A Proteomic-Based Approach for Detection of ... - ACS Publications

A Proteomic-Based Approach for Detection of Chicken in Meat Mixes Miguel A. Sentandreu,†,# Paul D. Fraser,† John Halket,‡ Raj Patel,‡ and Peter M. Bramley*,† Center for Systems and Synthetic Biology, School of Biological Sciences, and Specialist Bioanalytical Services Ltd., Royal Holloway, University of London, Egham TW20 0EX, U.K. Received October 5, 2009

A proteomic-based method has been developed for the detection of chicken meat within mixed meat preparations. The procedure is robust and simple, comprising the extraction of myofibrillar proteins, enrichment of target proteins using OFFGEL isoelectric focusing, in-solution trypsin digestion of myosin light chain 3, and analysis of the generated peptides by liquid chromatography-mass spectrometry/ mass spectrometry (LC-MS/MS). Using this approach, it was possible for example to detect 0.5% contaminating chicken in pork meat with high confidence. Quantitative detection of chicken meat was done by using AQUA stable isotope peptides made from the sequence of previously selected speciesspecific peptide biomarkers. Linearity was observed between the amount of the peptide biomarker and the amount of chicken present in the mixture; further independent replication is required now to validate the method. Apart from its simplicity, this approach has the advantage that it can be used effectively for the detection of both raw and cooked meat. The method is robust, reliable, and sensitive, representing a serious alternative to methods currently in use for these purposes. It is amenable to highly processed foods which can be particularly problematic, as the tertiary protein structure is often affected in processed food precluding immunoassays. In addition, this proteomic analysis will permit the determination of definitive discriminatory sequence, unlike the DNA PCR based methods used presently. The present article also demonstrates the translation of the technology to routine mass spectrometry equipment, making the methodology suitable for public analysts. Keywords: meat authentication • mass spectrometry • quantitation • AQUA • peptide biomarkers • OFFGEL fractionation

Introduction Consumers demand clear and reliable information about the food they consume. This information is an important aspect of product choice. For example, a product may be chosen on the basis of religious or health concerns. Honest and accurate labeling of food is therefore an essential component of food safety and choice, especially in the case of processed food products where differentiation of the different constituents can be difficult. Legislation must protect the consumer against misdescription, which is commonly carried out in order to increase profit. To perform the objective robust, accurate, and sensitive methods must be established. In the case of meat products, the most common cases of adulteration deal with substitution of high quality meat with lower value meat, resulting in an increased profit for food producers. Different approaches have been used to determine meat authentication. A chemometrics approach establishing variance within the chemical composition between samples * Author for correspondence. Tel: (+44) 01784 443555; fax: (+44) 01784 414224; e-mail: [email protected]. † Center for Systems and Synthetic Biology. ‡ Specialist Bioanalytical Services Ltd. # Present address: Instituto de Agroquı´mica y Tecnologıá de Alimentos (CSIC), 7 Agustıń Escardino Avenue, 46980 Paterna (Valencia), Spain.

3374 Journal of Proteome Research 2010, 9, 3374–3383 Published on Web 05/03/2010

could be used. However, the robustness of this approach to routine analysis has not been substantiated due to natural variation, including factors such as species, breed, sex, growing environment, or age at the time of slaughter. An alternative to this approach utilizes species-specific biomarkers capable of providing information about the composition of food. This is the basis of methods involving marker proteins for meat speciation. This approach includes different chromatographic1 and electrophoretic2 methods, but the most widespread methodology for meat authentication based on protein detection relies on immunoassays. Once developed, immunoassays are easy to use, having high sensitivity and capacity to process a high number of samples in short times. However, in some cases there are some important limitations such as cross-reactivity between closely related species.3 There also can be important limitations when analyzing processed meat products, because processing can negatively influence the recognition of the target protein by the antibody (e.g., modification of the tertiary structure). The limitations in protein-based analysis have promoted in recent years the development of alternative methods for food speciation based on DNA analysis. This technology has a considerably higher discriminating power because it has the ability to identify species-specific DNA fragments even for 10.1021/pr9008942

 2010 American Chemical Society

Proteomic Detection of Chicken Meat closely related species. However, unambiguous discriminating sequence information cannot be routinely acquired unless cloning and sequencing are employed. The processing of food can also limit quantitative DNA-based methodologies due to the harsh conditions used and chemical modifications that result. For example, some processing conditions cook meat at temperatures higher that 100 °C. This can result in DNA degradation, increasing the chances of having nonspecific fragments.4 DNA extraction is a critical step because it greatly influences the analysis. As food matrices are very complex, it is difficult to define a standard extraction protocol and so extraction methods have to be optimized for each particular case in order to ensure that enough DNA is obtained for the analysis and that inhibitors of the PCR are reduced or eliminated. This problem becomes even more important in those assays where quantitation is required.5 Thus, the potential exists for non-DNA-based approaches with the potential for unambiguous discriminating power. Recent advances on mass spectrometry offer a feasible alternative to DNA-based analysis on foodstuffs. For example, in recent years proteomics has been successfully applied for the discrimination of fish species,6 or in the identification of soybean proteins added to processed meat products.7 In the present article, we describe the development of mass spectrometry-based methodologies for the specific detection of chicken in meat mixes. This has been done by selecting myosin light chain 3 as the adequate target protein capable of generating chicken-specific peptides. The method is robust, reliable, and sensitive, and it can be applied indistinctively to either raw or cooked meats.

