Protein content.

Due to the availability of only small amounts (mg) of GPT, 1 mg of the lyophilized GPT were dissolved in 1 mL of ethanol/water (GPT isolated from prolamin fractions) or glutelin solution (GPT isolated from glutelin fractions), filtered (0.45 μm), injected (20 μL) into the analytical RP-HPLC system (prolamin and glutelin gradient) and the protein concentrations were calculated from external calibration with PWG-gliadin (2.9–46.6 μg) as described above. This re-chromatography also allowed verifying the purity and identity of the isolated GPT by comparing their retention times with those determined previously during the analyses of the corresponding prolamin and glutelin fractions.

SDS-PAGE.

SDS-PAGE was carried out according to Lagrain et al. [31] using a homogeneous NuPAGE 10% polyacrylamide - Bis-Tris gel (Invitrogen, Carlsbad, CA, USA) and a MOPS-Tris running buffer (pH 7.7) containing DTT (5 mmol/L) added to the inside chamber. The isolated protein fractions or GPT (1.5 mg) were dissolved in 1 mL of extraction buffer under reducing conditions (DTT, 50 mmol/L), incubated for 24 h, heated to 60°C for 10 min while shaking and centrifuged (5000 × g, 5 min, 22°C). Per sample, 2–5 μL were applied to the slots. A mixture of seven proteins (M_r 6 500–200 000) was used as marker. The running time was 40 min at 200 V and 115 mA. After the run, the proteins were fixed for 30 min in 12% trichloroacetic acid, stained for 30 min with Coomassie Brilliant Blue R-250 and destained twice [31]. The gels were scanned, the images converted to grayscale, the lanes of interest plotted as x/y-diagrams and the peaks integrated using ImageJ open source software (National Institute of Mental Health, Bethesda, MD, USA) [32].

N-terminal sequence analysis.

Isolated GPT were dissolved in acetonitrile/water (30/70, v/v) containing 0.1% (v/v) TFA. The amount of protein applied onto the polyvinylidene difluoride membrane was between 50 and 100 pmol. Sequencing was carried out by automated Edman degradation on a protein sequencer Procise 492 (Applied Biosystems, Carlsbad, CA, USA) running in the pulsed-liquid mode with ten degradation cycles [33].

LC-ESI-QTOF-MS.

An ESI-QTOF-MS (microTOF-Q, Bruker Daltonics, Bremen, Germany) coupled with an UltiMate 3000 HPLC system (Dionex, Idstein, Germany) was used for LC-MS experiments [34]. The stationary phase was an XBridge Protein BEH C₄ column (3.5 μm, 30 nm, 2.1 × 150 mm, Waters, Milford, MA, USA). The mobile phase was (A) water/TFA (999/1, v/v) and (B) acetonitrile/TFA (999/1, v/v) with a linear elution gradient from 0–0.4 min 0% B, 0.5 min 24% B, 20 min 56% B, 20.1–32 min 90% B and 32–33 min 0% B at a flow rate of 0.2 mL/min and a temperature of 30°C. Isolated GPT (2–5 mg) were dissolved in 1 mL acetonitrile/water (30/70, v/v) acetonitrile containing 0.1% (v/v) TFA and 20 μL were injected. The MS was operated in the positive ionization mode (capillary voltage: -4000 V, end plate offset: -500 V). Nitrogen was used as drying (8.0 L/min, 180°C) and nebulizing gas (0.13 MPa). The scan range was m/z 750–3200 (quadrupole ion energy: 5.0 eV). Analysis of the LC-MS data was performed using the software DataAnalysis 3.4 (Bruker Daltonics). M_r were calculated with related-ion deconvolution (mass range: 5000–100 000, maximum charge: 100, envelope cut-off: 75%, M_r agreement: 0.05%) and maximum entropy deconvolution (mass range: 5000–100 000, instrument resolution power: 10 000).

Untargeted LC-MS/MS of chymotryptic GPT hydrolyzates.

