Journal of the Science of Food and Agriculture
J Sci Food Agric 81:485±496 (online: 2001)
HPLC flavonoid profiles as markers for the botanical origin of European unifloral honeys Francisco A Toma´s-Barbera´n,1* Isabel Martos,1 Federico Ferreres,1 Branka S Radovic2 and Elke Anklam2 1
Laboratorio de Fitoquı´mica, Departamento de Ciencia y Tecnologı´a de Alimentos, CEBAS (CSIC), PO Box 4195, E-30080 Murcia, Spain Food Products and Consumer Goods Unit, Institute for Health and Consumer Protection, Joint Research Centre Ispra, Commission of the European Union, I-21020 Ispra, Italy 2
Abstract: The HPLC phenolic pro®les of 52 selected uni¯oral honey samples produced in Europe were analysed to detect possible markers for the ¯oral origin of the different honeys. Lime-tree (®ve markers), chestnut (®ve markers), rapeseed (one marker), eucalyptus (six markers) and heather (three markers) honeys had speci®c markers with characteristic UV spectra. In addition, the ¯avanone hesperetin was con®rmed as a marker for citrus honey, as well as kaempferol for rosemary honey and quercetin for sun¯ower honey. Abscisic acid, which had been reported to be a possible marker for heather honey, was also detected in rapeseed, lime-tree and acacia honeys. Ellagic acid in heather honey and the hydroxycinnamates caffeic, p-coumaric and ferulic acids in chestnut, sun¯ower, lavender and acacia honeys were also detected. The characteristic propolis-derived ¯avonoids pinocembrin, pinobanksin and chrysin were present in most samples in variable amounts. # 2001 Society of Chemical Industry
Keywords: honey; botanical markers; ¯oral markers; plant origin; quality; HPLC analysis; ¯avonoid; phenolic compounds
INTRODUCTION
In the last few years there has been increasing interest in ®nding objective analytical methods that could complement pollen analysis in the determination of the ¯oral origin of honey. In this context it has been suggested that the next step in this type of research will be an attempt to correlate ¯oral source with the presence of certain compounds originating either in the nectar or in some biochemical modi®cations of nectar compounds carried out by the bee.1 Volatile compounds,2 aromatic and degraded carotenoid-like substances,3±5 amino acids,6,7 degradation products of phenylalanine,8 aromatic aldehydes and heterocyclics,9 aromatic acids and their esters10 and phenolic compounds11±15 have been found in honey and have been related to the ¯oral origin. In fact, the ¯avonoid hesperetin proved to be a useful marker for the ¯oral origin of citrus honey16 and showed some advantages over methyl anthranilate, another biochemical marker of the ¯oral origin of citrus honey.14 In addition, the ¯avonol kaempferol has been found to be a marker for rosemary honey,17 abscisic acid for heather honey18 and homogentisic acid for strawberry-tree (Arbutus unedo) honey.19 The available studies on honey ¯oral markers are generally restricted to individual mono¯oral honeys,
and studies in which extracts of honey samples from different ¯oral origins are compared are very scarce. A comprehensive approach should be achieved to detect those mono¯oral honeys that contain speci®c markers that could be useful in objective botanical origin determinations. The present study aims at the identi®cation of characteristic ¯avonoid (and other phenolic) HPLC pro®les of selected, economically relevant, uni¯oral honeys produced in Europe. This work will identify possible markers, the isolation of which will be the purpose of further studies.
EXPERIMENTAL Samples
Honey samples were collected from different localities in Europe and stored at ÿ20 °C until analysed. Their botanical origin was con®rmed by pollen analysis20 and other classical quality determinations (HMF, moisture, mineral (ash) content). In addition to pollen analysis, the fructose/glucose ratio was measured for acacia honey and the electrical conductivity and protein content for heather honey.21 The honey samples analysed are listed in Table 1.