Materials and Methods Authentic meat samples were obtained from an EC-approved plant operated by Acacia Foods Ltd./PDM group. After removing appreciable adipose tissue, they were frozen at -80 °C until needed. Ammonium bicarbonate, bromophenol blue, dithiotreitol (DTT), trifluoroacetic acid (TFA), Tris(hydroxymethyl)methylamine (Tris-HCl), glycerol, sodium dodecyl sulfate (SDS), and Proteo Silver protein staining kit were from Sigma (Poole, UK). Formic acid (HPLC grade) was from Romil (Cambridge, UK). Ultrapure protogel and concentrated 10× Tris/glycine/ SDS (electrophoresis grade) were from National Diagnostics (Hessle, UK). MALDI-grade 2,5-dihydroxybenzoic acid (DHB) was from Bruker Daltonik GmbH (Bremen, Germany). C18 ZipTip pipet tips were from Millipore (Watford, UK). Syringe filters (0.45 µm) were from Fisher Scientific (Loughborough, UK). HPLC-grade acetonitrile (ACN) was from VWR (Poole, UK). The Bradford protein assay was from Bio-Rad (Hemel Hemstead, UK). Modified trypsin (sequencing grade) was from Roche Diagnostics (Lewes, UK). Stable isotope labeled peptides were synthesized by Thermo Fisher Scientific (Ulm, Germany). Cooking of Meats. Frozen meat samples were cut in portions of about 50 g and were thawed at +4 °C overnight. Each sample was placed in individual aluminum foil trays with the capacity to hold the meat exudates. The different samples were then cooked in an oven at 180 °C for 1 h. After that, samples were taken out from the oven and left to cool at room temperature before use. Extraction of Meat Proteins. One gram of either raw or cooked meat sample was homogenized in 10 mL of 50 mM Tris-HCl buffer, pH 8.0, by using a vortex for 2 min. The homogenate was then centrifuged at 10000g for 20 min at 4 °C. The supernatant was collected, constituting the sarcoplas-

research articles mic extract in which all the soluble proteins were contained. The pellet was resuspended in 10 mL of 50 mM Tris-HCl, pH 8.0, containing 6 M urea and 1 M thiourea and homogenized in a vortex for 5 min in order to solubilize the myofibrillar proteins. The homogenate was then centrifuged at 10000g for 10 min, collecting the supernatant. This extract constituted the myofibrillar extract. Both sarcoplasmic and myofibrillar extracts were filtered through 0.45 µm membrane filters prior to use. Protein concentration of the extracts was determined by the method of Bradford,8 using the Biorad protein assay kit and bovine serum albumin as the protein standard. Protein OFFGEL Fractionation. Myofibrillar extracts (2.5 mg of total protein) were fractionated by isoelectric focusing using an Agilent 3100 OFFGEL Fractionator (Palo Alto, CA), following manufacturer’s instructions. Proteins were focused based on their pI using 24-cm long Immobiline DryStrip (GE Healthcare, Chalfont, UK) with a linear pH gradient ranging from 4.0 to 7.0. A total of 140 µL of sample was loaded in each well. After focusing, samples were directly collected and stored at -20 °C for further use. SDS-PAGE. OFFGEL protein fractions (30 µL) were mixed in equal volume with sample buffer solution (0.5 M Tris-HCl pH 6.8, 50% v/v glycerol, 10% w/v SDS, 0.2 M DTT and 0.05% bromophenol blue), then heated at 90 °C for 4 min. Appropriate volumes were loaded in order to always have 5 µg of protein per lane. Different acrylamide percentages were assayed, ranging from 8 to 15% w/v according to the method of Laemmli,9 and using the Sigma Silver staining kit (PROTSIL1KT). Bio-Rad standard protein kit was used as the protein standard (ref 161-0303). In-Gel Digestion of Protein Bands. Stained bands corresponding to the proteins of interest were excised by using a scalpel, then introduced into Eppendorf (0.5 mL) tubes and washed three times for 10 min with 50 mM ammonium bicarbonate, pH 8.0 (50 µL). After that, gel pieces were dried three times with acetonitrile (50 µL) for 10 min. Once the gel pieces shrank and turned opaque, 10 µL of 12.5 ng/µL trypsin dissolved in 50 mM ammonium bicarbonate, pH 8.0, were added. An additional 15 µL of 50 mM ammonium bicarbonate was added to each tube in order to cover the gel pieces. The tubes were incubated at 37 °C overnight; then, the supernatant was transferred to a clean Eppendorf tube. The gel pieces were washed with 25 µL of ACN/0.1% TFA (50:50), sonicated for 10 min and the supernatants were combined. This solvent was evaporated using a GeneVac Ez-2 Plus rotatory evaporator (Ipswich, U.K.) and reconstituted in 5 µL of 0.1% v/v TFA. Samples were then desalted by using Zip-tip according to manufacturer’s instructions. In-Solution Protein Digestion. Targeted OFFGEL protein fractions were also directly subjected to trypsin digestion. Twenty microliters of each OFFGEL fraction were mixed with 20 µL of 0.1% TFA. After homogenization, samples were desalted using Zip-tips. Protein elution was made with 20 µL of 70% ACN/30% 0.1 TFA. The solvent was evaporated using a GenVac evaporator and reconstituted with 15 µL of 50 mM ammonium bicarbonate, pH 8.0, followed by addition of 15 µL of 12.5 ng/µL trypsin dissolved in the same buffer. Samples were incubated at 37 °C overnight and then evaporated using the GenVac evaporator. Reconstitution was made with an appropriate volume of 0.1% v/v TFA in order to analyze the tryptic peptides by either matrix assisted laser desorption/ ionization-time of flight (MALDI-TOF) or liquid chromatograJournal of Proteome Research • Vol. 9, No. 7, 2010 3375