The isolated GPT (1 mg) were reconstituted in 1 mL of Tris-HCl buffer (0.1 mol/L, pH 7.8, 2 mol urea/L) containing α-chymotrypsin (TLCK treated, ≥ 40 unit/mg protein, Sigma-Aldrich, Steinheim, Germany) at an enzyme/substrate ratio of 1/200 (w/w). After incubation for 24 h at 37°C, the digestion was stopped by addition of 3 μL TFA. The resulting peptide mixtures were subjected to solid phase extraction on Supelco DSC-C₁₈ tubes (Sigma-Aldrich). The tubes were conditioned with methanol (1 mL) and equilibrated with TFA (0.1%, v/v, 1 mL). After loading the peptide mixtures, the tubes were washed with water containing TFA (0.1%, v/v, 5 x 1 mL) and the peptides were eluted with methanol (2 mL). The eluate was dried using a vacuum centrifuge (40°C, 6 h, 800 Pa), reconstituted in 500 μL formic acid (FA) (0.1%, v/v), filtered (0.45 μm) and analyzed by ion trap LC-MS/MS [35]. An UltiMate 3000 HPLC system (Dionex) was coupled to an HCTultra PTM ion trap MS (Bruker Daltonics) with collision-induced dissociation (CID). The peptides were separated on an Aeris PEPTIDE XB-C₁₈ column (3.6 μm, 10 nm, 2.1 × 150 mm, Phenomenex) and water/FA (999/1, v/v) (A) and acetonitrile/FA (999/1, v/v) (B) as solvents with a flow rate of 0.2 mL/min, a column temperature of 30°C, an injection volume of 10 μL and a linear gradient: 0–5 min 0% B, 45 min 30% B, 55–60 min 90% B, 62–77 min 0% B. The ESI interface was operated using the following parameters: mode: positive, capillary voltage: -4000 V, capillary exit voltage: -1500 V, skimmer voltage: 40 V, drying gas: nitrogen (8.0 L/min, 325°C), nebulizing gas: nitrogen (207 kPa). The MS instrument settings were: scan: standard enhanced, m/z range: 500–2000, scan speed: 8.1 m/z/s, smart target value: 300 000, maximum acquisition time: 100 ms, MS/MS setting: Auto-MS(n), collision gas: helium, absolute threshold: 10 000, relative threshold: 0.5%, fragmentation amplitude: 0.4 V. Data analysis was carried out with the software DataAnalysis 3.4 and BioTools 3.2 (Bruker Daltonics). A Mascot generic file (*.mgf) was generated from the MS/MS data file, which was exported to the MS/MS ions search module of the Mascot software (Matrix Science, London, UK) using the National Center for Biotechnology Information non-redundant (NCBI) database (U.S. National Library of Medicine, Bethesda, MD, USA) of February 2014. Peptides were searched within the taxonomy Viridiplantae with peptide mass tolerance: ± 5 Da, fragment mass tolerance: ± 0.5 Da, mass value: monoisotopic, peptide charges: +1, +2, +3, enzyme: chymotrypsin, maximum number of missed cleavages: 2 and variable modification: ammonia-loss. Peptide ion scores were calculated by the software as -10 × log(P), with P: probability for the observed match being a random event. Peptide scores > 40 were considered to indicate identity or extensive homology (p < 0.05) [36] and scores between 15 and 40 were additionally verified manually [37]. Protein scores (maximum number of protein hits: 30) were derived from peptide scores as sum of the highest ions score for each particular protein sequence, excluding the scores of duplicate matches.

Wheat (gliadins and glutenins).

