In inflammatory diseases, release of oxidants leads to oxidative damage to biomolecules. HOCl (hypochlorous acid), released by the myeloperoxidase/H2O2/Cl− system, can cause formation of phospholipid chlorohydrins, or α-chloro-fatty aldehydes from plasmalogens. It can attack several amino acid residues in proteins, causing post-translational oxidative modifications of proteins, but the formation of 3-chlorotyrosine is one of the most stable markers of HOCl-induced damage. Soft-ionization MS has proved invaluable for detecting the occurrence of oxidative modifications to both phospholipids and proteins, and characterizing the products generated by HOCl-induced attack. For both phospholipids and proteins, the application of advanced mass spectrometric methods such as product or precursor ion scanning and neutral loss analysis can yield information both about the specific nature of the oxidative modification and the biomolecule modified. The ideal is to be able to apply these methods to complex biological or clinical samples, to determine the site-specific modifications of particular cellular components. This is important for understanding disease mechanisms and offers potential for development of novel biomarkers of inflammatory diseases. In the present paper, we review some of the progress that has been made towards this goal.
HOCl (hypochlorous acid) as a damaging oxidant
HOCl is an oxidizing and chlorinating compound that can be generated by activated phagocytes as an antimicrobial agent. The enzyme responsible is MPO (myeloperoxidase), which is particularly abundant in neutrophils. MPO is a haem enzyme that uses hydrogen peroxide to oxidize halides and pseudohalides to the respective hypohalous acids (e.g. Cl− is oxidized to HOCl) , and can also oxidize nitric oxide to nitrogen dioxide and other nitrating species . Although, under physiological conditions, HOCl is not the sole or even main product of MPO, it is nevertheless a specific product of this enzyme, as other peroxidases, such as eosinophil peroxidase, have very low affinities for chloride and generate little HOCl. As HOCl was the first product identified for MPO, there has been much interest in its formation and biological effects.
HOCl is generated under inflammatory conditions when phagocytes are activated and MPO is released and, consequently, there has been considerable interest in its potential role in inflammatory diseases [3–5]. There is substantial evidence that MPO and HOCl play a role in atherosclerosis, and HOCl-modified proteins have been observed in atherosclerotic tissue . Other indicators of HOCl-induced damage have also been found to increase in inflammatory diseases, such as free chlorotyrosine and α-chloro fatty aldehydes [7,8]. HOCl reacts readily with a variety of biological molecules, and its ability to chlorinate as well as oxidize biomolecules is furthermore of interest because it offers the potential for the discovery of biomarkers of MPO involvement in pathology, and possibly for disease progression . The reactions of HOCl with carbohydrates have received relatively little attention to date, whereas reactions with nucleic acids have been more extensively studied and have been reviewed previously . This review will focus on the products of HOCl attack on lipids and proteins, and specifically on their detection by soft-ionization MS methods.
HOCl reactions with lipids
Any living cell contains many hundreds of different types of lipid, many of which are expected to be susceptible to damage by HOCl. However, attention has focused mainly on the more abundant of these species, such as cholesterol and unsaturated phospholipids. HOCl can add across double bonds by electrophilic attack to form chlorohydrins , which are adjacent chlorohydroxy compounds (Figure 1), and the formation of these products has been well studied with unsaturated free fatty acids and phosphatidylcholines containing esterified unsaturated fatty acids. For example, oleate can form two isomers of the 9,10-monochlorohydrin, whereas linoleate can yield both mono- and bis-chlorohydrins, depending on the ratio of HOCl to lipid used. Polyunsaturated fatty acyl groups such as arachidonate can be multiply modified and a variety of positional isomers are possible, but it is also known that extensive modification of polyunsaturated phosphatidylcholines tends to result in hydrolysis of the modified fatty acyl chain, yielding lysolipids . The formation of a family of chlorohydrins of cholesterol [6-β-chloro-cholestane-(3β,5α)-diol, 5-α-chloro-cholestane-(3β,6β)-diol and 6-α-chloro-cholestane-(3β,5β)-diol] has been observed following reaction with HOCl, as well as a dichlorinated product, 5,6-dichloro-cholestane-3β-ol [12,13]. Glycerophospholipids in which one of the fatty acyl chains is attached to the glycerol at sn-1 by a vinyl ether bond (plasmalogens) are susceptible to attack by HOCl at the ether linkage; cleavage of the vinyl ether yields an α-chloro fatty aldehyde plus a lysolipid . For example, reaction of HOCl with the plasmalogen 1-O-hexadec-1′-enyl-sn-glycero-3-phosphocholine results in formation of 2-chlorohexadecanal (Figure 1). The sn-1 vinyl ether has been found to react more readily with HOCl than an unsaturated sn-2 fatty acyl chain, although, at higher HOCl/lipid ratios, the alkene is a secondary target and lysophosphatidylcholine chlorohydrins can be formed .
