Introduction

Methionine (Met) is an essential amino acid for humans and livestock. In livestock farming, it is one of the limiting amino acids for monogastric animals such as pigs and poultry due to its small abundance in many plant proteins used for animal feeding [1]. Methionine can be easily oxidized by several oxidants including hydrogen peroxide, intermediates of glycation and lipid peroxidation, and also UV light [2,3,4,5]. Together with histidine, tyrosine, and tryptophan, methionine is among the amino acids most sensitive to photooxidation [6]. The primary oxidation product of protein-bound Met is methionine sulfoxide (MetSO) (Fig. 1). Free methionine may also react to dehydromethionine (DHMet) by forming a characteristic N–S bond at pH values > 7, which occur under physiological conditions as well as in many foodstuffs, but the product is converted to MetSO at pH values above 9 [7, 8]. MetSO may be oxidized to methionine sulfone (MetSO2) by strong oxidants such as performic acid, which is sometimes used for analyzing the concentration of Met [9].

Fig. 1
figure 1

Pathways of photochemical degradation of free methionine, literature see text

Full size image

When milk is exposed to direct UV light in clear bottles, it evolves a burnt-feather flavor. This is called “sunstruck” or “sunlight” flavor and emerges due to riboflavin-sensitized photooxidation. In the past, research mainly focused on the volatile compounds originating from this reaction. The Strecker aldehyde methional, which emerges from Met, was made responsible for the off-flavor in early works [10]. However, free Met has only a very small concentration in milk and milk products [11, 12], and it was later suggested that light-induced breakdown products of methional, such as methanethiol (MT), dimethyl sulfide, and acrolein, were responsible for the off-flavor [13]. Dimethyl sulfide and acrolein may be formed from methional by retro-Michael reaction [14]. Methional and acrolein were identified as reaction products of the photooxidation of methionine in the presence of riboflavin. These were found toxic against a variety of microorganisms [15]. A photo-induced degradation was also described for the development of a skunky flavor in beer. This involves thiyl radicals originating from the degradation of sulfur-containing amino acids, which can also be found in the degradation compounds of hops [16]. The riboflavin-dependent oxidation of free Met to methional with the subsequent formation of MT from methional was also described in wine [17]. The research on the reaction products of methionine that remain in the protein after scission of bonds to form volatile compounds has been disregarded, even though these are the products that are actually consumed, whereas the volatile aroma compounds may leave the food products.

Native milk proteins give rise to the sunlight flavor in the presence of riboflavin. As methional is a very potent flavor substance that is detectable at 50 ppb in milk, it was suggested that already a small degree of proteolysis may account for this effect [10]. Basically, reactions of protein-bound amino acids may be different from those of free amino acids, since the amino group is not available as a reaction partner. As it was not possible to extract methional from sunlight-exposed milk, its contribution to the sunlight flavor has been questioned [13].

Several works investigated the chemical fate of Met during photooxidation and the characterization of the specific reaction products. Photooxidation of thioethers in general was described for preparative formation of sulfoxides through irradiation with visible light (λ = 450 nm) in the presence of riboflavin and its stable derivative tetra-O-acetylriboflavin [18, 19]. It is generally believed that photooxidation of Met leads mainly to MetSO. MetSO was also reported as the sole oxidation product during type I photosensitized oxidation in the presence of pterins [20]. Photoreduction of MetSO was reported to occur under anaerobic conditions [21].

The European Food Safety Authority (EFSA) has commented on the safety of UV irradiation, suggesting to further elucidate the reaction pathways due to this treatment [22]. Therefore, the main goal of this work is to identify the non-volatile reaction products that result from photooxidation of peptide- or protein-bound methionine and estimate the relevance of the respective reaction pathways. Peptide-bound 2-amino-4-oxobutyric acid (formyl alanine, β-aspartic semialdehyde, (β-ASA)) and homocysteine (Hcy) were identified as quantitatively outstanding products of riboflavin-sensitized photooxidation of peptide-bound Met.

