Jana Hajšlová, Milena Zachariasova and Monika Tomaniova

Jana Hajšlová, Milena Zachariasova and Monika Tomaniova II International Congress Food Technology, Quality and Safety October 28 29, 2014, Novi Sad, Serbia

TALK SUMMARY Food scandals, fraud Traceability needs Food integrity concept Authentication Metabolomics, laboratory tools Case study: saffron Conclusions

Recent food fraud incidents 2008 Melamine in Chinese milk products 54,000 babies hospitalised, 6 deaths High volume low price Long term fraud due to deficient analytical methods 2012 Czech Republic methanol in spirits 42 deaths in Eastern Europe Short term crude fraud- easy to detect but high profits 2013 Horsemeat in Europe No food safety issues, relatively easy to detect High volume low price, Long term fraud(?) due to lack of intelligence/surveillance?

Establishing FOOD FRAUD system Annual report 2013

Traceability required by EU legislation Artile 3, paragraph 15 Traceability means the ability to trace and follow a food, feed, food-producing animal or substance intended to be, or expected to be incorporated into a food or feed, through all stages of production, processing and distribution

TRACEABILITY - questions to be answered: Is the sample typical of the material which it claims to be (compliance with label / certificate?) Is the material typical for the batch of product from which it came? From which farming system is the product coming and what is its processing history? What was the distribution / location of the product after delivery? Tracing food commodities in Europe

Design of traceability system 1. General considerations 2. Choice of objectives 3. Products to achieve the objectives 4. Identify suppliers and customers 5. Flow of materials under its control 6. Information requirements 7. Establish procedures documenting the flow of products and information 8. Establish documentation needed to achieve the objectives

FOOD INTEGRITY Primary Production Storage Processing Packaging Distribution Retail Product Consumption Verification procedures + Supply Chain Management, Risk management procedures

FOOD INTEGRITY: a comprehensive concept

Safety issues of concern (EFSA, 2013) marine and non-marine bio-toxins, novel food ingredients, pesticides, hydrocarbons, persistent organic pollutants, plasticisers, residues of medical products, Heavy metals.and cocktails thereof

Emerging risks definition (EFSA, 2007) (i) New hazard + High exposure (ii) Known hazard + Newly identified high exposure NEW RISKS (iii) Known hazard + Increased sensitivity + High exposure

FOOD AUTHENTICITY AND FRAUD Important food quality parameter Most of valued food commodities are subject to fraud Substitution or extension by cheaper product / ingredient / raw material Geographic origin misdeclaration Use of undeclared technology / processing ECONOMIC ASPECTS CONSUMERS DECEPTION THREAT TO CONSUMERS HEALTH

RESEARCH INTERESTS Journal of Food Science Vol. 77, Nr. 4, 2012

RESEARCH INTERESTS The top 7 ingredients represented more than 50 % of the scholarly records in the database and included: olive oil milk honey SAFFRON orange juice coffee apple juice

METABOLOMICS: analytical definition METABOLOMICS is a comprehensive analysis of the metabolome, focused on the broadest possible range of small molecules (<1200 Da) without a particular bias to specific groups of metabolites: METABOLOMIC FINGERPRINTING: NON-TARGET analysis with minimum sample preparation. (Example: wine). Metabolomic profiling: analysis of a SPECIFIC GROUP of metabolites. (Example: polyphenols in wine).

Application scope of metabolomics 2008 This review focuses on the recent trends and potential applications of metabolomics in four areas of food science and technology: (1) food component analysis; (2) Food quality / authenticity assessment (3) food consumption monitoring (4) physiological monitoring in food intervention or diet challenge studies.

