From top-down to bottom-up: Time-dependent monitoring of proteolytic protein degradation by LC-MS

Journal of Chromatography B, 1015 (2016) 111–120

Contents lists available at ScienceDirect

Journal of Chromatography B journal homepage: www.elsevier.com/locate/chromb

From top-down to bottom-up: Time-dependent monitoring of proteolytic protein degradation by LC-MS Joanna Tucher, Tomas Koudelka, Jana Schlenk, Andreas Tholey ∗ AG Systematische Proteomforschung & Bioanalytik, Institut für Experimentelle Medizin, Christian-Albrechts-Universität zu Kiel, Germany

a r t i c l e

i n f o

Article history: Received 14 December 2015 Received in revised form 11 February 2016 Accepted 14 February 2016 Available online 17 February 2016 Keywords: Proteolysis Cleavage product Lysozyme LC-MS Disulfide bond Intermediates

a b s t r a c t The understanding of proteolytic processes includes manifold aspects, ranging from the characterization of proteases and their corresponding substrates to the localization of cleavage sites. The analysis of protease-catalyzed reactions at a single time-point in many cases excludes the identification of intermediate cleavage products of potential biological function. To overcome this problem, proteolysis has to be monitored over time. For that purpose, we established a straight-forward two-step approach. First, Tricine-SDS-PAGE separation of the proteolytic products is applied to survey the proteolytic reaction. In the second step, the reaction mixture is analyzed by an LC-MS set-up. An optimized chromatographic separation coupled to electrospray Orbitrap mass spectrometry allowed the simultaneous monitoring of intact substrates, intermediates and cleavage products of lower molecular weight. The applicability of the strategy was shown on the example of the gastric protease pepsin and its physiologically relevant substrate hen egg white lysozyme, one of the major egg allergens. While lysozyme-derived cysteine-free peptides were cleaved comparatively fast, disulfide bonds protected connected peptides from rapid peptic proteolysis. Two previously identified potential IgE-binding motifs were observed as disulfide-linked cleavage products. In summary, the presented approach is not only ideally suited for the simulation of gastro-intestinal digestion, which is of high interest in food research, but can be transferred to any protease-substrate pair of interest. Furthermore the strategy can be exploited to deduce the effect of post-translational modifications on proteolysis. © 2016 Elsevier B.V. All rights reserved.

1. Introduction The protease-catalyzed hydrolysis of peptide bonds represents a key event in a variety of biological processes, such as blood-clotting, tissue development, immune responses and gastrointestinal digestion. Beside the knowledge concerning the nature and identity of a protease and its corresponding substrates, the identification of cleavage sites is one of the major goals in degradomics [1] as these determine the molecular properties of the cleavage products and hence the biological function thereof. If there is more than one cleavage site within a protein, new questions regarding intermediate proteolytic products with kinetically preferred cleavage sites arise as these can exhibit other biological

∗ Corresponding author at: AG Systematische Proteomforschung & Bioanalytik, Institut für Experimentelle Medizin, Christian-Albrechts-Universität zu Kiel, Niemannsweg 11, 24105 Kiel, Germany. E-mail address: [email protected] (A. Tholey). http://dx.doi.org/10.1016/j.jchromb.2016.02.021 1570-0232/© 2016 Elsevier B.V. All rights reserved.

activities. To unravel this issue it is necessary to monitor proteolytic reactions over time. Several studies have performed time-course analyses of proteolytic reactions in the context of highly complex samples employing N-terminomics approaches. Three time-points of human granzyme B catalyzed proteolysis on Jurkat cell lysates were investigated employing SILAC (stable isotope labeling by amino acids in cell culture) and the N-terminal COFRADIC (combined fractional diagonal chromatography) protocol [2]. Caspase-dependent proteolysis on Jurkat cell lysates was studied by SRM (selected reaction monitoring) of approximately 1000 peptides [3]. An 8plex-iTRAQ-TAILS (terminal amine isotopic labeling of substrates) approach was applied to explore degradation kinetics of MMP-10 (matrix metalloproteinase-10) on fibroblast secretomes [4]. However, besides long and error-prone sample preparation steps, which include chemical derivatization or subsequent enrichment strategies to reduce sample complexity, N-terminomics approaches do not allow for the identification of full length cleavage products as only N-terminal peptides are monitored after digestion of cleavage

112

J. Tucher et al. / J. Chromatogr. B 1015 (2016) 111–120

Fig. 1. Analytical strategy. The protease and substrate of interest are incubated at particular reaction conditions. After sampling at specific time-points, the protease is irreversibly inactivated and samples are subjected to Tricine-SDS-PAGE (non-reducing and reducing conditions) for a first survey. Relevant samples are further analyzed by LC-MS (disulfide linkages maintained and reduced).