The CP contents of the lyophilized wheat fractions were 93.5 ± 0.4% for gliadins and 82.8 ± 0.2% for glutenins, showing that the extraction procedure from the flour followed by dialysis and lyophilization yielded gluten fractions with very high protein contents comparable to that of PWG-gliadin [15]. The isolated GPT (ω5-, ω1,2-, α- and γ-gliadins and HMW- und LMW-GS) separated from the fractions by preparative RP-HPLC also had very high protein contents ranging from 81.3 ± 2.1% for LMW-GS to 100.8 ± 1.1% for ω1,2-gliadins (Table 2). Re-chromatography of the isolated GPT by analytical RP-HPLC confirmed the identity of each GPT (S1 Fig), because the characteristic retention times matched those in Figs 1A and 2A and there were essentially no impurities visible at 210 nm. SDS-PAGE of the wheat flour, gliadin and glutenin fractions and wheat GPT revealed that all GPT had been obtained in high purity (Fig 4A). The characteristic bands for each GPT were observed at the corresponding M_r ranges of 80 000–120 000 for HMW-GS, 60 000–68 000 for ω5-gliadins, 43 000–60 000 for ω1,2-gliadins and 32 000–45 000 for α- and γ-gliadins and LMW-GS, as reported before [31]. Minor traces of HMW-GS (≈ 2.8%, determined by semiquantitative image analysis of the SDS-PAGE gel using ImageJ) were observed in the wheat gliadin fraction, but these disappeared in the HPLC-purified GPT.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. SDS-PAGE of flours, prolamin and glutelin fractions and isolated gluten protein types.

(A) Wheat. M: marker, 1: wheat flour, 2: wheat prolamin fraction, 3: ω5-gliadins, 4: ω1,2-gliadins, 5: α-gliadins, 6: γ-gliadins, 7: wheat glutelin fraction, 8: high-molecular-weight glutenin subunits (HMW-GS), 9: low-molecular-weight glutenin subunits (LMW-GS). (B) Rye. M: marker, 10: rye flour, 11: rye prolamin fraction, 12: ω-secalins, 13: γ-75k-secalins, 14: γ-40k-secalins, 15: rye glutelin fraction, 16: HMW-secalins. (C) Barley. M: marker, 17: barley flour, 18: barley prolamin fraction, 19: γ/B-hordeins, 20: C-hordeins, 21: barley glutelins, 22: B/γ-hordeins, 23: D-hordeins. (D) Oats. 24: oat prolamin fraction (avenins), 25: oat flour.

