oxidized calmodulin by the 20S proteasome is enhanced through oxidation (Ferrington et al., 2001). Mass spectrometric analysis of the degradation process revealed that secondary structure but not surface hydrophobicity control proteolysis in a nonprocessive manner. However, no data are presently available with regard to the selectivity of the 20S proteasome towards specifically oxidized isoforms.

In general, the age-dependent accumulation of methionine sulfoxide on calmodulin would suggest a decline in the activity of the methionine sulfoxide reductase system. Such an age-dependent decline of both MsrA and a member of the MsrB family (hCBS-1) was experimentally observed during the replicative senescence of human WI-38 fibroblasts (Picot et al., 2004).

B. Sarco/endoplasmic Reticulum Ca-ATPase (SERCA)

SERCA plays a major role in muscle relaxation through the ATP-dependent transport of cytosolic calcium into the lumen of the sarcoplasmic reticulum (Møller, Juul, & le Maire, 1996; MacLennan, Rice, & Green, 1997). In general, calcium transport into the SR or ER constitutes an important part in the maintenance of intracellular calcium homeostasis together with calcium transport across the plasma membrane and into the mitochondria (Trump & Berezesky, 1995). The ca. 110 kDa SERCA is classified as a P-type ATPase, which forms a covalent aspartyl-phosphoryl enzyme intermediate (at Asp351) during the calcium transport cycle through γ-phosphoryl transfer from ATP (Møller, Juul, & le Maire, 1996; MacLennan, Rice, & Green, 1997). It is expressed in three major isoforms: the fast-twitch skeletal isoform (SERCA1), the cardiac/slow-twitch isoform (SERCA2a), and its alternatively spliced gene product, the smooth/non-muscle isoform (SERCA2b), and an additional non-muscle isoform (SERCA3) (Lytton et al., 1992; Wu & Lytton, 1993). In vivo, the SERCA2a isoform is regulated by phospholamban, which is present in slow-twitch and cardiac but not in fast-twitch muscle (Cantilina et al., 1993; Kimura et al., 1996).

