Food and Chemical Toxicology 47 (2009) 547–554

Contents lists available at ScienceDirect

Food and Chemical Toxicology

journal homepage: www.elsevier .com/ locate/ foodchemtox

Antioxidant effects of flavonoid from Croatian Cystus incanus L. rich bee pollen

Ana Šaric a, Tihomir Balog a,*, Sandra Sobocanec a, Borka Kušic a, Višnja Šverko a, Gordana Rusak b,Saša Likic b, Dragan Bubalo c, Barbara Pinto d, Daniela Reali d, Tatjana Marotti a

a Division of Molecular Medicine, Rudjer Boškovic Institute, Bijenicka 54, 10000 Zagreb, Croatiab Faculty of Science, University of Zagreb, 10000 Zagreb, Croatiac Faculty of Agriculture, University of Zagreb, 10000 Zagreb, Croatiad Universyta di Pisa, Dipartimento di Patologia Sperimentale, Biotecnologie Mediche, Infettivologia ed Epidemiologia, Via San Zeno 37, Pisa 56127, Italy

a r t i c l e i n f o

Article history:Received 29 July 2008Accepted 8 December 2008

Keywords:AntioxidantsCystus incanus L. bee pollenFlavonoidsGene expressionLipid peroxidation

0278-6915/$ - see front matter � 2008 Elsevier Ltd. Adoi:10.1016/j.fct.2008.12.007

Abbreviations: AOE, antioxidant enzymes; CAT, catisoenzyme; Gpx, glutathione peroxidase; LPO, lipid pdismutase; tSOD, total superoxide dismutase; TBARSsubstances.

* Corresponding author. Tel.: +385 1 456 11 72; faxE-mail address: balog@irb.hr (T. Balog).

a b s t r a c t

Oxidant/antioxidant status, estrogenic/anti-estrogenic activity and gene expression profile were studiedin mice fed with Cystus incanus L. (Cistaceae) reach bee pollen from location in Central Croatia’s Dalmatiacoast and offshore islands. Seven phenolic compounds (out of 13 tested) in bee pollen sample weredetected by high performance liquid chromatography (HPLC) analysis. Phenolics detected in C. incanusL. bee pollen belong to flavonol (pinocembrin), flavanols (quercetin, kaempferol, galangin, and isorham-netin), flavones (chrysin) and phenylpropanoids (caffeic acid). Bee pollen as a food supplement (100 mg/kg bw mixed with commercial food pellets) compared to control (commercial food pellets) modulatedantioxidant enzymes (AOE) in the mice liver, brain and lysate of erythrocytes and reduced hepatic lipidperoxidation (LPO). Bee pollen induced 25% of anti-estrogenic properties while no estrogenic activity wasfound. Differential gene expression profile analyses after bee pollen enriched diet identify underex-pressed gene Hspa9a, Tnfsf6 (liver) and down-regulated gene expression of Casp 1 and Cc121c (brain)which are important in the apoptosis pathway and chemotaxis. These results indicate that used bee pol-len possess a noticable source of compounds with health protective potential and antioxidant activity.

� 2008 Elsevier Ltd. All rights reserved.

1. Introduction taminant dioxin (Depeint et al., 2002; Moon et al., 2006; Ashida

The plants of genus Cystus (Cystus incanus, Cystus monspeliensis,Cystus ladanifer, etc.) is recognized as medicament in folk medicinebased on its antidiarrheics, anti-inflammatory or photoprotection(skin) properties (Attaguile et al., 2000; Kullavanijaya and Lim,2005), which stimulates bone calcification (Hamamoto et al.,2006) or prevent bone lost caused by increasing age (Yamaguchiet al., 2006).

Bee pollen as apicultural product is focused for human diet be-cause of its nutritional value. It contains carbohydrates, lipids, fats,vitamins, ashes, and minerals, which contribute to the health ef-fects and flavonoids which are regarded as protective agent. Flavo-noids have different structural features and show several biologicalactivities. It appear that they may strongly influence antioxidantactivity, gene expression, drug-metabolizing enzymes, such as cellsignaling or cytochrome P450 (CYP) enzymes, express phytoestro-genic potential, protect against toxicity of the environmental con-

ll rights reserved.

alase; CYP, cytochrome P450eroxidation; SOD, superoxide, thiobarbituric acid reactive

: +385 1 456 10 10.

et al., 2000). The interaction of these natural antioxidants withreactive oxygen species through their free radical scavenging prop-erties are implicated in inflammation leading to a profound effectin defense processes and angiogenesis-related diseases such asrheumatoid arthritis or psoriasis (Krishnamachary et al., 2002),hearth disease or cancer. It has also been proposed that dietaryphytoestrogens (Tham et al., 1998) (isoflavonoids, flavonoids,coumestans, and mammalian lignans) through their estrogenicand anti-estrogenic effect may prevent or alleviate menopausalsymptoms, post-menopausal osteoporosis (Strauss et al., 1998)and can be utilized in the prevention or treatment of breast cancerin women (Oh and Chung, 2006). Dietary components undergoseveral transformations during digestion. These reactions affecttheir bioavailability including the absorption, metabolism, trans-port to the target organs and potential biological effects of the ac-tive molecules. The differences in flavonoid metabolite formationin the organism may have a higher or lower bioavailability thanparent compound, and results in a change of the overall protectiveresponse too. Therefore, the results obtained by in vitro experi-ments do not necessarily reveal the effect of flavonoid from beepollen in vivo. Recent studies have shown that flavonoids obtainedfrom pollen of different geographical or botanical origin containcompounds with various nutritional relevance. The free radicalreactions and scavenging capacities to reactive oxygen species of

548 A. Šaric et al. / Food and Chemical Toxicology 47 (2009) 547–554

the pollen may be due to factors such as differences in atmosphericor environmental conditions, soil or physiology of plant. Moststudies on the biological activity flavonoid from bee pollen havebeen made in vitro in isolated systems.