* Correspondence to: Francisco A Toma´s-Barbera´n, Laboratorio de Fitoquı´mica, Departamento de Ciencia y Tecnologı´a de Alimentos, CEBAS (CSIC), PO Box 4195, E-30080 Murcia, Spain E-mail:
[email protected] Contract/grant sponsor: European Commission (DG Enterprise); contract/grant number: 12672-97-02 AICA ISP B (Received 28 January 2000; revised version received 15 September 2000; accepted 22 November 2000)
# 2001 Society of Chemical Industry. J Sci Food Agric 0022±5142/2001/$30.00
485
FA TomaÂs-BarberaÂn et al Sample 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52
Botanical origin
Geographical origin
Acacia (Robinia pseudoacacia) Acacia Acacia Acacia Acacia Acacia Acacia Chestnut (Castanea sativa) Chestnut Chestnut Chestnut Chestnut Chestnut Chestnut Chestnut Chestnut Citrus (Citrus sinensis) Citrus Citrus Eucalyptus (Eucalyptus camaldulensis) Eucalyptus Eucalyptus Eucalyptus Heather (Erica spp) Heather Heather Heather Heather Heather Heather Heather Heather Bell-heather Lavender (Lavandula spp) Lavender Lime-tree (Tilia europaea) Lime-tree Lime-tree Lime-tree Rapeseed (Brassica spp) Rapeseed Rapeseed Rapeseed Rapeseed Rapeseed Rosemary (Rosmarinus of®cinalis) Rosemary Rosemary Sun¯ower (Helianthus annuus) Sun¯ower Sun¯ower Sun¯ower
Germany Germany Italy Italy Italy Italy France Italy Italy Italy Italy Italy Spain France Germany Germany Italy Spain Italy Italy Italy Spain Portugal Germany Germany France The Netherlands The Netherlands England England England England France France Portugal Germany The Netherlands The Netherlands Germany Germany Germany Denmark Denmark France England Spain Spain Portugal Italy Italy France France
Table 1. Honey samples
Extraction of flavonoids from honey
The different honey samples (100 g each) were mixed with ®ve parts of water (pH 2 with HCl) until completely ¯uid and ®ltered through cotton to remove solid particles. The ®ltrate was then passed through a column (25 cm 2 cm) of Amberlite XAD-2 (Fluka Chemie; pore size 9 nm, particle size 0.3±1.2 mm).22 The phenolic compounds remained in the column, 486
while sugars and other polar compounds eluted with the aqueous solvent, ¯avonoid recovery being higher than 95%.22,23 The column was washed with acid water (pH 2 with HCl, 100 ml) and subsequently with distilled water (ca 300 ml). The whole phenolic fraction was then eluted with methanol (ca 300 ml) and taken to dryness under reduced pressure (40 °C). The residue was re-dissolved in 5 ml of water and J Sci Food Agric 81:485±496 (online: 2001)
HPLC ¯avonoid pro®les as uni¯oral honey markers
extracted with diethyl ether (5 ml 3).22 The ether extracts were combined, concentrated under reduced pressure and re-dissolved in 0.5 ml of methanol for HPLC analysis. Samples were stored under N2 until analysed. Repeatability of the HPLC analysis was 5% (n = 5).
HPLC analysis of honey flavonoids
All HPLC analyses were achieved using a Merck± Hitachi L-6200 liquid chromatograph with a Shimadzu SPD-M6A diode array detector and a Merck± Hitachi A-2000A autosampler. Data were stored and processed with the DAD-Manager software (Merck). The column used was a Lichrochart RP-18 (Merck, Darmstadt, Germany; 12.5 cm 0.4 cm, particle size 5 mm). Elution was with water/formic acid (19:1 v/v) (solvent A) and methanol (solvent B). The elution was accomplished with a solvent ¯ow rate of 1 ml minÿ1, starting with 30% methanol, remaining isocratic for 15 min, then using gradient elution to reach 40% methanol at 20 min, 45% methanol at 30 min, 60% methanol at 50 min and 80% methanol at 52 min. The system then remained isocratic with 80% methanol up to 60 min. The ¯avonoids were detected with the diode array detector to obtain the UV spectra of the different phenolic compounds, and the chromatograms were recorded at 340 and 290 nm.