Materials and methods

Chemicals

Boric acid and formaldehyde (37% aqueous solution) were purchased from Carl Roth (Karlsruhe, Germany). Benzoyl methionine was purchased from Bachem (Bubendorf, Switzerland). Sulfuric acid and disodium hydrogen phosphate dodecahydrate were purchased from Merck (Darmstadt, Germany). (−)‑Riboflavin, 2,4-dinitrophenylhydrazine (DNPH), and sodium dihydrogen phosphate dihydrate were purchased from Sigma-Aldrich (Steinheim, Germany). The tetrapeptides glycylalanylmethionylglycine (GAMG) and glycylmethionylalanylglycine (GMAG) were purchased from ProteoGenix (Schiltigheim, France). Iron(II) sulfate × 1.5 H2O was purchased from Riedel-de-Haën (Seelze, Germany). Potassium tris(oxalato)ferrate(III) trihydrate and 1,10-phenanthroline monohydrate were purchased from Alfa Aesar (Kandel, Germany). Sodium acetate was purchased from Fluka (Buchs, Switzerland). HPLC-grade acetonitrile and hydrochloric acid were from VWR (Darmstadt, Germany). Benzoyl methionine sulfoxide was self-synthesized according to Hellwig et al. [4]. Nanopure water (Arium pro, Sartorius AG, Göttingen, Germany) was used to prepare all solutions except for formaldehyde analysis, where double-distilled water (Destamat Bi 18E; QCS GmbH, Maintal, Germany) was used. For the LC–MS/MS measurements, LC–MS-grade water and acetonitrile (Honeywell Specialty Chemicals, Seelze, Germany) as well as formic acid and trifluoroacetic acid (Fisher Scientific, Schwerte, Germany) were used.

Photooxidation experiments

A solution containing 1 mM benzoyl methionine and 0.1 mM riboflavin in 0.1 M sodium phosphate buffer (1 mL, pH 7.4) in a capillary electrophoresis (CE) vial was placed in a photo reactor (PhotoRedoxBox, Interchim, Montluçon, France) and irradiated by UV light (lamp, EvoluChem LED LG 18W) at λ = 366 nm for up to 240 min. In further experiments, the photooxidation reaction was performed with varying amounts of riboflavin (10 mol% or 3 mol% relative to Bz-Met). Moreover, Bz-Met was replaced by Bz-MetSO and the tetrapeptides GAMG and GMAG. Samples were withdrawn during the incubation period after 1, 5, 15, 30, 60, 120, and 240 min of irradiation, while irradiation was interrupted. Prior to CE analysis, 100 µL of the reaction mixtures was diluted with 100 µL of the CE buffer as described below. For the LC–MS/MS analysis, the reaction mixtures were diluted with 100 µL of the HPLC solvent A as described below. During the photooxidation of the tetrapeptides and for quantification of formaldehyde, only 10 mol% riboflavin was used, and samples were only taken after 60 min of irradiation.

Actinometry

Based on literature methods [23, 24], 1 mL of a 0.16 M solution of potassium tris(oxalato)ferrate(III) in 0.1 M H2SO4 was irradiated in a CE vial in the PhotoRedoxBox for 5 min. A sample of 100 µL was withdrawn and immediately added to 2.5 mL of water. Then, 100 µL of this solution was added to 750 µL of a mixture consisting of the same volumes of (1) 0.1% (w/v) o-phenanthroline in water, (2) a buffer solution made of 9 mL 1 M H2SO4 and 15 mL 1 M sodium acetate filled to 25 mL with water, and (3) water. After gently shaking the mixtures for 30 min in the absence of light, the absorbance was measured at 510 nm using a V-750 Spectrophotometer (Jasco, Pfungstadt, Germany). For calibration, solutions of iron(II) sulfate × 1.5 H2O (0.3–30 µM) were prepared in 0.1 M H2SO4 and treated like the irradiated samples. With the potassium ferrioxalate actinometer, a photon flux of (41.6 ± 1.1) nEinstein/s/mL was determined for the experimental setup used in this work.