SERCA is characterized by a relatively long half-life in the cell (ca. 14–15 days) and is turned over even more slowly in aged rats (Ferrington, Krainev, & Bigelow, 1998). Specifically, for homogenates from fast-twitch skeletal muscle fibers a significant age-related decrease of the rate of Ca2+-uptake and loading capacity of the SR has been reported (Larsson & Salviati, 1989), rationalized by a potential inactivation of the SERCA. However, in isolated SR vesicles the SERCA displayed no difference irrespective of age (Gafni & Yuh, 1989; Ferrington et al., 1997; Viner et al., 1997). An increased rate of inactivation of the “old” protein was only observed when SR vesicles were exposed to mild heating at 37°C (Gafni & Yuh, 1989; Ferrington et al., 1997) or 40°C (Viner et al., 1997). It was initially suggested that alterations of the membrane environment cause the higher sensitivity of the “old” SERCA to heat inactivation (Gafni & Yuh, 1989). Subsequent studies demonstrated indeed a slightly different phospholipid composition of “old” as compared with “young” membranes (Krainev et al., 1995). However, these differences affect neither the physical properties of bulk and protein-associated lipids nor the rotational dynamics of the protein (Krainev et al., 1995), membrane properties that directly influence SERCA activity. Narayanan et al. (1996) reported that the effects of aging on skeletal muscle function are muscle specific with a significant age-dependent change of ATP-supported calcium uptake activity for slow-twitch but not for fast-twitch muscle. Potentially because of an age-dependent ca. 50% reduction of the abundance of calmodulin-dependent protein kinase II, cardiac SERCA2a displayed ca. 40% lower phosphorylation levels in aged compared to young heart (Xu & Narayanan, 1998). A more recent chemical analysis of the fast-twitch, SERCA1, isoform of the SR Ca-ATPase isolated from skeletal muscle of old rats revealed a significantly lower content of reduced Cys (ca. 1.5 mol Cys/mol SERCA1) (Viner et al., 1997). However, comparable initial activities of “old” and “young” SERCA, when tested under optimum assay conditions, would suggest that the modified Cys residues are not critically important for protein activity. Specifically the slow-twitch, SERCA2a, isoform of the SR Ca-ATPase accumulates 3-nitrotyrosine as a result of biological aging (Viner et al., 1996, 1999). The presence of a nitrated protein in aged muscle tissue suggests the involvement of nitric oxide-derived species in age-related modifications, especially as these species are also known for their high reactivity towards Cys residues (Radi et al., 1991). Mass spectrometric analysis of SERCA2a isolated from aged slow-twitch skeletal muscle revealed that Tyr nitration occurred predominantly on Tyr294, Tyr295, and Tyr753 while negligible Tyr nitration in vivo was detected at the more accessible positions Tyr122 and Tyr130. These results were supported by parallel amino acid analysis on tryptic fragments containing the respective modified Tyr residues. In contrast, in vitro incubations of SERCA1 with peroxynitrite resulted predominantly in the nitration at position Tyr122, supported by tandem MS analysis of the nitrated peptide Glu121-Tyr(NO2)-Glu-Pro-Glu-Met-Gly-Lys (Sharov et al., 2002). This apparent discrepancy between nitration selectivity in vitro and in vivo may point to the importance for protein turnover “controlling” the apparent selectivity of post-translational modification in vivo. If covalent protein modifications result in significant changes of protein conformation and/or hydrophobicity/hydrophilicity recognizable on the protein surface, such modified proteins may be easily recognized for turnover (cf. the parameters controlling the degradation of calmodulin by the 20S proteasome, vide supra). On the other hand, if only subtle conformational changes occur, which do not dramatically modulate the secondary structure on the surface of a protein; such covalent modifications may not be recognized for accelerated protein turnover. In SERCA2a, the nitrated Tyr residues at positions Tyr294 and Tyr295 are located on helix M4 at the membrane-lumen interface of the sarcoplasmic reticulum (Toyoshima et al., 2000). Nitration in this position may affect calcium transport as both Tyr residues are located close to the calcium-binding residue Glu309 on helix M4, and to the contact area with helix M5 harboring the calcium-binding residues Asn768 and Glu771 (Fig. 4).

Figure 4.

Location of calcium-binding residues and the nitration targets Tyr294 and Tyr295 on helix M4 of sarco/endoplasmic reticulum Ca-ATPase (SERCA). Structure taken from the protein databank (pdb accession code 1EUL; Toyoshima et al. (2000)).

A slight nitration-dependent translocation of helix M4 relative to helix M5 may be sufficient for protein inactivation. However, such conformational change may not sufficiently alter secondary structures recognizable on the surface so that SERCA2a nitrated at Tyr294 and Tyr295 may not be recognized for accelerated protein turnover. In contrast, Tyr122 is located at the N-terminus of the more accessible cytosolic domain of SERCA. Hence, nitration of Tyr122in vivo may mark the protein for accelerated turnover. Model studies with several synthetic peptides have shown that Tyr nitration can result in a lowering of the phenolic pKa to values around 7 (Yee et al., 2003), i.e., Tyr nitration may cause the introduction of additional negative charge into the affected peptide regions within a protein.