products by a secondary (work) protease. An alternative method suitable for time-course analyses, PROTOMAP (protein topography and migration analysis platform) [5], is based on SDS-PAGE separation of protease treated samples of different incubation times, in-gel digestion and subsequent LC-MS analysis. The sequence coverage of substrates and cleavage products is displayed in peptographs including semi-quantitative data obtained by spectral counting. Although this strategy is able to identify full-length cleavage products, the assignment of N- and C-termini remains challenging as target protease specificities might be unknown and bottom-up approaches usually result in a loss of information. Topdown proteomics, based on LC separation and mass spectrometric analysis at the level of intact proteins, is an upcoming approach suitable to circumvent some of the aforementioned pitfalls. In particular, proteins up to around 25 kDa can be routinely analyzed with sufficient sensitivity [6]. In the present work, we developed a strategy for the simultaneous LC-MS analysis of the intact protein substrate and the cleavage products formed at different time-points: a combined top-down and bottom-up LC-MS approach without any derivatization steps. In order to determine the most relevant time-points, the LC-MS analysis is preceded by gel electrophoresis. The strategy offers the direct identification of cleavage sites and enables to monitor the

influence of post-translational modifications, such as disulfide linkages, on proteolysis. The applicability of the method was shown on the example of pepsin and hen egg white lysozyme (LYZ); a protease-substrate pair of biological significance. Pepsin, a protease located in the human stomach, catalyzes the first step of food protein degradation under acidic conditions. The proteolytic products subsequently enter the small intestine and are further decomposed by pancreatic proteases, such as trypsin and chymotrypsin. Hen egg white lysozyme is a 14.3 kDa hydrolytic enzyme of 129 amino acids stabilized by four disulfide bonds. It is not only found in hen egg white, but variants of this enzyme are also present in different biological fluids, such as human milk, saliva and tears and in numerous other organisms [7]. The enzyme decomposes the peptidoglycan layer of bacterial cell walls, in particular of Gram-positive bacteria, and consequently possesses antimicrobial activity. This feature is not only exploited in nature but also in food technology, e.g. as preservative on cheese rind. Additionally, LYZ was shown to contain antimicrobial peptide motifs [8–10]. On the other hand the protein is one of the major egg allergens, next to ovalbumin or ovomucoid, and particularly infants suffer from allergic reactions thereof [11,12]. However, in most cases the allergy disappears with growing age. One reason is the drop in pH of the stomach from approximately 4 to 2 [9], which increases the susceptibility of the allergen to peptic

J. Tucher et al. / J. Chromatogr. B 1015 (2016) 111–120

113

Fig. 2. LC-UV analysis (214 nm) of a five component mixture of a synthetic peptide (Peptide, KLVNRRSEFSALTPASR), insulin (Ins), cytochrome c (CytC), LYZ and ␤-lactoglobulin (bLG) at 60 ◦ C using a (A) PepSwift monolithic, (B) C-4 or (C) C-8 trapping column in combination with a ProSwift monolithic analytical column. *Denotes that the assignment of insulin and cytochrome c was not possible due to retention time shift of single components.

114

J. Tucher et al. / J. Chromatogr. B 1015 (2016) 111–120

digestion. Hence, the probability of the allergen or large proteolytic fragments thereof reaching the intestinal mucosa decreases. Accordingly sensitization of the immune system and IgE-binding, which mostly mediates allergic reactions, is prohibited. The example of pepsin and LYZ investigated here shows the high potential of the presented analytical approach in the context of food allergen research, as the monitoring of proteolysis over time enables the investigation of food allergen stability and allows for an identification of proteolytic products. 2. Material and methods 2.1. Chemicals Acrylamide-Bisacrylamide (30%), TEMED, Tris base, Tris-HCl, tricine as well as SDS were obtained from Carl Roth (Karlsruhe, Germany). The 2 kDa peptide (KLVNRRSEFSALTPASR) contained in the self-assembled protein mass standard for Tricine-SDS-PAGE and the peptide ITASVNCAKKIVSDGNGMNA were synthesized by Biosyntan (Berlin, Germany). All other chemicals and proteins, including hen egg white LYZ (L6876), porcine pepsin (P6887, 4220 U/mg), bovine catalase, bovine carboanhydrase, bovine ␤-lactoglobulin, monellin (Dioscoreophyllum cumminsii), equine cytochrome c and bovine insulin, were purchased from SigmaAldrich (Steinheim, Germany). Deionized water (18.2 M/cm) was prepared by the arium611 VF system from Sartorius (Göttingen, Germany). 2.2. Optimization of the LC column combination For the simultaneous separation of analytes of higher and lower molecular weight a monolithic analytical column (ProSwift RP4H, 100 ␮m × 250 mm, Thermo Fisher Scientific, Bremen, Germany) was used. As no trapping column of the identical stationary material was commercially available, three different trapping columns were tested in combination with the analytical column: (i) a monolithic (PepSwift, 200 ␮m i.D. × 5 mm, Thermo Fisher Scientific), a C-4 (Acclaim PepMap C-4, 300 ␮m i.D. × 5 mm, 5 ␮m, 300 Å, Thermo Fisher Scientific) and a C-8 (Acclaim PepMap C-8, 300 ␮m i.D. × 5 mm, 5 ␮m, 100 Å, Thermo Fisher Scientific) pre-column. Each column combination was evaluated by the separation of 0075 ␮g of a 1:1:1:1:1 (w/w) mixture of a 2 kDa synthetic peptide, bovine insulin, equine cytochrome c, hen egg white LYZ and bovine ␤lactoglobulin on a U3000 nano-LC-UV-system (Dionex, Idstein, Germany). The mixture was injected three times on the same column set-up to monitor reproducibility. Additionally 0.015 ␮g of each single analyte was injected. The column oven temperature was set to 60 ◦ C and the UV trace was monitored at a wavelength of 214 nm. Samples were first desalted on the pre-column for 4 min at a flow rate of 15 ␮L/min (i) or 30 ␮L/min (ii, iii), respectively, with 3% ACN/0.1% TFA. Afterwards analytes were separated on the analytical column applying eluent A (0.05% TFA) and B (80% ACN/0.04% TFA) at a flow rate of 500 nl/min: 4–4.1 min, from 5 to 20% B; 4.1–30 min, from 20 to 80% B; 30–31 min, from 80 to 100% B; 31–47 min, constant at 100% B; 47–48 min, from 100 to 5% B; 48–60 min constant at 5% B. 2.3. Peptic degradation of LYZ LYZ was dissolved in 30 mM NaCl to a concentration of 2 mg/mL. Two aliquots were adjusted to pH 2.0 or pH 4.0 applying 1.0 M or 0.1 M HCl, respectively. The pH was verified by a micro electrode (InLab Micro Pro, Mettler Toledo, Columbus, USA). Porcine pepsin (844 U/mg LYZ) was dissolved in 30 mM NaCl of pH 2.0 or 4.0 and was accordingly added to the LYZ samples of pH 2.0 or 4.0 to an enzyme-to-protein ratio (w/w) of 1:5. Samples and corresponding

