**2.1. Proteomics in plant glycation research**

are still reactive and can be easily involved in further reactions [25]. In the past few decades, several AGEs were comprehensively characterized (Figure 2). Among lysine-derived modifi‐

**Figure 2.** AGEs detected *in vivo*: GD-HI, glyoxal-derived dihydroxyimidazolidine; MGD-HI, methylglyoxal-derived di‐ hydroxyimidazolidine; MG-H, methylglyoxal-derived hydroimidazolone; Glarg, glyoxal-derived hydroimidazolone;

formyl-5-hydroxymethyl-pyrrolyl)-*L*-norleucine (pyrraline) [28] and glyceraldehyde-derived pyridinium compound (GLAP) [29] are the best-characterized. Not less attention was paid to the modifications of arginine. Thus, Schwarzenbolz and coworkers reported 1-(4-amino-4 carboxybutyl)2-imino-5-oxo-imidazolidine (Glarg) as a product of the reaction of arginine with

Methylglyoxal was shown to form isomeric methylglyoxal-derived hydroimidazolones (MG-

isomer [32]. Hydrolysis of 2-amino-5-(2-amino-4-hydro-4-methyl-5-imidazolon-1-yl)pentano‐ ic acid (MG-H3) yields carboxyethyl-*L*-arginine (CEA) [33]. Sequential modification of arginine

Upon their absorption in human intestine, AGEs interact with endothelial and macrophage pattern recognition receptors for AGEs (e.g. RAGEs) and trigger NF-κB-mediated expression

tetrahydropyrimidine-2-yl)-*L*-ornithine (tetra-hydroargpyrimidine, TH-Argpyr) [35].








298 Abiotic and Biotic Stress in Plants - Recent Advances and Future Perspectives

cations, *N*<sup>ε</sup>

CML, N<sup>ε</sup>

Hs) with *N*<sup>δ</sup>

glyoxal [30] and yielding *N<sup>δ</sup>*


with two methylglyoxal molecules results in *N*<sup>δ</sup>

ornithine (argpyrimidine, Argpyr) [34] and *N*<sup>δ</sup>

Recently, using liquid chromatography-tandem mass spectrometry (LC-MS/MS)-based approach, Bechtold and co-workers estimated the total levels of individual AGE classes in *Arabidopsis thaliana* leaf tissues [16]. Their methodology relied on the tandem mass spectro‐ metric (MS/MS) analysis of individual amino acids in tissue hydrolyzates by multiple reaction monitoring (MRM) [45]. Though this method provides high sensitivity and specificity, this strategy could not provide information about the identity and modification patterns of individual glycated proteins. However, this question can be addressed by the methods of LC-MS-based proteomics – the technique based on the analysis of protein enzymatic hydrolyzates [46]. Indeed, in its classical implementation, unbiased LC-MS approach combines high separation efficiency of the high-performance or ultra-high-performance liquid chromatogra‐ phy (HPLC or UPLC, respectively) and outstanding resolution and mass accuracy of highresolution mass spectrometry – Orbitrap- and quadrupole-time-of-flight (QqTOF)-MS [47,48]. It dramatically increases the analytical resolution (i.e. the number of analytes, annotated in one experiment) of the whole method. Moreover, due to the superior separation, low-abundant post-translationally modified peptides are less amenable to the ion suppression effects, accompanying electrospray ionization (ESI) [49]. Obviously, this analytical strategy seems to be the optimal for the study of such heterogenic and low-abundant set of modifications as AGEs.