The aim of the present work was to characterize in vivo antiox-idant, and in vitro estrogenic/anti-estrogenic activity of the C. inc-anus L. bee pollen. This activity was compared with profile ofphenolic compounds present in bee pollen of C. incanus L. The geneexpression profile of 96 genes indicative for stress and toxicity(SuperArray Inc.) in both liver and brain were performed too. Forthis purpose mice were fed either with commercial food pellet(control group) or with commercial food pellet mixed with bee pol-len from location in Central Croatia’s Dalmatia coast and island ex-posed to sun.

2. Materials and methods

2.1. Chemicals

Horse heart cytochrome C (Type VI) and human blood CuZnSOD (Type I, lyophi-lized powder, 2400 U/mg protein), bovine serum albumin, hydrogen peroxide (30%),bovine liver catalase and xantine, xantine oxidase, 2-thiobarbituric acid, dodecyl sul-fate sodium salt, and flavonoids (naringenin, quercetin, taxifolin, pinocembrin, gen-istein) as well as Folin Ciocalteu reagent, and 1,1,3,3-tetrametoxypropane werepurchased from Sigma, St. Louis, MO, USA. Other flavonoids and phenolics (galangin,isorhamnetin, myricetin, chrysin, kaempferol, luteolin, daidzein, phenylpropanoidcaffeic acid) were purchased from Fluka, Switzerland. Glutathione peroxidase(Gpx) (RANSEL) and superoxide dismutase (SOD) (RANSOD) assay kit was obtainedfrom RANDOX, San Diego, CA, USA. HPLC-grade solvents were purchased from Merck,Germany. Cellulose acetate filters Schleicher & Schnell (0.2 lm pore) were used. Sal-mon sperm DNA was obtained from Invitrogen, Carlsbad, CA, USA; biotin-16-dUTPfrom Roche, Mannheim, Germany; RNeasy Mini kit from Quiagen, Hilden, Germany;MMLV reverse transcriptase from Promega, Madison, Wi and linear polymerase reac-tion was created using AmpoLabeling-LPR kit from SuperArray. GEArray Q seriesMouse Stress and Toxicity PathwayFinder gene array kit and CDP-Star chemilumi-nescent substrate supplied in Chemiluminescent Detection kit were obtained fromSuperArray Inc. (Bethesda, MD). VersaDoc imager was obtained from Bio-Rad Labo-ratories, Hercules, CA. GEArray Expression Analysis Suite (GEASuite) was accessibleat GEAsuite.superarray.com. All other chemicals were of analytical grade.

2.2. Bee pollen samples

In our study, we used a bee pollen obtained from apiaries located in naturallypreserved part of coastaline area exposed to sun and offshore islands of Midlle Dal-matia, Croatia. A voucher specimen is deposided et Herbarium Croaticum (ZA) un-der collection No. 338-1. Pollen of subshrub plant C. incanus L. was collected by beesduring flowering from April until June. Fresh bee pollen catched in especially capterof the beeskep were taken and frozen at �18 �C. This bee pollen contained largeamount of moisture (20–30%) and thus was dried under coercion ventilation at60 �C keeping dry the amount of moisture at 8–10%. After being dried up bee pollenwas pulverized through an original treatment by firm HEDERA d.o.o. (Stobrec, Croa-tia) and was performed with no chemical refinement.

Melissopalynological analysis showed botanical origin of used bee pollen (C.incanus L. 30.4 ± 2.3%; Quercus ilex 47.2 ± 3%; Quercus spp. 17.7 ± 1%; Asphodelusspp. 1% and Brassicaceae 1.5 ± 0.5%).

Chemical analysis of bee pollen, using standard methods, showed that it con-tained 12.50% water, 6.92% fat (Beythien and Diemair, 1963), carbohydrates58.61%, cellulose 1.83%, 18.42 protein (Kjeldahlu, 1960), 1.72% ashes, and the fol-lowing minerals (Gorsuch, 1970): lead (Pb) 0.060 mg/kg, iron (Fe) 99.90 mg/kg, cop-per (Cu) 7.330 mg/kg, mercury (Hg) 0.013 mg/kg, zinc (Zn) 74.70 mg/kg, manganese(Mn) 31.50 mg/kg, chrom (Cr) 0.379 mg/kg, calcium (Ca) 997,0 mg/kg, cadmium(Cd) 0.060 mg/kg, magnesium (Mg) 564.0 mg/kg, selen (Se) 0.999 mg/kg, molybden(Mo) >0.005 mg/kg, kalium (K) 4.006 mg/kg, natrium (Na) 192.0 mg/kg, chlor (Cl)654.0 mg/kg, phosphor (P) 3.084 mg/kg, arsen < 0.01 mg/kg. The vitamin (Songet al., 2000) content in this bee pollen was as follows: vitamin E 12.80 mg/100 g,a caroten 11.90 mg/100 g, b caroten 11.91 mg/100 g, vitamin C 3.44 mg/100 g, vita-min B1 1.10 mg/100 g, B2 0.44 mg/100 g, and vitamin B6 0.30 mg/100 g.

2.3. HPLC analyses

Qualitative and quantitative chromatographic analysis of phenolics was per-formed on a HPLC system (Agilent 1100 Series) equipped with a quaternary pump,multiwave UV/Vis detector, autosampler, and fraction collector. The column usedwas a 5 lm Zorbax RX-C18 (250 � 4.6 mm, Agilent Technologies). Injection volumewas 200 lL, flow rate 1.0 mL/min, and temperature was 45 �C. The bee pollen wasfractionated. Fractions 1 (tR 9.3–10.8 min), 2 (tR 12.9–14.3 min), 3 (tR 18.4–

19.7 min), 4 (tR 22.5–23.7 min), 5 (tR 24.2–25.9 min), 6 (tR 27–28 min), 7 (tR28.3–29 min), 8 (tR 33–34 min), 9 (tR 34.6–36.8 min) were obtained using elutionprofile consisting of solvent A (5% formic acid) and solvent B (methanol). Linear gra-dient from 10% to 90% B within 45 min was used.

The identification of phenolic compounds was made by comparing their reten-tion times and UV spectra with those of the standards and in the end by rechroma-tography of extracts with authentic standards. Fractions 1, 2, 3, 4, 5, 6, 7, and 8 wereanalyzed using solvents A (5% formic acid) and B (acetonitrile) on linear gradientsfrom 5% to 53% solvent B within 30 min. Fractions 8 and 9 were analyzed using iso-cratic separation on 33.5% solvent B in 30 min.