Fig. 3. Tricine-SDS-PAGE. LYZ incubated with pepsin at pH 2 under non-reducing (A) and reducing (B) conditions, and at pH 4 under non-reducing (C) and reducing (D) conditions. Abbreviations: lysozyme, L; pepsin, P; protein mass standard, M.

J. Tucher et al. / J. Chromatogr. B 1015 (2016) 111–120

controls of LYZ and pepsin were incubated at 37 ◦ C (LYZ concentration: 1.5 mg/mL). After 1 min, 5 min, 15 min, 30 min, 1 h, 2 h, 4 h and 24 h of incubation, 160 ␮L of the samples were added to 75 ␮L of 160 mM sodium carbonate (pH 12.0) to irreversibly inactivate pepsin at room temperature. To analyze effects of disulfide bonds on the proteolysis, LYZ was reduced in 25 mM TCEP-HCl at 56 ◦ C for 1 h prior incubation with pepsin at pH 2 for 1 h and 24 h at 37 ◦ C. Pepsin was inactivated as described above and samples were analyzed by LC-MS. 2.4. Tricine-SDS-PAGE Tricine-SDS-PAGE was carried out according to a previously published protocol [13,14]. Briefly, 16% gels were employed to resolve low molecular weight proteolytic fragments of LYZ under non-reducing or reducing conditions. A mixture of bovine catalase, carboanhydrase, ␤-lactoglobulin, hen egg white LYZ, monellin (Dioscoreophyllum cumminsii) and a 2 kDa synthetic peptide was prepared as a protein mass standard. The protein mass standard (5 ␮g per lane) and LYZ samples (10 ␮g per lane, vacuumconcentrated to 5 ␮L) were mixed 1:1 (v/v) with reducing or non-reducing sample buffer (100 mM Tris-HCl, pH 6.8; 1% (w/v) SDS; 0.02% (w/v) Coomassie Brilliant Blue G; 24% (w/v) glycerol and 4% (v/v) ␤-mercaptoethanol (under reducing conditions)). After incubation at 70 ◦ C for 15 min, samples were cooled on ice for 5 min and then applied into the wells of the stacking gel. Electrophoresis was carried out at a constant voltage of 125 V. Protein and peptide bands were stained with 0.25% (w/v) Coomassie Brilliant Blue R in 10% acetic acid. 2.5. LC-MS Based on the results obtained by Tricine-SDS-PAGE, only LYZ samples that were incubated with pepsin at pH 2 were subjected to LC-MS analysis. Samples were measured under two conditions: non-reduced (samples diluted to 0.01 mg/mL with 0.01% TFA) and reduced, to facilitate the identification of disulfide-linked cleavage products. Disulfide bonds were reduced at pH 7.0 (50 mM ammonium bicarbonate) using 25 mM TCEP-HCl for 1 h at 37 ◦ C. Afterwards samples were diluted to 0.01 mg/mL with 0.25% TFA.

115

Controls of LYZ and pepsin were only analyzed without a prior reduction step. For LC separation the Dionex U3000 nano-LC system (Dionex) was equipped with a C-8 trapping column (see above) and a monolithic analytical column (see above). The column oven temperature was set to 60 ◦ C and the UV trace was monitored at a wavelength of 214 nm. All samples were injected once with the exception of the reduced sample-set of LYZ incubated with pepsin at pH 2.0, which was measured in duplicate to monitor reproducibility. Prior to analyte separation on the monolithic column, 0.025 ␮g of sample was desalted on the trapping column at a flow rate of 30 ␮L/min with 3% ACN/0.1% TFA. After 4 min, analytes were eluted with a flow rate of 500 nL/min onto the analytical column applying the following conditions: eluent A (0.05% formic acid (FA)) and B (80% ACN/0.04% FA); 4.0–4.1 min, from 5 to 10% B; 4.1–45 min, from 10 to 70% B; 45–46 min, from 70 to 95% B; 46–56 min, constant at 95% B; 56–57 min, from 95 to 5% B; 57–70 min, constant at 5% B. The LC-system was coupled online to a LTQ Orbitrap Velos mass spectrometer equipped with LTQ Tune Plus 2.7.0 and XCalibur 2.2 (all Thermo Fisher Scientific). In the tune file two micro-scans and a maximum injection time of 1000 ms were defined for MS and MSn acquisition. MS data was recorded between 4 and 70 min applying four scan events. A MS full scan between 500 and 2000 m/z was acquired at a resolution of 100,000 in the Orbitrap mass analyzer. Following, the three most intense precursors with charge states ≥ 2+ were subjected to HCD fragmentation (normalized collision energy, 35%; isolation width, 3 m/z; activation time, 0.1 ms). Unassigned charge states were not excluded from fragmentation. Fragment ions were recorded in the Orbitrap mass analyzer at a resolution of 60,000. Dynamic exclusion with the following settings was applied for fragmentation: repeat count, 2; repeat duration, 30 s; exclusion duration, 180 s; precursor mass tolerance, 10 ppm. As 1+ charged precursor ions were observed during data acquisition, the reduced 2 h as well as the 24 h sample (pH 2.0) were remeasured without the exclusion of 1+ charged ions. 2.6. Data interpretation LC-MS data were analyzed using Proteome Discoverer 1.4 (Thermo Fisher Scientific) and the search algorithm SEQUEST for