Capillary electrophoresis (CE) with UV detection

CE analyses were performed on an Agilent Technologies 7100 Capillary Electrophoresis system (Agilent Technologies, Santa Clara, CA, USA). Chem Station software B.04.03-SP2[108] (Agilent Technologies) was used for instrument control and data acquisition. A polyamide-coated fused silica capillary (Optronis, Kehl, Germany) with the following dimensions was used to perform the separation: 50 µm i.d. × 375 µm o.d. × 63 cm total length, 55 cm length to the detector. The separation was performed in a borate buffer (150 mM). The pH was adjusted to 10.5 with sodium hydroxide solution (1 M) using a Five Easy Standard pH meter (Mettler Toledo, Columbus, OH, USA). The running voltage was adjusted to + 25 kV, and the capillary was cooled to 12 °C, while the sample tray remained uncooled. Before each injection, the capillary was flushed (180 s, 50 mbar) with buffer. For injection, 35 mbar pressure was applied for 10 s.

High-performance liquid chromatography with UV detection (HPLC–UV)

For these works, a Hitachi Elite LaChrom high-performance liquid chromatograph consisting of a pump (L-2130), an autosampler (L-2200), and a diode array detector (L-2455) was used. A stainless-steel column (150 mm × 4.6 mm, 5 µm) filled with TC-C18 material (Agilent Technologies, Waldbronn, Germany) was used at room temperature. Solvent A was 0.1% (v/v) aqueous trifluoroacetic acid (TFA), and solvent B was a mixture of acetonitrile and water (90/10, v/v). A gradient was formed (0 min, 5% B; 2 min, 5% B; 35 min, 80% B; 37 min, 80% B; 40 min, 5% B; 45 min, 5% B) at a flow rate of 0.8 mL/min. The injection volume was 20 µL. The absorbance was read at 365 nm, and spectra were recorded between 220 and 400 nm during runs. Before analysis, photooxidized samples (100 µL) were mixed with 100 µL of a 0.1% (m/v) solution of 2,4-dinitrophenylhydrazine (DNPH) in 2 M hydrochloric acid. After mixing for 30 s, the samples were further incubated at room temperature for 5 min. Then, 300 µL acetonitrile and 500 µL of a mixture of double-distilled water, acetonitrile, and TFA (90/10/0.1, v/v/v) were added. After mixing, samples were filtered (0.2 µm) and analyzed directly.

High-performance liquid chromatography with electrospray ionization time of flight mass spectrometry (HPLC-ESI-ToF–MS/MS)

The LC–MS/MS analyses were performed on a UHPLC system (Agilent Technologies, Santa Clara, CA, USA) connected with a timsTOF system (Bruker Daltonik, Bremen, Germany). DataAnalysis software 5.3 (Bruker) was used for instrument control and data acquisition. An RP-18 column (Eurospher II, 100–3, 2 × 100 mm; Knauer, Berlin, Germany) was used. The injection volume was 1 µL, and the column flow was 0.2 mL/min. UV spectra were recorded between 210 and 450 nm. The mass spectrometer worked in the positive scan mode (m/z 20–1300; scan time 500 ms; dry gas flow 10 L N2/min; dry temperature 220 °C; nebulizer pressure 2.2 bar; capillary voltage 4500 V). Fragmentation spectra were recorded at collision energies between 20 and 50 eV.

