Identification of yeast oxidized proteins

Journal of Chromatography A, 1141 (2007) 22–31

Identification of yeast oxidized proteins Chromatographic top-down approach for identification of carbonylated, fragmented and cross-linked proteins in yeast Hamid Mirzaei, Fred Regnier ∗ Department of Chemistry, Purdue University. West Lafayette, IN 47907, USA Received 5 September 2006; received in revised form 1 November 2006; accepted 2 November 2006 Available online 22 December 2006

Abstract The effects of oxidative stress on the yeast proteome were studied using hydrogen peroxide as the stress agent. Oxidized proteins were isolated by (1) biotinylation of oxidized proteins with biotin hydrazide, (2) affinity selection using monomeric avidin affinity chromatography, and (3) further fractionated by reversed-phase liquid chromatography (RPLC) on a C8 column. Oxidized protein fractions from RPLC were then trypsin digested and the peptide cleavage fragments identified by tandem mass spectrometry (MS/MS). Slightly over 400 proteins were identified. Sites of carbonyl formation were found in roughly one fourth of these proteins. Oxidation on other amino acids in carbonylated peptides was seen in 32 cases while carbonylation was absent in 96 of the oxidized proteins observed. Although there are large numbers of potential oxidation sites, oxidation seemed to be restricted to a small area in most of the proteins identified. Sometimes multiple amino acids in the same tryptic peptide were oxidized. A second trend was that more than 8% of the proteins identified appeared in more than one of the RPLC fractions. Based on the position of the peptides identified in the primary structure of protein candidates derived from databases it was concluded that this occurred by fragmentation of a parent protein. It is not clear from the data whether the fragmentation process was of enzymatic or oxidative origin. Finally, peptides from two or more proteins occurred together in more than one reversed phase fraction with 2% of the proteins identified. This data was interpreted to mean that this was the result of protein cross-linking. © 2006 Published by Elsevier B.V. Keywords: Proteomics; Oxidative stress; Hydrogen peroxide; Yeast; Biotin hydrazide; Affinity chromatography; Protein fragmentation; Protein cross-linking; Mass spectrometry; Specific sites of oxidation; Tandem mass spectrometry; Carbonylation

1. Introduction Energy production in aerobically grown cells is achieved by oxidation of nutrients with molecular oxygen (O2 ) and is accompanied by the production of hydrogen peroxide (H2 O2 ). H2 O2 is rapidly degraded by cells in a variety of tightly regulated processes. When these regulatory processes are overwhelmed or fail because of insufficient supplies of critical enzymes, genetic defects, environmental factors, or disease, the resulting H2 O2 can produce reactive oxygen species (ROS) via the Fenton reaction that are damaging to cells [1,2]. Production of uncontrolled amounts of ROS leads to a phenomenon known as oxidative stress [3]. One of the more negative

∗

Corresponding author. E-mail address: [email protected] (F. Regnier).

0021-9673/$ – see front matter © 2006 Published by Elsevier B.V. doi:10.1016/j.chroma.2006.11.009

outcomes of oxidative stress is that the structure of proteins, DNA, and RNA can be permanently altered [4]. Some of these damages can be repaired [5], whereas other forms of damage are irreversible. Among the irreparable forms of protein modification, oxidation of amino acids side chains to the carbonyl level and cleavages in the protein backbone are the most serious [6]. Aldehyde and ketone groups generated in proteins during this oxidative process either lead to protein degradation by proteosomes in the best case or cross-linking with other proteins in the worst case. Accumulation of aggregates from protein cross-linking is a serious problem because protein aggregates can be more difficult to degrade and cause long-term damage to cells [7,8]. In either case, the formation of carbonyl groups is a hallmark of severe oxidative stress. Protein carbonyl groups resulting from oxidation have been quantified in a number of ways. Derivatization with

H. Mirzaei, F. Regnier / J. Chromatogr. A 1141 (2007) 22–31

2,4-dinitrophenyhydrazine [9], fluorescinamine (FINH2 ) [10], digoxigenin-hydrazide [11] or reductive tritiation [12] have all been used to detect and quantify protein carbonyl formation in a wide range of biological studies. Derivatization with biotin hydrazide is another way to tag carbonyl groups in oxidized proteins. Subsequent selection of biotinylated proteins with an avidin affinity chromatography column provides a simple means to isolated oxidatively damaged proteins. Oxidized proteins from mouse brain have been selected in this way and identified by LC/MS/MS [13]. Biotinylated proteins are also easily recognized in 2-D gel electrophoresis by using avidin–fluorescein isothiocyanate (avidin-FITC) as a staining agent [14]. Twenty oxidized proteins from yeast (Saccharomyces cerevisiae) have been identified in this manner, but detection limits of in-gel methods suffer from loading capacity limitations [15]. The objective of the study described here was to develop a method suitable for large-scale oxidative stress studies in which identification of carbonylated proteins and their oxidation sites was the primary goal. Yeast was selected based on its rate of growth and ability to recover from oxidative stress [16]. Aldehyde and ketone containing proteins were biotinylated with 5 mM biotin hydrazide immediately after cell lysis. Oxidized proteins were then avidin affinity selected and further fractionated with RPLC. Isolated protein fractions from RPLC were tryptic digested and the tryptic peptides sequenced by MS/MS. MS/MS spectra of the tryptic peptides were subjected to a database search using the Mascot search engine.

