Quality assessment of saffron (Crocus sativus L.) extracts via UHPLC-DAD-MS analysis and detection of adulteration using gardenia fruit extract (Gardenia jasminoides Ellis)

Accepted Manuscript Quality assessment of saffron (Crocus sativus L.) extracts via UHPLC-DADMS analysis and detection of adulteration using gardenia fruit extract (Gardenia jasminoides Ellis) Moras Benjamin, Loffredo Loïc, Rey Stéphane PII: DOI: Reference:

S0308-8146(18)30452-7 https://doi.org/10.1016/j.foodchem.2018.03.025 FOCH 22568

To appear in:

Food Chemistry

Received Date: Revised Date: Accepted Date:

12 October 2017 3 March 2018 7 March 2018

Please cite this article as: Benjamin, M., Loïc, L., Stéphane, R., Quality assessment of saffron (Crocus sativus L.) extracts via UHPLC-DAD-MS analysis and detection of adulteration using gardenia fruit extract (Gardenia jasminoides Ellis), Food Chemistry (2018), doi: https://doi.org/10.1016/j.foodchem.2018.03.025

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Quality assessment of saffron (Crocus sativus L.) extracts via UHPLC-DAD-MS analysis and detection of adulteration using gardenia fruit extract (Gardenia jasminoides Ellis) MORAS Benjamina, LOFFREDO Loïc b, REY Stéphanea Activ’Inside, route de Beroy, ZA Grand Cazau, 33750, BEYCHAC ET CAILLAU, Bordeaux, Area,

a

France b

Botanicert, 4 traverse Dupont, Espace Jacques Louis LIONS, 06130 Grasse, FRANCE

Corresponding authors: [email protected] (MORAS Benjamin) E-mail address: [email protected] (MORAS Benjamin) [email protected] (LOFFREDO Loïc) Highlights A new separation of saffron metabolites with ultra-high performance is proposed Picrocrocin derivative characterization is presented Iridoid analysis from gardenia should be used to detect frauds

Abstract A new UHPLC-DAD-MS method based on a Core-Shell particles column was developed to realize the rapid separation of saffron stigma metabolites (Crocus sativus L.). A single separation of 35 compounds included cis and trans-crocetin esters (crocins), cis-crocetin, trans-crocetin, kaempferol derivatives, safranal, and picrocrocin from pure saffron stigmas. This method permitted the detection of 11 picrocrocin derivatives as the typical group of compounds from saffron as well as the detection of gardenia-specific compounds as typical adulterant markers. The metabolite concentration in a Standardized Saffron Extract (SSE) was determined using the method described herein and by comparison to the ISO3632 conventional method. The safranal content was 5 to 150 times lower than the value of 2% that was expected via ISO3632 analyses. Using the same Core-Shell separation, geniposide detection appeared to be a relevant approach for detecting the adulteration of saffron by using gardenia.

1

1. Introduction Saffron, which is the most expensive spice in the world, refers to the dried stigma from Crocus sativus L. The manual picking of its three stigmas and the low mass produced by each plant explain its high cost. The use of saffron is very ancient: its earliest representation appeared approximately 4000 years ago in some paintings and ceramics of the Minoan civilization in the region of Crete (Dewan, 2015). Currently, Iran is known to be the most productive country, with more than 90% of the world’s production. Saffron is used traditionally for cooking or perfume, but it appears in several medical indications, as mentioned in recent works focused on in vitro, in vivo and clinical studies. Several potential health benefits of saffron consumption have been described, including mental health, wellbeing-promoting effects with anti-depressant and anti-anxiety effects, premenstrual syndrome, and men’s sexual health and fertility (Agha-Hosseini, Kashani, Aleyaseen, Ghoreishi, Rahmanpour, Zarrinara, et al., 2008; Akhondzadeh, Tahmacebi-Pour, Noorbala, Amini, Fallah-Pour, Jamshidi, et al., 2005; Ghadrdoost, Vafaei, Rashidy-Pour, Hajisoltani, Bandegi, Motamedi, et al., 2011; Heidary, Vahhabi, Reza Nejadi, Delfan, Birjandi, Kaviani, et al., 2008; Mazidi, Shemshian, Mousavi, Norouzy, Kermani, Moghiman, et al., 2016; Noorbala, Akhondzadeh, Tahmacebi-Pour, & Jamshidi, 2005; Papandreou, Tsachaki, Efthimiopoulos, Cordopatis, Lamari, & Margarity, 2011; Shiri, Koskimaki, Tammela, Hakkinen, Auvinen, & Hakama, 2007). The conventional method used to determine the quality of saffron is the international ISO3632/TS (2010). This official method describes the water solubilization of saffron metabolites, followed by three measures at 257 nm, 330 nm, and 440 nm for picrocrocin, safranal and crocin, respectively. To obtain the concentration of the different metabolites in dried stigmas or saffron extracts, this method is frequently used with calibration to a crocin or safranal standard. This method is commonly adapted to quantify the saffron components in Standardized Saffron Extract (SSE) for its standardization. Since Noorbala described a recommended dosage at 0.30–0.35 mg safranal for 15 mg of saffron extract (2% w/w), it has been used to standardize SSE (in reference clinical studies) at a 2% safranal content (Noorbala, Akhondzadeh, Tahmacebi-Pour, & Jamshidi, 2005). However, the analysis of safranal in SSE is not well described. To reach a 2% safranal content, the UV spectrometric method ISO3632 is commonly used. However, this method can lead to an overestimation of the safranal concentration, 2