Examples of oxidation products of HOCl-specific attack on phospholipids or proteins
In addition to electrophilic addition, HOCl can also carry out N-halogenation reactions at amines in the headgroups of phosphatidylethanolamines and phosphatidylserines, yielding chloramines [16,17]. These initial products are unstable, and tend to decay to non-chlorinated species such as aldehydes. Phosphatidylethanolamine readily forms a dichloramine that decomposes to an N-centred radical, although the final product of the headgroup oxidation is unknown. In contrast, phosphatidylserine can be converted via a monochloramine into phosphatidylglycoaldehyde .
Detection of oxidized phospholipids by MS
Phospholipids can readily be detected by soft-ionization MS methods, such as ESI (electrospray ionization) or MALDI (matrix-assisted laser-desorption ionization). Most work has been done on phosphatidylcholines and sphingomyelins in positive-ion mode, as choline carries a constitutive positive charge, but phosphatidylethanolamines can also be monitored in positive-ion mode, and negative ionization is suitable for phosphatidylserines [18–20]. The formation of oxidation products such as chlorohydrins can be monitored by the change in mass-to-charge ratio (m/z) of the ionized molecule: addition of HO35Cl to any unsaturated phospholipid results in an increase in m/z of 52 Da, and, as the 37Cl isotope exists at a relative abundance of ∼25%, there will also be a significant signal at +54 Da [21,22]. For example, the M+ ion of 1-palmitoyl-2-linoleoyl-3-glycerophosphocholine has an m/z of 758, which on reaction with HOCl is converted into monochlorohydrins with signals at m/z 810 for the 35Cl isotope and m/z 812 for the 37Cl isotope, and bis-chlorohydrins at m/z 862, 864 and 866 for 35Cl2, 35Cl37Cl and 37Cl2 respectively. The chlorohydrins are formed by initial attack of Cl+ on the π-electrons to generate a carbenium ion, followed by nucleophilic attack of water-derived OH−, demonstrated by experiments with H218O and HO37Cl, but the second step can be replaced by the elimination of hydrogen from the carbenium ion to from a chlorinated alkene . This corresponds to the exchange of hydrogen in the native lipid for Cl, with a corresponding increase in mass of 34 Da for 35Cl. The formation of some of these species is shown in Figure 2; their structures have been elucidated by the application of MS/MS (tandem MS), facilitated by the use of precursor-ion-scanning and neutral-loss experiments. Chlorinated lipids can be identified by the loss of 36 Da corresponding to the loss of HCl, as shown in Figure 2, and chlorohydrins of phosphatidylcholine have also been reported to yield a product ion at m/z −95 resulting from loss of HCl together with fragmentation of the choline headgroup .
