Methionine sulfoxides in serum proteins as potential clinical biomarkers of oxidative stress

Oxidative stress contributes to the pathophysiology of a variety of diseases, and circulating biomarkers of its severity remains a topic of great interest for researchers. Our peptidomic strategy enables accurate and reproducible analysis of circulating proteins/peptides with or without post-translational modifications. Conventional wisdom holds that hydrophobic methionines exposed to an aqueous environment or experimental handling procedures are vulnerable to oxidation. However, we show that the mass spectra intensity ratio of oxidized to non-oxidized methionine residues in serum tryptic proteins can be accurately quantified using a single drop of human serum and give stable and reproducible results. Our data demonstrate that two methionine residues in serum albumin (Met-111 and Met-147) are highly oxidized to methionine sulfoxide in patients with diabetes and renal failure and in healthy smokers versus non-smoker controls. This label-free mass spectrometry approach to quantify redox changes in methionine residues should facilitate the identification of additional circulating biomarkers suitable for predicting the development or progression of human diseases.

Oxidative stress is a common component of the pathophysiology of a variety of human conditions and diseases, including aging 1,2 , cancer 3 , diabetes 4 and neurodegenerative 5 and cardiovascular diseases 6 . Oxidative modifications of biomolecules, such as lipids, proteins and DNA, are postulated to contribute to the development and/or progression of the listed conditions. Determination of in vivo oxidative stress status requires quantification of either reactive oxygen/nitrogen species or damaged biomolecules. Because the former have very short half-lives and show high chemical reactivity, free radical processes have been monitored by detecting the latter. The identification of clinical biomarkers that accurately evaluate the severity of oxidative stress remains an important unmet challenge 7 .
Methionine (Met) oxidation is a mechanism by which proteins perceive oxidative stress and function in redox signaling 8 . Met residues are highly susceptible to modification by mild oxidants 8 and can be oxidized spontaneously during common experimental procedures 9 . In vitro Met oxidation is a reversible process 10,11 and is dependent upon solvent accessibility 12,13 and structural determinants 14,15 . Met oxidation can also modify the physicochemical properties of the whole protein and therefore modulate its function 14,16 . However, the biological implications of the presence of oxidized Met in specific disease-related proteins have only been studied in limited numbers of human pathologies, including diabetes 17 , skin disease 10,11 , Alzheimer's disease 18 and Parkinson's disease 19 . Mass spectrometry has emerged as a powerful tool to identify protein modifications in biological samples, but no adequate proteomic methodologies exist to characterize the exact sites and extent of oxidation in the peripheral circulation.
We have established a series of new technologies to enrich and analyze circulating low-molecular-weight peptides suitable for mass spectrometry analysis and have comprehensively identified native and tryptic peptides in the human peripheral circulation [20][21][22][23] . Among the large number of native peptides we recently identified, ~3% contained post-translationally oxidized Met residues. These results prompted us to develop a new label-free mass spectrometry approach that would enable identification of tryptic peptides containing Met residues that had undergone oxidative modification in serum. Such analysis of post-translational modification has been thought to be extremely challenging, but we demonstrate that this method is applicable to the assessment of whole-body oxidative stress status using a single drop of human serum.

Results
Selection of serum tryptic peptides containing an oxidized methionine. Because Met residues are reportedly vulnerable to oxidation in many experimental settings, we first examined whether unoxidized Met residues in tryptic peptides spiked into human serum can be oxidized into methionine sulfoxide (Met(O)) during the processes of reductive alkylation, trypsin digestion and subsequent analysis by liquid chromatography-mass spectrometry (LC-MS). We quantified signal intensities of peptides with Met(O) ([Met(O)]) and those with corresponding unoxidized Met ([Met]) using extracted ion chromatograms of LC-MS analysis and found that there were no significant differences in [Met(O)]/[Met] ratios when quantified before and after processing of human serum samples containing tryptic E. coli β -galactosidase peptides (Fig. 1).