according to recent research performed on saffron stigmas from different origins (García-Rodríguez, López-Córcoles, Alonso, Pappas, Polissiou, & Tarantilis, 2017). It appears essential to compare ISO3632 with an appropriate chromatographic analysis to more precisely standardize the concentrations of saffron metabolites. Different studies have described the identification and quantification of saffron metabolites using HPLC (High Performance Liquid Chromatography), including of safranal, crocins (crocetin esters), picrocrocin, kaempferol and its derivatives (Manuel Carmona, Zalacain, Sánchez, Novella, & Alonso, 2006; D'Archivio & Maggi, 2017; Han, Wanrooij, van Bommel, & Quye, 2017; Masi, Taiti, Heimler, Vignolini, Romani, & Mancuso, 2016). However, these compounds have been analysed using separate methods and a common reverse phase C18, which gave limited separation performance, especially for picrocrocin derivatives. In addition, the time consumption, cost, and limitation to examining a single family of compounds in each HPLC analysis are cumbersome. Similar to other botanical sources, the use of high-quality saffron powder in pharmaceutical preparations or dietary supplements has increased drastically. The physical characterization of botanical origin, spectrometry characterization and DNA analysis allow the certification of saffron stigma quality. However, the most frequently used physical form is a fine extract powder of stigma, in which some of the natural cell structures are destroyed. The lack of DNA (from saffron or adulterant extract) due to the processing conditions can limit the authentication of the species that are present in the extract. Along with milk, honey and olive oil, saffron is one of the most adulterated food products (Moore, Spink, & Lipp, 2012). HPLC methods aid in the identification of adulterant markers such as synthetic dyes (i.e., tartrazine). Extracts from gardenia fruit (Gardenia jasminoides Elis) can be used as a natural adulterant in saffron because this fruit contains many of the same crocins that saffron does, so the two plants have very similar chromatographic profiles. It has been demonstrated that the differentiation of gardenia and saffron crocins cannot be achieved, and research efforts should be devoted to improving the identification of gardenia extract adulterants (Manuel Carmona, Zalacain, Sánchez, Novella, & Alonso, 2006). Adulteration using gardenia extract will thus likely increase because of its low cost and availability. 3

Considering all quality requirements, this study presents a new, fast UHPLC-DAD/MS separation method for saffron compounds that uses a Core-Shell particle column to estimate the concentration of the main metabolites, identify the typical picrocrocin derivatives, which are not well described by classical LC separation (C18 phase with particle sizes greater than 3 µm), and detect adulteration with gardenia. In the first step, the identification of saffron metabolites is presented. Using the same chromatographic conditions, the identification of typical monoterpenoids gardenia and iridoids is proposed because they could represent markers of adulteration by gardenia.

2. Materials and Methods 2.1. Reagents and materials Dried saffron stigma samples, grade I, were obtained from Iranian producer in the Khorasan Region. Controls were realized to assess organoleptic, macroscopic, and microscopic properties and to check the absence of dyes according to ISO3632-2. Dried gardenia, saffron stigma and SSE samples were obtained from Iranian and French producers from a nutraceutical market. SEE n°1 to n°11 originated from nutraceutical suppliers (standardized at 2% safranal by ISO3632). In addition, adulterated saffron (extract n°12) was specifically designed for the study by adding 15% of Gardenia crocin extract to the saffron products. Others concentrations were also used to assess the minimum proportion detectable of gardenia crocin extract (0% to 100% w/w). Safranal was purchased from Sigma Aldrich (Ref W338907, purity ≥ 90%); β-cyclocitral from Sigma Aldrich (Ref 16976. purity ≥ 97%); crocin (trans-crocetin-4-GG) from Phytolab (Ref 80391, purity 99.04%), kaempferol-3-O-glucoside from Extrasynthese (Ref 1243 S batch purity 99%), transcrocetin from Toronto Research (Ref C794950, purity 95%). Two UHPLC systems were used: Waters™ H-Class™ with PDA, ELSD, and SQD1 detectors, with the Empower 3 software, and Thermo Scientific™ Ultimate™ 3000 Rapid Separation with a DAD detector. 2.2. UV spectrometry conventional analysis: ISO3632 For each saffron product, 0.5 g of powder was placed in a 1-L brown glass volumetric flask and stirred for 1 h at ambient temperature (20°C) with 900 mL of distilled water. After magnetic stirring, the flask

4

was completed, and the solution was mixed. The water extract was filtered through a 0.45-µm cellulose acetate filter. If necessary, the extract was diluted before measurement. The absorbances at 257 nm (1%) (λ max of picrocrocin), 330 nm (λ max of safranal) and 440 nm (λ max of crocins) were recorded. These compounds correspond, respectively, to the bitterness, flavour and colour strength of the raw material. Commonly, the quantity index of crocin

permits saffron to be categorized

in one of four levels. The highest Category I must have a value higher than 190. The following equation is used to determine the index quantity:

where factor;

is the absorbance at a specific wavelength (330 nm, 440 nm or 257 nm); is the weight of the tested sample (in g); and

is the dilution

is the content of water and volatiles, as

determined via desiccation at 105°C (in %). Standard calibrations were performed using the following correlations:

(for safranal) according to another study (García-Rodríguez, López-

Córcoles, Alonso, Pappas, Polissiou, & Tarantilis, 2017); and crocin);

(for β-cyclocitral), where

is the absorbance and

(for is the

concentration in mg/L in water. A conversion factor was used to convert the β-cyclocitral equivalent expression into picrocrocin according to their respective molecular weights. 2.3. Sample extraction before UHPLC After grinding to achieve a particle size less than 300 µm, 100 mg of dried stigma powder was placed in a 20-mL volumetric flask (or directly introduced into the volumetric flask for the extracts) and extracted using MeOH 80% in distilled water. The powder was first dispersed for 5 min via sonication at 30°C and extracted for 1 h via magnetic stirring, followed by a second sonication for 5 min at 30°C. This procedure permits the extraction of different metabolites during the same extraction compared with other MeOH concentrations. The solution was filtered through a 0.22-µm PTFE membrane before injection. For SEE, 100 mg of dried powder was directly used with the same protocol.

5

2.4. UHPLC-DAD-MS analysis The amount of extracted compounds was determined via the UHPLC-DAD analysis. A 1-μl sample was injected into a chromatographic column containing a Core-Shell silica reverse phase, Kinetex C18, Phenomenex 2.6 µm, 150 × 2.1 mm. Because ultra-high performance was obtained, the term UHPLC was kept according to the system requirement used and back pressure observed. The mobile phase was composed of 0.01% formic acid in water (A) and acetonitrile (B). Elution was performed using a 21-min multi-step gradient, starting with 1 min at 5% B; 1 min to 9 min from 5% to 40% of B; 9 min to 15 min from 40% to 100% of B; 15 min to 17 min at 100% of B; 17 min to 17.1 min from 100% to 5% of B; and 5% followed by an equilibrium plate at 5% of B. The eluent flow rate was 0.6 ml/min, and UV absorbance was measured at 250 nm, 310 nm, 350 nm, and 440 nm, respectively, for picrocrocin and its derivatives, safranal, flavonoids, and crocetin and ester derivatives (crocins). Complete UV spectra were determined for each peak using DAD acquisition. Mass detection was performed with Waters ACQUITY® SQD Electro Spray Ionization in positive mode. Capillary temperature and voltage were 350°C and 4 kV respectively. Nebuliser pressure was 65 psi and nitrogen flow rate 11 ml/min.

2.5.

Analytical standards for routine analysis

The differents compounds families and the analytical standards used are presented in Figure 1. Crocin analytical standard from gardenia has been used in some studies, but its purity is very low, as reported by Hadizadeh and others (Hadizadeh, Mohajeri, & Seifi, 2010). To our knowledge, only trans-4-GG crocin and trans-3-Gg crocin have been developed by analytical standard suppliers with high purity (≥95%) and were presented in recent works (Lautenschlager, Lechtenberg, Sendker, & Hensel, 2014). Because the cis forms of crocins are not available, only absorption coefficients have been mentioned by previous authors (Maggi, Sánchez, Carmona, Kanakis, Anastasaki, Tarantilis, et al., 2011). Cis and trans crocetin esters have shown similar absorption rates as those of aglycon crocetins. Pure trans-4-GG was used to quantify the crocetin ester family (crocins). Pure transcrocetin was used to quantify aglycon crocetins. For picrocrocin, a high-purity analytical standard is 6

not available. Therefore, cinnamic acid has been recently used as an equivalent compound (Masi, Taiti, Heimler, Vignolini, Romani, & Mancuso, 2016). Picrocrocin standard can be produced in-house (Cossignani, Urbani, Simonetti, Maurizi, Chiesi, & Blasi, 2014). However, it is still difficult to check the purity due to the high cost of analytical methods needed (NMR or infrared). To overcome this lack of analytical standard, β-cyclocitral is often used to estimate the picrocrocin content in routine use because of its similar structure . β-cyclocitral is assumed to be close to picrocrocin and its derivatives, which were estimated via β-cyclocitral calibration (considering the molecular weight due to the sugar moiety). The kaempferol derivative content was estimated using kaempferol-3-O-glucoside, and safranal was quantified using its proper standard.

3.

Results 3.1. UHPLC-DAD-MS profile of stigmas of Crocus sativus L

After optimizing the chromatographic conditions, the ultra-high performance separation of a large range of compounds from saffron was achieved using the Core-Shell particle column. All compounds were separated between 2 and 13 min of elution (Figure 2). The mass fragmentations and UV spectra were compared to the literature data (Manuel Carmona, Zalacain, Sánchez, Novella, & Alonso, 2006; Valle Garcia-Rodriguez, Serrano-Diaz, Tarantilis, Lopez-Corcoles, Carmona, & Alonso, 2014). The 15 main crocetin ester structures proposed by Carmona were also identified herein and corresponded to trans-crocetin or cis-crocetin esters containing 1 to 5 sugar moieties. In addition, the aglycon structures trans-crocetin and cis-crocetin were identified at the end of the run (Figure 2). The chromatographic conditions used in this study permitted 35 compounds to be separated in the same run. The separation of picrocrocin derivatives (or aroma precursors) appeared between 2 and 5 minutes (P1 to P11 in Figure 2). The 10 new structures proposed by Carmona, as aroma precursors, were also found in this study but with one more compound and slight differences. In addition, the Core-Shell approach proposed herein permitted kaempferol derivatives and safranal to be separated in the same run, in addition to crocin and picrocrocin derivatives (Manuel Carmona, Zalacain, Sánchez, Novella, & Alonso, 2006; Straubinger, Bau, Eckstein, Fink, & Winterhalter, 1998; Winterhalter & Straubinger, 2000). 7