ESI mass spectra of HOCl-modified PAPC (1-palmitoyl-2-arachidonoyl-3-glycerophosphocholine)
Analysis of chlorohydrin or chlorinated lipid formation in biological samples is greatly facilitated by the use of LC (liquid chromatography)–MS, as the oxidized and chlorinated products formed often have similar masses to species of native lipids that may be present in the sample. Phospholipid classes can be separated using normal-phase columns, whereas reverse-phase chromatography with either C8 or C18 columns has been used to separate phospholipids based on chain length and degree of unsaturation; oxidized phospholipids are eluted earlier than the corresponding native lipid in this system . Although LC–MS and LC–MS/MS have been used to analyse oxidized phospholipids in biological or clinical samples (reviewed in ), there has been less success with chlorinated phospholipids, although a few studies have demonstrated the formation of chlorinated lipids in biological samples subjected to oxidative stress in vitro. For example, monochlorohydrins of palmitoyl-linoleoyl-, palmitoyl-oleoyl- and stearoyl-arachidonoyl-phosphatidylcholines were observed in human LDL (low-density lipoprotein) incubated with HOCl or MPO/H2O2/Cl− , and chlorohydrins formed from phosphatidylcholines containing oleate or linoleate could also be detected in cultured mammalian cells that had been treated with HOCl . In contrast, α-chloro fatty aldehydes have been detected in human atherosclerotic plaque samples and in rat heart tissue following myocardial infarction, but using GC–MS, as opposed to LC–MS [8,24].
HOCl reactions with proteins
It has been suggested that one reason chlorinated phospholipids have not been detected in tissue samples from inflammatory models or clinical samples is that HOCl reacts much more readily with other biomolecules than with unsaturated lipids. Specifically, it was found that HOCl reacts very rapidly with the sulfur-containing side chains of methionine and cysteine residues and, at slightly lower rate constants, with amines and other nitrogen-containing residues. Even the reaction of tyrosine with HOCl is substantially faster than that of 5-pentenoic acid, which was used as a model unsaturated fatty acid . Thus any HOCl formed during inflammatory situations that is not scavenged by antioxidants is more likely to attack proteins than unsaturated phospholipids and, consequently, protein oxidation products may offer more potential as biomarkers of MPO- and HOCl-induced damage.
There has been interest in the reactions of HOCl with proteins and amino acids dating back many years [10,25,26]. As mentioned above, at physiological pH, HOCl reacts preferentially with cysteine, cystine and methionine residues, but amines such as lysine and histidine residues are also good targets. The initial reaction products are chlorinated (e.g. chloramines, sulfenyl chloride) (Figure 1), but they are relatively unstable and tend to decompose with loss of the chlorine to generate more stable oxidation products, such as aldehydes or higher oxidation states of sulfur (sulfones, sulfenic acid or cysteic acid). The tryptophan oxidation product 2-hydroxyindole is also a stable product of HOCl attack on proteins. Many of these products can also be formed by free-radical-mediated oxidation of proteins, and are thus not specific for HOCl attack. In contrast, reaction of HOCl with the aromatic residue tyrosine generates 3-chlorotyrosine (Figure 1), which is more stable and specific for HOCl; thus, although the initial reaction with tyrosine is comparatively slow, it offers a better choice as a biomarker of HOCl oxidation. Free chlorotyrosine has already been used as a biomarker and detected in various inflammatory diseases [9,27], but this does not give information on the protein that was oxidized, which is likely to be valuable for understanding the overall process of disease pathology.
Detection of protein oxidation by MS
As with lipid oxidation, protein oxidation can be monitored readily by ESI–MS, owing to the intrinsic ionizability of proteins or peptides, and alterations in the mass as the side chains are oxidatively modified. Various approaches have been utilized in this area, although relatively few of these studies have focused specifically on HOCl-induced damage. For example, many MS studies aimed at understanding the mechanisms and products of protein oxidation have been carried out using purified proteins in order to simplify the interpretation of the spectra. Alternatively, some work aimed at identifying the occurrence of particular amino acid modifications as markers in complex biological or clinical samples has adopted the approach of total protein digestions, followed by multiple reaction monitoring to detect the presence of oxidized residues. Other studies have aimed to identify specific proteins that have undergone oxidation in vivo, using a proteomics approach in which certain oxidations can be labelled and identified in two-dimensional gel electrophoresis. All of these approaches have provided valuable information, but they also have limitations in terms of their application.