2. Materials Biotin hydrazide, ultralinked immobilized monomeric avidin, d-biotin, sodium cyanoborohydride and trifluoroacetic acid (TFA), Slide-A-Lyzer dialysis cassettes and Coomassie blue (Bradford) protein assay kits were purchased from Pierce (Rockford, IL, USA). Iodoacetamide, dithiothreitol (DTT), trypsin, glycerol, 2-mercaptoethanol, IGEPAL CA630 non-ioninc detergent and N-␣-tosyl-l-lysine chloromethyl ketone (TLCK) were obtained from Sigma (St. Louis, MO, USA). Sodium phosphate, urea, sodium chloride and calcium chloride were purchased from Mallinckrodt (St. Louis, MO, USA). Protease inhibitor cocktail was purchased from Roche Diagnostics (Indianapolis, IN, USA). ZipTip pipette tips were purchased from Millipore (Bedford, MA, USA). Vydac 208TP54 reversed-phase C8 column were purchased from W. R. Grace & Co. (Columbia, MD, USA). Coated nanospray tips were purchased form New Objective (Woburn, MA. USA). The affinity selection and RPLC analyses were done on an Integral Micro-Analytical Workstation (PE Biosystems, Framingham, MA, USA) and Dionex LC packing capillary LC instrument (Dionex, Sunnyvale, CA, USA). Mass spectral analyses were done using a PE Sciex QSTAR hybrid LC/MS/MS, a triple quadruple time of flight (QqQ-TOF) mass spectrometer. All spectra were obtained in the positive ion mode.

23

3. Methods 3.1. Yeast strain and culture conditions The method of Yoo and Regnier [14] was followed in growing S. cerevisiae wild-type strain SM1058 [17]. The cells were grown at 37 ◦ C in yeast extract–peptone–dextrose (YPD) medium (1% yeast extract, 2% peptone, 2% glucose) using a shaking incubator at 200 rpm. Cell growth was determined by absorbance (A) measurements at 600 nm. For processing a midlog phase culture, the overnight cultured cells were inoculated in fresh medium to a cell density of 0.2–0.3 A600 . 3.2. Preparation of total protein from yeast treated with hydrogen peroxide Exponentially growing cells at a density of 2.4 A600 were treated with 5 mM hydrogen peroxide to induce oxidative stress. Cells from 50 or 500 mL cultures were harvested and then washed twice with cold water by centrifugation at 3000 rpm for 10 min at 4 ◦ C. The pellet was resuspended in lysis buffer (pH 7.4) containing 3.8 mM NaH2 PO4 ·6H2 O, 49.4 mM Na2 HPO4 ·6H2 O, 48.4 mM NaCl, 5 mM KCl, 20% glycerol, 1% 2-mercaptoethanol, 0.3% IGEPAL CA-630 (Sigma) nonionic detergent, Complete-Mini protease inhibitor and 5 mM biotin-hydrazide. After 30 min an equal volume of 30 mM sodium cyanoborohydride in lysis buffer was added to reduce C N bonds. Cells were broken by repeated vortexing at 4 ◦ C for 10 min with an equivalent volume of glass beads (0.6 mm diameter; Sigma, G-8772). Supernatant was collected by centrifugation at 14,000 rpm for 10 min at 4 ◦ C. Protein concentration was measured by the Bradford method using a Coomassie protein assay kit (Pierce). The concentration of control and oxidized lysate was then adjusted to 2 mg/mL. The final supernatant was kept for further processing. 3.3. Avidin affinity selection Ultralinked immobilized monomeric avidin was packed into a stainless steel column (100 mm × 4.6 mm I.D., 1.7 mL volume) at 100 psi. The packed column was washed with ten column volumes of phosphate-buffered saline (PBS) (0.1 M sodium phosphate, 0.15 M NaCl, pH 7.4) and 5 column volumes of Biotin Blocking and Elution buffer (2 mM d-biotin in PBS) to block any non-reversible biotin-binding site on the column. Biotin from reversible biotin binding sites was removed by washing with 5 column volumes of regeneration buffer (0.1 M glycine, pH 2.8). Finally, the column was re-equilibrated with 10 column volumes of PBS. Five hundred microlitres of sample (2 mg/mL) followed by 0.25 mL of PBS was loaded into the column. The column was incubated at room temperature for 1 h and washed with 10 column volumes of PBS to remove all unbound proteins. Biotinylated proteins were eluted with 10 column volumes of Biotin Blocking and Elution buffer, and the column was regenerated with 10 column volumes of column regeneration buffer followed by 10 column volumes of PBS.

24

H. Mirzaei, F. Regnier / J. Chromatogr. A 1141 (2007) 22–31

3.4. Reversed-phase separation of biotinylated proteins A Vydac 208TP54 reversed-phase C8 column (250 mm × 4.6 mm I.D.) was used to desalt and fractionate biotinylated proteins. The reversed-phase column was equilibrated with 5 column volumes of buffer A (99.5% deionized H2 O (dI H2 O), 0.5% acetonitrile (ACN) and 0.1% TFA). 6 M urea was added to further denature the selected proteins before application to the reversed-phase column. After a 5 column volume wash a linear 60 min gradient was applied from 100% buffer A to 60% buffer B (5% dI H2 O, 95% ACN and 0.1% TFA) to elute proteins from the column. A total of 27 fractions were collected. Collected fraction were vacuum dried and stored for digestion. 3.5. Proteolysis Six molar urea and 10 mM dithiothreitol were added to proteins fractions. After a 1 h incubation at 65 ◦ C, iodoacetamide was added to a final concentration of 10 mM and reaction allowed to proceed for an additional 30 min at 4 ◦ C. Samples were then diluted six fold by addition of 50 mM 4-(2hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES) buffer (pH 8.0) and 10 mM CaCl2 . Sequence grade trypsin (2%) was added and the reaction mixture incubated at 37 ◦ C for at least 8 h. Proteolysis was stopped by addition of tosyl lysine chloroketone (TLCK) (trypsin:TLCK ratio of 1:1, w/w).