research articles phy with electrospray ionization coupled to tandem mass spectrometry (LC-ESI-MS/MS). MALDI-TOF MS. Sample aliquots (1.5 µL) were spotted onto a 600 µm hydrophilic Anchor (AnchorChip600, Bruker, Coventry, UK), then the same volume of a 5 mg/mL solution of DHB (dissolved in ACN/0.1% TFA 1:2) was added. MALDI-TOF spectra were generated with a Bruker MALDI-TOF Reflex III (Coventry, UK) operated in the positive reflectron mode with an acceleration voltage of 20 kV. Peptide ions were generated by a nitrogen laser emitting at 337 nm. The mass spectrometer was calibrated in the m/z range from 1000 to 3000 using a peptide calibration standard (code 206195, Bruker). Each mass spectrum constituted the sum of 300 shots. FlexAnalysis 2.4 software (Bruker) was employed for data analysis. Protein identification was done by direct interrogation of the peptide mass list obtained for each of the protein digests into the MASCOT 2.2 search engine (Matrix Science, London, UK) against NCBInr 20100102 or UniprotKB/Swiss-Prot 57.12 database, using a peptide tolerance of 100 ppm and the option Chordata (vertebrates and relatives) as taxonomy restriction parameter. LC-ESI-MS/MS. Peptide separation and identification was carried out using an AS3000 autosampler, a Finniganmat P4000 LC system (Thermo Electron Corp. San Jose, CA) directly coupled to a LCQ Deca ion trap instrument (Thermo Electron Corp.). Samples (30 µL) were injected into the LC-MS system by using the autosampler. Chromatographic separation was made using a 5 µm CLIPEUS C18 column (150 × 0.5 mm i.d., Higgins Analytical Inc., Mountain view, CA), using the following conditions: Flow rate 18 µL/min. Isocratic gradient with 0.1% v/v FA in H2O (solvent A) during 30 min; then, linear gradient from 0 to 40% v/v 0.1% FA in ACN (solvent B) for 120 min. Operating conditions for the ion trap mass spectrometer were ESI positive mode, capillary temperature 220 °C, collision energy normalized to 33%, source voltage 4.5 kV, and capillary voltage 3.0 V. First scan even was full MS scan from m/z values 100 to 2000. The second scan event was a dependent MS/MS scan of the most intense ion enabling dynamic exclusion (after three scans of the most intense ion). Data acquisition was done using the Xcalibur v1.2 software. Peptide identification was done by interrogating the generated .dta files containing MS/ MS data into the MASCOT search engine (http://www.matrixscience.com), against the NCBInr 20100102 database or UniProtKB/Swiss-Prot 57.12 database, using a MS/MS tolerance of 0.6 Da. The option Chordata (vertebrates and relatives) was selected as taxonomy restriction parameter. Peptide Quantification Using Stable Isotope Labeled Peptides (AQUA). Different amounts of cooked chicken meat were mixed with cooked pork meat in order to create meat mixes containing 0%, 0.5%, 1%, 2%, 5%, and 10% of chicken in pork meat. All these mixes were subjected to myosin light chain 3 (MLC-3) enrichment using procedures described above with the exception that the fractions were spiked with 0.1% v/v TFA (10 µL) containing the two synthesized stable isotope labeled peptides (AQUA peptides, 5 pmol) corresponding to the selected chicken biomarker peptides: DQGTFEDF(13C9,15N)VEGLR and AL13C6,15N)GQNPTNAEINK. The mixture was desalted by using Zip-tips, eluting the content of the tip with 20 µL of ACN/0.1% TFA (70:30). After drying, samples were subjected to in-solution trypsin digestion as already described. Reconstitution of digested samples was made in 50 µL of 0.1% TFA. Thirty microliters of this volume were automatically injected on the Thermo LC-ESI-MS/MS 3376

Journal of Proteome Research • Vol. 9, No. 7, 2010

Sentandreu et al.

Figure 1. 12% SDS-PAGE protein profile obtained for (A) TrisHCl and (B) urea/thiourea extracts prepared from the indicated animal species for both raw and cooked meat samples. Lanes 1, 3, and 5: raw meats; lanes 2, 4, and 6: cooked meats. Arrows indicate the position of myosin light chain isoforms in the gel: MLC-1: myosin light chain 1; MLC-2: myosin light chain 2; MLC3: myosin light chain 3.

system described above. Data analysis of the obtained chromatograms was done by using Thermo Qual Browser software from Xcalibur v2.0.6. Integration of the chromatographic peaks corresponding to the different native biomarkers and AQUA peptides was performed manually, using the mass range option for both singly and doubly charged forms of each peptide. Since the initial amount of the heavy peptide was known, comparison of peak area values between the native peptide and that of the labeled peptide counterpart allowed calculating the amount of the native peptide in the sample.

Results Selection of the Target Proteins. An important feature of the peptide-based detection method is its amenability to both cooked and raw material. In order to assess the heat stability of peptide biomarkers, SDS-PAGE analysis of sarcoplasmic and myofibrillar extracts were carried out for both fresh and highly cooked meats from three different animal species. Thus, beef, pork, and chicken meat cuts were cooked in an oven at 180 °C during 1 h. Figure 1 shows the protein profile obtained for the

research articles

Proteomic Detection of Chicken Meat

Table 1. Species-Specific Peptides Identified after Trypsin Digestion of Common Protein Bands Found in Both Fresh and Cooked Urea/Thiourea Soluble Extracts of Beef, Pork, and Chicken Meats in the Region 15-25 kDa (see Figure 1B)a protein

myosin light chain 1


species (protein accession no.) b

beef (NP_001073046) chickenc (P02604) porkb (ABK55642) beefb (NP_001069115)

mass [M + H]

modified mass [M + H]

modification

48-62

1

QQQDEFKEAFLL FDR

1369.642

79-91

0

ALGQNPTNAEINK

1514.6820

123-135

0

DQGSYEDFVEGL R

2621.23

9-32

1

RAAAEGGSSSVF SMFDQTQI QEFK

10-32 120-138 74-93 74-91 120-131 9-30

0 1 1 0 0 1

AAAEGGSSSVFS MFDQTQIQ EFK FLEELLTTQCDRF SQEEIK EASGPINFTVFLNMFGEKLK EASGPINFTVFLN MFGEK FLEELLTTQCDR RAAEGSSNVFSMFDQTQIQEFK