A new keyword in food characterization 2013

Instrumental platforms for metabolomics AMS-HRMS LC HRMS NMR ICP MS GC-HRMS METABOLOME IR / Raman

What is saffron? Dried stigma (~20 mm long) Dried stigma of Crocus sativus Crocus sativus (ULCM) The most expensive spice in the world (10 15 /g) Parts of saffron flower (ULCM) Unique properties characteristic color, taste & alluring aroma Swedish Saffron buns Cooking French Bouillabaisse Italian Milanese Risotto Persian dessert Sholezard Spanish Paella Indian Biryani Anticarcinogenic Antimutagenitc Immunomodulation Antioxidant u Traditional medicine Coloring Perfumery Buddhist monks

Saffron production PRODUCTION 2004 Worldwide = 170 tones Europe = 6800 kg = 4% WW SAFFRON TREATMENT Spain Morocco Italy Greece Iran India 1. Harvesting 4. Cleaning 3. Drying 2. Separation 5. Packaging

Fraud practices on saffron The use of other parts: petals or leaves. Diluted saffron using other materials. The use of other products containing crocetin (Gardenia jasminoides). The adulteration using dyes (synthetic). False information about the country of origin (most valuable saffron is produced in PDO e.g. La Mancha, less valuable saffron from Iran or India).

Saffron fraud in newspaper This article described the real situation in Spain. ONLY 1% of saffron, which was sold in the Spanish market was cultivated in Spain. So, where is this saffron coming from?

Saffron authentication tools: past and current Former trends Current trends Physico-chemical & Biochemical measurements FINGERPRINTING PROFILING Single or only a few markers TARGET analysis of specific metabolites misses a large part of molecular information Comprehensive information on sample composition NONTARGET metabolome fingerprinting methods overcome these limitations

Case study: classification of 44 saffron samples set PDO, La Mancha & Aragon Saffron Packed in Spain Turkey PDO, Greek India

Sample preparation (final method) Homogenization Crushed by pressure in paper envelope UHPLC-ESI(+)/(-) QTOFMS Extraction 50 mg of sample + 5 ml EtOH/H 2 O (70/30; v/v) 1 h ultrasonic bath E X T R A C T Centrifugation

Metabolomic fingerprinting workflow MarkerView (AB sciex) DATA PROCESSING DATA PRE-TREATMENT Peak picking Filtering Alignment Normalization PCA OPLS-DA Cross-Validation Pareto scaling MULTIVARIATE ANALISIS (SUPERVISED & UNSUPERVISED ANALISIS) MarkerView & SIMCA INTERPRETATION CONFIRMATION OF STRUCTURE ELEMENTAL COMPOSITION, IDENTIFICATION MARKER IDENTIFICATION IDA Explorer PeakView (Formula Finder) ONLINE DATABASES HRMS MS/MS PeakView

MarkerView MarkerView software can process data acquired from non-classified workflows using Principal Components Analysis (PCA) or Principal Component Analysis-Discriminant Analysis (PCA-DA) ( PLS-DA ). DATA PROCESSING: Select samples (44 samples) Peak picking (saffron metabolome (0.4-14 min)) Filters and Alignment (RT and m/z values)

Multivariate Data Analysis of Saffron: PCA (ESI+) PCA (unsupervised pattern) 1. The first step of the data analysis in order to detect patterns in the measured data. 2. Positive data, PC1 describes 35.6% of the variability in the data and PC2 11.9% of variability, therefore 47.5% of variability. Spain PDO La Mancha & Aragon

Data dimensionality reduction DATA DIMENCIONALITY: 2317 molecular features (MF) 69 Positive Data Matrix 2317 MF Monoisotopic Peaks 1620 MF Frequency of occurrence in the samples (50%) 563 MF t-test 69 MF

Multivariate data analysis of saffron: PCA (ESI+) Frequency of occurrence in the samples (50%) (Noise values)

Multivariate data analysis of saffron: PCA (ESI+) m/z = 798.6 Frequency of occurrence in the samples (50%) EXCLUDED

Multivariate data analysis of saffron: PCA (ESI+) These samples were removed. Their characteristic markers did not correspond to saffron metabolome. Spain PDO La Mancha & Aragon The statistical model (OPLS- DA, supervised model) was performed using Spanish samples and La Mancha- Aragon (PDO) samples

Multivariate data analysis of saffron: SIMCA, OPLS-DA, ESI+ Spain PDO The quality of the model, saffron origin, was evaluated by the goodness-of- Fit parameter (R 2 X=0.96), the proportion of the variance of the response variable that is explained by the model (R 2 Y=0.89) and the predictive ability parameter (Q 2 =0.86). Seven variables were used.