RESULTS
The honey samples analysed in the present study can be arranged into three groups regarding their constituents that could be used as markers for their botanical origin. The ®rst group of honeys is one in which the presence of possible characteristic markers has been detected, and which need to be identi®ed in further work, as well as in ¯oral nectar. This group includes lime-tree, chestnut, eucalyptus, rapeseed and heather honeys. The second group of honeys is one in which ¯oral markers have previously been reported; their possible use as markers is con®rmed. This group includes citrus, rosemary and sun¯ower honeys. The third group of honeys is one in which no markers have been detected. This group includes acacia and lavender honeys. Group 1. Unifloral honeys in which possible floral markers have been detected
Lime-tree honey Four lime-tree samples from two European countries (Germany, two; Netherlands, two) were analysed. All analysed samples showed a common HPLC pro®le (Fig 1) characterised by the presence of ®ve unknown compounds (L1±L5). These compounds were present in all samples and had characteristic UV spectra that are shown in Fig 2. All compounds have quite simple spectra with only one maximum (L1, 302; L2, 279; L3, 299; L4, 295; L5, 330 nm). Compound L1 is the
Flavonoid identification and quantification
The different honey and propolis ¯avonoids were identi®ed by chromatographic comparisons with authentic markers (commercial or previously isolated and identi®ed from honey)12,24 and by matching their UV spectra with those of the markers. Honey ¯avonoids were quanti®ed by the absorbance of their corresponding peaks in the chromatograms as reported previously.23 The ¯avanones (pinocembrin and pinobanksin) were measured against an external standard of pinocembrin detected at 290 nm, the ¯avones with an unsubstituted ring B (chrysin, galangin and tectochrysin) against chrysin at 340 nm and the rest of the ¯avonols and ¯avones (quercetin, kaempferol, 8-methoxykaempferol, etc) against quercetin at 340 nm. The hydroxycinnamic acid derivatives (caffeic, p-coumaric, ferulic and related compounds) were quanti®ed as caffeic, and ellagic acid as an authentic marker (Sigma, St Louis, MO, USA) at 340 nm. The abscisic acid isomers ABA2 (cis±transabscisic acid) and ABA1 (trans±trans-abscisic acid) were quanti®ed using a marker of cis±trans-abscisic acid (Sigma) at 290 nm. Several compounds were not identi®ed, and in this case their UV spectra were recorded. In the tables these compounds are quanti®ed as the % of the total absorbance of the chromatograms at 290 nm in order to provide an approximate ®gure of their presence in the analysed honeys. J Sci Food Agric 81:485±496 (online: 2001)
Figure 1. HPL chromatograms of lime-tree honey (sample 38): CAF, caffeic acid; PCO, p-coumaric acid; ABA2, cis–trans-abscisic acid; CAFD, unidentified caffeic acid derivative; PB, pinobanksin; 8MK, 8methoxykaempferol; KM, kaempferol; PC, pinocembrin; CHR, chrysin; L1– L5, unidentified markers.
487
FA TomaÂs-BarberaÂn et al
several maxima and/or shoulders (CH1, 330sh, 309, 283sh; CH2, 319sh, 309, 298sh; CH3, 336sh, 323, 310sh, 271; CH4, 350, 336, 321sh, 258sh, 243; CH5, 313 nm). In addition, the markers found in lime-tree honey (L1±L3 and L5) were also detected in most chestnut samples, and in fact L1 was the main peak in these chromatograms (Table 3). Both compound groups CH1±CH5 and L1±L5 could be used as markers for the ¯oral origin of chestnut honey. The content of propolis-derived compounds is in general very small, although caffeic and p-coumaric acids are present in signi®cant amounts in all samples (with the exception of sample 16). In future works, chestnut ¯oral nectar should be analysed to determine whether these compounds (CH1±CH5 and L1±L5) are present and con®rm their ¯oral marker status.