#### **2.2. Protein isolation and proteolysis**

Depending on the target protein fraction (soluble or total), aqueous (aq.), extraction can be performed in the absence and presence of phenol, respectively [50]. In the first case, however, even the extracts of green parts contain high amounts of soluble metabolites that might inhibit the activity of proteases used for digestion. Such incompleteness of proteolysis can be observed as well, when proteins are extracted from plant parts rich in anti-nutritive (i.e. protease inhibiting or denaturing) phenolics like insoluble condensed tannins in seeds [51]. Therefore, the extracts can be purified by gel filtration chromatography and/or ultrafiltration using Centricon or Vivaspin centrifugation devices [52] prior to the determination of protein concentration. Alternatively, the proteins can be isolated by phenol extraction. In this case, phenolics contaminants can be removed by addition of 1–5% of soluble or insoluble polyvi‐ nylpyrrolidone [50]. While purified aqueous extracts can be easily digested by proteases in the presence of only deoxycholate as a denaturizing agent [53], the dried proteins isolated with phenol (containing also the fraction of hydrophobic membrane proteins) can be reconstituted only in the presence of both chaotropic compounds (urea, thiourea) and strong detergents. Conventional detergents, such as sodium dodecyl sulfate (SDS) or 3-[(3-cholamidopropyl)di‐ methylammonio]-1-propanesulfonate (CHAPS), co-elute with proteolytic peptides and disturb ESI. This can be avoided by application of the detergents which do not impact protease activity and can be easily destroyed upon the digest. For example, *Rapi*GestTM SF Surfactant (Waters Corporation, Milford, US) or Progenta Protea Biosciences, Inc., Morgantown, US) can be applied [54]. The detergents can be dissolved in the lysis buffer (7 mol/L urea and 2 mol/L in 50 mmol/L Tris-HCl, pH 7.5) and destroyed directly after proteolysis at pH 2 for 20–45 min (Figure 3).

The completeness of the digestion can be controlled by the SDS-polyacrylamide gel electro‐ phoresis (PAGE). Performing SDS-PAGE additionally prior to proteolysis provides the opportunity to validate the Bradford assay results by relative quantification of total lane densities and, if necessary, normalize the results of the LC-MS-based label-free quantification. It is important to stress that the use of chaotropic agents, acid-labile detergents and strong acids for their cleavage ultimately require RP-based solid phase extraction (SPE) after com‐ pletion of the digest and verification of its completeness by SDS-PAGE.

**Figure 3.** Protocol for tryptic digestion of the total plant protein obtained by phenol extraction. The procedure em‐ ployed tryptic digestion in the presence of an acid-labile detergent ALLS II, required to ensure efficient solubilisation of membrane proteins.

#### **2.3. LC-MS analysis of glycated peptide mixtures**

[46]. Indeed, in its classical implementation, unbiased LC-MS approach combines high separation efficiency of the high-performance or ultra-high-performance liquid chromatogra‐ phy (HPLC or UPLC, respectively) and outstanding resolution and mass accuracy of highresolution mass spectrometry – Orbitrap- and quadrupole-time-of-flight (QqTOF)-MS [47,48]. It dramatically increases the analytical resolution (i.e. the number of analytes, annotated in one experiment) of the whole method. Moreover, due to the superior separation, low-abundant post-translationally modified peptides are less amenable to the ion suppression effects, accompanying electrospray ionization (ESI) [49]. Obviously, this analytical strategy seems to be the optimal for the study of such heterogenic and low-abundant set of modifications as

Depending on the target protein fraction (soluble or total), aqueous (aq.), extraction can be performed in the absence and presence of phenol, respectively [50]. In the first case, however, even the extracts of green parts contain high amounts of soluble metabolites that might inhibit the activity of proteases used for digestion. Such incompleteness of proteolysis can be observed as well, when proteins are extracted from plant parts rich in anti-nutritive (i.e. protease inhibiting or denaturing) phenolics like insoluble condensed tannins in seeds [51]. Therefore, the extracts can be purified by gel filtration chromatography and/or ultrafiltration using Centricon or Vivaspin centrifugation devices [52] prior to the determination of protein concentration. Alternatively, the proteins can be isolated by phenol extraction. In this case, phenolics contaminants can be removed by addition of 1–5% of soluble or insoluble polyvi‐ nylpyrrolidone [50]. While purified aqueous extracts can be easily digested by proteases in the presence of only deoxycholate as a denaturizing agent [53], the dried proteins isolated with phenol (containing also the fraction of hydrophobic membrane proteins) can be reconstituted only in the presence of both chaotropic compounds (urea, thiourea) and strong detergents. Conventional detergents, such as sodium dodecyl sulfate (SDS) or 3-[(3-cholamidopropyl)di‐ methylammonio]-1-propanesulfonate (CHAPS), co-elute with proteolytic peptides and disturb ESI. This can be avoided by application of the detergents which do not impact protease activity and can be easily destroyed upon the digest. For example, *Rapi*GestTM SF Surfactant (Waters Corporation, Milford, US) or Progenta Protea Biosciences, Inc., Morgantown, US) can be applied [54]. The detergents can be dissolved in the lysis buffer (7 mol/L urea and 2 mol/L in 50 mmol/L Tris-HCl, pH 7.5) and destroyed directly after proteolysis at pH 2 for 20–45 min