2.4. Phenolic standards

The following compounds, representative of various subclasses of phenolics,were used in our study: flavonol (quercetin, isorhamnetin, kaempherol, galangin,myricetin, and taxifolin), flavone (luteolin), flavanone (pinocembrin and naringe-nin), isoflavone (genistein and daidzein), flavone (chrysin) and phenylpropanoid(caffeic acid). All of the standards were dissolved in ethanol (96%, v/v) to give0.01 mg/mL solutions.

2.5. Animals and experimental design

Female CBA/Hr mice aged 4 months from a breeding colony at the Ruder Boš-kovic Institute (Zagreb, Croatia) were used. The animals were maintained underthe following laboratory conditions: four to a cage; light on from 06:00 to 18:00;22 �C ± 2 �C room-temperature; access to food pellets (Muchedola RF21, Italy),and tap water ad lib. Experimental and control group consisted of 10 mice each.Mice were fed 14 days before testing either with commercial food pellets (controlgroup) or with commercial food pellets mixed with bee pollen (100 mg/kg bw).Dose of 100 mg/kg of bee pollen were used to correspond the dose usually usedin humans with correction for mouse metabolism. This study was in compliancewith guidelines of the European Communities Council Directive of 24 November1986 (86/609/EEC) and approved by Ministry of Agriculture, Forestry and WaterManagement for the Republic of Croatia.

2.6. Botanical evaluation of pollen

Five 0.10 g pollen samples were analyzed according to the method of (Louveauxet al., 1978). Pollen grains morphometric was carried out by the Hund h 500 (Wetz-lar, Germany) light microscope attached to the digital camera (Motic m 1000) andcoupled to an image analysis system (Motic Images Plus Software). Pollen grainswere identified according to melissopalynological collection from NiedersächsischesLandesinstitut für Bienenkunde, Mediterranean melissopalynology collection fromUniversita degli Studi di Perugia, Facolta di Agraria and personal pollen collections.

2.7. Acquisition of samples

Mice were anesthetized with diethyl ether and bled from the jugular vein intoheparinized tubes. Whole blood was centrifuged (2000g for 10 min). Erythrocytes inthe remaining pellet were lysed in cold deionized water, and the AOE activitieswere determined in the lysate. Hemoglobin concentration was measured by Cell-Dyn 1600 (Abbot, USA). The liver and brain were removed immediately by manualdissection, blotted on filter paper, and weighed. A portion of the tissue was placedinto 1.15% KCl on ice and homogenized (1300 rpm; 10% kg/10�2 m3 for the liver and20% kg/10�2 m3 for brain) using an ice-packed Potter-Elvehjem homogenizer(Braun, Biotech. Int., Germany) for determination the formation of malondialde-hyde according to the presence of thiobarbituric acid reactive substances (TBARS).To determine the AOE a section of tissue was placed in 50 mM phosphate buffer (pH7.8), homogenized (10% kg/10�2 m3 for the liver and 20% kg/10�2 m3 for brain), son-icated on ice for 30 s in three 10 s intervals, and centrifuged at 4 �C (20,000g for15 min). LPO products in the liver and brain homogenates were determined imme-diately on the same day while aliquots of the resulting supernatant for determina-tion of AOE were stored in plastic tubes at �70 �C until assayed. The absorbance ofLPO product and AOE activity was monitored using a Camspec M330 UV–Vis spec-trophotometer equipped with M330 Camspec software package (Camspec Ltd.,Cambridge, UK). Other portion of the mice liver and brain tissues were quickly dis-sected on ice under sterile conditions. Samples of each tissue from three animalsper group were pooled and used for the determination of gene expression profilesof 96 genes indicative of mouse stress and toxicity (GEArray). Each analysis wasperformed at least three times.

2.8. Assay for LPO product concentration

The LPO in the liver and brain were estimated by the measurement of malondi-aldehyde levels on the base of malondialdehyde reacted with thiobarbituric acid at532 nm, according to the method as described in previous report (Ohkawa et al.,1979). Briefly, tissue homogenate was mixed with sodium dodecyl sulfate, acetatebuffer (pH 3.5), and an aqueous solution of thiobarbituric acid. After being heated at95 �C for 60 min, the red pigment produced was extracted with n-butanol–pyridine

A. Šaric et al. / Food and Chemical Toxicology 47 (2009) 547–554 549

mixture and estimated by the absorbance at 532 nm. The results were expressed asnmol/mg of protein in the liver and brain tissue according to a standard curvewhich was prepared with serial dilutions of standard 1,1,3,3-tetramethoxypropane.

2.9. Determination for SOD activity

SOD activity in the liver and brain was assayed by the inhibition of xantine/xan-tine oxidase mediated reduction of cytochrome c as previous described method(Flohe and Ötting, 1984). Total superoxide dismutase (tSOD) activity in the lysateof erythrocytes was determined by RANSOD assay kit. One unit of SOD activity inthe liver and brain was defined as the amount of enzyme required to give 50% inhi-bition in the typical calibration curve obtained with standard SOD and was ex-pressed as IU/mg protein. In the lysate of erythrocyte, tSOD activity wasexpressed as U/mg Hb.

2.10. Determination of catalase (CAT) activity

CAT activity in lysate of erythrocytes and tissues was determined according tothe method as described in previous work (Aebi, 1984), by measuring absorbancechanges in the reaction mixture using the final concentrations of 10 mM H2O2 and50 mM phosphate buffer (pH 7.0) at 240 nm during the time interval of 30 s aftersample addition. The CAT activity was expressed as IU/mg Hb in lysate of erythro-cytes and as IU/mg protein for activity in tissues. Every sample was analyzed withappropriate blanks without H2O2. One unit of CAT was defined as the amount of en-zyme which liberates half the peroxide oxygen from H2O2 solution in 30 s at 25 �C.