Multivariate analysis of saffron: PCA (ESI+) OPLS-DA Score plot Variable trend plot m/z 798.5719 RT 11.7 Marker PDO Spain Average

Identification of markers: workflow Selection of markers using MarkerView PeakView : Its mass (m/z), retention time (RT) and MS/MS Formula Finder (Molecular formula) IDA Explorer (MS/MS Pathway) Libraries: MassBank, METLIN, MMCD, CSFMetabolome, DrugBank, LMSD, PubChem, KEGG, BioCyc, MetaCyc, HumanCyc, Reactome

Marker identification: saffron (PDO) FORMULA FINDER: 1. MS (accurate mass) 2. Isotopic pattern (Theoretical vs. Experimental) 3. MS/MS data, fragment ions 47 Candidates C 44 H 80 NO 9 P OXIDIZED GLYCEROPHOSPHOLIPIDS

Marker identification: Saffron (PDO) ESI+ ESI- ESI+ ESI- PC 36:4 PC 18:2/18:2

Marker identification: Saffron (PDO) PCA Loading plot Oxidized glycerophospholipids C 44 H 80 NO 9 P Glycerophospholipids C 44 H 80 NO 8 P

Oxidations a consequence of drying process It is the most important and most delicate task during which the stigmas lose 20% of their initial weight and turn into the saffron spice. The drying methods vary slightly between regions: In Sardinia, a process called feidatura (Olive oil) and a constant temperature of almost 45 C. In Western Macedonia (Greece), the fresh stigmas are spread out in thin layers, placed on rectangular silk sieves and stored for 12 to 24 hours in a room with controlled temperature of approximately 25 and 30 C. In La Mancha, thin layers (2cm) of fresh stigmas are placed on silk or metal sieves and are exposed to higher temperatures such as butane gas fire, or vine coals and heaters or coal operated stoves. As far as drying time is concerned, they prefer the shortest time of about half an hour and a highest temperature of 70 C.

Tentative marker identification m/z RT Molecular Formula Observation 798.5674 11.7 C 44 H 80 NO 9 P 812.6164 12.6 C 46 H 86 NO 8 P 810.6057 12.4 C 46 H 84 NO 8 P 838.6363 12.7 C 48 H 88 O 8 NP 820.5855 12.4 C 47 H 82 O 8 NP 796.5504 11.6 C 44 H 78 NO 9 P PDO Oxidized PC 36:4 PC 18:2/18:2 PDO PC 38:3 PDO PC 38:4 PC 18:1/20:3 PDO PC 40:4 PDO PC 39:6 PDO Oxidized PC 36:5 353.2311 7.8 Unknown Spain

Screening records of saffron samples analysisi for dyes and addition of other spices AIM OF THIS STUDY Dyes Plants/Spices Adulteration Fraud SLE RETROSPECTIVE DATA SEPARATION (U)HPLC HRMS(/MS) IONIZATION ESI+/ESI- MS ANALYSIS Scan HRMS MS/MS

Fraud? (case No.1: Saffron from Czech e-shop) Target screening to artificial colorants In the most suspicious sample three artificial colorants were found O O S HO N NH O HO O S O S O HO O OH S O O O S OH O N NH O OH Azorubine (E122) HO O S O N N O OH N N Tartrazine (E102) O S OH O Ponceau 4R (E124)

Fraud? (case No.2: Saffron from Turkish market) Characteristic marker: m/z 611.1616 / 2.86 min BPC MS MS/MS Molecular formula (from exact mass): C 27 H 32 O 16 FORMULA FINDER Theoretical vs. measured isotopic profile

Identification of the turkish sample Searching in databases: Saffron Safflower Hydroxysafflor yellow A Yellow colorant of safflower (Carthamus tinctorius), also called bastard saffron.