MS/MS data were submitted to Mascot for database searches and identification of the corresponding peptides and proteins one at a time. Analyst software version 1.0 of Mascot was used to create the peak list from a raw data. No smoothing of the data, signal to noise ratio or peak de-isotoping was applied. Charge states were determined manually and specified during a Mascot search for each peptide. The centroid of MS/MS peaks was determined using the following parameters. The merge distance was set at 100 ppm with minimum and maximum widths of 10 and 500 ppm, respectively. The percentage height was set a 50%, The on-line version of Mascot was used for all searches. Peptides were searched individually. The following search parameters were used for peptide identification. Database: NCBInr, Taxonomy: S. cerevisiae (72,412 entries), Enzyme: Trypsin, Missed cleavages: 4, Fixed modification: Carboxymethylation of cysteine, Peptide tolerance ±1.2 Da, MS/MS tolerance ±0.6 Da, Peptide charge was specified for each peptide, Monoisotopic peaks were used for identification. The Mascot scoring system was used as a measure of identification certainty. A detailed description of the peptide scoring system used by Mascot can be found at http://www.matrixscience.com/help/scoring help. The score threshold was adjusted to a 5% rate of false positives. Any protein that was identified by the search engine based on a single peptide was not reported, even if the peptide score was high. When 5 or more different peptides from a single protein were identified in a fraction the protein was considered to be positively identified, even when the peptide scores did not exceed the identity threshold.

3.6. LC/MS of biotinylated protein digest 4. Results Digested oxidized proteins were separated on a Agilent ZORBAX C18 column (150 mm × 0.5 mm I.D.) using a Dionex LC packing capillary LC instrument at 4 ␮L/min. Solvent A was 0.01% TFA in deionized H2 O (dI H2 O) and solvent B was 95% ACN/0.01% TFA in dI H2 O. The flow from the column was directed to the QSTAR workstation (Applied Biosystems, Framingham, MA) equipped with an ESI source. Flow from the HPLC was diverted to waste for 10 min after sample injection at 100% solvent A to remove salts, remaining derivatizing reagent GRP, and weakly adsorbed peptides. The QSTAR was then reconnected and peptides were separated in a 60 min linear gradient (from 0% B to 60% B). MS spectra were obtained in the positive ion mode at a sampling rate of one spectrum per second. 3.7. Mass spectrometry of digested fractions All mass spectral analyses were performed using a QSTAR workstation (Applied Biosystems, Framingham, MA). The nanospray ionization mode was used to examine column fractions. Trypsin digested fractions were desalted using Millipore ZipTips. Desalted digests were loaded into coated nanospray tips. Nanospray ionization was achieved at 1200 V with 25 units of curtain gas. After MS acquisition, peptides were manually detected from the MS spectrum and their M/Z value was recorded for subsequent MS/MS analysis. The collision energy was adjusted for each peptide according to its M/Z to yield comprehensive fragmentation. After acquisition of MS spectral data,

4.1. Analytical strategies Oxidized proteins can be examined in several ways as illustrated in Fig. 1. The first approach shows that following biotinylation of carbonyl groups in oxidized proteins, the entire proteome would be subjected to proteolysis before any type of fractionation. But only biotinylated peptides would be selected with the avidin affinity column for further characterization. One weakness of this approach would be that all other peptides that could be used to identify proteins are lost. It has been found that another limitation of this approach is that peptide fragments containing oxidation sites can be very large and/or have too many modifications for easy identification [18,19]. This is because lysine and arginine are major targets of oxidation and after oxidation they will not be cleaved by trypsin. Finally, single peptides from a protein may be selected in some cases. The uncertainty involved in trying to identify parent proteins based on a single peptide is well known [20,21]. The second approach in Fig. 1 shows that oxidized proteins will be selected before proteolysis and reversed phase chromatography of tryptic peptides. The advantage of this approach would be that all the peptides from oxidized proteins would be available for identification of oxidized proteins. The relatively sizable disadvantage would be that all tryptic peptides are obtained in a single pool, much as in “bottom-up” proteomics. Preliminary studies with this method led to the identification

H. Mirzaei, F. Regnier / J. Chromatogr. A 1141 (2007) 22–31

25

Fig. 1. The three candidate approaches for global proteomics of oxidized proteins. The preliminary data from approach one and experimental data acquired from approach two indicates certain limitations that could be overcome by utilizing approach three.

of 14 oxidized proteins the blood serum of rats treated with 2-nitropropane [22]. It is seen in Fig. 1 that the third approach addresses some of the problems of the other strategies by fractionating avidin selected oxidized protein mixtures before proteolysis. Extensive fractionation before proteolysis will give a substantial reduction in the complexity of peptide mixtures and assure that all the peptides from a protein are co-resident in a single fraction. The presence of un-oxidized peptides in a fraction would make it much easier to identify oxidized peptides. It also has the potential to identify cross-linked proteins as will be shown below [23]. Based on preliminary studies [17,18,21] using schemes 1 and 2, it was concluded that scheme 3 would be the most definitive method for studying protein oxidation.

the column was recycled by washing with 10 column volumes of regeneration buffer and then returned to the loading mobile phase. Columns were further eluted with 10 column volumes of PBS before reloading. The avidin affinity chromatogram of biotinylated proteins is shown in Fig. 2. The control is the same yeast culture before addition of hydrogen peroxide. The peak observed in the control represents naturally biotinylated carboxylase enzymes in addition to oxidized proteins naturally occurring in cells. The concentration of oxidized proteins is far higher in controls than that of natural biotinylated proteins [data not shown]. Since affinity selection was carried out in non-denaturing PBS buffer non-covalently associated proteins remain bound during affinity chromatography. This means any non-oxidized