Figure 2. UV spectra of lime-tree honey markers L1–L5.
main constituent in the HPL chromatograms of all the analysed lime-tree samples. In addition, variable amounts of propolis-derived compounds were detected, and in some cases different amounts of the ¯avonols kaempferol and 8-methoxykaempferol. In all samples, cis±trans-abscisic acid was also detected in variable amounts (50±500 mg per 100 g honey). Lime-tree honey constituents are listed in Table 2. The possible marker R1 of rapeseed honey (see below) was also detected in three of the samples. Sample 39 contained compounds CH1±CH4 which are characteristic markers of chestnut honey (see below), suggesting that this sample is a mixture of chestnut and lime-tree (Fig 3) and supporting the use of these markers as a tool for the detection of the botanical origin of honey. Chestnut honey Nine chestnut honey samples from four European countries (Spain, one; France, one; Germany, two; Italy, ®ve) have been analysed. All analysed samples have an identical HPLC pro®le (Fig 4), with the exception of sample 16. The majority of samples are characterised by the presence of ®ve unidenti®ed compounds (CH1± CH5). They have characteristic UV spectra (Fig 5). These spectra are generally quite complex, with 488
Eucalyptus honey Four eucalyptus honey samples from three European countries (Italy, two; Portugal, one; Spain, one) were analysed. The HPLC analyses revealed that the four samples had a common pro®le (Fig 6) characterised by the presence of six characteristic compounds (EU1± EU6) in all samples (Table 4). Compounds EU1± EU3 have quite simple UV spectra (Fig 7) characterised by the presence of just one maximum (EU1, 292sh, 250; EU2, 288; EU3, 291 nm), and compounds EU4±EU6 showed characteristic spectra of ¯avones (Fig 8) (EU4, 371, 296, 250; EU5, 352, 297sh, 265sh, 252; EU6, 351, 292, 265sh, 256 nm). The UV spectra show that EU4 is a 3-hydroxy¯avone and that EU5 and EU6 are either 3-methoxy¯avones
Table 2. HPLC analyses of lime-tree (Tilia europaea) honey constituents
Samples Compound
36
Unknown possible markers L1 (25.4) L2 (1.5) L3 (1.8) L4 (0.6) L5 (1.2) ABA2 539.5 R1 (1.3) CH1 Ð CH2 Ð CH3 Ð CH4 Ð Propolis-derived constituents p-Coumaric acid 130.7 Caffeic derivative Pinobanksin 8-Methoxykaempferol 86.7 Kaempferol Pinocembrin Chrysin 31.3
37
38
(16.7) (26.7) (0.8) (1.1) (1.1) (2.2) (1.0) (0.5) (4.5) (2.8) 290.6 173.6 Ð (0.9) Ð Ð Ð Ð Ð Ð Ð Ð 28.6 959.1 28.5 606.5 219.8
140.3 23.9 797.8 105.0 53.2 586.2 128.8
39 (32.5) (3.0) (1.0) (0.4) (0.2) 44.9 (3.7) (2.9) (8.1) (3.5) (0.4) 107.5 3.7 3.0
82.4 11.1
The unknown compounds L1±L5, CH1±CH4 and R1 are quanti®ed as the % of the total absorbance of the chromatogram at 290 nm (values in parentheses). The other compounds are quanti®ed as mg per 100 g honey.
J Sci Food Agric 81:485±496 (online: 2001)
HPLC ¯avonoid pro®les as uni¯oral honey markers
or ¯avones without oxygenation at position 3.25 In further studies these compounds should be isolated and identi®ed and their occurrence in eucalyptus ¯oral nectar should be explored. The ¯avonol quercetin is present in all samples in signi®cant amounts (Table 4). In addition, the characteristic propolis-derived compounds were present in variable amounts in all eucalyptus honey samples.
Figure 3. HPL chromatograms of lime-tree honey (sample 39) in which the markers of chestnut honey are also present (CH1–CH4): CAF, caffeic acid; PCO, p-coumaric acid; ABA2, cis–trans-abscisic acid; PB, pinobanksin; PC, pinocembrin; CHR, chrysin; L1–L5, unidentified markers; R1, rapeseed honey marker.
Figure 4. HPL chromatograms of chestnut honey (sample 14): CAF, caffeic acid; PCO, p-coumaric acid; CAFD, unidentified caffeic acid derivative; PB, pinobanksin; PC, pinocembrin; CHR, chrysin; CH1–CH5 and L1–L4, unidentified markers.