Figure 1.
Comparison of methionine oxidation profiles before and after spiking peptide into human serum, reductive alkylation and trypsinization. Trypsin-digested E. coli β -galactosidase (2.5 μ g/μ l serum) was added to human serum as a monitor peptide and subsequently processed by reductive alkylation and trypsinization. Representative extracted ion chromatograms corresponding to the peptides LAVMLR (a,b) and MSGIFR (c,d) of the tryptic E. coli β -galactosidase peptide before (a,c) and after (b,d) spiking, reductive alkylation and trypsinization, are presented. Smaller peaks with an earlier retention time, corresponding to LAVMLR or MSGIFR peptides containing methionine sulfoxide (designated as (O) above M) were magnified 10-fold and are shown above the original peaks. We next determined whether [Met(O)]/[Met] for serum tryptic peptides is stable and can be quantified reproducibly using the study workflow summarized in Fig. 2. Initially, serum samples obtained from healthy controls were digested overnight with trypsin and subjected to highly sensitive analysis using nano-flow LC-MS. This resulted in identification of 53 peptides containing Met(O). Using LC-MS incorporating conventional HPLC, we selected potential biomarker candidates for further analysis based on the following criteria: that they were tryptic peptides containing Met and/or the corresponding Met(O) and possessing sufficient mass spectrum intensity, that the mass spectra of both peptides were clearly separable from other peptides, that mis-cleavage products were not detected and that determination of [Met(O)]/[Met] after repeated measurements of 12 aliquots showed inter-assay coefficients of variation (CVs) of less than 30%. Five Met-containing tryptic peptides fulfilled these criteria at this initial assessment (Table 1). Of these, more than 99.9% of serum complement C3 (Met-1118) was Figure 2. Schematic overview of the label-free mass spectrometry approach for quantifying oxidized and unoxidized methionine residues. After serum separation, proteins were digested to peptides without the use of depletion columns, and subjected to liquid chromatography-mass spectrometry (LC-MS) analysis to identify abundant tryptic peptides containing methionine residues. Tryptic peptides with unoxidized methionines and those containing methionine sulfoxide showed distinct retention times, as shown in the lowest panel. "M" within a blue curve denotes a methionine residue in a serum tryptic peptide and a red circle above the "M" represents the sulfoxide bond on a methionine residue.
oxidized to Met(O) in all samples examined and therefore this peptide was excluded from further analysis. The remaining four tryptic peptides, albumin (Met-111), albumin (Met-147), immunoglobulin (Ig)γ 1 chain C region (Met-135) and α 1-antitrypsin (Met-409), were subjected to subsequent detailed analyses for potential application as clinical biomarkers.
Optimization of a method for quantifying methionine oxidation. To accurately quantify the frequency of Met(O) residues relative to their corresponding unoxidized residues, complete trypsin digestion is mandatory prior to LC-MS analysis. Trypsin treatment reduced the number of undigested products as a function of time, at least during the initial period (data not shown), but from 18 hours onwards, mis-cleavage products were undetectable and did not significantly affect [Met(O)]/[Met] in any of the four candidate serum tryptic peptides (Fig. 3a). Therefore, all serum samples were trypsin-digested for 24 hours in the experiments performed thereafter.
We next tested whether the rate of spontaneous oxidation of Met residues in serum proteins would be affected by the length of clotting time necessary to obtain serum, given that withdrawn blood samples are left for at least 30-60 min at room temperature for a clot to form and residual cellular elements and other contaminants to be removed. To investigate this, blood samples were either immediately centrifuged after blood withdrawal, or allowed to clot at room temperature for 30 min to 6 hours before centrifugation, and the sera obtained were processed for reductive alkylation, trypsin digestion and subsequent LC-MS analysis. We found that [ of the four tryptic peptides were very similar, irrespective of the time the blood samples were left at room temperature (Fig. 3b).
We Because high glycemic variability induces more oxidative stress than continuous high glucose 24