Fig. 4. Map of 86 identified cleavage products of LYZ in the reduced sample-set after incubation with pepsin at pH 2 (data of all time-points combined). Cysteines connected via disulfide bonds are highlighted in the same color. In total 43 unique cleavage sites of pepsin were identified as a result of the broad cleavage site specificity of the enzyme.

116

J. Tucher et al. / J. Chromatogr. B 1015 (2016) 111–120

peptide identification. Data from non-reduced samples were also searched by SEQUEST to identify peptides without disulfide linkages. The FASTA database consisted of the reviewed, canonical protein sequences of Gallus gallus (downloaded 15.01.15, UniProt) and of porcine pepsin (2263 sequences in total). Search parameters were set as follows: enzyme specificity, no enzyme; precursor mass tolerance, 10 ppm; fragment mass tolerance, 0.02 Da; no fixed modification; variable modification, oxidation (M). Only peptides of high confidence (1% FDR) and a search engine rank of 1 were considered. In addition only peptides with at least three peptide spectrum matches over all time-points were included. The identity of the most intense signals of the non-reduced sample-set at pH 2.0 was determined by manual MS/MS spectra interpretation. To monitor the development of signal intensities of reduced and non-reduced cleavage products over time, the maximum intensity of the most abundant charge state for every time-point was extracted from the mass spectra manually. Hereby, the charge state was fixed over time and the most intense signal (denoting the most intense isotopic peak of a certain peptide) was extracted as the maximum signal intensity. Signal intensities of the reduced samples, measured in two technical replicates per time-point, were averaged. Relative signal intensities were calculated and data was visualized as heatmaps by conditional formatting in Excel. 3. Results and discussion 3.1. Analytical strategy For the time-dependent monitoring of proteolytic processes we developed an analytical strategy composed of two major steps (Fig. 1). After the incubation of the protease and substrate of interest for certain time intervals and the subsequent inactivation of the protease, the first step encompasses a fast screening of samples by Tricine-SDS-PAGE. This separation technique enables the resolution of low molecular weight analytes down to approximately 1 kDa and permits the identification of relevant samples for LCMS analysis. The LC-MS set-up was optimized to allow the parallel measurement of the intact substrate as well as of higher, medium and smaller sized cleavage products. 3.2. Optimization of the LC column combination

Fig. 5. Development of relative signal intensities over time of intact LYZ and 86 cleavage products identified in the reduced sample-set of LYZ incubated with pepsin at pH 2. The color code allows for an estimation of the most abundant cleavage products according to the maximum absolute signal intensity identified over time. [M+H]+ refers to the monoisotopic mass. *Denotes the [M+H]+ of the theoretical highest peak within the isotope pattern as the monoisotopic peak can usually not be assigned for higher molecular weight analytes. The heatmap shows an early appearance of large proteolytic fragments, which get sequentially degraded into smaller cleavage products over time.

For the simultaneous nano-LC separation of peptides and proteins (below 25 Da) the choice of suitable LC stationary phases is of fundamental importance. We chose a monolithic ethylvinylbenzene/divinylbenzene analytical column (ProSwift) as it was previously shown, that monolithic columns deliver excellent separation of both peptides and proteins [15,16]. In order to simplify the whole analytical set-up, in particular to allow a direct analysis of proteolytic reaction mixtures without additional cleaning steps, the use of pre-columns is mandatory. To our knowledge, a trapping column of the same stationary material and degree of cross-linking as the applied analytical column, which would have been the usual set-up, was not commercially available. Hence, three trapping columns with other stationary materials were tested: (i) a monolithic trapping column of a higher degree of crosslinking compared to the analytical column, (ii) a C-4 and (iii) a C-8 trapping column. We tested the three column combinations on a nano-LC-UV-system by separating a 1:1:1:1:1 (w/w) mixture of a synthetic peptide (2 kDa), insulin (6 kDa), cytochrome c (12 kDa), LYZ (14 kDa) and ␤-lactoglobulin (18 kDa). Samples were first desalted and concentrated on the trapping column and were subsequently eluted and separated on the analytical column with an identical gradient. Triple injections of the peptide/protein mixture

J. Tucher et al. / J. Chromatogr. B 1015 (2016) 111–120

117

Fig. 6. Identified cleavage products of the non-reduced sample-set of LYZ digested with pepsin at pH 2. Disulfide-linked peptides (A) Cys-6 → Cys-127, (B) Cys-30 → Cys-115 and (C) Cys-64 → Cys-80, Cys-76 → Cys-94. (D) Peptides without disulfide bonds. (E) Monitoring of relative signal intensities over time; peptide numbers refer to numbering in (A)–(D); color code according to the maximum absolute signal intensity identified over time. The heatmap (E) indicates a slower peptic degradation for disulfide-linked cleavage products (A–C); whereas cysteine-free peptides (D) are degraded comparatively fast.