J Sci Food Agric 81:485±496 (online: 2001)
Rapeseed honey Six samples of rapeseed honey from four European countries (England, one; France, one; Germany, two; Denmark, one) were analysed. The samples are characterised by the presence of the two abscisic acid isomers ABA2 (cis±trans-abscisic acid) and ABA1 (trans±trans-abscisic acid) in amounts ranging from 25 to 250 mg per 100 g honey, and by larger amounts of kaempferol, 8-methoxykaempferol and quercetin (Table 5). In addition, one unidenti®ed marker (R1) was present in all the samples and still needs to be identi®ed. This compound was the main one in the HPLC chromatograms of the analysed rapeseed honeys (Fig 9) and showed a rather simple UV spectrum with a maximum at 276 nm (Fig 10). The samples also contain variable amounts of propolisderived constituents (Table 5).
Figure 5. UV spectra of chestnut honey markers CH1–CH5.
489
FA TomaÂs-BarberaÂn et al Table 3. HPLC analyses of chestnut (Castanea sativa) honey constituents
Sample Compound Unknown possible markers CH1 CH2 CH3 CH4 CH5 L1 L2 L3 L5 Propolis-derived constituents Caffeic acid p-Coumaric acid Caffeic derivative Pinobanksin Pinocembrin Chrysin
8 (1.6) (2.9) (1.4) (0.7) (0.6) (9.5) (4.0) (1.1) (0.3) 108.7 23.7 15.8 112.3 17.4
9
10
11
12
13
14
15
16
(2.8) (1.7) (2.5) (0.9) (1.1) (32.5) (1.7) (0.6) (0.5)
(0.7) (3.2) (0.2) nd (0.6) (48.7) (1.1) (1.0) (2.2)
(1.6) (5.4) (2.1) (1.0) (0.7) (27.7) (3.9) (3.3) (0.1)
(1.5) (6.3) (1.0) (2.2) (0.6) (45.7) (1.3) (0.5) (1.0)
(1.2) (8.7) (1.4) (2.0) (0.7) (21.0) (4.6) nd (0.2)
(1.4) (3.9) (2.8) (1.1) (0.7) (40.5) (2.1) (1.4) (0.1)
(2.5) (9.5) (2.1) (1.8) (0.2) (29.9) (2.5) nd (0.8)
(6.0) (8.6) nd nd nd (11.4) nd nd (13.8)
152.3 30.2 40.8 961.9 52.0 31.7
70.5 12.8 17.5 944.3 270.1 57.8
72.0 17.9 15.1 98.6 6.8
29.3 11.7 5.6 113.1
117.5 39.8 30.8 303.1 42.5
83.6 105.7 nd 198.3 133.0 27.6
nd 0.8 nd
9.2
59.4 31.8 22.9 198.5 100.0 31.3
19.2
The unknown compounds CH1±CH5 and L1±L5 are quanti®ed as the % of the total absorbance of the chromatogram at 290 nm (values in parentheses). The other compounds are quanti®ed as mg per 100 g honey.
Heather honey In European heather honey, including Erica-type (bellheather or heath) and Calluna-type (ling), some ¯avonoids, ellagic acid and the plant hormone abscisic acid have been suggested as useful markers for the determination of ¯oral origin.13,18,26 The samples
analysed in the present work were from Erica spp. The HPLC analysis showed that the major compounds were not ¯avonoids. In fact, ¯avonoids were minor constituents of these extracts. The majority of samples showed a common HPLC pro®le (Fig 11) characterised by the occurrence of the unknown compounds H1±H3. The only sample showing a different pro®le was sample 33, which was labelled as `bell-heather' and contained a speci®c compound of this sample that was named BH1 (Table 6). Compounds H1±H3 seem to be characteristic of this honey type (Table 6) and do not correspond to any of the markers already available. These compounds have characteristic UV spectra (Fig 12) with distinctive maxima and spectral shape (H1, 307sh, 264; H2, 285; H3, 341sh, 317, 305sh, 269 nm). Other common constituents in the analysed samples are the two isomers of abscisic acid (ABA1 and ABA2) and ellagic acid that have previously been reported as heather honey markers (Table 6). The content of abscisic acid ranges from 400 to 1800 mg per 100 g honey, these values being smaller than those previously found in heather honey samples from Portugal (2500±16 600 mg per 100 g honey).18 The ellagic acid values range between 300 and 1200 mg per 100 g honey. These values are larger than those previously reported for heather honey samples produced in Portugal (100±600 mg per 100 g honey).18,26 Group 2. Honeys in which phenolic floral markers have previously been reported
Figure 6. HPL chromatograms of eucalyptus honey (sample 21): EU1– EU6, unidentified markers; PB, pinobanksin; QU, quercetin; PC, pinocembrin; CHR, chrysin.