https://doi.org/10.1371/journal.pone.0172819.g004

N-terminal sequencing of the wheat GPT was used as an additional confirmation of the purity and identity of the isolates (Table 2). Typical N-terminal sequences were determined for all wheat GPT [17,47,48,49], including EGEASGQLQC characteristic of HMW-GS Ax1, Bx7 and Bx6, EGEASEQLQC characteristic of HMW-GS Dx2 and Dx5, and EGEASRQLQC characteristic of HMW-GS By9, Dy10 and Dy12 [31], all of which were present in the wheat flour mixture. For LMW-GS, both the s-type and the m-types [50] were detected by N-terminal sequencing. The M_r of the GPT determined by LC-ESI-QTOF-MS were in good agreement with reference sequences in the NCBI database (Table 2). Only one mass signal was detected for HMW-GS, because this GPT was very hard to solubilize. Compared to SDS-PAGE, the M_r obtained by LC-ESI-QTOF-MS for the isolated GPT were about 30% lower. This overestimation of M_r by SDS-PAGE, which was observed for all GPT studied here, is ascribed to a stretched conformation of the proline-rich sequence domains in the presence of SDS and has frequently been reported before [27,31,51]. Untargeted LC-MS/MS of chymotryptic digests of the single wheat GPT resulted in the identification of 157 characteristic peptides in total. In the hydrolysates of each GPT from wheat, 6 peptides were identified in ω5-gliadins, 24 in ω1,2-gliadins, 31 in α-gliadins, 11 in γ-gliadins, 43 in HMW-GS and 42 in LMW-GS (S1 Table). These peptides matched 12 protein sequences for ω5-gliadins in the NCBI database, 25 for ω1,2-gliadins, 63 for α-gliadins, 28 for γ-gliadins, 64 for HMW-GS and 82 for LMW-GS, all of which had a protein score above 63 (S5 Table), which is the threshold calculated by the Mascot software for a protein identification to be significant. The NCBI accession given in Table 2 as reference sequence for each GPT is the best match considering correct N-terminal sequence, a M_r within the range detected by LC-ESI-QTOF-MS and the highest protein score calculated by the Mascot software after untargeted LC-MS/MS analysis considering the type and number of identified peptides. As shown in S5 Table, other protein sequences with higher scores were also assigned to the pool of detected peptides from each GPT, but these either had alternative N-terminal sequences from the main one(s) determined by N-terminal sequencing (e.g., NIQVDPSGQV in AFX69682.1 for γ-gliadins or MENSHIPGLE in ACA63873.1 for LMW-GS) or the sequences in the database were only fragments and not complete protein sequences (e.g., AGK83348.1, AGK83148.1 and AGK83270.1 for LMW-GS). The analysis of the chymotryptic GPT hydrolysates was important to confirm the identities of the isolated GPT, identify characteristic peptides and check for possible impurities. No peptides from other wheat GPT were detected in HMW-GS, ω5-gliadins and ω1,2-gliadins, reconfirming the results of analytical RP-HPLC, SDS-PAGE and N-terminal sequencing. Four and 3 peptides from LMW-GS were detected within the isolated α-gliadins and γ-gliadins, which were assigned to 5 and 9 LMW-GS protein sequences, respectively. Vice-versa, 5 α-gliadin peptides were detected within the LMW-GS isolate that corresponded to 3 α-gliadin accessions (S1 and S5 Tables). Due to their similar M_r and RP-HPLC retention times (15 – 20 min, S1C and S1F Fig), α-gliadins and LMW-GS can only be separated according to solubility during sequential extraction of wheat flour. Three extraction steps were shown to yield ≈ 95% of the gliadins [25], but the co-extraction of alcohol-soluble oligomeric HMW gliadins (13–20% of total gliadins) could not be avoided. HMW gliadins consist of ≈ 50% LMW-GS, so that 7–10% of total gliadins are estimated to actually be LMW-GS [32]. To avoid this slight impurity, further pre-fractionation by gel-permeation HPLC would be necessary prior to RP-HPLC, a step which was deemed expendable in the present study after thoroughly weighing benefits (obtaining α-gliadins with > 95% purity as opposed to ≈ 90%) and costs (labor-, material- and time-intensive). Untargeted LC-MS/MS also revealed some additional information, e.g., that LMW-GS of the i-type were also present (e.g., BAB78763.1), which had not been detected by N-terminal sequencing, probably because the i-type occurs in smaller amounts compared to the s- and m-types [50].

Rye (secalins).

The lyophilized rye prolamin and glutelin fractions had a CP content of 89.4 ± 0.1% and 53.7 ± 0.8%, respectively. The HPLC-purified GPT (HMW-, ω-, γ-75k- and γ-40k-secalins) contained 74.3 ± 3.7% (HMW-secalins) to 96.9 ± 4.8% (ω-secalins) protein (Table 2) and had their identities and purities confirmed by re-chromatography on the analytical RP-HPLC system (S2 Fig) and comparison to Figs 1C and 1D and 2B. The typical M_r ranges determined by SDS-PAGE (Fig 4B) were 95 000–105 000 for HMW-secalins, 68 000–75 000 for γ-75k-secalins, 43 000–50 000 for ω-secalins and 35 000–40 000 for γ-40k-secalins, which corresponds well to earlier studies [27]. As already seen with RP-HPLC, the prolamin fraction contained all four secalin types, whereas ω- and γ-40k-secalins were missing from the glutelin fraction. The N-terminal sequence of HMW-secalins was identical to one of the two of wheat x-type HMW-GS, because of the close botanical relationship of wheat and rye [49]. All N-terminal sequences (Table 2) matched those reported in the literature [24,49,51]. LC-ESI-QTOF-MS revealed the M_r ranges of all rye GPT (Table 2) and these agreed well with reference sequences for HMW-, γ-75k- and ω-secalins. Fig 5A shows the m/z-scans within the peak eluting at 8.9 min that were used for maximum entropy deconvolution to calculate the M_r of 39 453.8 of this specific ω-secalin.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5.