2.11. Determination of Gpx activity

The Gpx activity in lysate of erythrocytes and tissues was measured by Gpx as-say kit (RANSEL) based on the method previous described (Paglia and Valentine,1967). Gpx activity was determined with RANSEL kit by measuring the rate ofNADPH oxidation to NADP+, which is accompanied by a decrease in absorbance at340 nm per minute. One Gpx unit is proportional to the amount of NADPH con-sumed in nmol per minute at 37 �C and pH 7.2 at 25 �C. To obtain the linearity ofthe assay, if the absorbance change per minute exceeded 0.1, the sample was di-luted with the diluting agent. Gpx activity was expressed as IU/mg Hb in lysateof erythrocytes and IU/mg protein in tissues.

2.12. Determination of protein concentration

Protein concentration in the tissue samples (mg/g) was estimated by the meth-od as described in previous report (Lowry et al., 1951) using bovine serum albuminas the standard.

2.13. GEArray analysis

Studies based on genomic approaches were performed on the liver and brain ofcontrol and mice fed with bee pollen which allowed to identify genes belonging todifferent families: (1) gene involved in oxidative metabolism, (2) gene involved ininflammatory and immune response, (3) gene related to apoptosis, (4) gene partic-ipating in oxidative stress response, and (5) gene participating in signaltransduction.

Total RNA was extracted from pool of three individual mouse livers and brainsin each group using TRIzol reagent according to the manufacturer’s instructions andpurified by Rneasy Mini kit. Three micrograms of total RNA was reverse-transcribedusing the MMLV reverse transcriptase into complementary DNA (cDNA) probes la-beled with biotin-16-dUTP in a 30 cycle linear polymerase reaction using AmpoLa-beling-LPR kit. GEArray Q series Mouse Stress and Toxicity PathwayFinder genearray kit was used to compare gene expression profiles of 96 genes indicative ofmouse stress and toxicity. The arrays were handled following the manufacturer’sprotocol. The SuperArray membranes were prehybridized at 60 �C for no less than4 h with 100 lg/mL salmon sperm DNA and then were hybridized with denaturedbiotin-labeled cDNA probes overnight at 60 �C. After hybridization, membraneswere washed twice in 2 � SSC with 1% SDS and in 0.1 � SSC with 0.5% SDS, furtherincubated with alkaline phosphatase-conjugated streptavidin and finally developedwith CDP-Star chemiluminescent substrate supplied in Chemiluminescent Detec-tion kit. Images of the membranes were obtained after exposing the membranesto a CCD camera for 20 minutes using VersaDoc imager and loaded into the GEArrayExpression Analysis Suite which was used to obtain expression profile of each genelater utilized in gene expression analysis.

2.14. Estrogenic/anti-estrogenic activity of bee pollen

Estrogenic activity of bee pollen was tested using Saccharomyces cerevisiae yeaststrain (RMY 326 ER-ERE) containing the human estrogen receptor a (hERa) and aXenopus laevis vitellogenin ERE sequence linked to the reporter gene lacZ encodingfor the enzyme b-galactosidase (Liu and Picard, 1998). The induction of transcrip-tion of the reporter gene by the complex receptor-agonist was detected and quan-tified by spectrophotometry.

The yeast method was previously described (Pinto et al., 2004). Briefly, recombi-nant yeast cells were stored at �80 �C and grown at 30 �C on selective agar plates.Yeast colonies were sampled from agar and incubated at 28 �C for 7 h in orbital sha-ker (210 rpm). After incubation OD600 nm of colonies was measured and adjusted to<0.1 nm by diluting with fresh medium. After dilution yeast culture were incubated(17 h) overnight in the presence of 17b estradiol (10 nM) as a positive control, sol-vent (negative control) and increasing concentrations of pollen from C. incanus L.alone, for estrogenic activity or together with 1 nM 17b estradiol for anti-estrogenicactivity. Estradiol and Cystus pollen were dissolved in DMSO and added to the yeastculture so that the concentration of solvent did not exceed 1% (v/v). The enzymaticreaction was started by adding o-nitrophenyl b-D-galactopyranoside (ONPG) andincubating at 30 �C for 5 min. The reaction was stopped by adding Na2CO3 and theabsorbance was measured at 420 nm. The b-gal activity was normalized to the num-ber of cells assayed and expressed as Miller units (Miller, 1972) using the formula:

b-gal units = (1000 � OD420 nm)/(t � V � OD600 nm)t = time of incubation (min)V = volume of culture in the assay (mL)

The results are expressed as percent of the b-gal activity obtained with 17bestradiol (E2). Each value is presented as the mean ± SD.

2.15. Statistical analysis

The data were analyzed using the statistical package SPSS for Windows(v.11.00) and were presented as mean ± SE. The significance of the differences be-tween control and bee pollen treated groups were compared with one way analysisof variance (ANOVA) followed by Scheffe’s post hoc test. A mean difference was sig-nificant at the 0.05 level.

3. Results

3.1. Phenolics concentration of bee pollen

In Fig. 1 quantitative determination of phenolics present in non-hydrolyzed and hydrolyzed extracts of C. incanus L. pollen are pre-sented. Out of 13 phenolics determined, six were not detectedneither in nonhydrolyzed or hydrolyzed extracts (myricetin, luteo-lin, daidzein, genistein, naringenin, and taxifolin) while isorhamne-tin and quercetin were observed only in hydrolyzed samples.Caffeic acid and kaempherol were present in both extracts in sim-ilar amounts while the concentrations of chrysin and pinocembrinwere much higher in nonhydrolyzed extracts. In nonhydrolyzedextracts pinocembrin (1.418 lmol/g pollen) was the most abun-dant phenolic followed by chrysin (1.351 lmol/g pollen) and gal-angin (0.859 lmol/g pollen). In hydrolyzed samples isorhamnetin(6.705 lmol/g pollen) was followed by quercetin (3.25 lmol/g pol-len), kaempherol (1.563 lmol/g pollen), and chrysin (0.786 lmol/gpollen).

3.2. Effect of bee pollen on TBARS level in the liver and brain

Bee pollen in a dose of 100 mg/kg bw significantly decreasedTBARS concentration in the liver (p < 0.0001) but was without ef-fect in brain (Fig. 2).

3.3. Effect of bee pollen on AOE activity in the liver and brain

In the liver and brain dose of 100 mg/kg bw bee pollen signifi-cantly decreased SOD activity (Fig. 3A) (p < 0.003), without effect-ing Gpx activity (Fig. 3B).We observed increased CAT activity(Fig. 3C) in the liver (p < 0.007) with no effect in brain of mice trea-ted with 100 mg/kg bw of bee pollen.