Different fragmentations were identified corresponding to the osidic moieties linked to crocetin aglycones ([M+H]+ at 329 m/z, [M+Glucose+Na]+ at 513 m/z, [M+2Glucose+Na]+ at 675 m/z. [M+3Glucose+Na]+ at 837 m/z. [M+4Glucose+Na]+ à 999 m/z and [M+5Glc+Na]+ at 1161 m/z, 329 m/z). These typical fragments were found in the fragment patterns of crocetin ester derivatives (crocins). All of the flavonoids characterized in the saffron sample were kaempferol derivatives. They were integrated using UV data (at 350 nm) to identify the family using MS data. For all of the kaempferol derivatives, 287 m/z was a relevant fragment. The 6 derivatives were identified as glycoside derivatives with between 1 and 3 glucoses. Few analytical works have paid attention to the identification of picrocrocin derivatives. The name picrocrocin derivatives can also be given to safranal derivatives or aroma precursors with the specificity to present one or more sugar moiety. The compounds detected at 250 nm between 2 min and 5 min were assumed to be picrocrocin derivatives (Figure 2). According to this hypothesis and the works of Carmona and Straubinger, we propose a structural characterization of the family from the MS and UV characterization (Figure 3). The 250-nm wavelength seems to be characteristic of these metabolites. By comparing the analyses from Carmona and others with those determined in the present study, the picrocrocin derivatives’ profile is significantly different. Compound P1 (C16H24O8 with 5 unsaturations) showed UV λmax at 257 nm, which can be explained by the ring closure (Figure 3). A close structure has been described by Carmona and others but with an additional hydroxyl group, whereas Winterhalter and others described the same structure as that of compound P1, which is presented in (Figure 3) (Winterhalter & Straubinger, 2000). The aglycon form of this compound (TDOI) has been already described as a product degradation of crocins and a safranal precursor (M. Carmona, Zalacain, Salinas, & Alonso, 2006). The fragmentation patterns at m/z 345 corresponded to [M+H]+, m/z 367 to [M+Na]+, m/z 711 to [2M+Na]+ and m/z 183 for the loss of a sugar moiety [MC6H10O5+H]+. Compound P2 (C16H28O8 with 3 unsaturated bonds) showed a typical UV spectrum at 245 nm for a conjugated carbonyl or aldehyde and presented a signal at m/z 371 for [M+Na]+ as well as signals at m/z 169 and 185 that corresponded to the opening of the aglycon ring (according to the unsaturated bond number). Compound P3 (C20H30O13 with 6 unsaturated bonds) showed a particular 8

UV spectrum at 288 nm that was probably due to the linear configuration according to Carmona and others as well as signals at m/z 501 [M+Na]+ and m/z 155 [M- 2C6H10O5+H]+ that corresponded to the loss of two glucose units. The number of unsaturated bonds corresponded to 2 rings and 4 conjugated double bonds. Compound P4 (C16H26O4 with 4 unsaturations) presented a 250-nm UV spectrum and m/z 347 for [M+H]+, m/z 369 for [M+Na]+, m/z 715 for [2M+H]+ and m/z 185 for the loss of glucose [M-C6H10O5+H]+. Compound P5 (C15H24O7 with 4 unsaturated bonds) showed a 239 nm UV spectrum corresponding to the conjugated carbonyl in the ring. The mass fragmentation signal gave m/z 339 as [M+Na]+, m/z 633 as [2M+H]+ and m/z 155 corresponding to the loss of a sugar moiety [M-C6H10O5+H]+. Compound P6 (C15H24O7) is an isomer of compound P5. Compound P7 (C16H26O4 with 4 unsaturations) seems to be an isomer of compound P4. Compound P8 presents a UV spectrum similar to those of compounds P6 and P7, but no relevant structural information has been obtained. Compound P9 (C22H36O12 with 5 unsaturations) presents a UV spectrum at 251 nm that is associated with conjugated aldehyde and shows mass fragmentation signals at m/z 515 for [M+Na]+ and m/z 169 for the loss of two glucose units [M- 2C6H10O5+H]+. Compound P10 was identified as picrocrocin, with a mass fragmentation signal of the aglycon structure [M-C6H10O5+H]+ at m/z 169 and [M+H]+ and [M+Na]+, respectively, at m/z 331 and 353. Compound P11 seems to be identified as HTCC (4-hydroxy-2.6.6-trimethylcyclohex-1-enal) (Maggi, et al., 2011). This compound is a degradation product of picrocrocin, according to ageing process studies (M. Carmona, Zalacain, Salinas, & Alonso, 2006). For each compound, a structure has been suggested based on the literature, mass fragmentation and UV absorption (Figure 3). The identification of these compounds revealed the presence of a group of readily detectable metabolites that should be analysed in saffron powder to determine the origin of the saffron. In addition, adulterant detection was studied using the same chromatographic separation. Others compounds between C1 and C2 or between C5 and C6 have been not identified and should be more studied. The mass fragments were not well established: I1 had a λmax at 249 nm like picrocrocin derivatives, I2 may correspond to a trans-crocin and I3 had a particular λmax but could be tentatively related to carotenoid UV spectra.