For the analysis of Bz-Met and its degradation products, the mobile phases were 0.1% formic acid in water (solvent A) and 0.1% formic acid in a mixture of acetonitrile and water (90/10, v/v, solvent B). The proportion of solvent B was 20% at the beginning, which was increased to 67% within 17 min, and finally to 100% in 8 more min. It remained at 100% for 4 min and got back to 20% within 1 min, remaining at 20% for another 4 min prior to the next injection. Separations were performed at 40 °C. For the analysis of the tetrapeptides and their degradation products, the mobile phases were 0.1% formic acid in water (solvent A) and acetonitrile (solvent B). The proportion of solvent B was 2% at the beginning, which was increased to 15% during 18 min, then to 50% within 5 min and further to 100% in 2 min. It remained at 100% for 5 min and got back to 2% within 1 min, remaining at 2% for another 6 min prior to the next injection. Separations were performed at 25 °C.

Results

Stability of methionine and benzoyl methionine during irradiation

During irradiation with UV light of the amino acid methionine (Met) in the presence of riboflavin, Met is reported to be mainly degraded to methionine sulfoxide (MetSO) [7]. Furthermore, several volatile compounds are formed from methionine during photooxidation [25]. The present work is intended to focus entirely on reactions taking place at the thioether side-chain of methionine. The amino acid methionine has a free amino group that can participate in reactions that may not occur in protein-bound methionine. Therefore, the N-terminally protected derivative benzoyl methionine (Bz-Met) was chosen as a substrate for photooxidation reactions. The UV-active benzoyl group enables quantification of the compound by UV detection after different separation methods such as capillary electrophoresis (CE) or HPLC. In a first experiment, the dye-sensitized photostability of Bz-Met was assessed by coincubation with riboflavin in a sodium phosphate buffer at pH 7.4 at a wavelength of λ = 366 nm. Figure 2 shows the electropherograms at a detection wavelength of 230 nm after irradiation of Bz-Met with 10% riboflavin. Figure 2a shows the start of the irradiation (t = 0 min). Riboflavin could no longer be detected with CE after more than 5 min irradiation, even in a blank experiment, when only riboflavin was irradiated. The electropherogram in Fig. 2b shows the sample after 30 min of irradiation with three peaks that newly appeared after irradiation. The peak with a migration time of 15.4 min could be assigned to Bz-MetSO by comparison with a standard. The peaks at 7.7 min and 14.4 min could not be assigned to degradation products in the CE analysis. The products were later identified as benzoyl homocysteine (Bz-Hcy, product A) and benzoyl β-aspartic semialdehyde (Bz-β-ASA, product B) with LC–MS/MS (see “Identification of the photoreaction products”). In a blank experiment, where only Bz-Met was irradiated without riboflavin, no degradation of Bz-Met was observed.

Fig. 2
figure 2

Electropherograms recorded at 230 nm after a 0 min and b 30 min UV irradiation of Bz-Met with 10% riboflavin

Full size image

In the photooxidation experiments, Bz-Met was used as a model for protein-bound methionine. It was irradiated with UV light (λ = 366 nm) together with different concentrations of riboflavin for 4 h. Samples were taken during the irradiation and analyzed with CE. The detection wavelength was 230 nm, which is the absorption maximum of the benzoyl group at a pH value of 10.5. The detection takes place during separation (“on column”). Therefore, the peak areas have to be corrected by the migration time, as slower migrating compounds remain longer in the detection window and cause higher peak areas. For degradation kinetic analysis, the percentages refer to the corrected peak areas with 100% equal to the amount of Bz-Met prior to UV irradiation (t = 0 min). The degradation of Bz-Met and the formation of different oxidation products are shown in Fig. 3. Bz-Met (Fig. 3a) was degraded exponentially over about 1 h with a slight increase afterward, independent of the riboflavin concentration. Bz-MetSO (Fig. 3b) was formed rapidly from Bz-Met at the beginning of the photooxidation experiment. The formation slowed down during the irradiation and a slight decrease was observed at the end of the irradiation. The maximum amount of Bz-MetSO of 28–30% relative to the initial Bz-Met concentration was found between 1 and 2 h irradiation.