4.2. Affinity selection of biotinylated proteins Avidin shows such an enormous binding affinity that recovery of biotinylated proteins from immobilized avidin columns is poor [24]. For this reason an attenuated form of immobilized avidin was chosen in these studies. Biotinylated proteins were isolated from yeast lysate by passing samples over an immobilized avidin column (scheme 3 of Fig. 1). Samples were loaded on avidin affinity columns at a linear velocity of 2 mm/s using PBS as mobile phase. No sample pretreatment was applied beyond removing excess biotin hydrazide by dialysis prior to affinity selection. Subsequent to the capture of biotinylated proteins, columns were eluted with at least 10 column volumes of PBS to remove un-bound proteins. Biotinylated proteins were eluted from the affinity column by switching to a mobile phase containing Biotin Blocking and Elution Buffer at the same flow rate. After proteins were eluted

Fig. 2. Avidin affinity selection of oxidized proteins from yeast cultures stressed by hydrogen peroxide. The control sample was derived from the yeast culture in the absence of hydrogen peroxide. The peak observed in the control represents naturally biotinylated proteins along with naturally occurring oxidized proteins.

26

H. Mirzaei, F. Regnier / J. Chromatogr. A 1141 (2007) 22–31

result of basic studies conducted to examine protein recovery from wide pore silica media [25]. 4.4. Peptide fractionation prior to electrospray ionization (ESI) mass spectrometry

Fig. 3. C8 RPLC profile of affinity selected proteins. Fractions were collected in such a way as to best isolate proteins in the peaks. A total of 27 fractions were collected and individually trypsin digested prior to MS/MS analysis.

proteins associated with oxidized proteins could have been recovered in the avidin selected fraction. Even though the use of a mild denaturing condition could have probably prevented the selection of such associated proteins, it was avoided because the attenuated form of avidin lacks the robustness of tetrameric avidin and cannot withstand mild denaturing conditions without loss of binding capacity. Loss of avidin binding capacity could have resulted in serious loss of oxidized-biotinylated proteins during affinity selection and was detrimental to our experiment. 4.3. Reversed-phase separation of affinity selected biotinylated proteins Fig. 3 shows the reversed-phase chromatogram of biotinylated proteins separated on a Vydac C8 column. A total of 27 fractions were collected in such a manner as to best isolate the eluted proteins from adjacent peaks. More fractions were taken in the middle of the elution curve because of the greater presence of proteins in this region. Hydrogen peroxide, an exogenous source of ROS, randomly generates oxygen radicals as it enters cell. Because Hydrogen peroxide generated ROS are randomly distributed throughout the cell all proteins are expected to be targeted equally regardless of their size or location in the cell. Hence any bias toward lower or higher molecular weight proteins is an indication of a bias in the method. Avidin affinity selection and mass spectrometry are not known to have any bias toward lower molecular weight proteins while there are indications of such bias in RPLC. To examine potential bias in protein recovery from reversed-phase column molecular weight distribution of identified proteins was compared with the molecular weight distribution of all yeast proteins (data not shown). No clear bias in size distribution of oxidized proteins was observed (MW distribution of oxidized and native population of yeast proteins were similar) indicating lack of biased recovery. These results are compatible with the

The 27 protein fractions seen in Fig. 3 were tryptic digested and analyzed by ESI-MS. A concern in this approach is that the degree of complexity in these fractions would be too high to allow recognition of all peptides. Suppression of ionization during ESI is a serious problem with complex mixtures of peptides that frequently results in a failure to recognize some peptides [26]. This problem can be avoided by increasing the resolution of peptide mixtures before ESI through HPLC fractionation of the mixture. The penalty for using HPLC to increase resolution is an increase in analysis time. Peptide mixtures obtained by proteolysis of fractions collected from the C8 RPLC columns were examined in two ways to determine whether mixture complexity was a problem. One was by further fractionation of the mixtures with C18 RPLC followed by direct ESI-MS analysis. The other was by introducing mixtures directly into the mass spectrometer after desalting with ZipTips. In the later case, samples were manually collected and introduced through a nanospray inlet. Fig. 4 shows the C18 RPLC elution profile of the tryptic digest from fraction 17. The number of peaks in the chromatogram indicates lack of extensive complexity. A comparison of the RPLC/MS profile of fraction 17 with that from the ZipTip/MS is seen in Fig. 5. Four more peptides were found with the RPLC/MS method. This means that RPLC fractionation of tryptic digests of protein fractions is only slightly better than the ZipTip approach. In this study 400 proteins were identified by MS/MS analysis from 27 fractions which means an average of 15 proteins per fraction. Number of identified proteins per fraction indicates that fractions were not too complicated for direct infusion. Even though direct infusion has problems such as suppression

Fig. 4. C18 RPLC profile of fraction 17 tryptic digest. The low level of complexity eliminates the real need for use of further separation steps. The whole digest was desalted via ZipTips before MS analysis.