C. Cu,Zn Superoxide Dismutase

Recent data on superoxide dismutase (SOD) will provide an example on how mass spectrometry was utilized to discard any age-dependent accumulation of oxidized variants of this specific protein. SOD catalyzes the dismutation of superoxide to hydrogen peroxide. Much effort has been spent to characterize potential age-dependent changes of SOD, but few systematic trends have evolved as different authors report increased, unchanged, and decreased activities of SOD in various aged tissues (Warner, 1994). It became known that the cytosolic SOD, Cu,ZnSOD (also known as SOD1) suffers oxidative modification by its product, hydrogen peroxide, a reaction potentially involving site-specifically formed hydroxyl radicals at the active site containing redox-active CuII (Uchida & Kawakishi, 1994). This process leads to the formation of 2-oxo-His (structure 5), a characteristic product identified also for the in vitro metal-catalyzed oxidation of other proteins (Lewisch & Levine, 1995; Zhao et al., 1997). Interestingly, recent tandem mass spectrometry data show that for Cu,ZnSOD not the His residue bridging the redox-active CuII and the structurally important ZnII ion is oxidized, but His residues only binding CuII represent the predominant oxidation targets (Kurahashi et al., 2001). Earlier, the heterogeneity of Cu,ZnSOD isolated from rat liver was interpreted in terms of oxidative modifications (Mavelli, Ciriola, & Rotilio, 1983), and Santa Maria et al. (1995) reported that Cu,ZnSOD isolated from liver but not lungs (Santa Maria, Ayala, & Revilla, 1996) showed an age-dependent loss of His, monitored by the diethylpyrocarbonate method. However, these findings could not be confirmed for Cu,ZnSOD isolated from the liver of Fisher 344 rats (Ghezzo-Schöneich et al., 2001). The ESI-MS analysis of Cu,ZnSOD from the liver of 3, 12, and 26 months old rats showed no accumulation of oxidized Cu,ZnSOD when compared to an authentic Cu,ZnSOD standard prepared by the oxidation of the protein with hydrogen peroxide. As an important control experiment, we injected oxidized bovine Cu,ZnSOD into the tissue homogenates, and reisolated ca. 70% of the oxidized protein, demonstrating that oxidized Cu,ZnSOD is not totally degraded under our isolation conditions. Hence, we conclude that oxidized Cu,ZnSOD does not accumulate in aged tissue. Consistent with this, we detected no age-dependent changes in the specific activity of Cu,ZnSOD.

Thumbnail image of

XI. CONCLUSION

The application of modern mass spectrometry methods to aging research can yield a wealth of information ranging from relative protein expression to the functional characterization of post-translational modifications. Important goals for the future will be the enrichment and characterization of proteins of very low abundance and/or short half-lives to obtain a complete analysis of proteomes of aging organisms.

XII. ABBREVIATIONS

ALICE

acid-labile isotope-coded extractant

BEMAD

β-elimination followed by Michael addition with dithiothreitol

CBP

calmodulin-binding protein

DIGE

differential in-gel electrophoresis

DNP

2,4-dinitrophenyl

ESI

electrospray ionization

4-HNE

4-hydroxy-2-nonenal

IBTP

(4-iodobutyl)triphenylphosphonium

ICAT

isotope-coded affinity tag

IMAC

immobilized metal affinity chromatography

MaDMaT

mass-deficient mass tag

MALDI

matrix-assisted laser desorption/ionization

MS

mass spectrometry

MsrA/B

methionine sulfoxide reductase A/B

MudPIT

multidimensional protein identification technology

NRP

native reference peptide

PDI

protein disulfide isomerase

RNS

reactive nitrogen species

ROS

reactive oxygen species

SCX

strong cation exchange chromatography

SELDI

surface-enhanced laser desorption/ionization

SERCA

sarco/endoplasmic reticulum Ca-ATPase

TAP

tandem affinity purification

TOF

time-of-flight

Acknowledgements

Support of our laboratory by the NIH (PO1AG12993) is gratefully acknowledged.

Biographical Information

Dr. Christian Schöneich received his Ph.D. in Chemistry in 1990 from the Technical University Berlin, Germany. Between 1987 and 1991 he worked in the Department of Radiation Chemistry at the Hahn-Meitner Institut in Berlin. He then did post-doctoral research in the Department of Pharmaceutical Chemistry at the University of Kansas, USA, where he is now Professor. His research focuses on oxidation reactions of proteins in vivo and in vitro, and their potential consequences for the development of stable protein pharmaceuticals, biological aging, and age-related pathologies.

Ancillary