The completeness of the digestion can be controlled by the SDS-polyacrylamide gel electro‐ phoresis (PAGE). Performing SDS-PAGE additionally prior to proteolysis provides the opportunity to validate the Bradford assay results by relative quantification of total lane densities and, if necessary, normalize the results of the LC-MS-based label-free quantification. It is important to stress that the use of chaotropic agents, acid-labile detergents and strong acids for their cleavage ultimately require RP-based solid phase extraction (SPE) after com‐

pletion of the digest and verification of its completeness by SDS-PAGE.

AGEs.

(Figure 3).

**2.2. Protein isolation and proteolysis**

300 Abiotic and Biotic Stress in Plants - Recent Advances and Future Perspectives

The proteolytic digests are, typically, complex mixtures. Therefore, for successful detection and identification of their components (i.e. peptides), adequate chromatographic and mass separation techniques need to be applied. The selection of a strategy for LC-MS analysis (in terms of throughput, sensitivity, separation efficiency and reproducibility) depends on the aims of particular research. In the absolute majority of cases, LC-MS analyses rely on so-called data-dependent acquisition experiments (DDA) [55]. These complex experiments comprise survey MS scan (typically performed in Orbitrap- or TOF mass analyzer), and multiple dependent ones – MS/MS, relying on linear ion trap (LIT) or QqTOF analysis [56,57]. Modern instruments provide a possibility for introduction of the second dependent scan. Thus, hybrid LTQ-Orbitrap instruments support multi-step activation (MSA) experiments, comprising an additional MS/MS scan with a low-energy collision-induced dissociation (CID) [58]. The modification-specific neutral losses, appearing under these conditions, trigger a high-energy dependent MS/MS scan providing rich structural information. In the past decade, these experiments were successfully applied to glycated tryptic digests using Amadori-specific losses of two and three waters, as well as additional formaldehyde molecule [59].

In the most straightforward way, DDA experiments are performed without additional analytical procedures preceding a RP-HPLC separation – so-called shotgun proteomics. However, as well as other PTMs, glycative and glycoxidative modifications are low abundant. Hence, their ionization might be suppressed by highly abundant species. To avoid this, early glycated tryptic peptides can be selectively enriched by boronic acid affinity chromatography (BAC) before LC-MS/MS analysis [60]. However, such enrichment is not possible for the whole fraction of advanced glycated peptides due to their structural heterogeneity. Therefore, these low-abundant species need to be directly detected in conventional data-dependent acquisition (DDA) experiments. As the DDA algorithm relies on the MS/MS analysis of the most intense signals in each time segment, this type of experiments suffers from so-called undersampling, i.e. missed fragmentation of low-abundant quasi-molecular ions [61]. Because of this reason, shotgun proteomics is not a desired strategy for the analysis of PTMs.

Thus, the number of co-eluting peptides in DDA analyses needs to be reduced to increase the coverage of the AGE-modified proteome. This can be addressed by three approaches: *(i)* introduction of enrichment or pre-fractionation step, *(ii)* gas phase fractionation (GPF) on the MS level and *(iii)* use of retention time-based exclusion lists. In all cases, for reliable identifi‐ cation of AGE-containing peptides, multiple DDA experiments are required. A special attention needs to be paid on the number of dependent MS/MS scans and the total duration of cycle. Thus, the number of MS/MS scans should not be too high, as the quasi-molecular ions of AGE peptides have relatively low intensities and require, therefore, longer fill times in LIT and accumulation time TOF. The cycle times typically need to be shorter, than in conventional shotgun DDA experiments, as the peak widths of low abundant peptides are smaller.