H. Mirzaei, F. Regnier / J. Chromatogr. A 1141 (2007) 22–31

27

Fig. 5. The top mass spectrum is the average ion intensity spectrum of fraction 17 tryptic digest RPLC profile. The bottom panel is the direct nanospray mass spectrum of fraction 17 that was ZipTip desalted. Four peptides found in the HPLC/MS profile were not found in the ZipTip/MS approach. This indicates that the analytical capacity of the mass spectrometer was not exceeded in either case.

of ionization in complicated samples, it has some advantages too. In direct infusion all peptides can be manually detected while in data dependent analysis (RPLC/MS/MS) usually the top three most abundant peptides are subjected to MS/MS. This only includes populations of peptides that exceed 5 ion counts while in direct infusion peptides with less intensity can also be analyzed as long as their intensity exceeds the accepted signal to noise ratio. In direct infusion the time and the collision energy for CID fragmentation can be manually adjusted to acquire comprehensive fragmentation while in data dependent acquisition the time and collision energy is fixed for all peptides. In addition, direct infusions are carry-over free while in RPLC/MS/MS there is always a percentage carry-over which can complicate the analysis. 4.5. Identification of biotinylation sites A total of 1703 peptides from 415 proteins were identified by this procedure (a complete list of identified peptides and their corresponding proteins is provided as Supplementary Tables S-1 and S-2). This is an average of 4.1 peptides per protein. Carbonylation sites were identified in 99 of these proteins with an average of 1.4 sites per protein. An average of 3.1 peptides (total number of 308) were identified per carbonylated protein. This is the first report of a proteomic approach that identified large

numbers of oxidation sites along with specific oxidation sites in proteins. Even though the reproducibility of the method was not examined directly by repeating the entire procedure, data were validated by comparing the results with previous data that were generated using the 2-D gel electrophoresis (2-DGE) approach. Although far fewer proteins were identified with S-DGE, 70% of proteins found with 2-DGE also were found using the chromatography approach described [14]. 4.6. Identification of biotinylated peptides which carry other oxidative modifications in addition to carbonylation Although protein identification based on a few peptides is sufficient, finding biotinylated peptides from the protein provides unequivocal evidence it was oxidized. Moreover, it could even be possible to identify the site of oxidation. The problem with identifying biotinylated peptides is that there are more than 20 possible amino acid side chain oxidations. In addition, modifications caused by backbone cleavage and cross-linking are possible as well. This means there are so many possibilities that search engines such as Mascot cannot differentiate between all the options. This problem was circumvented by trying to identify only certain types of oxidation. In the studies here, lysine, arginine,

28

H. Mirzaei, F. Regnier / J. Chromatogr. A 1141 (2007) 22–31

Table 1 Some examples of biotinylated peptides which carry other oxidative modifications and their corresponding proteins No.

Protein

Accession no.

Fraction no.

Identified peptide

1

Mitochondrial glutamyl-tRNA synthetase Mitochondrial glutamyl-tRNA synthetase

625182 6324540

5

(1) ANVVDDH7 LMGITH5 VIR3 (662.20)3+

2

Mcs1p

1236327

5

(1) K4 VE10 LPTDTQASTH7 KKNSLE10 K (1259.60)2+ (2) RKQHSGTCKSDVK (511.30)3+

3

G2882

1628449

7

(1) DVDNLL11 (711.10)1+ (2) DD10 DELVRDIGTNL (749.40)2+ (3) ISFLAW8 K4,11 (579.80)2+ (4) SHLSTD10 LVATR (611.30)2+ (5) GIFPGREIL (501.20)2+

4

Myosin-like protein

171959

9

(1) SDH5 DTPM6 ESIQNGENSD10 ER3 L (1296.70)2+

Identification of biotinylated peptides not only confirms the presence of carbonylation sites within the protein sequence but also revealed the actual site of carbonylation. All modifications are coded by number superscripted to the amino acids. The table of coded modifications is included in Appendix A (Table A.1).

proline and threonine were targeted. Subsequent to oxidation and biotinylation these amino acids have well defined structures that are easily recognized by search engines. Database searches for oxidation peptides were achieved with Mascot using modifications defined on the Unimod website [27]. Even though these limitations restricted searches it was possible to identify peptides with other forms of oxidation in addition to biotinylation (Table 1). For example, the peptide ANVVDDHLMGITHVIR from mitochondrial glutamyl-tRNA synthetase was found to have an oxidized and biotinylated arginine residue in addition to two other modifications. The first histidine in the sequence was oxidized to hydroxyl histidine and the second was oxidized to aspartic acid. Confidence in this identification is based on the fact that other peptides form glutamyl-tRNA synthetase were found in the same fraction. Mcs1p protein is another example. A lysine residue in the peptide KVELPTDTQASTHKKNSLEK was biotinylated and the histidine residue in position 13 was oxidized to aspartic acid. The other peptide found from this protein was un-modified. Another example was G2882 protein. Five signature peptides from this protein were identified. Four of them are native peptides and the fifth (ISFLAWK) was found to carry an oxidized and biotinylated lysine residue along with an oxidized tryptophan. A single peptide (SDHDTPMESIQNGENSDERL) from myosin-like protein was found to be biotinylated on arginine and had oxidized histidine and methionine residues. Of the 99 biotinylated proteins identified, 38 had peptides that carried other oxidative modifications in addition to carbonylation. These examples suggest that proteins undergo multiple modifications at certain sites, perhaps in domains that are more accessible to ROS. It also explains that failure to identify biotinylated peptides from many oxidized proteins is probably due to other oxidative modifications not predicted by the search engine. 4.7. Identification of peptides with oxidative modifications other than carbonylation The fact that biotinylated peptides were found to be oxidized on other amino acids means there is a high probability