Fig. 3
figure 3

Photosensitized degradation of Bz-Met during coincubation with riboflavin. a degradation of Bz-Met, b formation of Bz-MetSO, c formation of product A, and d formation of product B. Percentages refer to the initial Bz-Met concentration (means ± SD, n = 3)

Full size image

The slight increase in the concentration of Bz-Met during longer irradiation times is consistent with a control experiment using Bz-MetSO instead of Bz-Met. This experiment showed a decrease of Bz-MetSO during irradiation with UV light. Bz-Met was detected after irradiation of Bz-MetSO with riboflavin after 30 min (10% riboflavin) and 2 h (3% riboflavin). Up to 3.2% Bz-Met was found after 4 h irradiation with 10% riboflavin. This shows that photoreduction can also take place under the experimental conditions chosen in this study. Furthermore, the formation of Bz-Met and product A (up to 7.0%) from Bz-MetSO was observed.

Product A (Fig. 3c) was formed more slowly during the irradiation of Bz-Met with riboflavin and reached 16–20% after 4 h irradiation. Product B (Fig. 3d) showed the biggest differences between concentrations of riboflavin used. The formation of product B started rapidly, for both concentrations. With 10% riboflavin, the formation reached an average maximum of 35% after 30 min and decreased to 27% until the end of the experiment. In contrast to that, a maximum of 44% after 60 min was reached for irradiation with 3% riboflavin. Thirty-seven percent of product B was still found after 4 h, which is higher than the maximum of the irradiation with 10% riboflavin.

Identification of the photoreaction products

In the present work, it was not possible to connect CE directly to MS due to the non-volatile separation buffer. Reaction products were identified by LC–MS/MS measurements. Bz-Met has a monoisotopic molecular mass of 253.0773 Da. The peak of the protonated molecular ion was found at a retention time of 8.3 min ([M + H]+exp. = 254.0846, [M + H]+theor. = 254.0845, Δm/z = 0.4 ppm) and can be seen in both Figs. 4 and 5. The oxidation product Bz-MetSO with its monoisotopic molecular mass of 269.0722 Da eluted at a retention time of 2.7 min ([M + H]+exp. = 270.0796, [M + H]+theor = 270.0795, Δm/z = 0.4 ppm).

Fig. 4
figure 4

LC–MS/MS measurement of Bz-Met irradiated for 30 min with or without riboflavin. Extracted wavelength chromatograms (λ = 245 nm) of irradiated Bz-Met with (blue) and without (orange) riboflavin after 30 min irradiation

Full size image
Fig. 5
figure 5

LC–MS/MS measurement of Bz-Met irradiated for 30 min with or without riboflavin. MS and MS/MS spectra of a Bz-Met without riboflavin and, b Bz-MetSO, c Bz-Hcy, and d Bz-β-ASA with riboflavin

Full size image

The broad peak eluting after Bz-MetSO (tR = 3.1 min) was further characterized. The UV spectrum extracted at 3.0 min showed a UV maximum at 233 nm. A peak with an m/z of 222.0763 was noticed in the mass spectrum extracted at 3.1 min. The extracted ion chromatogram (EIC) of m/z 222.0763 displayed the same broad shape as the peak in the UV chromatogram with a shift of ca. 0.1 min, which is due to the distance between the two detectors. The m/z of 222.0763 is equivalent to the sum formula C11H11NO4 (M = 221.0688 Da, calculated [M + H]+ = 222.0761, Δm/z = 0.9 ppm). This sum formula corresponds to benzoyl β-aspartic semialdehyde (Bz-β-ASA).

A third reaction product eluted in a sharp peak (tR = 3.3 min) directly after the broad peak. This peak has an m/z ratio of 240.0690, which is equivalent to the sum formula C11H13NO3S (M = 239.0616, calculated [M + H]+ = 240.0689, Δm/z = 0.4 ppm). This sum formula equals benzoyl homocysteine (Bz-Hcy). Unfortunately, the three oxidation products could not be sufficiently separated via HPLC measurement (Fig. 4). Therefore, CE had been employed for the measurements of the time course of degradation.