490
Citrus honey The propolis-derived constituents dominate the HPLC pro®les of the three citrus honeys analysed. The citrus honey ¯oral marker hesperetin is present in all samples in amounts ranging between 50 and 100 mg J Sci Food Agric 81:485±496 (online: 2001)
HPLC ¯avonoid pro®les as uni¯oral honey markers Sample Compound Unidenti®ed possible markers EU1 EU2 EU3 EU4 EU5 EU6 Identi®ed possible marker Quercetin Propolis-derived constituents Pinobanksin Pinocembrin Chrysin Table 4. HPLC analyses of eucalyptus (Eucalyptus camaldulensis) honey constituents
20
21
22
23
(7.0) (11.3) (7.0) 171.6 790.1 283.1
(3.3) (4.9) (3.3) 330.2 832.5 418.3
(6.9) (10.9) (0.5) 100.5 289.7 111.1
(11.0) (4.5) (2.9) tr 108.1 143.7
283.4
485.7
165.0
68.3
1611.2 446.3 114.8
549.9 1066.2 394.4
34.0 46.4 11.1
1136.3 522.2 110.2
The unknown compounds EU1–EU3 are quanti®ed as the% of the total absorbance of the chromatogram at 290 nm (values in parentheses). The other compounds are quanti®ed as mg per 100 g honey. The unknown ¯avonols EU4±EU6 are quanti®ed as quercetin at 340 nm.
per 100 g honey. These values are in the concentration range previously found in other Spanish citrus honey samples (30±90 mg per 100 g honey)14 and con®rm that hesperetin is a suitable marker of the ¯oral origin of citrus honey. The results of the HPLC analyses are listed in Table 7.
Rosemary honey Three rosemary samples (Spain, two; Portugal, one) were analysed. The results of the analysis are summarised in Table 8. The analysed samples contain mainly propolis-derived compounds, and the only compounds related to the ¯oral origin are kaempferol, which has already been described as a ¯oral marker for
Figure 7. UV spectra of eucalyptus honey markers EU1–EU3. Figure 8. UV spectra of eucalyptus honey flavonoid markers EU4–EU6.
J Sci Food Agric 81:485±496 (online: 2001)
491
FA TomaÂs-BarberaÂn et al Sample Compound Unknown possible marker R1 Known possible markers ABA1 ABA2 Quercetin 8-Methoxykaempferol Kaempferol Propolis-derived constituents Pinobanksin Pinocembrin Chrysin Table 5. HPLC analyses of rapeseed (Brassica spp) honey constituents
40
41
42
43
44
45
(21.2)
(14.0)
(43.2)
(28.0)
(29.9)
(24.5)
48.5 64.3 54.0 94.4 233.3
107.3 136.5 87.6 126.9 327.0
44.7 45.9 12.3 31.6 259.7
18.5 42.9 122.7 49.3 501.4
14.8 9.5 72.8 56.3 235.5
23.7 36.3 67.4 307.0 269.2
606.8 480.7 104.0
344.6 457.6 52.8
204.8 157.9 33.5
209.1 343.9 41.4
783.2 422.6 127.0
165.6 121.0 26.0
The unknown compound R1 is quanti®ed as the % of the total absorbance of the chromatogram at 290 nm (values in parentheses). The other compounds are quanti®ed as mg per 100 g honey.