other peptides in biotinylated proteins will be oxidized as well. Since selection in this study targeted carbonylation at the protein level, peptides with other types of oxidation would be seen only if they were in proteins that had also been carbonylated. Some examples of peptides with oxidative modifications other than carbonylation are listed in Table 2. Growth regulation protein was identified based on 5 signature peptides, three of which were oxidized. The histidine at position 10 in the peptide TKAHDDLYNHPVEK was hydroxylated and the lysine residue at the C-terminus was oxidized and biotinylated. Another peptide was only biotinylated at arginine of position 2 (TRDDEVY). The third oxidized peptide (TKAHDDLYNHPVEKF) contained an aspartate residue at position 4 arising from histidine oxidation. Interestingly peptides 1 and 3 cover the same sequence but have totally different oxidative modifications. Ribosomal protein L19 was identified based on the presence of 26 peptides, 7 of which had oxidized amino acids. Only two of the seven oxidized peptides were biotinylated. The rest of the oxidized peptides carry other oxidative modifications such as hydroxylation of histidine and tryptophan. Four signature peptides from CTR9 protein were found, three of which were oxidized. Both histidine residues were oxidized in the peptide VKSIHAEDHR. The first histidine in the sequence was oxidized to hydroxyl histidine and the second was oxidized to aspartic acid. This peptide alone accounts for two oxidative modifications other than carbonylation. A second modified peptide was found in which lysine residues at positions 15 and 20 were oxidized. The third modified peptide carried a methionine sulphoxide at position 3 of the peptide and an aspartate residue at position 8 resulting from the oxidation of histidine. Protein kinase C-like protein was identified on the basis of 5 signature peptides, two of which were oxidized. None were biotinylated. Methionine was oxidized in the peptide YQIDGDQRSSSAAEGGAMESK and both methionine and tryptophan were oxidized in the peptide CKDEMWY. For 99 out of 392 identified proteins with missing biotinylated peptide other oxidative modifications were identified.

H. Mirzaei, F. Regnier / J. Chromatogr. A 1141 (2007) 22–31

29

Table 2 Examples of oxidized peptides carrying modifications other than carbonylation and their corresponding parent proteins No.

Protein

Accession no.

Fraction no.

Identified peptide

1

Growth regulation protein Protein required, with binding partner Psr1p

173189

11

(1) TKAHDD10 LYNH5 PVE10 K4 (984.50)2+ (2) TKAH7 D10 DLYNHPVEKF (908.40)2+ (3) TR3 D10 DEVY (560.30)2+ (4) KQQQQQQKIE10 KGSNSSSNTK (460.70)5+ (5) MSLPESLLLCL11 (1299.60)1+

6324617

2

Ribosomal protein L19

602897

Protein component of the large (60S) ribosomal subunit

6319444

11

(1) ALVEHIIQAK (1121.60)1+ (2) AASVVGVGK (787.40)1+ (3) AASVVGVGKR (472.30)2+ (4) ALVEHIIQAK (561.30)2+ , (374.50)3+ (5) ALVE10 HIIQAK (381.90)3+ (6) AVTVH5 SKSR (500.20)2+ (7) DALLKEDA (437.70)2+ (8) DPNETSEIAQANSR (766.40)2+ (9) HKR3 ALVEHIIQAK (581.30)3+ (10) HLYHVLYK (1072.60)1+ , (536.80)2+ , (358.20)3+ (11) KHKR3 ALVEHIIQAK (468.20)4+ (12) KVWLDPNETSEIAQANSR (515.50) 4+ , (686.70) 3+

(13) KVWLD10 PNE10 TSEIAQANSR (701.30)3+ (14) KVW5 LDPNETSEIAQANSR (692.30)3+ (15) KVW5 LD10 PNE10 TSEIAQANSR (706.30)3+ (16) KVW5 LD10 PNETSEIAQANSR (699.60)3+ (17) KVWLDPNE10 TSEIAQANSR (520.70)4+ (18) LAASVVGVGK (900.50) 1+ , (450.80)2+ (19) LAASVVGVGKR (528.80) 2+ , (352.90)3+ (20) LPSQVVWIR (1097.60)1+ , (549.30)2+ , (366.20)3+ (21) NGTIVKK (380.20)2+ (22) RALVEHIIQAK (426.60)3+ (23) RLAASVVGVGK (1056.60)1+ (24) VWLDPNETSEIAQANSR (644.00)3+ (25) VW5 LD10 PNETSEIAQANSR (656.30)3+ (26) VWLD10 PNETSEIAQANSR (651.30)3+ 3

4

CTR9 protein Cdp1p

2498269 2565014

Component of the Paf1p complex

6324427

Protein kinase C-like protein

172177

11

(1) EESIITF (838.40)1+ , (419.70)2+ (2) VKSIH5 AEDH7 R (1186.50)1+ (3) TLSDSDE10 D10 D10 D10 D10 VVKK4 PSHNK4,11 (715.40)4+ (4) E10 AM6 AISEH7 NVKD10 D10 SD10 LSD10 KD10 NE10 YD10 E10 E10 QP (887.90)4+

17

(1) QHD10 PIID10 KKIPL (487.30)3+ (2) YQIDGDQRSSSAAEGGAM6 ESK11 (1112.60)2+ (3) QETVSL (676.40)1+ (4) KVSLDNF11 (845.40)1+ (5) CKDEM6 W13 Y11 (1073.60)1+

Detection of these modifications adds to the pool of information gathered regarding the extent of oxidative damage to the yeast proteins. All modifications are coded by the number superscripted to the amino acids. The table of coded modifications is included in Appendix A (Table A.1).