9

3.2.

Iridoid as adulteration markers from Gardenia jasminoides L

Even if the chromatographic separation applied to gardenia fruit shows a high similarity between the chromatographic profiles at 440 nm (typical wavelength to quantify crocins), no clear distinctions can be obtained. The same crocins were found in gardenia: G11, G13, G14 and G15 are respectively the same crocins found in saffron as C3, C5, C7, and C8 respectively (Figure 4). According to the mass spectra, approximately 10 iridoids were identified after separation by using a core-shell column with the same conditions as for the saffron analysis. These compounds can be used as gardenia markers (“negative marker”) to authenticate saffron stigmas and extracts (Figure 4). In the case of a complete fraud by crude extract or gardenia fruit powder, the presence of these iridoids is a very simple proof of adulteration. However, in the case of gardenia crocin extract it is harder to detect.

10

3.3.

Determination of the saffron dietary supplement quality

According to the previous identifications, the method is applied for the first time to the SSE (Standardized Saffron Extracts with 2% safranal using the ISO3632 methodology) to determine the quality of extracts and verify the absence the detection of gardenia iridoids. The UHPLC-DAD-MS method presented in this study was applied to 11 SSEs and to a saffron extract that had been adulterated by gardenia crocins (Extract 12) with crocin extracts from gardenia, as described in the material and method and specifically designed for this study (Table 1). The concentrations of safranal obtained via the ISO3632 methodology ranged from 2.02% to 2.90%, except for extracts 5 and 11, which were very poor in safranal. The safranal content was more accurately determined by using the UHPLC-DAD-MS method, which gave a very different amount below 0.1%. This overestimation of safranal content using the ISO3632 methodology in the SSE was in accordance with a previous analysis of dried pure stigma (García-Rodríguez, López-Córcoles, Alonso, Pappas, Polissiou, & Tarantilis, 2017). These authors explained this difference by the absorption of ciscrocetin ester in the same wavelength of the λ max of safranal (approximately 310 – 330 nm). The highest concentration in safranal determined using UHPLC was obtained for n°1 and was 0.4% safranal. The significant differences observed from different SSE may be due to the origin of the stigma (D'Archivio & Maggi, 2017). These extracts, which are dedicated to the nutraceutical market, have been developed according to the first works based on the ISO3632 quantification (Noorbala, Akhondzadeh, Tahmacebi-Pour, & Jamshidi, 2005). For reference in clinical studies, the determination of safranal by using the ISO3632 method could be the first indicator for determining the SSE quality. However, to determinate the specific metabolite contents, the UHPLC method or HPLC equivalent method should be recommended. A very low content of crocins compared to other standardized extracts was observed in the extracts n°5 and n°11. The low amounts of crocins obtained using both the UHPLC-DAD-MS and ISO3632 methods can be correlated with the yellow colour of the two powders. The strong red colour was associated with a high content of crocins, even if crocins from gardenia were added.

11

Kaempferol-3-sophoroside and kaempferol-3-sophoroside-7-glucoside were detected and quantified in significant proportions in all SSE. The typical picrocrocin derivatives P1, P3, P9, P10 (picrocrocin) and P11 detected from pure stigmas were also detected in saffron extracts. The quantification of picrocrocin derivatives, crocins, kaempferol derivatives and safranal demonstrates the predominance of saffron in the extract and it could be sufficient to assess the quality of the SSEs, but the adulteration by gardenia is the most difficult fraud to detect. For the adulterated extract (n°12), the ISO3632 methodology showed an apparently high quality with a similar concentration as that observed for other saffron extracts. The crocin and picrocrocin derivative profiles were also very similar, according to the chromatograms where extract n°12 correspond to the 15% adulteration (Figure 5 A). Other levels of adulteration has been tested and presented in the second chromatogram (Figure 5 B) from 0% to 100% adulteration by mixing saffron with gardenia crocin. The presence of geniposide, was confirmed via the mass spectra, and the UV analysis permitted to detect the adulteration from 5% to 100% addition of crocin gardenia. For a complete or large substitution of saffron powder by gardenia, the presence of picrocrocin derivatives permits fraud to be detected. If a small amount of crocin extract from gardenia had been added, this method is able to detect a minimum addition of 5% (w/w). In addition to recommended other method detecting adulteration (synthetic or natural dyes), this method permits to quantify a very large range of hydroalcoholic soluble compounds and to detect the adulteration by crocin gardenia extract.

Conclusion A new UHPLC-DAD-MS separation obtained via the Core-Shell particle column permitted to identify and quantify more than 35 components from saffron, such as safranal, kaempferol derivatives, and crocetin ester (crocins), to be identified and quantified by using a single and fast analysis. The typical markers of saffron, named picrocrocin derivatives, were also characterized. Gardenia iridoids can be analysed at the same time as the marker of adulteration. This method allowed the determination of quality of raw material and saffron extracts. Standardized saffron extracts with 2% safranal using ISO3632 showed poor contents after analysis using UHPLC, and they were mostly less than 0.1%. 12

This method can be used in addition to spectrometric methods, DNA analysis and analysis of dyes.