proved reproducible and peak widths were comparable between the three set-ups (Fig. 2). The retention times of all mixture components shifted towards later time-points using trapping columns (i) to (iii) (Fig. 2A–C), which is caused by the increasing percentage of hydrophobic solvent B needed for elution from the pre-column to the analytical column . The data also indicate a comparably low analyte retention by the monolithic trapping column, which could be problematic concerning peptides of low molecular weight. The best separation, especially regarding insulin, cytochrome c and LYZ, was achieved in combination with the C-8 trapping column. In addition single injections of the five components revealed instable retention times regarding the monolithic and C-4, but not the C-8 trapping column (data not shown). This observation might be explained by matrix effects, e.g. peptide/protein interactions which alter the retention times of single components in the context of additional analytes. In conclusion, we opted for the C-8 trapping column, which is further supported by the theoretical suitability of C-8 reversed phase stationary material and the size of analytes expected in the peptic digest of LYZ. 3.3. Monitoring of proteolysis by Tricine-SDS-PAGE Tricine-SDS-PAGE is a helpful tool to quickly assess the extent of proteolytic degradation. A slightly modified and comparatively simple version [13,14] of the original protocol [17,18] was applied for that purpose. Pepsin and LYZ were incubated at an enzyme-tosubstrate ratio of 1:5 (w/w) at pH 2 and 4 to compare proteolytic degradation of LYZ under conditions present in the adult and infant stomach, respectively. After 1 min, 5 min, 15 min, 30 min, 1 h, 2 h, 4 h and 24 h aliquots of the reaction mixture were taken and pepsin was irreversibly inactivated by increasing the pH above 8 [19]. Tricine-SDS-PAGE gels show a clear difference between peptic proteolysis of LYZ at pH 2 and pH 4 (Fig. 3). Whereas LYZ is completely degraded after 4 h at pH 2, only a small amount of proteolytic fragments was detected after 24 h at pH 4. This observation is in

agreement with the low proteolytic activity of pepsin at pH 4, which has its pH optimum around 2 [19]. Nevertheless it was shown, that porcine pepsin releases antimicrobial peptides by the proteolytic processing of hen egg white LYZ at pH 4 [9]. The authors subsequently transferred the results to human LYZ, which is secreted in multiple tissues and fluids such as breast milk. They concluded that this mechanism protects the newborn from bacterial infections. However, our data could not confirm a significant and comparable degradation of LYZ at this pH. Another study [20] could likewise not detect proteolytic fragments of LYZ after peptic digestion at pH 3.2. On closer inspection of the gels obtained from the incubation of pepsin and LYZ at pH 2 (Fig. 3A, B) it can be observed that the overall molecular weight of proteolytic fragments decreases between 2 and 24 h. Three bands of approximately 2, 4 and 6 kDa, which appear after 5 min of incubation under non-reducing conditions, become increasingly blurred after 30 min, whereas under reducing conditions two bands of about 2 and 4 kDa stay largely defined until 24 h. However, after 24 h the intensity of the band at 4 kDa diminishes, whereas a band at about 1 kDa becomes more intense. It has to be noted, that gels under non-reducing conditions do not allow for precise molecular weight estimation. Disulfide bonds prevent the complete linearization of peptides or proteins and consequently can affect the electrophoretic migration through the gel. In a further experiment the influence of heat denaturation on the susceptibility to peptic digestion was studied. LYZ was heated to 90 ◦ C for 10 min prior to adjustment of the pH to 2 and 4. Besides a slight increase in the degradation of LYZ at pH 4, no major difference could be depicted at pH 2 compared to the digestion pattern observed for non-heated LYZ (Fig. A.1). In conclusion heating of LYZ, e.g. during food processing, at this specific condition, does not affect its digestibility. For ovalbumin it was shown that heating at 100 ◦ C for 30 min highly increases its susceptibility to pepsin, whereas this effect was not observed for ovomucoid [21].