CONCLUSIONS Metabolomic fingerprinting using LC-ESI-QTOFMS is a suitable tool for fast quality assessment of saffron. UHPLC-HRMS(/MS) fingerprinting analysis provides sufficient discrimination power and information to discriminate saffron origin. Metabolomic fingerprinting (ESI+), OPLS-DA, allowed real Spanish saffron to be distinguished between saffron cultivated in Spain and packaged in Spain. Glycerophospolipids and their lipid oxidation are the most significant markers. Relatively high number (around 15%) of samples seem to be adulterated, as well as label Spanish samples were from unknown origin.

Sample preparation HOMOGENIZATION & SAMPLE PREPARATION 50 mg Grinding Weighting SOLID PHASE MICROEXTRACTION (SPME) SPME Fiber: Incubation: Extraction: Desorption: 100 µm PDMS 50/30 µm DVB/CAR/PDMS 60 µm PDMS/DVB 40 C (10 min), 50 C (5 min) 40 C (20 min), 50 C (5 min) 250 C (1 min) in injection port

GC QTOFMS (Quadrupole-Time of Flight) Agilent 7200 GC-QTOFMS MODES OF OPERATION TOF single MS QTOF tandem MS/MS Electron ionisation (EI) Chemical ionisation (CI) - Positive (PCI) - Negative (NCI) TOF (MS) Full spectral information mass spectra library (NIST) search High resolution & mass accuracy low mass error identification of unknowns QTOF (MS/MS) Accurate mass product ion spectra identity confirmation High selectivity & sensitivity (ultra)trace analysis

PCA: country of origin (packaging) Compounds influencing the distribution

PCA: country of origin (packaging) Italy Greece Spain Morocco Iran India 23

Significance analysis: PDO others Entities are filtered based on their p-values from statistical analysis. 2,5,5-Trimethylcyclohex-2-enon Ethanone, 2-(formyloxy)-1-phenyl- 2-Hydroxy-3,5,5-trimethyl-cyclohex-2-enone 5-Heptenal, 2,6-dimethyl- Compounds influencing the distribution Cyclopentane, 1,1-dimethyl- 25

Saffron adulteration x10 7 1.2 1 0.8 0.6 0.4 0.2 0 x10 1 0.8 0.6 0.4 0.2 0 7 VV5/14 Saffron (whole stigma) PDO La Mancha Spain Bought in 2014 VV11383 Powdered material Unknown origin Bought in 2013 EI; TIC m/z 30 400 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 Time (min) 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 Time (min) 28

Adulteration by curcuma CURLONE IUPAC name: (6S)-2-Methyl-6-[(1S)-4-methylene-2-cyclohexen-1-yl]-2-hepten-4-one Molecular formula: C 9 H 14 O Natural occurrence: Curcuma Ar-TUMERONE Molecular formula: C 12 H 22 O IUPAC name: (1 R, 6S)-2-methyl-6-(4-methylcyclohexa-2,4-dienyl)hept-2-en-4-one Formation: Dehydrogenation of curlone Properties: Insecticidal and repellent effects 29

Identification of curlone VV11383 - Powdered material, country of origin Turkey, bought in 2013 x10 7 1 0.8 0.6 0.4 0.2 EI; TIC m/z 30 400 Pure integration 130 peaks 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 Time (min) 1. DECONVOLUTION (EI) Deconvolution 191 peaks x10 6 5 4.5 4 3.5 3 2.5 2 1.5 1 0.5 0 Cpd 170 22.2 22.4 22.6 22.8 23 Time (min) 30