3.4. Effect of bee pollen on AOE activity in lysate of erythrocytes

Contrary to the liver and brain, the activity of tSOD, Gpx andCAT in the lysate of erythrocytes was significantly increased(p < 0.001; p < 0.006; p < 0.025, respectively) with bee pollen(100 mg/kg bw) diet (Table 1).

A

0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

Caffe

ic a

cid

Gal

angi

n

Isor

ham

netin

Kaem

pfer

ol

Myr

icet

in

Que

rcet

in

Lute

olin

Chry

sin

Daid

zein

Gen

iste

in

Narin

geni

n

Pino

cem

brin

Taxi

folin

Conc

entra

tion

(µm

ol/g

)

ND ND NDND NDND ND ND

B

0

1

2

3

4

5

6

7

8

Caffe

ic a

cid

Gal

angi

n

Isor

ham

netin

Kaem

pfer

ol

Myr

icet

in

Que

rcet

in

Lute

olin

Chry

sin

Daid

zein

Gen

iste

in

Narin

geni

n

Pino

cem

brin

Taxi

folin

Con

cent

ratio

n (µ

mol

/g)

NDNDNDNDND ND

Fig. 1. Concentration of nonhydrolyzed (A) and hydrolyzed (B) form of phenolics of bee pollen from coastal area and offshore islands of Midlle Dalmatia, Croatia.ND = nondetectable.

LIVER

0

0.2

0.4

0.6

0.8

1

1.2

1.4

C CY C CY

TB

AR

Snm

ol/m

g pr

otei

n

BRAIN

a

Fig. 2. TBARS concentration in the liver and brain of mice fed with commercial foodpellets (control, C) and commercial food pellets mixed with bee pollen (100 mg/kg bw; CY). Experimental and control group consisted of 10 mice each. The data aremean values ± SE. ap < 0.0001, compared with the control group.

550 A. Šaric et al. / Food and Chemical Toxicology 47 (2009) 547–554

3.5. Gene expression profile in the liver and brain of mice fed with beepollen

Differentially expressed genes of known function havebeen identified in the liver and brain after feeding withbee pollen (Table 2). Heat-shock protein (Hspa9a) and tumornecrosis factor (ligand) superfamily, member 6 (Tnfsf6) weresignificantly underexpressed (p < 0.027; p < 0.038, respectively)in the liver. In the brain after feeding with bee pollenexpression of two genes, caspase 1 (Casp1) and chemokine(C–C motif) ligand 21c (leucine) (Ccl21c) was significantlydownregulated (p < 0.037; p < 0.013, respectively). However,other genes involved in oxidative metabolism (cytochromeP450 family) in the liver (CYP7a1, CYP7b1, and CYP4a10),or genes involved in antioxidant (Sod2) or cell-mediatedimmunity (Il18) processes tended to be underexpressed butwithout statistical significance (SOD2; control 45536 ± 878:bee pollen 41547 ± 1811 and Il18; control 41909 ± 383: beepollen 36708 ± 1221).

0

1

2

3

4

5

6

7

8

SOD

(IU

/ mg

prot

ein)

LIVER BRAIN

a

a

A

C

0

10

20

30

40

50

60

70

80

C CY C CY

CA

T(I

U/m

g p

rote

in)

0

0.2

0.4

0.6

0.8

1

1.2 b

B

0

0.5

1

1.5

2

Gp

x(I

U/m

g p

rote

in)

Fig. 3. SOD (A), Gpx (B), and CAT (C) activity in the liver and brain of mice fed withcommercial food pellets (control, C) and commercial food pellet mixed with beepollen (100 mg/kg bw; CY). Experimental and control group consisted of 10 miceeach. The data are mean values ± SE. ap < 0.003, compared with the control group;bp < 0.007.

A. Šaric et al. / Food and Chemical Toxicology 47 (2009) 547–554 551

3.6. Estrogenic/anti-estrogenic activity of bee pollen

Using the S. cerevisiae yeast assay, bee pollen was tested for itsestrogenic and anti-estrogenic activity at various concentrations.As shown in Fig. 4 at final concentration of 330 lg/mL bee pollenshowed a weak estrogenic activity (6.4% E2 ± 0.02). In general,

Table 1tSOD, Gpx and CAT activity in the lysate of erythrocytes after feeding with bee pollen(100 mg/kg bw).

tSOD Gpx CAT

Control 709.8 ± 37.8 0.378 ± 0.009 13.07 ± 0.70Bee pollen 1290.6 ± 151.3a 0.455 ± 0.015b 15.49 ± 0.72c

Antioxidative effect of bee pollen (100 mg/kg bw mixed with commercial foodpellets) on tSOD, Gpx, and CAT activities in lysates of the mice erythrocyte. Controlmice were fed with commercial food pellets (control). Experimental and controlgroup consisted of 10 mice each. The data are mean values ± SE and are expressedas IU/mg Hb.

a p < 0.001, compared with the control group.b p < 0.006 compared with the control group.c p < 0.025 compared with the control group.

bee pollen inhibited the estradiol-induced reporter gene activityin a concentration dependant manner. Anti-estrogenic activitywas higher (47.4% inhibition of b-gal activity of E2) at the higherbee pollen concentration tested (330 lg/mL).

4. Discussion

In vivo data on the biological effects, bioavailability or metabo-lism from different dietary sources or isolated phenolics after oralfeeding are contradictory and scarce (Rice-Evans et al., 1997). Iden-tification of phenolic from food is important since their nature, sizeand structure influence their metabolism, absorption, distributionand excretion in animals and humans. Although flavonoids arepoorly absorbed in the body after their absorption into the bloodthey are rapidly metabolized in the intestines and liver and as suchbecome active as antioxidant or antiradical preventing free radicaltoxicity, oxidative stress and pathophysiology of various diseases.