As depicted in Fig. 5, Bz-Met, Bz-MetSO, Bz-Hcy, and Bz-β-ASA have a common mass fragment with the molecular mass of 105.0329, which is the benzoyl fragment (M+  = 105.0327, Δm/z = 1.9 ppm). Therefore, all the three products must be degradation products of the methionine residue. For Bz-Met, Bz-β-ASA, and Bz-Hcy, no other main fragments were found in the MS/MS spectra. For Bz-MetSO, another fragment with an m/z of 206.0813 was found. This equals to a fragmentation between γ-C and S (M+ = 206.0812, Δm/z = 0.5 ppm). Fragments with a neutral loss of 64 Da have been reported as “diagnostic” for sulfoxides [26, 27]. The difference of 63.9984 Da seen in our experiment corresponds to the molecular formula CH4SO ([M + H]+theor = 63.9983, Δm/z = 0.4 ppm) and verifies the loss of methanesulfenic acid suggested previously [26].

The peak area ratio between Bz-β-ASA and Bz-Hcy in the UV chromatogram is about 3:1. This equals the ratio between product B and A from the CE measurements. Therefore, the broad peak of Bz-β-ASA can be assigned to product B from the CE experiments and the sharp peak of Bz-Hcy to product A. The thiol group of Bz-Hcy is deprotonated at the pH value of 10.5 during CE analysis. Thus, the compound is twice negatively charged and, therefore, must migrate to the detector in about half the time as compared to the singly charged components Bz-Met, Bz-β-ASA, and Bz-MetSO during electrophoresis.

When Bz-β-ASA is formed from Bz-Met by loss of the thiomethyl group, this implies that an aldehyde, namely formaldehyde, must result when Bz-Hcy is formed. In a preliminary experiment, formaldehyde was indeed detected in irradiated solutions of Bz-Met. When 1 mM Bz-Met was irradiated in the presence of 0.1 mM riboflavin, an amount of 0.15 ± 0.06 mM formaldehyde had been formed after 1 h (Fig. S1), equivalent to a conversion of ca. 15%. This is only slightly lower than the ca. 18% of product A (Hcy) that is formed in the same incubation after 1 h (Fig. 3). Riboflavin was degraded rapidly during irradiation with UV light. In the CE measurements, riboflavin could only be detected for up to 5 min irradiation, while it was detected for up to 30 min with LC–MS/MS (Fig. S2a). The main degradation product found was lumichrome. It was formed during the first 15 min of irradiation and degraded afterward (Fig. S2b).

Formation of β-aspartic semialdehyde (β-ASA) and homocysteine (Hcy) during UV irradiation of tetrapeptides

These experiments were performed to show that the observations from Bz-Met can be transferred to intact peptides and proteins. Two different tetrapeptides were used as models for the photooxidation of proteins. In these tetrapeptides, glycine residues were placed at the ends of the chain while methionine and alanine are in the center. The two tetrapeptides glycylalanylmethionylglycine (GAMG) and glycylmethionylalanylglycine (GMAG) were irradiated with UV light and samples were taken at 0 min and after 60 min of irradiation. LC-TOF–MS was used to search for the degradation products that are analogous to those from the irradiation of Bz-Met. Figures 6 and S3 show the extracted ion chromatograms of the respective ions at the beginning and after 60 min irradiation. MS and MS/MS spectra of those peaks are depicted in Fig. 7 for GAMG and Fig. S4 for GMAG. The theoretical and measured molecular masses can be found in Table 1. The results depicted in Fig. 7c and d indicate that methionine can also be oxidized to aspartic semialdehyde and homocysteine in peptides via irradiation with UV light.