0 1 0 0

2242.1576 2000.9786 2520.1823 2364.0812 2074.0161 1987.9833 1894.8792

2386.1595 2258.1525 2016.9735 1524.7362 2536.1772

Cys_CAM: 129 MSO: 87 MSO: 87 Cys_CAM: 129 MSO: 20

2380.0761

MSO: 20

2003.9782 1910.8741

MSO: 85 MSO: 138

10-30 92-111 72-89 137-153

2523.2184

Cys_CAM: 128

90-104 59-71 154-165 92-104 41-50 51-58 118-137

1 1 0 0 1 0 1

AAEGSSNVFSMFDQTQIQEFK GADPEDVIMGAF KVLDPDGK EASGPINFTVFLT MFGEK NMWAAFPPDVA GNVDYK LKGADPEDVIMG AFK LNVKNEELDAMIK NICYVITHGEDK GADPEDVIMGAF K DGIIDKDDLR ETFAAMGR HFLEELLTTQCDR FSQEEIK

1789.8901 1661.7951 2332.173

Cys_CAM: 128 Cys_CAM: 128 MSO:119 or 128

117-130 118-130 110-129

1 0 1

KHFLEELLTTQCD R HFLEELLTTQCDR HVLATLGEKMTE EEVEELMK

acetyl(K): 49 MSO: 119,128

81-93 13-25 50-62 37-49 119-129

0 1 1 0 0

DQGTFEDFVEGL R EAFLLFDRTGDA K ILGNPSKEEMNA K ALGQNPTNAEINK

acetyl(K): 62/63

57-63

1

1590.8195 1516.8039 1391.6623 1349.6405 1159.5953 882.4138 porkb (NP_001006592)


chickenb (NP_001038097) 1512.72 1482.766 1430.718 1369.7

d

peptide sequence

1913.923

2465.1289

chickenc (P02609)

MCd

position

1411.7 1383.696 1399.731 891.5

a MSO: oxidation of methionine; Cys_CAM: carbamidation of cysteine; acetyl (K): acetylation of Lys. Missed cleavages.

Tris-HCl (A) and urea/thiourea (B) soluble extracts prepared from fresh and cooked meat samples of the three animal species tested. The Tris-HCl derived soluble extracts indicated that following cooking most of the proteins bands observed in fresh meat extracts (Figure 1A, lanes 1, 3, and 5) disappeared (Figure 1A, lanes 2, 4, and 6). This observation reflects the fact that cooking of meat is a very aggressive process, giving rise to denaturation/insolubilization of most of the soluble proteins. From these results, it was concluded that Tris-HCl extracts are not suitable sources to initate the identification of speciesspecific peptide biomarkers for meat authentication due to the poor protein complement in the case of highly cooked meats. The protein profile obtained for urea/thiourea extracts also showed differences between fresh and cooked meat extracts. As can be observed in Figure 1B, in this case cooked meat extracts (lanes 2, 4, and 6) contained more protein bands than fresh extracts (lanes 1, 3, and 5). This can be explained by looking at results obtained for Tris-HCl extracts (see Figure 1A). It is evident that part of the sarcoplasmic proteins that got insoluble due to denaturation during the cooking process was

b

MTEEEVEELMK EEMNAKK

NCBI database. c UniprotKB/Swiss-Prot database.

subsequently recuperated in the urea/thiourea extracts (Figure 1B). The resulting myofibrillar extract was then more complex, with overlapping protein bands especially above 30 kDa. Interestingly, the region between 15-25 kDa (black line rectangle) contains some intense well-resolved protein bands common to both raw and cooked extracts. These proteins appeared to be better targets for obtaining species-specific peptide biomarkers to develop a MS approach for identification of both fresh and cooked meats. In-gel digestion of these bands, followed by peptide mass fingerprinting analysis using MALDI-TOF MS, identified these common protein bands as the different isoforms of myosin light chain. Peptide mass lists corresponding to the identification of these proteins are shown in Tables S1-S7 of the Supporting Information (myosin light chains 1, 2, and 3, as indicated in Figure 1B). Table 1 shows the species-specific peptides identified from these isoforms in both fresh and cooked meats. These results illustrate the high stability of peptides under extreme processing conditions and support the use of myosin light Journal of Proteome Research • Vol. 9, No. 7, 2010 3377

research articles

Sentandreu et al.

Figure 2. 12% SDS-PAGE corresponding to the first nine fractions (lanes 1-9) retrieved after OFFGEL fractionation of a mixture containing 1% cooked chicken in cooked pork meat. Focusing was made in the pH range 4-7 by loading 2.5 mg of total protein. Positions in the gel for myosin light chains (MLC) are indicated by arrows.