118

J. Tucher et al. / J. Chromatogr. B 1015 (2016) 111–120

Controls of LYZ and pepsin showed no major contaminants or proteolytic fragments by self-digestion after 1 min or 24 h of incubation (Fig. A.2). In summary, Tricine-SDS-PAGE allows for a fast and simple screening of proteolytic reactions and for the identification of relevant samples for LC-MS analysis, hereby restricting the measurement time. In respect to the example of LYZ and pepsin, the hydrolysis rate was significantly more pronounced at pH 2 compared to pH 4; therefore, LC-MS measurements concentrated on proteolysis at pH 2. 3.4. Time-dependent analysis of peptic LYZ degradation at pH 2 by LC-MS All time-points of the LYZ incubation at pH 2 were analyzed by high resolution LC-ESI LTQ Orbitrap MS. As we only intended to analyze the degradation of a comparatively small protein, we did not include procedures known to improve top-down MS (e.g. the addition of charge reducing agents [22]) and instead applied a standard set-up widely adopted in bottom-up proteomics. The LC-MS analysis of the control of pepsin incubated for 24 h at pH 2 identified ten peptides derived from autoproteolysis of the protease. However, the signals of these peptides were of low intensity, showing that autoproteolysis was only a minor side reaction, which is in agreement with the results obtained by Tricine-SDSPAGE (Fig. A.2). We investigated the degradation products which were formed upon action of pepsin on non-reduced LYZ in order to monitor the influence of the presence of disulfide bonds on proteolysis. The LCMS analysis of the protein digestions was carried out in two series: in the first one, a reduction but no alkylation step was included prior to LC-MS analysis. In the second series reaction products were analyzed without reduction. In the reduced sample-set a total of 86 peptides with at least three peptide spectrum matches of high confidence were identified during the time-course analysis, which covered the whole sequence of LYZ (Fig. 4). The high number of proteolytic products is attributed to the broad cleavage site specificity of the enzyme. Peptides with molecular weights ranging from 738 (23 YSLGNW28 ) to 9277 Da (1 KVFGR· · ·PCSAL83 ) were observed. It is to be expected that more peptides are formed, in particular at later incubation times, which are not detected as they fall below the lower limit of the mass range (500 m/z). In total 43 unique cleavage sites were identified (Fig. A.3A). At these cleavage sites, alanine (19%), followed by asparagine, glycine and leucine were most frequently observed at the P1 position (Nterminal amino acid of the cleavage site) (Fig. A.3B). Regarding the P1 site (C-terminal amino acid of the cleavage site), alanine with 19% was slightly preferred, followed by tryptophane, asparagine, serine and valine. The MEROPS database of peptidases [23] reported over 400 pepsin cleavage events: here, phenylalanine (21%) followed by leucine (16%) are favored at the P1 site, whereas at P1 leucine (12%), alanine (12%) and valine (10%) are slightly preferred (Fig. A.3C). Both datasets prove the broad cleavage site specificity of pepsin. We extracted the retention time for each identified peptide for every time-point in the reduced sample-set. The average standard deviation of the retention time of a peptide in this study was 7 s. The high reproducibility enables an easy read-out and identification (together with the peptide mass) of the same peptide reoccurring at different time-points. In order to visualize the occurrence of identified peptides over time, relative signal intensities were displayed in a heatmap (Fig. 5). The time-course profile of the proteolytic events shows an early appearance of rather large fragments, such as 1 KVFGR· · ·PCSAL83 (9277 Da) or 57 QINSR· · ·NGMNA107 (5437 Da), which are degraded into smaller products at later time-points.

From the overall pattern it appears predominantly as if once a cleavage at a preferred site occurs, pepsin sequentially releases amino acids from the N- and C-terminal parts of the remaining peptide. This exopeptidase-like activity of pepsin is in particular noticeable for disulfide-containing cleavage products starting from the N-terminal positions 57, 63, 88, 90, 91, 108 and 109. To rule out acid catalyzed hydrolysis of peptide bonds, the synthetic LYZ peptide 88 ITASVNCAKKIVSDGNGMNA107 was incubated under the same conditions (pH 2, 37 ◦ C) without the addition of pepsin. No degradation products were observed after 24 h (Fig. A.4). In order to identify disulfide-linked proteolytic products, we analyzed the same time-course profile but without a prior reduction step. The most abundant signals over time were interpreted manually and relative signal intensities were displayed in a heatmap (Fig. 6). The most intense signals of the non-reduced sample-set derived from peptides connected via disulfide bonds between Cys-6 and Cys-127 (Fig. 6A), Cys-30 and Cys-115 (Fig. 6B) and proteolytic fragments containing simultaneously one intrapeptide (Cys-64 and Cys-80) and one inter-peptide (Cys-76 and Cys-94) disulfide linkage (Fig. 6C) as well as peptide sequences without cysteine residues (Fig. 6D). The manual spectra interpretation of disulfide-linked peptides was facilitated by the knowledge of the peptides identified in the reduced sample-set giving a first clue about probable peptide combinations. Furthermore, similar retention times often indicated structural commonalities, e.g. for peptides 1, 2, 3 and 4, which are characterized by the sequential loss of three alanines from the Cterminus of peptide 1 KVFGRCELAAA11 (Fig. 6A). Spectra of peptides connected by the same disulfide bond often showed similar HCD fragmentation patterns, which once more facilitated data interpretation. For instance, the same a- and b- or y-ions were detected in the lower molecular weight region, and characteristic dehydroalanine or persulfide species of the disulfide-linked peptides were observed (Fig. A.5). It has to be noted, that not all of the linked counterparts were identified in the reduced sample-set as some of them were out of the acquired m/z range. In general, pepsin is recommended for the determination of unknown disulfide bonds as the low pH prevents disulfide scrambling. Nevertheless, the protease exhibits a broad specificity profile, which increases the amount of possible peptide combinations compared to a highly specific protease such as trypsin [24]. However, knowledge of sequential cleavage of several amino acids from the N- or C-terminus of disulfide-linked peptides over time, as it was observed for LYZ, can facilitate data interpretation. Database searches of non-reduced samples identified cysteine-containing peptides such as 1 KVFGRCELAA10 or 91 SVNCAKKIVSDGNGMNA107 , which was surprising as these peptides should theoretically be connected to other cysteine peptides via disulfide bonds. While sample carry-over could be excluded, a closer inspection of the XIC of the according m/z values led to an interesting observation. The peptides mentioned above artificially emerged in the mass spectra during the elution of their disulfide-linked counterparts; a similar phenomenon was observed earlier in MALDI MS resulting from in-source decay [25]. The development of the TIC over time clearly shows increasing and decreasing signals, e.g. the disappearance of intact LYZ after 1 h (Fig. 7E) or the emergence of peptide 11 (28 WVCA31 → 111 WRNRCKGTD119 ) after 4 h (Fig. 7G). The timeresolved monitoring of relative signal intensities clearly depicted the sequential degradation of larger fragments to smaller proteolytic products at 24 h for disulfide-linked peptides (Fig. 6E). In contrast cleavage products without disulfide bonds were degraded comparatively fast. To confirm the protective effect of disulfide bonds on proteolysis we digested prior reduced LYZ with pepsin at the same conditions. After 1 h only small peptides (mainly 1+