13

3632, I. S. I. (2010). Spices - Saffron (Crocus sativus L.) - Part 2: Test methods. In ISO International standard, vol. ISO 3632-2 (pp. 1-42). Agha-Hosseini, M., Kashani, L., Aleyaseen, A., Ghoreishi, A., Rahmanpour, H., Zarrinara, A. R., & Akhondzadeh, S. (2008). Crocus sativus L. (saffron) in the treatment of premenstrual syndrome: a double-blind, randomised and placebo-controlled trial. BJOG, 115(4), 515-519. Akhondzadeh, S., Tahmacebi-Pour, N., Noorbala, A. A., Amini, H., Fallah-Pour, H., Jamshidi, A. H., & Khani, M. (2005). Crocus sativus L. in the treatment of mild to moderate depression: a double-blind, randomized and placebo-controlled trial. Phytother Res, 19(2), 148-151. Carmona, M., Zalacain, A., Salinas, M. R., & Alonso, G. L. (2006). Generation of saffron volatiles by thermal carotenoid degradation. J Agric Food Chem, 54(18), 6825-6834. Carmona, M., Zalacain, A., Sánchez, A. M., Novella, J. L., & Alonso, G. L. (2006). Crocetin Esters, Picrocrocin and Its Related Compounds Present in Crocus sativus Stigmas and Gardenia jasminoides Fruits. Tentative Identification of Seven New Compounds by LC-ESI-MS. Journal of Agricultural and food chemistry, 54(3), 973-979. Cossignani, L., Urbani, E., Simonetti, M. S., Maurizi, A., Chiesi, C., & Blasi, F. (2014). Characterisation of secondary metabolites in saffron from central Italy (Cascia, Umbria). Food Chem, 143, 446451. D'Archivio, A. A., & Maggi, M. A. (2017). Geographical identification of saffron (Crocus sativus L.) by linear discriminant analysis applied to the UV-visible spectra of aqueous extracts. Food Chem, 219, 408-413. Dewan, R. (2015). Bronze Age Flower Power: The Minoan Use and Social Significance of Saffron and Crocus Flowers. Chronika, Volume V, 42. García-Rodríguez, M. V., López-Córcoles, H., Alonso, G. L., Pappas, C. S., Polissiou, M. G., & Tarantilis, P. A. (2017). Comparative evaluation of an ISO 3632 method and an HPLC-DAD method for safranal quantity determination in saffron. Food Chemistry, 221, 838-843. Ghadrdoost, B., Vafaei, A. A., Rashidy-Pour, A., Hajisoltani, R., Bandegi, A. R., Motamedi, F., Haghighi, S., Sameni, H. R., & Pahlvan, S. (2011). Protective effects of saffron extract and its active constituent crocin against oxidative stress and spatial learning and memory deficits induced by chronic stress in rats. Eur J Pharmacol, 667(1-3), 222-229. Hadizadeh, F., Mohajeri, S. A., & Seifi, M. (2010). Extraction and purification of crocin from saffron stigmas employing a simple and efficient crystallization method. Pak J Biol Sci, 13(14), 691698. Han, J., Wanrooij, J., van Bommel, M., & Quye, A. (2017). Characterisation of chemical components for identifying historical Chinese textile dyes by ultra high performance liquid chromatography – photodiode array – electrospray ionisation mass spectrometer. Journal of Chromatography A, 1479, 87-96. Heidary, M., Vahhabi, S., Reza Nejadi, J., Delfan, B., Birjandi, M., Kaviani, H., & Givrad, S. (2008). Effect of saffron on semen parameters of infertile men. Urol J, 5(4), 255-259. Lautenschlager, M., Lechtenberg, M., Sendker, J., & Hensel, A. (2014). Effective isolation protocol for secondary metabolites from Saffron: Semi-preparative scale preparation of crocin-1 and trans-crocetin. Elsevier, 6. Maggi, L., Sánchez, A. M., Carmona, M., Kanakis, C. D., Anastasaki, E., Tarantilis, P. A., Polissiou, M. G., & Alonso, G. L. (2011). Rapid determination of safranal in the quality control of saffron spice (Crocus sativus L.). Food Chemistry, 127(1), 369-373. Masi, E., Taiti, C., Heimler, D., Vignolini, P., Romani, A., & Mancuso, S. (2016). PTR-TOF-MS and HPLC analysis in the characterization of saffron (Crocus sativus L.) from Italy and Iran. Food Chemistry, 192, 75-81. Mazidi, M., Shemshian, M., Mousavi, S. H., Norouzy, A., Kermani, T., Moghiman, T., Sadeghi, A., Mokhber, N., Ghayour-Mobarhan, M., & Ferns, G. A. (2016). A double-blind, randomized and placebo-controlled trial of Saffron (Crocus sativus L.) in the treatment of anxiety and depression. J Complement Integr Med, 13(2), 195-199.