chain isoforms as adequate target proteins to obtain speciesspecific peptide biomarkers. Enrichment of the Target Protein/Peptides. Since the same peptide biomarkers were found in both fresh and cooked meat (Table 1), subsequent work was focused on the detection of peptide biomarkers obtained from myosin light chains of cooked meats. As a model case, we aimed at detecting the presence of contaminating chicken in pork meat. Thus, different amounts of cooked chicken meat (1%, 5%, and 10%) were added to cooked pork meat. Preparation of urea/thiourea extracts, followed by SDS-PAGE protein separation, in-gel trypsin digestion of myosin light chains, and MALDI-TOF MS analysis allowed us to detect two of the previously identified chicken-specific peptides (Table 1) when chicken meat was present at 10% in the mixture. A first attempt to lower this detection limit was carried out by overloading SDS-PAGE sample lanes from 5 to 15 µg of total protein content. However, this approach failed to detect lower amounts (1 and 5%) of contaminating chicken in pork meat (data not shown). Thus, an additional enrichment step of the target proteins/peptides was necessary in order to lower the detection limit. To achieve this, we carried out the fractionation of proteins contained in the urea/thiourea meat extracts by using an Agilent 3100 OFFGEL fractionator. This apparatus performs isoelectric focusing of proteins on immobilized pH gradients gel strips. It differs from conventional gel isoelectric focusing in that fractionated proteins do not remain trapped in a gel but are recovered from a liquid phase, allowing an easier transfer to further separation and/or analysis. Another important advantage is its high protein loading capacity, contributing to reach an optimal enrichment of the target proteins. To assess the feasibility of this approach as an efficient enrichment step, we carried out the fractionation of different myofibrillar cooked meat extracts, including pork and chicken meats, together with mixes containing 1% and 10% chicken in pork meat. The protein separation profile obtained for either pork or chicken meat extracts was similar in the two species (data not shown) and is also similar to the separation profile obtained for the mix of meats containing 1% chicken in pork shown in Figure 2. Myosin light chains, previously selected as potential target proteins for obtaining peptide biomarkers, were focused in the 3378


Figure 3. Comparative zoom-in for m/z range 1490-1550 of MALDI-TOF MS analysis of MLC-3 coming from 100% pork, 100% chicken, and a mixture of 1% chicken in pork meat. Chicken biomarker peptide DQGTFEDFVGLR (M + H+ 1512.6965) originated from trypsin digestion of enriched myosin light chain 3 using OFFGEL fractionation is indicated with a black arrow. Sequence differences with the bovine peptide (M + H+ 1514.683) are also shown.

closest fractions to anode, corresponding to the most acidic pH values (fractions 1-9). It is also remarkable to observe that OFFGEL fractions 3-5 contained myosin light chain 3 but few other protein bands (Figure 2), suggesting that this approach can be an efficient enrichment step to detect low levels of chicken in meat mixes. From these results, we selected MLC-3 as the protein source to obtain chicken-specific peptides. Detection of Chicken-Specific Peptides in Mixtures Containing Low Amounts of Chicken Meat. The band of fraction 3 obtained after OFFGEL fractionation of the mix containing 1% w/v chicken in pork (Figure 2, lane 3) was cut from the gel and subjected to in-gel digestion with trypsin. This approach was used because the fraction was principally enriched in chicken myosin light chain 3 with respect to pork MLC-3, which tends to be focused in fraction 4 (data not shown). This can be explained due to the slightly lower pI of chicken MLC-3 with respect to the porcine protein. MALDITOF MS analysis of peptides obtained after trypsin digestion of this protein band revealed the presence of two chickenspecific peptides in the mixture. Figure 3 illustrates the chicken+ specific tryptic peptide DQGTFEDFVGLR (M + H 1512.6965) obtained from MLC-3 of this animal species. As can be observed, this peptide was detected in a 100% chicken meat extract but also in the mix containing as low as 1% chicken in pork meat. This result confirmed OFFGEL fractionation as an efficient enrichment/purification step in the development of a proteomic approach for meat authentication. Apart from this one, a second chicken biomarker peptide was detected in the mixture containing 1% chicken. As shown in Figure 4, a well-


Figure 4. Comparative zoom-in for m/z range 1380-1450 for MALDI-TOF MS of MLC-3 coming from 100% pork, 100% chicken, and 1% chicken in pork meat. Chicken biomarker peptide ALGQNPTNAEINK (M + H+ 1411.485, acetylated) originated from trypsin digestion of enriched myosin light chain 3 using OFFGEL fractionation is indicated with a black arrow.

defined signal was observed at m/z 1411.485, corresponding to the acetylated form of the peptide ALGQNPTNAEINK. Confirmation of the sequence for these two peptide biomarkers was accomplished by in-solution trypsin digestion of

research articles MLC-3 previously enriched by OFFGEL fractionation of a mix containing 10% chicken in pork meat. This was followed by analysis of the generated peptides by LC-ESI-MS/MS using a Thermo Surveyor LC system coupled to a LCQ Deca ion trap. Figure 5 shows the fragmentation pattern obtained for the peptide DQGTFEDFVEGLR, which was detected by MALDI-TOF + MS as M + H 1512.6965 (Figure 3). Fragmentation of peptide ALGQNPTNAEINK, previously detected by MALDITOF MS as + M + H 1411.485 (Figure 4), is shown in Figure 6. The quality of fragmentation spectra was good for the two peptides, matching most of the b and y fragment ions in both cases. Determination of the Chicken Meat Amount within a Mixed Meat Product Using Species-Specific Peptide Biomarkers. The workflow associated with a procedure developed for the identification of meat species using a proteomic approach is summarized in the diagram of Figure 7. The preparation of samples prior to peptide identification by mass spectrometry is relatively simple, involving a number of intuitive steps. To demonstrate the utility of the procedure as a tool for low level quantification, chicken meat was used as the contaminant in product composed of pork. In this instance, quantification was carried out using the AQUA (“Absolute QUAntitation”) approach using selected internal standard peptides. Implementation of the AQUA strategy was done by synthesizing heavy peptides to the previously selected chicken-specific peptides. For the peptide biomarker DQGTFEDFVEGLR (Figure 5), the heavy peptide was synthesized by placing one heavy phenylalanine in the 8th N-terminal position, resulting in a mass difference of +10 Da with respect to the native peptide. For the peptide ALGQNPTNAEINK (Figure 6), the stable isotope homologous was synthesized by placing one heavy leucine in the N-penultimate position (+7 Da). Quantification was tested using mixes having 0%, 0.5%, 1%, 2%, 5%, and 10% cooked chicken in cooked pork meat, respectively. As an example, Figure 8 shows the peptide

Figure 5. Ion trap MS/MS spectrum of peptide 81DQGTFEDFVEGLR93 (757.012+), obtained from trypsin digestion of chicken myosin light chain 3 (UniProtKB/Swiss-Prot entry P02605). Matched b and y fragment ions are indicated. Journal of Proteome Research • Vol. 9, No. 7, 2010 3379

research articles

Sentandreu et al.