4.8. Protein isoforms Multiple chromatographic peaks in Fig. 3 contained peptides derived from the same protein, as will be shown below. As a matter of fact more than 8% of the proteins identified appeared in more than one of the RPLC fractions (for a complete list see the table of proteins found in multiple fractions provided as Supplementary Table S-3). A study was undertaken to more fully understand the origin of these peaks. It has been noted above that proteins can be oxidatively modified on multiple amino acids. Perhaps these oxidative variants of proteins were separated by

the C8 RPLC and could account for this multiple peak phenomenon. A more radical possibility would be that oxidative alterations could cause bond cleavage in the primary structure of these proteins [28]. Chain cleavage could either be range or targeted. Random cleavage would produce fragments of variable chain length and hydrophobicity that would be spread across the RPLC chromatogram in Fig. 3, i.e. there would be no distinct chromatographic peaks. If these peaks are due to chain cleavage, backbone scission must be occurring at specific sites. The tryptic peptides YGSSLRR, LREMVEA, TVAGGAYTVSTAAAATVR, and KTVAGGAYTVSTAAAATVR from

30

H. Mirzaei, F. Regnier / J. Chromatogr. A 1141 (2007) 22–31

Table 3 An example of protein fragments identified in different fractions with common peptides Protein

Accession no.

Protein component of the large (60S) ribosomal subunit

6322554

Fraction no.

Peptides (1) TVAGGAYTVSTAAAATVR (834.40)2+ , (556.30)3+ (2) YGSSLRR (419.20)2+

7 11

(1) KTVAGGAYTVSTAAAATVR (898.40)2+ , (599.00)3+ (2) TVAGGAYTVSTAAAATVR (833.90)2+ , (556.30)3+

13

(1) TVAGGAYTVSTAAAATVR (833.90)2+ , (556.30)3+

17

(1) LREMVEA (424.30)2+ (2) KLEIQQHAR (374.90)3+ (3) KTVAGGAYTVSTAAAATVR (897.50)2+ (4) TVAGGAYTVSTAAAATVR (833.90)2+

18

(1) TVAGGAYTVSTAAAATVR (833.90)2+ (2) GAAGIWTCSCCKK (749.40)2+

19

(1) KTVAGGAYTVSTAAAATVR (898.50)2+ (2) TVAGGAYTVSTAAAATVR (833.90)2+

20

(1) TVAGGAYTVSTAAAATVR (833.90)2+ , (556.30)3+

The presence of common peptides indicated site-specific fragmentation. Perhaps fractions 17, 18, 19 and 20 contain the same fragment. All modifications are coded by number superscripted to the amino acids. The table of coded modifications is included in Appendix A (Table A.1). Table 4 An examples of protein fragments that do not have common peptides Protein

Accession no.

Transmembrane regulator of KAPA/DAPA transport

6324384

Fraction no.

Peptides KH7 W5 H7 VF (826.40)1+

5

AFVGIDSATH7M6 IDEVGYSK (1018.50)2+ AAILPNSSGGSFW8 (670.30)2+

11 20

This assumption can be confirmed by peptide mapping. All modifications are coded by number superscripted to the amino acids. The table of coded modifications is included in Appendix A (Table A.1).

a protein component of the large (60S) ribosomal subunit were found in fraction 17 seen in Fig. 3. But the peptides TVAGGAYTVSTAAAATVR, and KTVAGGAYTVSTAAAATVR were also found in fractions 7, 11, and 13 in Fig. 3. Based on this evidence it is likely that one or more of these later three fractions contain a cleavage fragment of the ribosomal protein that differs in primary structure from that in fraction 17 (Table 3). The structural relationship between the forms of the ribosomal protein in fractions 7, 11, and 13 is unclear. Peptides from the transmembrane regulator protein responsible for KAPA/DAPA transport were also found in multiple fractions (5, 15, and 20) as seen in Table 4. The fact that each of these fractions contained a completely different peptide from the same protein leads to the assumption that they are the result of oxidative cleavage. One of the peptides spanned positions

100–113 in the primary structure of the protein. A second came from amino acid residues 201–207. The third from fraction 20 was derived from amino acids 293–312 in the primary structure. The fact that these fragments appeared in distinct peaks suggests that cleavage of the backbone was relatively specific. A total of 32 proteins were identified in multiple non-adjacent peaks. 4.9. Co-eluting, multiple peaks of the same proteins After tabulation of the proteins detected in multiple RPLC peaks, it was noted that some were identified co-eluted in more than one fraction (Table 5). Such proteins were identified in 8 different groups. Three different protein components of the 60S ribosomal subunit were found in two different fractions (7 and 18).

Table 5 Examples of cross-linked proteins No.

Proteins co-eluted in multiple fractions

gi|

Co-elution fractions

1

(1) Protein component of the large (60S) ribosomal subunit

6322554

7 18

(2) Protein component of the large (60S) ribosomal subunit (1) Protein component of the large (60S) ribosomal subunit

6322847 6320128

Cross-linked fragments co-elute in different fractions. The identification of cross-linked proteins is only possible if they co-elute in more than one fraction. That requires further modification of cross-linked proteins after they are cross-linked. The higher the number of fragmented cross-linked proteins the lower the probability of random co-elution.