Mass spectra of isolated (A) ω-secalins and (B) γ-hordeins. The spectra show the average of scans under the peak with retention times (A) 8.9 min and (B) 12.5 min from the respective base peak MS chromatograms after LC-ESI-QTOF-MS analysis of the isolated ω-secalins and γ-hordeins, respectively. The insets show the mass spectra simulated by maximum entropy deconvolution.

https://doi.org/10.1371/journal.pone.0172819.g005

In case of γ-40k-secalins, there are only 4 reference sequences in the whole NCBI database (S6 Table), one of which is complete (AEW46799.1), whereas the other 3 are fragments. Even this complete sequence is only predicted and, to the best of our knowledge, there is no reliable reference sequence available, neither in the NCBI nor the UniProtKB database, because the curation status of rye gluten proteins is generally low and there are no reviewed entries (as of December 12, 2016). Some database entries (e.g., ADP95517.1, 75k gamma secalin from T. aestivum) also appeared to be somewhat imprecisely named, because gluten proteins from T. aestivum are called gliadins or glutenins, but not secalins. The M_r of the γ-40k-secalin entry (21 377, Table 2) was too low compared to the M_r determined by LC-ESI-QTOF-MS (≈ 32 300), but this sequence identified by untargeted LC-MS/MS of the chymotryptic γ-40k-secalin digest was the only database match available and identified based on 7 characteristic peptides (γ40k.1, .2, .5, .9, .11, .13 and .15, S2 Table). In total, 78 characteristic rye gluten peptides were identified in the chymotryptic hydrolysates of the single isolated rye GPT, 34 of which were from HMW-secalins, 11 from γ-75k-secalins, 18 from ω-secalins and 16 from γ-40k-secalins (S2 Table). These peptides allowed the identification of 14 protein sequences for HMW-secalins, 65 for γ-75k-secalins, 16 for ω-secalins and 4 for γ-40k-secalins (S6 Table). The extensive homology of wheat and rye gluten proteins [27,49] was again evident from the fact that 3 peptides occurred in both HMW-GS and HMW-secalins (designated HG+HS) and 12 peptides in both ω1,2-gliadins and ω-secalins (designated ωg+ωs). In contrast, the γ-75k- and γ-40k-secalin peptides appeared to be unique to rye. As described above and expected from Gellrich et al. [27], 3 peptides from HMW-secalins (corresponding to 4 protein sequences) were detected in the ω-secalin isolate, but the amount of HMW-secalins is expected to be negligible, because SDS-PAGE of ω-secalins revealed no visible band with an M_r of ≈ 100 000 (Fig 4B). No peptides from other rye GPT were detected within HMW-secalins and γ-75k-secalins, but 2 peptides from γ-75k-secalins were detected within the γ-40k-secalin isolate. Due to the virtual lack of reference sequences for γ-40k-secalins, it was impossible to determine whether these 2 peptides (SQLEVVRSL and ASIVTGIVGH) were truly from γ-75k-secalins (which is unlikely, because the RP-HPLC retention times were clearly separated, S2C and S2D Fig), or whether these could also occur in γ-40k-secalins themselves, because both types share the same evolutionary origin [52]. The alignment of both protein sequences (AEW46799.1 and ADP95479.1) using the “Align” tool (UniProtKB) revealed an identity of 30.7%, with homologous sections close to the C-terminus of this γ-75k-secalin sequence. Two very similar peptides (AQLEVIRSL and ASTVAGIGGQ) also occur in this γ-40k-secalin sequence, substantiating the assumption that these peptides could also be present in yet unidentified sequences of γ-40k-secalins, because single to multiple amino acid substitutions occur very frequently within gluten proteins.