CONCLUSIONS GC QTOF MS Unknown compound identification: Full spectral information Mass spectral library search Exact mass Molecular ion (PCI) Product ion spectra Identity confirmation Agilent 7200 GC-QTOFMS SAFRON AUTENTICATION Metabolomics fingerprinting Differentiation according to Country of origin Protected destination of origin Form of material Country of purchase 35

Ambient mass spectrometry employing DART ion source Direct Analysis in Real Time DART TOFMS DART orbitrapms AccuTOF LP (Jeol) Time-of-flight mass spectrometer ~ 5000 7000 fwhm Exactive (Thermo Scientific) Orbitrap mass spectrometer ~ 10,000 100,000 fwhm

Relative Abundance DART MS PRINCIPLE MASS SPECTRUM of entire sample VV_9047-1_9052-2_pos #287-337 RT: 1.02-1.19 AV: 51 SB: 38 0.92-0.98, 1.23-1.29 NL: 1.12E6 T: FTMS + p NSI Full ms [50.00-1000.00] 133.0645 100 90 80 70 60 50 205.1946 354.3357 382.3669 40 30 20 10 0 221.1894 298.2732 337.3091 109.1011 240.2316 426.3932 179.1426 81.0700 409.3767 149.1321 478.4605 502.4603 439.3558 518.4553 593.4913 100 150 200 250 300 350 400 450 500 550 600 m/z excited He 200-400 C THERMO DESORPTION, APCI ionization Is it really organic? BioFach 2014

Does contain this food supplement healthy sea buckthorn oil? SAMPLES: 1. Sea Buckthorn Oil (Reference material) 2. Sunflower Oil (Reference material) 3. Sea Buckthorn Oil Pills (Commercial sample)

DART HRFMS mass spectra of the methanol water extracts at 250 C (+) Sea Buckthorn oil Sunflower oil SEA BUCKTHORN PILLS

DART TOFMS mass spectra of oils diluted with toluene 1:50 (v/v) at 450 C (+) Sea Buckthorn oil SEA BUCKTHORN PILLS Sunflower oil Sea Buckthorn oil

DART- HRMS FINGERPRINTS Sea Buckthorn oil Sunflower oil SEA BUCKTHORN PILLS

DART m/z: 537,4455 β-carotene identification MS SEA BUCKTHORN PILLS MS/MS 78,2 % of matches were found between theoretical and observed pathway..

Conclusion SUNFLOWER OIL + monoglycerolss (emulsifiers) + beeta-carotene (provitamin A mix)

Join us for discussion of future challenges!

7th FP EU project Ensuring the Integrity of the European food chain

Overall strategy

WP11: Dissemination & Knowledge Transfer WP leader: Jana Hajslova, ICT Prague Kick-off meeting FoodIntegrity 25-26 February 2014, FERA, York, UK

Training activities (4) TRAINERS: Well established experts institutes will act as key trainers centers and also contribute to preparation of other training materials: IAEA/FAO, Vienna, Austria CRA-W, Gembloux, Belgium RIKILT, Wageningen UR, The Netherlands QUB, Belfast, UK VSCHT, Prague, Czech Republic FiBL, Switzerland FERA, York, UK BfR, Berlin, Germany NOFIMA, Norway Barrila, Italy

Approach for development of appropriate training program To identify needs / priorities of industry and other stakeholders To exploit both existing intelligence and knowledge generated by the project To develop training program for interested stakeholders to fulfill their expectations and needs For transfer of achieved outcomes / new generated knowledge to end-users several concepts will be applied Commodity based concept Analytical methodologies based concept Other concept(s) (consumer issues, traceability, chemometrics) A network of competent laboratories / intelligence owners (European network of competence for analytical techniques in food authentication OR Food Authenticity / Fraud European Training Labs Network) will be established to provide training in specific technologies enabling food quality assessment and authentication

Thank you for your kind attention jana.hajslova@vscht.cz