Melissopalynological analysis of bee pollen used in our experi-ments showed that the most abundant entomophilous plant in ourbee pollen is C. incanus L. (30.4 ± 2.3%). The rest of bee pollen is pol-len from anemophilous plant (Q. ilex 47.2 ± 3%; Quercus spp.17.7 ± 1%) which does not produce nectar. On the other hand it isknown that entomophilous plant bee pollen is several time largerand have more nutritional value than pollen from anemophilousplants, so we calculate that the more that half of weight percentageof used bee pollen is pollen from C. incanus L.

Phenolics concentration of bee pollen analysis showed thatflavonoids were present as nonhydrolyzed, hydrolyzed or in bothforms. Flavanone pinocembrin (followed by flavanone galanginand flavone chrysin) was the main phenolic compound detectedin nonhydrolyzed form although in small amount chrysin andpinocembrin was present as hydrolyzed extracts. In both, hydro-lyzed and nonhydrolyzed extracts, bee pollen contain caffeic acidand kaempherol. Many authors following phenolic dietary inter-ventions showed contradictory results in bioavailability exertionor pharmacological effects between aglycones or glycosides dueto its metabolites (free radical scavenging, antioxidant, apoptosis,antibacterial, antiviral, anti-inflammatory, and antitumor) (Meyerset al., 2008; Izumi et al., 2000; Setchell et al., 2001; Zubik and Mey-dani, 2003; Gutzeit et al., 2005). Phenolics such as quercetin, isorh-amnetin, galangin, chrysin or pinocembrin have been shown toserve as protective defense against oxidative damage (Meyerset al., 2008; Bao and Lou, 2006; Zielinska et al., 2001; Supratimet al., 2008).

Results from in vitro and in vivo studies indicate that galanginwith antioxidative and free radical scavenging activities is capableof modulating enzyme activities and suppressing the genotoxicityof chemicals (Moon et al., 2001). Flavonoid metabolites are detect-able in the circulation soon after consumption, albeit the gastroin-testinal tract is exposed in much higher local concentrations ofboth the parent compounds and metabolites following feeding.The evidence for in vivo antioxidant effects of flavonoids is confus-ing but it seems beneficial towards in vitro studies. Namely, cells inculture may behave differently from cells in vivo. The studies (Longet al., 2000; Halliwell, 2003) reported that the addition of flavo-noids to tissue culture media generated high levels of H2O2 andthat all of the cellular effects of flavonoids observed represent an‘artifact’ of cell culture due to oxidative stress on cells. On the otherhand, even after extensive flavonoid intake, flavonoid concentra-tion in plasma may be insufficient to exert systemic antioxidant ef-fects in vivo, probably as a result of flavonoid metabolites(methylated, sulphated or glucuronidated forms) which tend tohave decreased antioxidant activity and must be tested within vivo study. The results of the present examination showed thatthe antioxidant potential in the lysate of erythrocytes of mice

Table 2Gen expression profile in the liver and brain of mice fed with bee pollen (100 mg/kg bw).

Symbol GeneBank Description Liver Brain P

Controlmean ± SE

Bee pollenmean ± SE

Controlmean ± SE

Bee pollenmean ± SE

Hspa9a NM_010481 Heat-shock protein, A 49015.8 ± 1509.5 41106.0 ± 989.5 0.027Tnfsf6 NM_010177 Tumor necrosis factor (ligand) superfamily,

member 641650.9 ± 430.1 34678.0 ± 1132.0 0.038

Casp 1 NM_009807 Caspase 1 40534.6 ± 663.6 35875.7 ± 959.6 0.037Ccl21c NM_023052 Chemokine (C–C motif) ligand 21c (leucine) 55957.8 ± 183.1 50190.3 ± 1088.1 0.013

Comparative gene expression profile analyses between control mice and mice fed with bee pollen (100 mg/kg bw mixed with commercial food pellets) in the liver and brain.Control mice were fed with commercial food pellets (control). Samples of each tissue were pooled and consisted of three mice per group. Each analysis was performed at leastthree times. The data are mean values ± SE.p represent values significantly different from control at p < 0.05.

Estrogenic activity

0

2

4

6

8

330333.30.33

β-G

AL

AC

TIV

ITY

(%

OF

10n

M E

2)

Antiestrogenic activity

0

10

20

30

40

50

60

70

80

90

330333.30.33

CONCENTRATION (µg/ml)

β-G

AL

AC

TIV

ITY

(%

OF

1nM

E2)

Fig. 4. Estrogenic and anti-estrogenic activity of bee pollen 0.33 lg/ml, 3.3 lg/ml,33 lg/ml, and 330 lg/ml in reaction mixture. The results are expressed as percentof maximal b-gal activity using 17b estaradiol (positive control). The results arerepresentative of three consecutive experiments.

552 A. Šaric et al. / Food and Chemical Toxicology 47 (2009) 547–554

was elevated through increased activity of SOD, CAT and Gpx lev-els. In the liver bee pollen diet decreased LPO process throughreduction in TBARS production which is paralleled with decreasedSOD activity and increased CAT activity. In the brain the decreasedSOD activity does not seem to be related with TBARS concentra-tion. The difference in antioxidant status observed between theerythrocyte lysate, liver and brain might be associated with differ-ent absorbtion mechanisms, metabolism and diverse uptake ofmetabolic products of phenolics after bee pollen diet. There is evi-dence that flavonol glycosides might be responsible for antioxidantcapacity of blood plasma (Hollman and Katan, 1999). Namely, fla-vonoid classes determine their absorption and their metabolism.Since metabolism is able to transform one class of flavonoid into