Figure 6. Ion trap MS/MS spectrum of peptide 37ALGQNPTNAEINK49 (685.392+), obtained from trypsin digestion of chicken myosin light chain 3. Matched b and y fragment ions are indicated.

Figure 7. Schematic summary of the method developed in this work for quantitative detection of meat species within meat mixtures.

mapping profile obtained after LC-ESI-MS/MS analysis of the sample containing 1% chicken in the mixture. Despite the complexity of this peptide profile, the peak resolution was good; this prevented coelution of the biomarkers with other peptides having the same mass and also contributed to the sensitive detection at low amounts of the biomarker peptides. Figure 9A shows the elution profile for m/z 685.35 found in the 1% chicken mixture, corresponding to the doubly charged form of the biomarker peptide ALGQNPTNAEINK. Figure 9B shows the ESI full mass spectrum for the retention time of this peptide. As expected, an m/z value of 685.49, corresponding to this peptide, was clearly observed; the corresponding m/z value of 688.79 was also observed for its heavy peptide counterpart AL(13C6,15N)GQNPTNAEINK. The mass difference of +7 Da is seen as +3.5 Da in these doubly charged forms, 3380


allowing a correct calculation of the peak areas of these two peptides, even if they coeluted in a single peak. Peak areas corresponding to selected peptide biomarkers and their corresponding heavy counterparts were obtained from LCESI-MS/MS analysis of the different meat mixes in order to calculate their amount. The amount of each peptide was plotted as a function of the percentage of chicken meat added to the mixture. Data obtained for the biomarker peptide DQGTFEDFVEGLR are plotted in Figure 10. A good linear correlation resulted between the percentage of chicken meat added to pork meat and the amount of chicken biomarker peptide detected by LC-ESI-MS/MS using the developed approach (see Figure 7). The results obtained for chicken peptide biomarker ALGQNPTNAEINK are shown in Figure 11. In this case, a good linear relationship was also obtained for this chicken biomarker, but the slope of the curve was notably higher in this latter case, possibly because the ionization properties are considerably better for peptide ALGQNPTNAEINK than for DQGTFEDFVEGLR. In the present work, we did not carry out biological replicates of the meat mixes prepared by the addition of the different percentages of chicken meat in pork meat. However, a range of percentages were used to demonstrate the trend and linearity. In this way, the feasibility of the approach could be assessed and act as the foundations for future work to create an independent validated method. Both peptides were detected on LC-ESI-MS/MS even when only 0.5% chicken meat was present in pork meat, indicating the good sensitivity and efficiency of this MS approach using a conventional ion trap, which is one of the most affordable MS equipment currently available in many laboratories.

Discussion In this work, a robust, accurate, and sensitive methodology has been developed for detecting the presence of chicken meat in other meat species through the identification of speciesspecific peptide biomarkers. The procedure combines protein enrichment strategies, identification, and discriminating se-


research articles

Figure 8. Typical LC peptide separation profile obtained after in-solution trypsin digestion of enriched myosin light chain 3 coming from a mixture containing 1% chicken in pork meat. Separation was done on a Thermo Surveyor LC system directly coupled to a Thermo LCQ Deca ion trap instrument. Separation was carried out on a CLIPEUS C18 column (150 × 0.5 mm i.d.).

quence from selected peptides by mass spectrometry. An objective of the study was robustness and ease of use to ensure applicability to routine analysis. Once optimized, the whole protocol has the potential to be carried out within 3-4 days, including LC-MS/MS analysis. Further optimization and developments in MS will potentially reduce analysis time. At the present level, it is difficult to estimate the cost of each individual analysis with precision, but it is envisaged that the methodology will be competitive with other peptide biomarker assays and cheaper than radioimmunoassay once the capital investments in the MS hardware have been made. Potentially this procedure constitutes an alternative to the current methods used for meat speciation such as immunoassays or DNA-based assays. The discriminating power of this proteomic approach goes further than present DNA assays because discriminating sequences can be ascertained routinely. In the method presented here, myosin light chain 3 was chosen as target protein in view of its properties for enrichment using OFFGEL fractionation. This has proved to be essential for detection of low amounts of chicken meat. Application of the procedure illustrated in Figure 7 allowed us to detect the presence of 0.5% contaminating chicken in pork meat, which is comparable to the PCR-based analysis method developed by Jańosi et al.10 for detection of poultry meat in different mixtures with pork meat. More recently, Koppel et al.11 developed a quantitative PCR assay for the simultaneous identification of pork, beef, turkey, and chicken, reaching comparable levels of sensitivity. However, these authors reported that quantification of the different meats in real samples can be difficult because DNA can be degraded during food processing or during storage. Apart from this, the

efficiency in the DNA extraction from the different food matrices can be also a problem for quantitative DNA-based assays.5 Immunoassays have been also widely employed for meat speciation because they are easy to use and have high throughput capability. Martin et al.12 developed a sandwich ELISA test for detection of chicken meat in different meat mixtures using a monoclonal antibody, reaching a detection limit of 1% w/v working with fresh meat mixes, but no information about detection of chicken meat in cooked meat samples was reported. On the contrary, the sandwich ELISA developed by Liu et al.13 allowed detecting very low amounts of porcine meat (0.05-0.1% v/v) in both raw and heatprocessed meat samples. Even if this assay was highly sensitive and specific, it is not always possible to reach this level of performance for all immunoassays. In some cases, it can be difficult to find a heat-stable antigen for the production of species-specific antibodies. The use of monoclonal antibodies can greatly increase the specificity of the assay. However, this technology can be expensive and quite time-consuming, and is not exempted from potential cross-reactivity. Former methods based on protein detection, such as electrophoresis, chromatography, or immunoassays, lack good resolving power in closely related species because they are not based on differences at the sequence level. In addition, they are more affected by denaturation of proteins and so some of them do not work on identification of highly cooked meat, for example.14 The proteomic-based method developed in this work is not affected by denaturation of proteins because it is based on identification of the primary structure of specific Journal of Proteome Research • Vol. 9, No. 7, 2010 3381

research articles

Sentandreu et al.