J. Tucher et al. / J. Chromatogr. B 1015 (2016) 111–120

119

Fig. 7. TIC of the non-reduced sample-set of LYZ digested with pepsin at pH 2: (A) 1 min, (B) 5 min, (C) 15 min, (D) 30 min, (E) 1 h, (F) 2 h, (G) 4 h and (H) 24 h. Peak labels refer to the peptide numbering assigned in Fig. 6. The TIC show the sequential degradation of intact LYZ and the occurrence and disappearance of cleavage products over time. The TIC of the reduced sample-set of LYZ digested with pepsin at pH 2 ((I) 1 h (J) 24 h) and the TIC of prior reduced LYZ digested with pepsin at pH 2 ((K) 1 h, (L) 24 h) show a different proteolytic pattern caused by the presence or absence of disulfide bonds during proteolysis. TIC of LYZ incubated without pepsin at pH 2 for 24 h (M). Abbreviation: lysozyme, L.

and 2+ charged) remained and a completely different proteolytic pattern was observed (Fig. 7 K, L, Fig. A.6). The identified proteolytic products and cleavage sites in our study are mostly in agreement with the results of Jiménez-Saiz et al. [20]. In this earlier study, peptides with masses between 803 and 3077 Da were identified after reduction at a time-point of 60 min and an enzyme-to-protein ratio of 1:20 (w/w). After Tricine-SDS-PAGE of the digest and subsequent Western-blotting, the authors identified three possible IgE-binding peptide sequences out of a single gel-band at approximately 4 kDa by in-gel digestion and MALDI MS analysis: (i) 11 AMKRHGLDNYRGYSLGN27 (1951 Da), (ii) 57 QINSRWWCNDGRTPGSRNLCNIPCSAL83 (3060 Da) and (iii) 108 WVAWRNRCKGTDVQA122 (1789 Da). Hence, from these data it could not unambiguously be deduced which of these peptide motifs is responsible for the IgE-binding. Interestingly, we noticed intense signals for peptides (ii) and (iii) over a broad time span in the reduced samples (Fig. 5). In addition, these peptides were identified as disulfide-linked species (peptides 5, 7, 14 and 15 in Fig. 6). Maximum relative intensities for these sequences were observed after approximately 1 h. In contrast to peptides (ii) and (iii), peptide (i) does not contain a cysteine residue and hence no disulfide bond, which might be the reason for its fast degradation (peptide 18 in Fig. 6). In general, the question arises if the IgE-binding capacity is retained after further cleavage of the peptide stretches by pepsin, which has to be investigated further.

In general, time-course analyses or alternatively monitoring proteolysis at different protease concentrations can help to identify preferred cleavage sites which can be caused by different accessibilities due to three-dimensional protein structures or by the presence of post-translational modifications. Using the example of the protease-substrate pair of pepsin and LYZ we observed a protective effect of disulfides against proteolysis at sites close to the disulfide bonds. This is in agreement with other predominantly gel-based studies, which showed the influence of disulfide bridges on the gastro-intestinal degradation of the egg allergen ovomucoid [26] and the peanut allergen Ara h 2 [27]. The exact reason for the masking of a particular proteolytic cleavage site in the presence of nearby disulfide bonds cannot be directly deduced from our data. Certainly, steric effects contribute to this behavior, e.g. by restricting the access of the protease active site to the cleavage site. The presented analytical strategy is not only attractive for the simulation of gastro-intestinal digestion and food research, but can be transferred to any protease-substrate pair of interest with limitations regarding the overall molecular weight of the analyte. Furthermore it can be exploited to investigate the impact of other post-translational modifications such as phosphorylation on proteolytic processing. The presented method will benefit from future progress in top-down proteomics, e.g. improvements in terms of LC separation and MS sensitivity, which will extend its applicability. Conflict of interest

4. Conclusions The authors declare no conflict of interest. The monitoring of educts and products of proteolytic protein degradation over time requires methods that allow for the parallel measurement of analytes spanning a wide range of molecular weights. For this purpose we developed a combined top-down and bottom-up LC-MS approach that is preceded by Tricine-SDS-PAGE for an initial screening of reaction mixtures.

Acknowledgments This work was supported by the SFB877 “Proteolysis as a Regulatory Event in Pathophysiology” (project Z2) and the Cluster