Fig. 6
figure 6

LC–MS/MS measurement of GAMG irradiated with 10% riboflavin. EIC of GAMG (blue), GA-MetSO-G (orange), GA-Hcy-G (green), and GA-β-ASA-G (red) after a 0 min and b 1 h of irradiation

Full size image
Fig. 7
figure 7

LC–MS/MS measurement of GAMG photooxidized in the presence of riboflavin. MS and MS/MS spectra at 0 h of photooxidation for a GAMG and after 1 h photooxidation for b GA-MetSO-G, c GA-Hcy-G, and d GA-β-ASA-G

Full size image
Table 1 Theoretical and measured molecular masses of GAMG (Fig. 8) and GMAG (Fig S1) and their degradation products
Full size table

Discussion and conclusion

Methionine is degraded to only a small degree in the absence of photosensitizing compounds. Different degradation reactions have been described in the presence of riboflavin. As depicted in Fig. 8, triplet-state riboflavin can invoke the formation of a sulfuranyl cation from methionine by electron transfer in a type-1 photosensitized reaction [16, 28]. Alternatively, riboflavin can convert O2 to singlet oxygen (1O2), and 1O2 can react at the sulfur atom of the thioether group (type-2 photosensitized reaction, [16]).

Fig. 8
figure 8

Pathway of photosensitized degradation of methionine as proposed by Cardoso et al. [16] with the formation of β-aspartic semialdehyde and homocysteine

Full size image

Intermediates from both types of reactions may stabilize under the formation of hydroperoxides. A hydroperoxide at the γ-C atom of methionine was suggested as an intermediate in the light-induced off-flavor formation in milk. In addition to the formation of dimethyl disulfide and other small-molecular volatiles such as dimethyl sulfide and methanethiol, the formation of protein-bound β-ASA as the protein-bound degradation product of methionine was proposed [16]. A sulfur smell was recognizable during the irradiation experiments with Bz-Met in the present study. After UV irradiation of Bz-Met in the presence of riboflavin, the highest peak in the electropherograms (product B) is assigned to the mass of Bz-β-ASA (Figs. 2, 3, 5). This reaction is not restricted to Bz-Met, but also applies to methionine in peptides as models for protein-bound methionine (Figs. 6, 7, S3 and S4).

The formation of the hydroperoxide may also occur at the methyl-C atom (Fig. 8). The respective reaction products of this reaction would be homocysteine (Hcy) and formaldehyde [28]. The exact mass of Bz-Hcy is assigned to a peak in the UV chromatogram during LC–MS/MS measurements, which corresponded to ca. 12% of Bz-Met degradation after 30 min. Therefore, product A in the electropherograms (Fig. 2) is assigned to Bz-Hcy, because the peak area ratios between Bz-Hcy and Bz-β-ASA in the extracted wavelength chromatogram and between product A and B are equal (Figs. 2, 4). Furthermore, the thiol group of Bz-Hcy should be deprotonated at the pH of capillary electrophoresis. In comparison with Bz-Met and Bz-β-ASA which are singly negatively charged, this should result in half the migration time of the compound, because it is twice negatively charged. Using 2,4-dinitrophenylhydrazine (DNPH) derivatization, we also were able to quantify formaldehyde resulting from the reaction in selected samples. Formaldehyde and Bz-Hcy were formed in similar amounts (Fig. S1). Apparently less Bz-Hcy was formed in the oxidation experiments compared to Bz-β-ASA, (Fig. 2). We conclude that the reason for this product distribution is the lower stability of a primary radical as compared to a secondary radical that are generated from the sulfuranyl cation (Fig. 8).

Early reports on the light-struck flavor of milk suggested methional as the flavor compound behind this reaction [10]. However, this assumption must be disputed, because the production of methanethiol precedes the formation of methional in the course of UV irradiation [25]. Moreover, methional can only be formed from free, but not protein-bound methionine, because the peptide backbone would have to break. Methional could not be isolated from irradiated milk [13]. Peptide backbone cleavage may occur during UV irradiation and can be enhanced adjacent to methionine residues [29, 30], but today, it is predominantly suggested that volatiles from the degradation of side chains are responsible for off-flavor formation [16]. Volatiles occurring in the headspace of UV-treated milk include hydrogen sulfide, methanethiol, dimethyl sulfide, dimethyl disulfide (DMDS), and many others, while DMDS is the compound that is mainly responsible for the burnt off-flavor [25, 31,32,33].