Figure 11. Changes in the amount of chicken biomarker peptide ALGQNPTNAEINK with increasing percentages of chicken meat in a mixture with pork meat.

Figure 9. (A) LC elution profile for the chicken peptide biomarker ALGQNPTNAEINK present in the mixture containing 1% chicken in pork meat after in-solution trypsin digestion of enriched myosin light chain 3. (B) ESI full mass spectrum obtained at the retention time of this peptide (84.02 min) for the m/z range 640-730.

Figure 10. Changes in the amount of chicken biomarker peptide DQGTFEDFVEGLR with increasing percentages of chicken meat in a mixture with pork meat.

peptide fragments derived from trypsin hydrolysis of a target protein. This means that the same biomarker peptides can be used for detection of both fresh and cooked chicken meat and so the same protocol can be applied in any case. Chassaigne et al.15 developed a proteomic approach for detection of the main allergens in peanuts, identifying some peptide biomarkers specific for the allergenic proteins. As in this case, the authors highlighted that the high stability of the selected peptides allowed their detection indistinctively in both raw and roasted peanuts. A proteomic approach also has been successfully 3382


employed by Leitner el al.16 in identifying the addition of soybean proteins to processed meat products. Protein separation strategies combined with identification of tryptic peptides by mass spectrometry have allowed the detection of small amounts of genetically modified soya (0.9% w/v) in a mixture with nonmodified soya.17 This report also described the use of AQUA strategy as a way to quantity different additions of transgenic soya to normal soya. They proved the suitability to use stable isotope peptides to make the assay quantitative, reaching a detection level of 0.5% transgenic soya in the mixture.18 In the present work, however, the extraction/ enrichment procedure is considerably simpler (Figure 7), which contributed to reduce the analysis time. The use of OFFGEL fractionation, followed by in-solution trypsin digestion without the use of SDS-PAGE, also contributed to simplification and more reproducible procedure. Despite these refinements, the same detection limit (0.5%) was achieved. Another important difference with respect to the work of Ocan ˜ a et al.18 is the equipment used for mass spectrometry analysis. While in that work authors analyzed the peptides contained in the different samples by a nanoLC-nanoESI system coupled to a QSTAR hybrid quadrupole QTOF, in the present work we employed a conventional LC-ESI system coupled to an ion trap mass spectrometer, this latter being a more robust and affordable device and so more amenable to implement the method in both industry and control laboratories. Methods based on mass spectrometry have been already applied to solve the problem of fish authentication. Mazzeo et al.19 developed a molecular profiling strategy based on the use of MALDI-TOF MS as a rapid screening technique to characterize the protein extracts of different fish species. They highlighted the importance of selecting suitable specific protein biomarkers capable of verifying product authenticity in both fresh and processed fish products. In another report, Carrera et al.6 carried out a classical proteomic approach comprising protein extraction, two-dimensional gel electrophoresis, in-gel digestion of the target protein bands, and analysis of the generated peptides by MALDI-TOF and LC-MS/MS to develop a routine method for discriminating between different fish species of the Merlucciidae family. Contrary to this, the use of mass spectrometry for the identification of meat species has been very limited to date. Taylor et al.20 highlighted the potential of mass spectrometry in identifying the different meats that can be present in a mixture in both raw and cooked

research articles

Proteomic Detection of Chicken Meat products. Here the differences existing in the mass value of hemoglobins and myoglobins of the different meat species were used as discriminative criterion. Using this approach, they were able to detect the presence of 10% horse hemoglobin in a mixture with beef hemoglobin. They reported the difficulty to lower this detection limit, together with working with meat mixtures instead of pure proteins. In the present work, it has been shown that the use of myosin light chain can provide an adequate source of peptide biomarkers. The use of the OFFGEL fractionation procedure provides a convenient reproducible means of enrichment, for which in-solution trypsin digestion and analysis by LC-ESI-MS/MS can be performed.

Conclusion The present manuscript describes and demonstrates the development of a robust, accurate, and sensitive method to detect low amounts of chicken meat in other types of meat mixtures using a proteomic approach. By implementing an OFFGEL enrichment fractionation step and MS detection by conventional LC-ion trap-MS/MS, it was possible to detect as low as 0.5% w/v contaminating chicken in pork meat with high confidence due to the acquisition of discriminating sequence information. The discriminating power of this approach is based on detection of chicken-specific peptides originated from trypsin digestion of previously enriched myosin light chain 3, being comparable to methods based on DNA analysis. This proteomic approach displays more robustness in addressing some of the major limitations of DNA-based methods, such as optimization of the extraction procedures according to the different matrices and recovery of samples with high DNA quality. From this perspective, the primary amino acid sequence of key peptide biomarkers used here will be considerably more resistant to food processing than DNA sequences. This makes the proteomic approach a serious alternative or addition to the current methods for meat speciation. Future work should take the experience gained in this work to develop new assays based on proteomic technology capable of identifying other meat species in processed meat products.