another, i.e. galangin has been sequentially transformed tokaempferol and then to quercetin (Silva et al., 1997) new pharma-cological activity might result. Furthermore, methylation of quer-cetin into isorhamnetin appears immediately after absorption(Olthof et al., 2000). The work of Manach et al. (1998) demon-strated that the antioxidant effect of aglycone could be differentfrom that of conjugated derivatives of quercetin which could exertphysiological effects, possibly in synergism with other compoundsfrom food rich in polyphenols. Flavonoids are conjugated in the li-ver, intestine or kidneys and excreted into bile or urine. The profileof metabolites excreted in urine is markedly different to that ofplasma (Mullen et al., 2006). The results of Hollman and Katan(1997) reveal that the sugar moiety of quercetin glycosides is a sig-nificant determinant of their bioavailability and absorption whichimply reactivity towards free radicals (Rice-Evans et al., 1997).Conjugation of quercetin with glucose (provided from onions) en-hances absorption from the small gut. Because of the long half-lives of elimination, daily consumption of flavonol- and flavanon-containing C. incanus L. pollen diet might cause accumulation ofquercetin in blood and increase the antioxidant capacity of blood.The majority of the literature exhibited in vitro mutagenicity ofquercetin (Vrijsen et al., 1999). However, quercetin DNA adductswere of transient nature (Van der Woude et al., 2005), suggestingchemical instability of the adducts, so genotoxicity observedin vitro has no impact in vivo or even more be antimutagenic (For-mica and Regelson, 1995) with beneficial effects on lung cancer(Knekt et al., 1997) or as part of supportive medical therapy forpeptic ulcers (Borrelli and Izzo, 2000; Cos�kun et al., 2004). The con-sumption of a flavonoid-rich bee pollen diet increased antioxidantcapacity in the lysate of erythrocytes and reduced the amounts ofliver TBARS concentration. Although the precise mechanismsunderlying the antioxidant effects of flavonoids have yet to beidentified, the research reported to date suggests that consumptionof flavanols in the diet can significantly augment the oxidative de-fense system.

Up to now, studies analyzing differential gene expression pro-files after feeding with C. incanus L. bee pollen have not be re-ported. We functionally annotated the list of genes whichidentify gene that were differentially expressed after feeding withbee pollen. Our GEArray chip among others detects genes (CYP7A1,CYP7B1, and CYP4A10) from the CYP family. Flavonoids may in-duce biosynthesis of several CYPs, modulate their enzymatic activ-ities or are metabolized by a number of CYPs. Thus, there has beeninterest in studying the interactions of flavonoids with CYP450,since this process has the potential to interfere with metabolismof various drugs (Hodek et al., 2002). It has been observed that C.incanus L. bee pollen did not modulate expression of CYP450 sys-tem genes (CYP7A1, CYP7B1, and CYP4A10).

In our study, we observed no toxic functions of free radicals inthe liver after feeding with C. incanus L. reach bee pollen since,

A. Šaric et al. / Food and Chemical Toxicology 47 (2009) 547–554 553

although statistically not significant, both SOD2 and IL18 geneexpression in the liver is decreased. Even more, determination ofgene expression profiles of 96 genes in liver indicative of mousestress and toxicity showed statistically significant downregulationof gene which may directly regulate stress-responsive signalingpathways or antagonize signaling cascade that result in apoptosis(Hspa9a, Tnfsf6). The same treatment in brain significantly down-regulated gene expression of caspase1 and Ccl21c important in theapoptosis pathway and chemotaxis.

A low estrogenic activity of C. incanus L. reach bee pollen wasfound by the assays utilized in our study. The anti-estrogenic activ-ity was evaluated what is currently at the forefront of research onhormone associated cancer risk reduction (breast, uterine, andprostate) as well as a modality for promoting healthy prostatefunction in older man. This result refers to possibility that bee pol-len treatment may serve for developing preventive and therapeuticagents for various estrogen-mediated diseases.

The present study reveal that be pollen reach with C. incanus L.apart from their natural inherent character have attracted a greatdeal of attention not only as nutritional supplement and functionalfood diets, but also may be recognized as a healthy food to patientswith various diseases (cancer, cardiovascular diseases, and diabe-tes) and can be applied to various fields of medicine and food.

Conflict of interest statement

The authors declare that there are no conflicts of interest.

Acknowledgements

This study was partially supported by Croatian Ministry of Sci-ence, Education, and Sport (Grant No. 098-0982464-1647). Theauthors thank Mrs. Vesna Matešic and Iva Pešun Medimorec fortechnical assistance.

References

Aebi, H., 1984. Catalase in vitro. In: Packer, L. (Ed.), Methods in Enzymology.Academic Press, Orlando, pp. 121–126.

Ashida, H., Fukuda, I., Yamashita, T., Kanazawa, K., 2000. Flavones and flavonols atdietary levels inhibit a transformation of aryl hydrocarbon receptor induced bydioxin. FEBS Lett. 476, 213–217.

Attaguile, G., Russo, A., Campisi, A., Savoca, F., Acquaviva, R., Ragusa, N., Vanella, A.,2000. Antioxidant activity and protective effect on DNA cleavage of extractsfrom Cystus incanus L. and Cystus monspeliensis L. Cell Biol. Toxicol. 16,83–90.

Bao, M., Lou, Y., 2006. Isorhamnetin prevent endothelial cell injuries from oxidizedLDL via activation of p38MAPK. Eur. J. Pharmacol. 547, 22–30.

Beythien, A., Diemair, W., 1963. Laboratoriumsbuch für den Lebensmittelchemiker,vol. 8. Verlag von Theodor Steinkopff, Dresden/Leipzig, pp. 28–29.

Borrelli, F., Izzo, A.A., 2000. The plant kingdom as a source of anti-ulcer remedies.Phytother. Res. 14, 581–591.

Cos�kun, Ö., Kanter, M., Armutçu, F., Çetin, K., Kaybolmaz, B., Yazgan, Ö., 2004.Protective effects of quercetin, a flavonoid antioxidant, in absolute ethanol-induced acute gastric ulcer. Eur. J. Gen. Med. 1, 37–42.

Depeint, F., Gee, J.M., Williamson, G., Johnson, I.T., 2002. Evidence for consistentpatterns between flavonoid structures and cellular activities. Proc. Nutr. Soc. 61,97–103.

Flohe, L., Ötting, F., 1984. Superoxide dismutase assays. In: Packer, L. (Ed.), Methodsof Enzymology. Academic Press, Orlando, pp. 93–104.

Formica, J.V., Regelson, W., 1995. Review of the biology of quercetin and relatedbioflavonoids. Food Chem. Toxicol. 33, 1061–1080.

Gorsuch, T.T., 1970. The destruction of organic matter. In: Belcher, R., Freiser, H.(Eds.), International Series of Monographs in Analytical Chemistry. PergamonPress Elmsford, New York, pp. 55–133.