14

Moore, J. C., Spink, J., & Lipp, M. (2012). Development and application of a database of food ingredient fraud and economically motivated adulteration from 1980 to 2010. J Food Sci, 77(4), R118-126. Noorbala, A. A., Akhondzadeh, S., Tahmacebi-Pour, N., & Jamshidi, A. H. (2005). Hydro-alcoholic extract of Crocus sativus L. versus fluoxetine in the treatment of mild to moderate depression: a double-blind, randomized pilot trial. J Ethnopharmacol, 97(2), 281-284. Papandreou, M. A., Tsachaki, M., Efthimiopoulos, S., Cordopatis, P., Lamari, F. N., & Margarity, M. (2011). Memory enhancing effects of saffron in aged mice are correlated with antioxidant protection. Behavioural Brain Research, 219(2), 197-204. Shiri, R., Koskimaki, J., Tammela, T. L., Hakkinen, J., Auvinen, A., & Hakama, M. (2007). Bidirectional relationship between depression and erectile dysfunction. J Urol, 177(2), 669-673. Straubinger, M., Bau, B., Eckstein, S., Fink, M., & Winterhalter, P. (1998). Identification of Novel Glycosidic Aroma Precursors in Saffron (Crocus sativus L.). Journal of Agricultural and food chemistry, 46(8), 3238-3243. Valle Garcia-Rodriguez, M., Serrano-Diaz, J., Tarantilis, P. A., Lopez-Corcoles, H., Carmona, M., & Alonso, G. L. (2014). Determination of saffron quality by high-performance liquid chromatography. J Agric Food Chem, 62(32), 8068-8074. Winterhalter, P., & Straubinger, M. (2000). Saffron - Renewed interest in an ancient spice. Food Reviews International, 16(1), 39-59.

15

1

2

(3’)

3

4

N°

(4’)

Saffron coumpounds familly

N°

Crocetin 1

Crocetin ester derivatives (crocins) (R1 and R2 for Glu or Gen or triglucoside)

Analytical standard used trans-crocetin (R1=R2=H)

1

trans-4-GG (R1=R2=Gentiobiose)

2

Safranal

2

Safranal

3

Picrocrocin and derivatives

(3’)

β-cyclocitral

4

Kaempferol derivatives (R1 ; R2 ; R3 for differents sugars)

(4’)

Kaempferol-3-O-glucoside

Figure 1. Saffron coumpounds analysed and analytical standards used

16

C3

PDA (210-450nm) C5

P10 C8

250 nm

P10

P11

P8

C7

P6

P7

P9

P5 P2

P1P3

P1

P4

F5

C6 F1

ID P1 P2 P3 P4 F1 P5 P6 P7 P8 P9 F2 F3 P10 F4 F5 P11 F6 C1 I1 I2 C2 C3 C4 C5 I3 C6 C7 C8 C9 C10 C11 C12 C13 C14 S C15 C16 C17

Rt (UV) 2.38 2.66 2.99 3.55 3.64 3.77 3.83 3.91 3.97 4.07 4.35 4.48 4.53 5.00 5.43 5.76 5.76 6.16 6.21 6.31 6.38 6.73 7.09 7.29 7.70 7.94 8.73 9.18 9.31 9.34 10.03 10.10 10.42 10.52 10.91 10.10 11.82 12.45

P3

P6 P5 P9 P4

F3 F2

C1

P11

P8 P2 P1

P7

C4 I3

C10 C9

F6

S C11 C12

I1I2C2

C14 C13

C15

C16

C17

F4

df

UV (λmax nm) 195. 256 244 288 250 265. 346 239 238 250. 314 238 251 348 265. 339 250 265. 315 265. 347 251 370 263. 442. 466 249 418. 437 261. 437. 466 261. 441. 464 261. 441. 464 261. 441. 464 239.413 261. 440. 464 263. 326. 434. 460 259. 434. 459 261. 326. 434. 457 261. 326. 434. 456 261. 326. 434. 458 258. 434. 458 260. 323. 428. 452 259. 322. 428. 452 313 259. 322. 428. 452 256. 428. 453 258. 319. 422. 446

Mw (g.mol-1) 344 348 478 346 773 316 316 346 492 610 772 330 610 610 168 448 1138 1138 976 976 814 652 976 652 814 814 652 490 652 652 150 490 328 328

Fragmentation pattern (m/z) 167. 183. 345. 367. 383. 711 169. 185. 371. 387 155. 453. 501 167. 185. 347. 369. 715 287. 449. 611. 773 155. 224. 339. 367. 471. 633 137. 155. 317. 339. 633 155. 167. 185. 369. 385. 715 155. 309. 547 169. 183. 367. 515 287. 449. 611 287. 449. 611. 773 169. 331. 353 287. 449. 611 287. 449. 611 169 287. 449 329. 635. 675. 837. 1161 329. 1161 329. 653. 675. 999 329. 999 329. 491. 653. 675. 837 329. 491. 653. 675 329. 511. 635. 653. 675. 797. 999 329. 653. 675 329. 837 329. 837 261. 326. 434. 458 329. 491. 513 329. 675 329. 653. 675 151 329. 513 329 329

Tentative identification

Compound

Picrocrocin derivative Picrocrocin derivative Picrocrocin derivative Picrocrocin derivative Flavonol Picrocrocin derivative Picrocrocin derivative Picrocrocin derivative Picrocrocin derivative Picrocrocin derivative Flavonol Flavonol Picrocrocin derivative Flavonol Flavonol Picrocrocin derivative Flavonol Carotenoid Picrocrocin derivative Carotenoid Carotenoid Carotenoid Carotenoid Carotenoid Carotenoid Carotenoid Carotenoid Carotenoid Carotenoid Carotenoid Carotenoid Carotenoid Carotenoid Carotenoid Monoterpene Carotenoid Carotenoid Carotenoid

C16H24O8 (P1) C16H28O8 (P2) C20H30O13 (P3) C16H26O8 (P4) Kaempferol-3-sophoroside-7-glucoside C14H20O8 (P5) C14H20O8 (P6) C16H26O8 (P7) C22H36O12 (P9) Kaempferol-dihexoside Kaempferol-3.7.4’-triglucoside Picrocrocin Kaempferol-dihexoside Kaempferol-3-sophoroside HTCC Kaempferol-hexoside trans-5-tG trans-5trans-4-GG trans-4 trans-3-Gg trans-2-gg cis-4-GG trans-2-G cis-3cis-3cis-2-gg trans-1-g cis-2cis-2Safranal cis-1-g trans-crocetin cis-crocetin

Figure 2. PDA chromatogram (200-450 nm) of saffron sample Crocus sativus with selected components and their identification by PDA – MS according to the available data. Magnification is presented for the absorption peaks at 250 nm only.