With the results of our study, it becomes evident that it is important to look at the structures that remain in a protein whose degradation has led to the formation of volatiles. β-ASA and Hcy are more reactive than Met and might lead to intra- and intermolecular cross-linking via amino-carbonyl reactions or thiol-disulfide exchange in irradiated proteins. Protein-bound β-ASA is a hitherto disregarded component of “protein carbonylation” assessed by the reaction of DNPH with protein-bound carbonyl groups. The high susceptibility of methionine to photo-induced degradation reactions with the intermediate formation of hydroperoxides makes it probable for the reaction to also play a role in the degradation of proteins and peptides for therapeutic use. Mainly oxidation to methionine sulfoxide and sulfone was discussed so far [34], but the formation of reactive degradation products such as β-ASA and Hcy shows that further reactions, e.g., intra- or intermolecular cross-linking, are possible. Oxidation of methionine to MetSO and MetSO2 in an eye’s lens was discussed as to contribute to changes in protein structure during cataract formation [21]. The relevance of other light-induced methionine oxidation products, especially β-ASA and Hcy due to their susceptibility to cross-linking, should be checked as well.

UV irradiation in the presence of riboflavin was suggested as a preparative reaction for the synthesis of sulfoxides from thioethers [19]. The formation of β-ASA and Hcy as the primary photooxidation products of N-α protected methionine may represent a similarly useful approach for large-scale and industrial syntheses of the latter compounds, and it should be worthwhile to further optimize the reaction conditions in this direction.

In this report, a massive formation of β-ASA from methionine was found. For every molecule of β-ASA one thiyl radical is formed which can dimerize to DMDS. This shows the potential of this reaction for the formation of volatiles impressively. However, the same applies to the formation of Hcy and formaldehyde. The EFSA has commented on the safety of UV irradiation in the treatment of milk and milk products stating that “the formation of protein oxidation products following UV treatment of milk is not of safety concern” [22]. A reduction in the content of riboflavin as well as an increase in the concentration of methionine sulfone during UV irradiation was observed, whereas neither an increase in DMDS nor in methional was detected [35]. Formaldehyde, however, was not considered in that study and should be investigated further. The formaldehyde concentration in milk is normally 0.3 mg/L at the most, as long as it is not fraudulently added [36, 37]. UV irradiation may represent a further source of formaldehyde in milk.

Formaldehyde also may be one of the compounds that elicit the toxic effects of photooxidation products of methionine. Methional and acrolein were identified as two reaction products of riboflavin-sensitized methionine photooxidation that are toxic for bacteria [15]. In the investigated bactericidal mixtures, however, much higher concentrations of methional and acrolein would have been necessary to achieve the respective toxicity. Thus, beneath “oxygen radicals” as stated by the authors [15], formaldehyde may be a further candidate to explain the toxicity of these mixtures.

Summed up, beyond the works published so far, we have now been able to demonstrate that besides MetSO, Hcy and β-ASA are quantitatively dominant reaction products in photosensitized oxidation of methionine. In oxidation experiments, Bz-β-ASA accounted for ca. 30% of Bz-Met degradation and Bz-MetSO for ca. 25%, whereas Bz-Hcy accounted for ca. 16–20% of the degradation. These results give important new insights into the reaction pathways of protein oxidation elicited during UV irradiation: First, β-ASA is formed in substantial amounts and may represent a significant part of protein carbonylation in UV-irradiated proteins. Second, both Hcy and β-ASA may participate in cross-linking reactions in irradiated proteins. These assumptions will be the basis of our further works.