Many studies have been carried out using model peptides or individual proteins subjected to oxidation by various oxidants, and different amino acid oxidation products have been catalogued. Although a comprehensive account is outside the scope of the present review, the type of information potentially available can be seen from the following examples. Methionine sulfoxide is a well-established marker of protein oxidation, with a mass increase of 16 Da compared with methionine that can be detected in intact proteins or peptides, and is characterized in MS/MS by a characteristic loss of 64 Da, corresponding to methanesulfenic acid (CH3SOH) [28,29]. The formation of dehydroalanine as a PTM (post-translational modification) of cysteine has been demonstrated in human serum albumin , although oxidation to sulfinic and sulfonic acids is more common . Cysteine residues can also be oxidized by NO to yield nitrosothiols (RSNO), with a mass increase of 29 Da, and their formation has been investigated by MALDI– or ESI–MS/MS in signalling proteins such as protein tyrosine phosphatases  or p21ras . Products of tryptophan oxidation following hydroxyl radical attack have been characterized by MS/MS  and have been observed in tryptic peptides of α-crystallin , whereas site-specific identification of nitrotyrosine in BSA has been achieved by LC–MS/MS . There has also been considerable interest in the development of MS methods for analysis of carbonyl adducts of proteins, e.g. by the lipid peroxidation product HNE (4-hydroxynonenal), and the modifications of histidine, lysine and arginine residues have been investigated in model proteins or human LDL [37,38]. Many of these studies have been carried out by peptide sequencing of tryptic digests using LC–MS/MS or MALDI–MS/MS to determine the site of modification; however, top-down approaches have also been applied, e.g. involving whole protein collision-induced dissociation , or Fourier-transform ion cyclotron resonance MS .
HOCl induces many modifications that are the same as those induced by radical attack, some of which have been investigated in specific proteins by soft-ionization MS. For example, HOCl oxidizes cysteine residues to sulfinic and sulfonic acids, or methionine to the sulfoxide or sulfone. These products have been monitored in GroEL treated with HOCl by LC–MS/MS ; the protein was subjected to trypsin digestion, and location of the susceptible residues was mapped by collisionally induced decomposition of the peptides. Other studies have focused on HOCl-specific oxidative damage. MS/MS has been used to characterize the formation of cysteine-derived sulfenamides, sulfinamides (Figure 1) and sulfonamides in synthetic peptides following oxidation with HOCl; these were formed by intramolecular reactions between sulfenyl, sulfinyl or sulfonyl chlorides and amine groups, but it has been suggested that these mechanisms could contribute to cross-linking of LDL . Similar products have been detected in the intact murine neutrophil protein S100A8 and were characterized by mapping of endoproteinase AspN digestion peptides . The best characterized marker of protein oxidation by HOCl is 3-chlorotyrosine; this has also been detected by LC–MS/MS in HOCl-oxidized GroEL, although it was a less abundant product than cysteic acid . LC–ESI–MS has been used to map the tyrosine residues susceptible to MPO/H2O2/Cl− and MPO/H2O2/NO3− attack in the Apo-AI (apolipoprotein AI) protein of human HDL (high-density lipoprotein), by trypsin digestion of purified protein and MS/MS of the resulting peptides . It was found that chlorination occurred preferentially at Tyr192, although some chlorotyrosine formation was also observed from four other tyrosine residues. The formation of chlorotyrosine is dependent on the proximity of lysine residues and on the tyrosine residue being in a hydrophilic location [43,44].
In terms of assessing the global occurrence of protein oxidation in a biological sample, the most straightforward approach is to subject the samples to total protein digestion (preferably enzymatic to avoid artefactual oxidation), and then use ESI–MS or LC–MS/MS to detect the presence of known oxidation products of various amino acid residues. Although this technique has largely been applied to the study of advanced glycation end-products, the retention times, parent ions, fragment ions and neutral losses for the oxidative stress markers methionine sulfoxide, dityrosine and 3-nitrotyrosine have also been characterized, and multiple reaction monitoring has been used to analyse these in rat models and human clinical samples [45,46]. A similar approach has been applied to the analysis of HNE-adducts of histidine . Using labelled internal standards, these methods allow the quantification of protein oxidation products, but have the disadvantage that they do not identify the proteins that were modified.