120

J. Tucher et al. / J. Chromatogr. B 1015 (2016) 111–120

of Excellence “Inflammation at Interfaces” funded by the DFG (Deutsche Forschungsgemeinschaft). Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.jchromb.2016. 02.021. References [1] C. López-Otín, C.M. Overall, Protease degradomics: a new challenge for proteomics, Nat. Rev. Mol. Cell Biol. 3 (2002) 509–519. [2] K. Plasman, P. Van Damme, D. Kaiserman, F. Impens, K. Demeyer, K. Helsens, M. Goethals, P.I. Bird, J. Vandekerckhove, K. Gevaert, Probing the efficiency of proteolytic events by positional proteomics, Mol. Cell. Proteomics 10 (2011) (M110.003301). [3] N.J. Agard, S. Mahrus, J.C. Trinidad, A. Lynn, A.L. Burlingame, J.A. Wells, Global kinetic analysis of proteolysis via quantitative targeted proteomics, Proc. Natl. Acad. Sci. U. S. A. 109 (2012) 1913–1918. [4] P. Schlage, F.E. Egli, P. Nanni, L.W. Wang, J.N. Kizhakkedathu, S.S. Apte, U. auf dem Keller, Time-resolved analysis of the matrix metalloproteinase 10 substrate degradome, Mol. Cell. Proteomics 13 (2014) 580–593. [5] M.M. Dix, G.M. Simon, B.F. Cravatt, Global mapping of the topography and magnitude of proteolytic events in apoptosis, Cell 134 (2008) 679–691. [6] P.D. Compton, L. Zamdborg, P.M. Thomas, N.L. Kelleher, On the scalability and requirements of whole protein mass spectrometry, Anal. Chem. 83 (2011) 6868–6874. [7] L. Callewaert, C.W. Michiels, Lysozymes in the animal kingdom, J. Biosci. 35 (2010) 127–160. [8] Y. Mine, F. Ma, S. Lauriau, Antimicrobial peptides released by enzymatic hydrolysis of hen egg white lysozyme, J. Agric. Food Chem. 52 (2004) 1088–1094. [9] H.R. Ibrahim, D. Inazaki, A. Abdou, T. Aoki, M. Kim, Processing of lysozyme at distinct loops by pepsin: a novel action for generating multiple antimicrobial peptide motifs in the newborn stomach, Biochim. Biophys. Acta 1726 (2005) 102–114. [10] H.R. Ibrahim, K. Imazato, H. Ono, Human lysozyme possesses novel antimicrobial peptides within its N-terminal domain that target bacterial respiration, J. Agric. Food Chem. 59 (2011) 10336–10345. ˜ E. Molina, Immunoreactivity of hen egg [11] G. Martos, R. López-Fandino, allergens: influence on in vitro gastrointestinal digestion of the presence of other egg white proteins and of egg yolk, Food Chem. 136 (2013) 775–781.

˜ Egg proteins as [12] S. Benedé, I. López-Expósito, E. Molina, R. López-Fandino, allergens and the effects of the food matrix and processing, Food Funct. 6 (2015) 694–713. [13] S.R. Haider, H.J. Reid, B.L. Sharp, Modification of tricine-SDS-PAGE for online and offline analysis of phosphoproteins by ICP-MS, Anal. Bioanal. Chem. 397 (2010) 655–664. [14] S.R. Haider, H.J. Reid, B.L. Sharp, Tricine-SDS-PAGE, Methods Mol. Biol. 869 (2012) 81–91. [15] H. Toll, R. Wintringer, U. Schweiger-Hufnagel, C.G. Huber, Comparing monolithic and microparticular capillary columns for the separation and analysis of peptide mixtures by liquid chromatography-mass spectrometry, J. Sep. Sci. 28 (2005) 1666–1674. [16] J. Mohr, R. Swart, M. Samonig, G. Bohm, C.G. Huber, High-efficiency nano- and micro-HPLC-high-resolution Orbitrap-MS platform for top-down proteomics, Proteomics 10 (2010) 3598–3609. [17] H. Schägger, G. von Jagow, Tricine-sodium dodecyl sulfate-polyacrylamide gel electrophoresis for the separation of proteins in the range from 1 to 100 kDa, Anal. Biochem. 166 (1987) 368–379. [18] H. Schägger, Tricine-SDS-PAGE, Nat. Protoc. 1 (2006) 16–22. [19] D.W. Piper, B.H. Fenton, pH stability and activity curves of pepsin with special reference to their clinical importance, Gut 6 (1965) 506–508. [20] R. Jiménez-Saiz, S. Benedé, B. Miralles, I. López-Expósito, E. Molina, R. ˜ Immunological behavior of in vitro digested egg-white López-Fandino, lysozyme, Mol. Nutr. Food Res. 58 (2014) 614–624. [21] G. Martos, I. Lopez-Exposito, R. Bencharitiwong, M.C. Berin, A. ˛ Mechanisms underlying differential food allergy response Nowak-Wegrzyn, to heated egg, J. Allergy Clin. Immunol. 127 (2011), 990–997, e1–2. [22] D. Lemaire, G. Marie, L. Serani, O. Laprévote, Stabilization of gas-phase noncovalent macromolecular complexes in electrospray mass spectrometry using aqueous triethylammonium bicarbonate buffer, Anal. Chem. 73 (2001) 1699–1706. [23] N.D. Rawlings, M. Waller, A.J. Barrett, A. Bateman, MEROPS: the database of proteolytic enzymes, their substrates and inhibitors, Nucleic Acids Res. 42 (2014) D503–509. [24] F. Liu, B. van Breukelen, A.J.R. Heck, Facilitating protein disulfide mapping by a combination of pepsin digestion electron transfer higher energy dissociation (EThcD), and a dedicated search algorithm SlinkS, Mol. Cell. Proteomics 13 (2014) 2776–2786. [25] V. Schnaible, S. Wefing, A. Resemann, D. Suckau, A. Bücker, S. Wolf-Kümmeth, D. Hoffmann, Screening for disulfide bonds in proteins by MALDI in-source decay and LIFT-TOF/TOF-MS, Anal. Chem. 74 (2002) 4980–4988. [26] J. Kovacs-Nolan, J.W. Zhang, S. Hayakawa, Y. Mine, Immunochemical and structural analysis of pepsin-digested egg white ovomucoid, J. Agric. Food Chem. 48 (2000) 6261–6266. [27] M. Sen, R. Kopper, L. Pons, E.C. Abraham, A.W. Burks, G.A. Bannon, Protein structure plays a critical role in peanut allergen stability and may determine immunodominant IgE-binding epitopes, J. Immunol. 169 (2002) 882–887.