The generalized analytical strategy might comprise both qualitative and quantitative ap‐ proaches, i.e. identification of glycated peptides in DDA experiments with their subsequent label-free quantification in additional full-scan MS experiments (Figure 4). For identification of glycation sites, early glycated peptides can be selectively enriched [60], while the analysis of AGE-containing species might rely on two-dimensional liquid chromatography (LC × LC) [62]. BAC is a well-established analytical tool to enrich Amadori and Heyns products from mammalian tissues (predominantly plasma) [56]. However, for application to the study of plant glycation, this method requires some optimization. Thus, the protein extract must be effectively washed prior to digestion (e.g. by ultrafiltration) to remove the co-extracted carbohydrate-related metabolites (mono-, oligosaccharides, sugar esters and glycosides of (poly)phenolics). Due to their *cis*-diol groups, these metabolites saturate the binding sites of the affinity column and reduce the peptide enrichment efficiency. Moreover, phenylpropa‐ noids, as well as their sugar esters and glycosides, are well-retained on reversed phase [63] and might co-elute with peptides in course of RP-HPLC separation, leading to essential ion suppression during the ESI process and shift of peptide signals in the lower part on the instrument dynamic range. Thus, the analysis of early glycation proteins in total extracts that can not be efficiently ultrafiltrated is a challenging task.

Glycation of Plant Proteins under Environmental Stress — Methodological Approaches, Potential Mechanisms... http://dx.doi.org/10.5772/61860 303

modification-specific neutral losses, appearing under these conditions, trigger a high-energy dependent MS/MS scan providing rich structural information. In the past decade, these experiments were successfully applied to glycated tryptic digests using Amadori-specific

In the most straightforward way, DDA experiments are performed without additional analytical procedures preceding a RP-HPLC separation – so-called shotgun proteomics. However, as well as other PTMs, glycative and glycoxidative modifications are low abundant. Hence, their ionization might be suppressed by highly abundant species. To avoid this, early glycated tryptic peptides can be selectively enriched by boronic acid affinity chromatography (BAC) before LC-MS/MS analysis [60]. However, such enrichment is not possible for the whole fraction of advanced glycated peptides due to their structural heterogeneity. Therefore, these low-abundant species need to be directly detected in conventional data-dependent acquisition (DDA) experiments. As the DDA algorithm relies on the MS/MS analysis of the most intense signals in each time segment, this type of experiments suffers from so-called undersampling, i.e. missed fragmentation of low-abundant quasi-molecular ions [61]. Because of this reason,

Thus, the number of co-eluting peptides in DDA analyses needs to be reduced to increase the coverage of the AGE-modified proteome. This can be addressed by three approaches: *(i)* introduction of enrichment or pre-fractionation step, *(ii)* gas phase fractionation (GPF) on the MS level and *(iii)* use of retention time-based exclusion lists. In all cases, for reliable identifi‐ cation of AGE-containing peptides, multiple DDA experiments are required. A special attention needs to be paid on the number of dependent MS/MS scans and the total duration of cycle. Thus, the number of MS/MS scans should not be too high, as the quasi-molecular ions of AGE peptides have relatively low intensities and require, therefore, longer fill times in LIT and accumulation time TOF. The cycle times typically need to be shorter, than in conventional

shotgun DDA experiments, as the peak widths of low abundant peptides are smaller.

The generalized analytical strategy might comprise both qualitative and quantitative ap‐ proaches, i.e. identification of glycated peptides in DDA experiments with their subsequent label-free quantification in additional full-scan MS experiments (Figure 4). For identification of glycation sites, early glycated peptides can be selectively enriched [60], while the analysis of AGE-containing species might rely on two-dimensional liquid chromatography (LC × LC) [62]. BAC is a well-established analytical tool to enrich Amadori and Heyns products from mammalian tissues (predominantly plasma) [56]. However, for application to the study of plant glycation, this method requires some optimization. Thus, the protein extract must be effectively washed prior to digestion (e.g. by ultrafiltration) to remove the co-extracted carbohydrate-related metabolites (mono-, oligosaccharides, sugar esters and glycosides of (poly)phenolics). Due to their *cis*-diol groups, these metabolites saturate the binding sites of the affinity column and reduce the peptide enrichment efficiency. Moreover, phenylpropa‐ noids, as well as their sugar esters and glycosides, are well-retained on reversed phase [63] and might co-elute with peptides in course of RP-HPLC separation, leading to essential ion suppression during the ESI process and shift of peptide signals in the lower part on the instrument dynamic range. Thus, the analysis of early glycation proteins in total extracts that

losses of two and three waters, as well as additional formaldehyde molecule [59].