In order to identify which proteins have been oxidized in complex samples, a proteomics approach is usually adopted, involving two-dimensional electrophoresis to separate the proteins together with the use of antibodies or tags specific for certain modifications, in order to label the proteins in the gel that contain the modification. Such methodologies are routinely used for detection of carbonyl groups on proteins using dinitrophenylhydrazine , free compared with modified thiols, such as S-nitrosylation  or nitrotyrosine . The gel spot can then be excised, and the proteins are digested and identified by peptide mass fingerprinting with MALDI–MS. However, these methods have some limitations. Peptide mass fingerprinting cannot pinpoint the site of oxidative modification on a protein identified as having been oxidized, and so post-source decay or further analysis by ESI–MS/MS is needed to sequence the peptides and identify the amino acid oxidized.
Methods for identifying specific oxidative modifications to proteins in complex mixtures
Although much has been learned about protein oxidation from the type of studies described above, our understanding of oxidative stress in physiological and disease processes would benefit substantially from the ability to map specific residue modifications in known proteins from complex mixtures, without the need for prior separation by two-dimensional electrophoresis. Advances in MS have facilitated analogous studies of the role of phosphorylation of specific proteins in the function of signalling pathways, and some current studies are developing such approaches for determination of oxidative PTMs. This requires the interfacing of LC to separate peptides, generated by the digestion of complex samples, to MS/MS or quadrupole–quadrupole–linear ion trap instruments capable of precursor-ion-scanning or neutral-loss experiments. For example, Roe et al.  have mapped sites of HNE attack at histidine residues in yeast lysate and yeast cells treated with HNE. The detection was facilitated by solid-phase capture to enrich HNE-modified peptides, but, in addition to collisionally induced decomposition for sequencing of tryptic peptides, neutral loss of 156.1 Da was used to identify HNE-containing peptides. The application of MS3 precursor-ion scanning for specific detection of 3-chlorotyrosine, 3-nitrotyrosine, hydroxytyrosine and hydroxytryptophan is currently being developed (L. Mouls, E. Silajdzic, N. Haroune, C.M. Spickett and A.R. Pitt, unpublished work). Although these modifications yield characteristic daughter ions in MS/MS precursor-ion scanning, such as m/z 170.1 for chlorotyrosine and m/z 181.1 for nitrotyrosine, the MS/MS approach is limited by a relatively high occurrence of false positives, as some peptides that do not contain these specific oxidative PTMs can generate fragments nearly isobaric in mass. However, this method can only be carried out using a quadrupole–quadrupole–linear ion trap instrument, and development of the technology is also currently hindered by software limitations.
The application of modern MS techniques has advanced our understanding of the oxidation products of both lipids and proteins, and several approaches are currently in use to analyse the occurrence of these products in biological systems and pathology. This is important, because, apart from the structural damage to cells or tissues resulting from biomolecule oxidation, oxidized lipids and proteins are known to have bioactive effects and roles in cell signalling and modulation of gene expression. There is a need to develop more and better biomarkers of specific types of oxidative stress, and this will necessitate further enhancement of MS approaches to allow the analysis of oxidized products in complex biological samples.
Bioanalysis in Oxidative Stress: A Biochemical Society Focused Meeting held at the University of Exeter, U.K., 2–3 April 2008. Organized and Edited by John Moody (Plymouth, U.K.) and Paul Winyard (Peninsula Medical School, Exeter, U.K.).
C.M.S. and A.R.P. gratefully acknowledge funding from the BBSRC (Biotechnology and Biological Sciences Research Council) (grants BBE0154841 and BBE015727) and A.R.P. acknowledges funding from BBSRC/EPSRC (Engineering and Physical Sciences Research Council) for the RASOR (Radical Solutions for Researching the Proteome) Consortium (BBC5115721).