Barley (hordeins).

The lyophilized barley fractions contained 87.3 ± 0.4% CP (prolamins) and 62.0 ± 0.5% CP (glutelins). The barley GPT (D-, C-, γ- and B-hordeins) isolated by preparative RP-HPLC had protein contents in the range from 85.3 ± 6.3% (B/γ-hordeins) to 99.7 ± 1.6% (γ/B-hordeins) (Table 2). The identities and purities of the GPT were again confirmed by re-chromatography (S3 Fig) and comparison to Figs 1E and 1F and 2C. The separation of the barley prolamin and glutelin fractions and types by SDS-PAGE showed the following M_r ≈ 100 000 for D-hordeins, 45 000–65 000 for C-hordeins and 32 000–40 000 for γ- and B-hordeins, which matched earlier investigations [41,53]. The major N-terminal sequences (Table 2) again corresponded to earlier reports [45,47,54] and more specifically matched γ3-hordein and B1-hordein. The M_r for γ-hordeins (Fig 5B) and B-hordeins determined by LC-ESI-QTOF-MS were in agreement with the reference sequences from the database, but the M_r for C-hordeins (≈ 45 000) was higher than any of the 6 protein sequences that were identified after chymotryptic digestion of the C-hordein isolate and untargeted LC-MS/MS (S3 and S7 Tables). Of those 6 sequences from the NCBI database, 3 were only fragments (P02864.1, P17991.1, AAA32942.1) and the one given as reference sequence (AAB28161.1) already had the highest M_r of the remaining 3 entries. Only 9 sequences in total are available for C-hordeins, 5 of which are fragments with a length of 105 amino acids or less (UniProtKB, as of December 12, 2016). The issue of incomplete proteomes within the Poaceae, and especially within Hordeum sp. and Secale sp. has been noted before and often results in unmatched peptide/protein identifications [55]. Overall, the reference sequences identified here for all barley GPT were very similar to those reported by Colgrave et al. [46]. Untargeted LC-MS/MS analyses led to the identification of 45 barley peptides in total, with 9 in the D-hordein, 11 in the C-hordein and 25 in the γ/B- and B/γ-hordein hydrolysates combined, of which 4 were specific for γ-hordeins (S3 Table). One peptide within D-hordeins was also identified in HMW-secalins (KVAKAQQL) and one within B-hordeins also in LMW-GS (LQPHQIAQL). All peptides identified within the C- and D-hordein hydrolysates were specific for that barley GPT, but, as discussed before, the isolation strategy applied here did not allow a separation of γ- and B-hordeins, because both GPT were present in γ/B- and B/γ-hordeins.

Oats (avenins).

Oat avenins were extracted from defatted oat flour with 60% ethanol and not further fractionated by preparative HPLC, because this fraction only contained 6 major protein peaks (Fig 1B). Furthermore, only the oat prolamin fraction is considered as oat gluten (avenins), because oat glutelins mostly contain 12S globulins [30] that are not considered as gluten. The CP content of the isolated avenins was 79.2 ± 0.6% and the N-terminal sequence TTTVQYDPSE (Table 2) was found to be similar to ones reported as avenins 5–7 by Anderson [56]. One alternative N-terminal sequence (TTTVQYNPSE) was also detected. The M_r range of the avenin fraction was 25 000–32 000 by SDS-PAGE (Fig 4D) and again lower (≈ 22 000–29 000) by LC-ESI-QTOF-MS. The two characteristic bands of α-globulins (≈ 35 000) and β-globulins (≈ 23 000) [30] seen in the oat flour on the SDS-PAGE gel were missing in the avenin fraction. A total of 37 avenin-specific peptides were detected in the chymotryptic hydrolysate by untargeted LC-MS/MS and assigned to 49 avenin protein sequences (S4 and S8 Tables). Globulin-specific peptides were not detected indicating the high purity of the avenin fraction.