H. Mirzaei, F. Regnier / J. Chromatogr. A 1141 (2007) 22–31

ROS-promoted formation of cross-linked proteins can proceed by at least 6 different mechanisms [28,29]. These processes are random and can occur between any two proteins that are in close proximity. In affinity chromatography, all the proteins that are strongly associated in any way will bind to the affinity column and co-elute in the single step-gradient elution process. On the contrary, the mobile phase used in RPLC is denaturing and a long, linear gradient is used for elution. This means that protein complexes will be dissociated during RPLC unless they are covalently cross-linked. The prospect that different, non-associated isoforms of two proteins from the 60S ribosomal subunit would coincidently coelute is very remote. Any time the same set of proteins co-elute in more than one fraction, they are most likely cross-linked. 5. Conclusions It can be concluded from data presented in this paper that under extreme oxidative stress 8% (400 proteins out of a total of 6000 proteins coded by yeast genome) were found oxidized or more of the proteins in a yeast culture can be irreversible oxidized and the culture will still survive. It is further concluded that a small percentage of the proteins in a cell are cleaved under extremes in oxidative stress. This, along with the formation of charge variant isoforms leads to the same protein appearing in multiple chromatographic fractions during the isolation of oxidized proteins. Finally, it is concluded that ribosomal proteins have a high probability of being oxidized and in some cases cross-linked to other ribosomal proteins. Although 8% of the proteins in cells were damaged, roughly 80% of proteins in the ribosome were oxidized. There was also evidence that some of these proteins were both fragmented and cross-linked. The high likelihood of damage to ribosomal proteins during oxidative stress is yet to be explained. The methods described in this paper make it possible to study chemical phenomena associated with the irreversible oxidation of proteins. Now it is necessary to determine how this very substantial number of differences in protein structures impact cellular dynamics. Systems such as yeast provide a model for the development of tools to study the wide vary variety of oxidative stress diseases in man. Acknowledgments This work was supported by grants from the National Institutes of Health (GM59996) and the University of Texas at San Antonio, Nathan Shock Aging Center (1P30-AG13319). Appendix A Appendix B. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.chroma.2006.11.009.

31

Table A.1 A list of all amino acid modifications 1 2 3 4 5 6 7 8 9 10 11 12 13

Thereonine oxidized and biotinylated (T) Proline oxidized and biotinylated (P) Arginine oxidized and biotinylated (R) Lysine oxidized and biotinylated (K) Hydroxylation of histidine and tryptophan (HW) Methionine oxidation (M) Histidine oxidation to aspartic acid (H) Tryptophan oxidation to formylkynurenin (W) Tyrosine oxidation to aminotyrosine (Y) Sodiation of aspartic acid and glutamic acid Sodiation at the C terminus Cysteine alkylated with iodoacetamide (C) Tryptophan oxidation to kynurenin (W)

Modifications in peptides are shown as superscripts for corresponding amino acids in the sequence. Further information regarding mass differences of the modified amino acids can be accessed from the Unimod website.

References [1] V.M.V. Costa, M.A. Amorim, A. Quintanilha, P. Moradas-Ferreira, Free Radic. Biol. Med. 33 (2002) 1507. [2] T. Ozben, Life and Behavioural Sciences, in: NATO Science Series, Series I, vol. 358, IOS Press, Netherlands, 2004, p. 99. [3] R.J. Bloomer, A.H. Goldfarb, Can. J. Appl. Physiol. 29 (2004) 245. [4] N. Sitte, Biol. Aging Modul. 1 (2003) 27. [5] T. Finkel, N.J. Holbrook, Nat. (Lond.) 408 (2000) 239. [6] I. Dalle-Donne, R. Rossi, D. Giustarini, A. Milzani, R. Colombo, Clin. Chim. Acta. 329 (2003) 23. [7] K.J.A. Davies, Biochimie 83 (2001) 301. [8] E.R. Stadtman, B.S. Berlett, Reactive Oxygen Species in Biological Systems: An Interdisciplinary Approach, Kluwer Academic/Plenum publishers, USA, 1999, p. 657. [9] J. Choi, C.A. Malakowsky, J.M. Talent, C.C. Conrad, R.W. Gracy, Biochem. Biophys. Res. Commun. 293 (2002) 1566. [10] G.T. Hermanson, Bioconjugate Techniques, Academic Press, USA, 1996. [11] J. Bautista, M.D. Mateos-Nevado, Biosci. Biotechnol. Biochem. 62 (1998) 419. [12] A.G. Lenz, U. Costabel, S. Shaltiel, R.L. Levine, Anal. Biochem. 177 (1989) 419. [13] B.A. Soreghan, F. Yang, S.N. Thomas, J. Hsu, A.J. Yang, Pharm. Res. 20 (2003) 1713. [14] B.-S. Yoo, F.E. Regnier, Electrophoresis 25 (2004) 1334. [15] S.P. Gygi, G.L. Corthals, Y. Zhang, Y. Rochon, R. Aebersold, Proc. Natl. Acad. Sci. U.S.A. 97 (2000) 9390. [16] K. Merker, N. Sitte, T. Grune, Arch. Biochem. Biophys. 375 (2000) 50. [17] S. Michaelis, I. Herskowitz, Mol. Cell. Biol. 8 (1988) 1309. [18] J.R. Requena, C.C. Chao, R.L. Levine, E.R. Stadtman, Proc. Natl. Acad. Sci. U.S.A. 98 (2001) 69. [19] J. Korczak, M. Hes, A. Gramza, A. Jedrusek-Golinska, Electron. J. Pol. Agric. Univ. 7 (2004). [20] D. Fenyo, J. Qin, B.T. Chait, Electrophoresis 19 (1998) 998. [21] M. Mann, M. Wilm, Anal. Chem. 66 (1994) 4390. [22] H. Mirzaei, F. Regnier, Anal. Chem. 77 (2005) 2386. [23] E.R. Stadtman, B.S. Berlett, Free Radic. Toxicol. (1997) 71. [24] F.M. Finn, G. Titus, D. Horstman, K. Hofmann, Proc. Natl. Acad. Sci. U.S.A. 81 (1984) 7328. [25] J.D. Pearson, N.T. Lin, F.E. Regnier, Anal. Biochem. 124 (1982) 217. [26] H. Liu, D. Lin, J.R. Yates III, BioTechniques 32 (2002) 898. [27] D.M. Creasy, J.S. Cottrell, Proteomics 4 (2004) 1534. [28] B.S. Berlett, E.R. Stadtman, J. Biol. Chem. 272 (1997) 20313. [29] E.R. Stadtman, B.S. Berlett, Chem. Res. Toxicol. 10 (1997) 485.