302 Abiotic and Biotic Stress in Plants - Recent Advances and Future Perspectives

shotgun proteomics is not a desired strategy for the analysis of PTMs.

can not be efficiently ultrafiltrated is a challenging task.

**Figure 4.** Experimental workflow for the analysis of glycated proteome. The analytical strategy comprises separation and MS/MS-based identification of modified peptides, as well as label-free quantification of corresponding glycated sites.

Pre-fractionation is usually introduced as an (relative to the RP) orthogonal separation procedure – most often cation exchange or hydrophilic interaction chromatography (HILIC) [62,64]. This approach essentially decreases sample complexity, facilitating, thereby, the fragmentation of low-abundant species. Similarly, fractionation can be applied on the MS level. Thus, simultaneously formed quasi-molecular ions (i.e. originating from the co-eluting species) can be fractioned by their *m/z* values in the quadrupole filter preceding LIT (GPF approach). In this case, however, repeated injections are necessary to cover the whole mass range that increases the overall analysis duration. For the most of applications, three GPF segments (*m/z* 400–600, 600–800 and 800–1500) are sufficient to significantly improve protein identification rates [57,64]. Additionally, based on the results of one DDA experiment, all unmodified peptides can be excluded from fragmentation in following ones on software level by generation of a corresponding "exclusion list" in the instrumental method [57]. However, the combination of these approaches is required for in-depth proteome analysis. The peptides can be identified by database search using SEQUEST or Mascot search engines.

Early glycated tryptic peptides can be annotated in high-resolution Orbitrap-MS experiments by the *m/z* value and charge of the corresponding quasi-molecular ion, and retention time in corresponding extracted ion chromatograms (XICs) as shown in Figure 5A [57]. Unambiguous identification of these peptides relies on characteristic series of b- and y-ions with consideration of the losses of three water molecules (pyrylium ions) and additional loss of formaldehyde (fulylium ions) [59] (Figure 5A). Identification and quantification of AGE-modified peptides are based on the same principle [65]. The only difference in comparison to the early glycation species is the absence of the neutral losses accompanying parent and fragment ions. Addi‐ tionally, the glycation state of peptides can be confirmed by characteristic fragments in the low *m/z* range [61,66]. This information can be further used for label-free quantification of indi‐ vidual AGE-modified peptides and, hence, specific glycation sites. The quantification typically relies on the integration of the annotated signals and qualitative comparisons of obtained peak areas. Thereby, the peak areas can be normalized to the signals of unmodified peptides in the same sample or in quality controls (QC). The peak integration can be performed by means of the vendor software packages – Xcalibur Quan Browser or LCquan.

**Figure 5.** Annotation of Amadori peptides by nanoUPLC-ESI-LTQ-Orbitrap-MS/MS DDA experiments. Reversed phase ESI-Orbitrap-MS total ion chromatogram and the segment of the extracted ion chromatogram (insert) for *m/z* 736.39 ± 0.02 in the tR range of 30–35 min (A) and an ESI-LTQ-MS/MS spectrum at *m/z* 736.4 corresponding to the [M +3H]3+ ion of the peptide VFDEFKAmPLVEEPQNLIK (B) acquired at 32.5 min of the same DDA experiment. Pyrylium b and y fragment ions are marked with asterisk [61].

The physiological role of glycation can be assessed by the system biology software tools. Thus, for the grouping of AGE-modified proteins by their functions, the mapping software MapMan (Max-Planck Institute of Molecular Plant Physiology, Potsdam-Golm, Germany, http:// mapman.gabipd.org) can be used. The functional annotation of proteins might give insight in biological effects of AGEs in plants and provide the material for future biological studies. Thus, the proteomic data can be complemented by the result of transcriptional analysis and deter‐ mination of enzymatic activities. Afterwards, the functional role of glycation in respect of particular proteins can be confirmed by the experiments with corresponding mutants.