Discussion
Met is readily oxidized to Met(O), and this reaction is believed to be reversibly catalyzed by the methionine sulfoxide reductases 26 present in nearly all organisms 27,28 . Met(O) can also be further oxidized to methionine sulfone, although this reaction occurs to a much lesser extent [29][30][31]   conditions. Thus, some Met residues in serum proteins may undergo oxidation far more easily than other residues in response to underlying diseases or conditions. Experience in many areas of redox research highlights how challenging it can be to identify clinical biomarkers reflecting endogenous oxidative stress status. The experimental and analytical performance of [Met(O)]/[Met] measurement was reproducible and robust, irrespective of the time of blood withdrawal, the time the sample was left at room temperature before centrifugation or repeated freeze/thaw cycles. Such results may be attributable to a previously reported feature of the ubiquitously expressed enzymes regulating Met(O) levels. Met(O) reductase, for example, may not function in the presence of circulating proteins 19,32 , and hence the oxidation status of serum proteins with intravascularly oxidized Met residues remains unchanged until the proteins are broken down or cleared, predominantly by the liver in the case of serum albumin.
Our LC-MS approach has overcome well-recognized technical difficulties associated with quantifying redox changes in circulating proteins. Incomplete digestion of serum proteins affected [Met(O)]/[Met] due to differential digestion efficiency between unoxidized Met and Met(O), but we have achieved complete trypsin digest by using phase-transfer surfactant plus Lys-C 33 . Stable and reproducible LC-MS analyses were achieved over a long run time    in diabetic patients with greater glycemic excursion than those with minimal variability. These results indicate that these Met residues are susceptible to oxidation by factors that are associated with the pathophysiology of such diseases. Accumulating evidence indicates elevated oxidative stress in diabetes [34][35][36] , and further increased oxidative stress and risk for cardiovascular disease when nephropathy is also present [37][38][39] . Acute fluctuations of glucose levels in diabetes is demonstrated to accelerate oxidative stress 24 . Cigarette smoke contains high concentrations of free radicals that can induce oxidative stress 40,41 and smokers have a higher prevalence of variety of diseases, including respiratory diseases, cardiovascular diseases and cancers 42  in any of the four peptides tested, longer term oxidative stress may be required to oxidize Met residues of proteins in the human peripheral circulation.
In conclusion, we have developed a novel label-free mass spectrometry approach to quantify Met oxidation in proteins present in the human peripheral circulation. Oxidized Met levels were accurately and reproducibly determined in a single drop of human serum. [Met(O)]/[Met] levels of two residues in serum albumin were elevated in type 2 diabetes, diabetes-associated chronic renal insufficiency and chronic smoking. This strategy could also be applied to identify other Met-containing serum proteins whose status reflects the severity of pathophysiological conditions or the likelihood of a patient developing a major disease in the future.

Methods
Subjects. The study population consisted of 18 healthy non-smokers (9 men and 9 women), 23 patients with type 2 diabetes (14 men and 9 women), 12 patients with non-dialyzed chronic renal failure associated with type 2 diabetes (9 men and 3 women), and 9 healthy smokers (9 men) ( Table 2). Type 2 diabetes was diagnosed in individuals who were confirmed to be insulin independent according to the criteria of the Japan Diabetes Society (patients with anti-glutamic acid decarboxylase autoantibody > 1.5 U/ml or serum C-peptide < 0.5 ng/ml were excluded) 43 . Chronic renal failure was diagnosed after confirming sustained elevation of serum creatinine levels for at least 6 months 44,45 . Clinical records were reviewed for all subjects and those with histories of acute inflammatory diseases, malignancies, recent attacks of cerebrovascular or cardiovascular accidents were excluded from the analysis. The protocol was approved by the Kitasato University Medical School Ethics Committee (B15-181) and informed written consent was obtained from all participants. All study methods were performed in accordance with the relevant guidelines and regulations of Kitasato University Medical School.