Acknowledgment. This project was financially supported by UK Food Standards Agency (Project Q01104). Special thanks to Enrique Sentandreu for assistance in the use of the ion trap and to Chris Gerrish for his valuable technical support. Supporting Information Available: Table of total peptide masses matched for beef myosin light chain I (Table S1), chicken myosin light chain 1 (Table S2), pork myosin light chain 1 (Table S3), beef myosin light chain 2 (Table S4), chicken myosin light chain 2 (Table S5), pork myosin light chain 2 (Table S6), and chicken myosin light chain 3 (Table S7) from bands of fresh and cooked urea/thiourea soluble extracts. This material is available free of charge via the Internet at http:// pubs.acs.org. References (1) Aristoy, M. C.; Soler, C.; Toldra, F. A simple, fast and reliable methodology for the analysis of histidine dipeptides as markers of the presence of animal origin proteins in feeds for ruminants. Food Chem. 2004, 84 (3), 485–491.

(2) Vallejo-Cordoba, B.; Gonzalez-Cordova, A. F.; Mazorra-Manzano, M. A.; Rodriguez-Ramirez, R. Capillary electrophoresis for the analysis of meat authenticity. J. Sep. Sci. 2005, 28 (9-10), 826– 836. (3) Hsieh, Y. H. P.; Sheu, S. C.; Bridgman, R. C. Development of a monoclonal antibody specific to cooked mammalian meats. J. Food Prot. 1998, 61 (4), 476–481. (4) Ebbehoj, K. F.; Thomsen, P. D. Species differentiation of heated meat-products by DNA hybridization. Meat Sci. 1991, 30 (3), 221– 234. (5) Woolfe, M.; Primrose, S. Food forensics: using DNA technology to combat misdescription and fraud. Trends Biotechnol. 2004, 22 (5), 222–226. (6) Carrera, M.; Can ˜ as, B.; Pin ˜ eiro, C.; Va´zquez, J.; Gallardo, J. M. De novo mass spectrometry sequencing and characterization of species-specific peptides from nucleoside diphosphate kinase B for the classification of commercial fish species belonging to the family Merlucciidae. J. Proteome Res. 2007, 6 (8), 3070–3080. (7) Leitner, A.; Castro-Rubio, F.; Marina, M. L.; Lindner, W. Identification of marker proteins for the adulteration of meat products with soybean proteins by multidimensional liquid chromatography Tandem mass spectrometry. J. Proteome Res. 2006, 5 (9), 2424– 2430. (8) Bradford, M. M. Rapid and sensitive method for quantitation of microgram quantities of protein utilizing principle of protein-dye binding. Anal. Biochem. 1976, 72 (1-2), 248–254. (9) Laemmli, U. K. Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature 1970, 227 (5259), 680– 685. ´ . Species-specific detection of (10) Jańosi, A.; Ujhelvi, G.; Gelencse´r, E poulty in meat model mixtures and commercial sausage-products by polymerase chain reaction-restriction fragment length polymorfism analysis. In Rapid Methods for Food and Feed Quality Determination; van Amerongen, A., Barug, D., Lauwaars, M., Eds.; Wageningen Academic Publishers: Wageningen, 2007; pp 189196. (11) Koppel, R.; Ruf, J.; Zimmerli, F.; Breitenmoser, A. Multiplex realtime PCR for the detection and quantification of DNA from beef, pork, chicken and turkey. Eur. Food Res. Technol. 2008, 227 (4), 1199–1203. (12) Martin, R.; Wardale, R. J.; Jones, S. J.; Hernandez, P. E.; Patterson, R. L. S. Monoclonal-antibody sandwich ELISA for the potential detection of chicken meat in mixtures of raw beef and pork. Meat Sci. 1991, 30 (1), 23–31. (13) Liu, L. H.; Chen, F. C.; Dorsey, J. L.; Hsieh, Y. H. P. Sensitive monoclonal antibody-based sandwich ELISA for the detection of porcine skeletal muscle in meat and feed products. J. Food Sci. 2006, 71 (1), M1–M6. (14) Owusu-Apenten, R. K. Speciation of meat proteins by enzymelinked immunosorbent assay. In Food Protein Analysis, CRC Press: New York, 2002; pp 247-280. (15) Chassaigne, H.; Norgaard, J. V.; Hengel, A. J. Proteomics-based approach to detect and identify major allergens in processed peanuts by capillary LC-Q-TOF (MS/MS). J. Agric. Food Chem. 2007, 55 (11), 4461–4473. (16) Leitner, A.; Castro-Rubio, F.; Marina, M. L.; Lindner, W. Identification of marker proteins for the adulteration of meat products with soybean proteins by multidimensional liquid chromatography Tandem mass spectrometry. J. Proteome Res. 2006, 5 (9), 2424– 2430. (17) Ocan ˜ a, M. F.; Fraser, P. D.; Patel, R. K.; Halket, J. M.; Bramley, P. M. Mass spectrometric detection of CP4 EPSPS in genetically modified soya and maize. Rapid Commun. Mass Spectrom. 2007, 21 (3), 319–328. (18) Ocan ˜ a, M. F.; Fraser, P. D.; Patel, R. K. P.; Halket, J. M.; Bramley, P. M. Evaluation of stable isotope labelling strategies for the quantitation of CP4 EPSPS in genetically modified soya. Anal. Chim. Acta 2009, 634 (1), 75–82. (19) Mazzeo, M. F.; De Giulio, B.; Guerriero, G.; Ciarcia, G.; Malorni, A.; Russo, G. L.; Siciliano, R. A. Fish authentication by MALDITOF mass spectrometry. J. Agric. Food Chem. 2008, 56 (23), 11071– 11076. (20) Taylor, A. J.; Linforth, R.; Weir, O.; Hutton, T.; Green, B. Potential of electrospray mass-spectrometry for meat pigment identification. Meat Sci. 1993, 33 (1), 75–83.

PR9008942

Journal of Proteome Research • Vol. 9, No. 7, 2010 3383