17

Figure 3. Proposed picrocrocin derivative structures

18

G8

G11

G7

G10

G9 G1

G14

G15

G12 G13

G3 G4 G2

G5 G6

ID

Rt (UV)

G1

1.31

234

Mw (g.mol-1) -

G2

1.76

237

404

207, 225, 427

Iridoid

G3 G4 G5 G6 G7 G8

2.08 2.28 2.53 2.87 3.63 4.06

237 195 238 241 240 240

404 404 346 550 388

207, 225, 427 207, 225, 427 185, 347, 369 209, 227, 573 209, 227, 411

G9

5.77

237, 327

-

Iridoid Unknown Iridoid Iridoid Iridoid Iridoid Iridoid derivative (with hydroxycinnamic acid)

G10

6.49

231, 312

696

147, 165, 309, 471, 697, 719

Iridoid

G11

6.71

329, 653, 675, 999

7.12

G13

7.27

Carotenoid Iridoid derivative (with hydroxycinnamic acid) Carotenoid

G14

8.66

G15

9.16

261, 441, 464 195, 230, 239, 327 261, 441, 465 263, 326, 434, 460 259, 435, 459

976

G12

UV (λmax) nm

814 976 652

Fragmentation pattern ESI+ (m/z) -

-

329, 491, 653, 675, 837 329, 511, 635, 653, 675, 797, 999 329, 653, 675

Tentative identification

Compound

Iridoid

Deacetyl-asperuloside acid methyl ester Gardenoside Scandoside methyl ester Genipin-1-β-D-gentibioside Geniposide 6"-O-trans-coumaroylgenipin gentibioside trans -4-GG (C3) trans -3-Gg (C5)

Carotenoid

cis -4-GG (C7)

Carotenoid

trans -2-G (C8)

Figure 4. Chromatogram (200-450 nm) obtained using the UHPLC-DAD-MS method applied to the gardenia fruit extract

19

A / 440 nm (Crocins)

B / 240 nm (Picrocrocin derivatives and iridoïds)

P10 Picrocrocin 100%

Geniposide

50%

P4 to P9 25% 15% 5%

Figure 5. Chromatograms of standardized saffron extract adulterated by gardenia crocin extract. (A) Signal overlay for the detection of crocins at 440nm. (B) Detection of geniposide in saffron extracts adulterated by gardenia crocin extract from 0% to 100% in final extract. The extract containing 15% of gardenia corresponds to n°12.

20

Table 1. Characterization of saffron metabolites in dietary supplement extracts by metabolite content, colour of the powder and detection of gardenia adulterant (geniposide) Flavonoids N°

Picrocrocin derivatives (%)

Crocin (eq. trans-4GG)

Safranal

Colour

(eq. kaempferol) SSE 1

UHPLC 2.05 ± 0.02

UHPLC 5.15 ± 0.20

UHPLC 0.29 ± 0.01

UHPLC 0.4233 ± 0.0010

ISO3632 2.44

2

4.10 ± 0.41

8.59 ± 0.86

0.72 ± 0.08

0.0650 ± 0.0010

2.71

3

1.24 ± 0.13

0.44 ± 0.05

0.31 ± 0.04

< LOQ

2.06

Yellow

4 5 6 7 8 9 10 11

2.91 ± 0.30 2.34 ± 0.24 1.06 ± 0.11 2.91 ± 0.30 1.69 ± 0.17 1.74 ± 0.18 2.95 ± 0.30 0.69 ± 0.07

7.08 ± 0.71 3.56 ± 0.36 2.17 ± 0.22 5.90 ± 0.59 4.36 ± 0.44 2.67 ± 0.27 6.22 ± 0.63 1.08 ± 0.11

0.61 ± 0.07 0.44 ± 0.05 0.23 ± 0.03 0.45 ± 0.05 0.36 ± 0.04 0.33 ± 0.04 0.49 ± 0.05 0.11 ± 0.02

0.0185 ± 0.0010 0.0040 ± 0.0002 0.0210 ± 0.0011 0.0093 ± 0.0005 0.0478 ± 0.0024 0.0375 ± 0.0019 0.0194 ± 0.0010 < LOQ

2.02 0.12 2.17 2.21 2.03 2.32 2.90 0.73

Orange/white points Orange Orange Orange Red Orange Red Yellow

12

0.92 ± 0.02

3.35 ± 0.13

0.11 ± 0.01

0.0243 ± 0.0010

2.01

Red

Red Orange/Brown points

21

Highlights A new separation of saffron metabolites with ultra-high performance is proposed Picrocrocin derivative characterization is presented Iridoid analysis from gardenia should be used to detect frauds

22