Serum sample collection. Venous blood samples were collected from healthy volunteers and patients
with type 2 diabetes with and without impaired renal function. Type 2 diabetic patients provided an additional serum sample at the time of scheduled blood sampling on the last morning of 72-h CGM recording using the CGMS system GOLD (Medtronic Minimed Inc. Northridge, CA) 46,47 . They also received our routine systemic evaluation protocols for diabetic patients covered by universal health coverage system in Japan 43,48 , which include laboratory tests, such as urinalysis, a complete blood count, serum biochemical analysis for more than 15 items, anti-glutamic acid decarboxylase antibody, glycated albumin, HbA1c, and fasting serum insulin. Blood was withdrawn from an antecubital vein into vacutainers containing pro-coagulant and allowed to clot at room temperature for the indicated time, before centrifugation at 2,000 × g for 15 min at room temperature. Aliquots were stored at − 30 °C until use.
Trypsin digestion of serum samples. Trypsin digestion of serum proteins was performed essentially as described 33 , with the following modification. Twenty μ L of 200 mM triethylammonium bicarbonate/12 mM sodium deoxycholate/12 mM sodium lauryl sulfate and 2 μ L 200 mM tris (2-carboxylethyl) phosphine hydrochloride/120 mM triethylammonium bicarbonate were added sequentially to thawed serum (2 μ l) and this was incubated at 50 °C for 30 min. Two μ l of 375 mM iodoacetamide was added, the mixture was incubated in a dark room for 30 min, then further incubated at 37 °C for the indicated times after addition of 2 μ l of 100 ng/μ l Lys-C and 2 μ l of 100 ng/μ l trypsin. 50 μ l of acetonitrile (ACN) and 50 μ l of 5% trifluoroacetic acid were then added to the digest, this was centrifuged at 19,000 g for 15 min, and the supernatant was subjected to LC-MS analysis.

Identification of methionine-containing peptides by LC-MS analysis.
Tryptic digests of serum samples were injected onto a C 18 0.075 × 20 mm trap column (Acclaim PepMap 100; GL Sciences, Tokyo, Japan) and then eluted onto a C 18 0.075 × 120 mm analytical column (Nano HPLC Capillary Column; Nikkyo Technos, Tokyo, Japan), configured to an EASY-nLC 1000 HPLC system (Thermo Fisher Scientific). The flow rate of the mobile phase was 300 nL/min; mobile phase A consisted of 0.1% formic acid (FA) and mobile phase B consisted of 0.1% FA/90% ACN. The mobile phase gradient was programmed as follows: 5-25% B (0-48 min), 25-50% B (48-58 min), 50-95% B (58-60 min), and 95% B (60-70 min). Separated peptides were introduced from the HPLC to a Q-Exactive (Thermo Fisher Scientific) operating in data-dependent mode, to automatically switch between full-scan MS and MS/MS acquisition. Full-scan MS spectra (m/z 350-1,000) were acquired in an Orbitrap instrument with a mass resolution of 70,000 at m/z 200, after accumulation of ions to a 1 × 10 5 target value. Fractions representing the 12 most intense full-scan peaks with charge state 2-4 were selected using an isolation window of 2.4 Da and fragmented in the high-energy collisional dissociation cell at a normalized collision energy of 27%. MS/MS spectra were acquired by the Orbitrap mass analyzer with a mass resolution of 17,500 at m/z 200, after accumulation of ions to a 1 × 10 5 target value. The ion selection threshold was 2 × 10 4 counts, and the maximum allowable ion accumulation times were 120 ms for full MS scans and 200 ms for MS/MS spectra. Typical mass spectrometric conditions were as follows: spray voltage 2 kV, no sheath or auxiliary gas flow, capillary temperature heated to 250 °C and dynamic exclusion time 15 s. Database searches were performed using the SEQUEST algorithm incorporated into Proteome Discoverer 1.4.0.288 software (Thermo Fisher Scientific). The search parameters were as follows: enzyme, trypsin; variable modification, oxidation of Met residue; variable modification, carbamidomethylation of Cys residue; peptide ion mass tolerance, 6 ppm; fragment ion mass tolerance, 0.02 Da; peptide charges, + 2 to + 8. The identified peptides were searched for in the decoy database and the false discovery rate was set as 0.01 using Percolator scoring with posterior error probability validation. Peptide quantitation was also performed using Proteome Discoverer