Gutzeit, H.O., Tokalov, S.V., Ludwig-M}uller, J., Rusak, G., 2005. Monitoring flavonoidmetabolism in human cells by exploiting fluorescence elicited upon quercetin/protein interactions. Croat. Chem. Acta 78, 337–342.

Halliwell, B., 2003. Oxidative stress in cell culture: an under-appreciated problem?FEBS Lett. 540, 3–6.

Hamamoto, R., Ishiyama, K., Hashimoto, K., Yamaguchi, M., 2006. Characterizationof the active component in bee pollen Cystus ladaniferus extract in stimulatingbone calcification and in inhibiting bone resorption in vitro. J. Health Sci. 52,607–612.

Hodek, P., Trefil, P., Stiborová, M., 2002. Flavonoids-potent and versatile biologicallyactive compounds interacting with cytochromes P450. Chem. Biol. Interact. 139,1–21.

Hollman, P.C.H., Katan, M.B., 1997. Absorption, metabolism and health effects ofdietary flavonoids in man. Biomed. Pharmacother. 51, 305–310.

Hollman, P.C.H., Katan, M.B., 1999. Dietary flavonoids: intake, health effects andbioavailability. Food Chem. Toxicol. 37, 937–942.

Izumi, T., Piskula, M.K., Osawa, S., 2000. Soy isoflavone aglycones are absorbedfaster and in higher amounts than their glucosides in humans. J. Nutr. 130,1695–1699.

Kjeldahlu, J., 1960. Officials methods of analysis of the association of officialagricultural chemists. J. Assoc. Off. Agric. Chem. 9, 12–13.

Knekt, P., Järvinen, R., Seppänen, R., Heliövaara, M., Teppo, L., Pukkala, E., Aromaa, A.,1997. Dietary flavonoids and the risk of lung cancer and other malignantneoplasms. Am. J. Epidemiol. 146, 223–230.

Krishnamachary, V., Levine, L.H., Pare, P.V., 2002. Flavonoid oxidation by the radicalgenerator AIBN: a unified mechanism for quercetin radical scavenging. J. Agric.Food Chem. 50, 4357–4363.

Kullavanijaya, P., Lim, H.W., 2005. Photoprotection. J. Am. Acad. Dermatol. 52, 937–958.

Liu, J.W., Picard, D., 1998. Bioactive steroids as contaminants of the common carbonsource galactose. FEMS Microbiol. Lett. 159, 167–171.

Long, L.H., Clement, M.V., Halliwell, B., 2000. Artifacts in cell culture: rapidgeneration of hydrogen peroxide on addition of (�)-epigallocatechin, (�)-epigallocatechin gallate, (+)-catechin, and quercetin to commonly-used cellculture media. Biochem. Biophys. Res. Commun. 273, 50–53.

Louveaux, J., Maurizio, A., Vorwohl, G., 1978. Methods of melissopalynology. BeeWorld 59, 139–153.

Lowry, O.H., Rosebrough, N.J., Farr, A.L., Randall, R., 1951. Protein measurement withfolin phenol reagent. J. Biol. Chem. 193, 256–275.

Manach, C., Morand, C., Crespy, V., Demigné, C., Texier, O., Régérat, F., Rémésy, C.,1998. Quercetin is recovered in human plasma as conjugated derivatives whichretain antioxidant properties. FEBS Lett. 426, 331–336.

Meyers, K.J., Rudolf, J.L., Mitchell, A.E., 2008. Influence of dietary quercetin onglutathione redox status in mice. J. Agric. Food Chem. 56, 830–836.

Miller, J.H., 1972. Assay for b-galactosidase. In: Miller, J.H. (Ed.), Experiments inMolecular Genetics, Cold Spring Harbor Laboratory. Cold Spring Hrbour Press,New York, pp. 352–355.

Moon, Y.H., Sohn, S.J., William, W.A., 2001. Anti-genotoxicity of galangin as a cancerchemopreventive agent candidate. Mutat. Res. 488, 135–150.

Moon, Y.J., Wang, X., Morris, M.E., 2006. Dietary flavonoids: effects on xenobioticand carcinogen metabolism. Toxicol. Vitro 20, 187–210.

Mullen, W., Edwards, C.A., Crozier, A., 2006. Absorption, excretion and metaboliteprofiling of methyl-, glucuronyl-, glucosyl- and sulpho-conjugates of quercetinin human plasma and urine after ingestion of onions. Brit. J. Nutr. 96, 107–116.

Oh, S.M., Chung, H., 2006. Antiestrogenic activities of Ginko biloba extracts. J. SteroidBiochem. Mol. Biol. 100, 167–176.

Ohkawa, H., Ohishi, N., Yagy, K., 1979. Assay for lipid peroxides in animal tissues bythiobarbituric acid reaction. Anal. Biochem. 95, 351–358.

Olthof, M.R., Hollman, P.C.H., Vree, T.B., Katan, M.B., 2000. Bioavailabilities ofquercetin-3-glucoside and quercetin-4-glucoside do not differ in humans. J.Nutr. 130, 1200–1203.

Paglia, D.E., Valentine, W.N., 1967. Studies on the quantitative and qualitativecharacterization of erythrocyte glutathione peroxidase. J. Lab. Clin. Med. 70,158–169.

Pinto, B., Picard, D., Reali, D., 2004. A recombinant yeast strain as a short termbioassay to assess estrogen-like activity of xenobiotics. Ann. Igiene 16, 579–585.

Rice-Evans, CA., Miller, NJ., Paganga, G., 1997. Antioxidant properties of phenoliccompounds. Trends Plant Sci. 2, 152–159.

Setchell, K.D., Brown, N.M., Desai, P., 2001. Bioavailability of pure isoflavones inhealthy humans and analysis of commercial soy isoflavone supplements. J. Nutr.131, 1362S–1375S.

Silva, D.I., Rodrigues, AS., Laires, A., Rueff, J., 1997. Metabolism of galangin by ratcytochromes P450 relevance to the genotoxicity of galangin. Mutat. Res. 393,247–257.

Song, W.O., Beecher, G.R., Eitenmiller, R.R., 2000. Chemical analysis. In: Song, W.O.,Beecher, G.R., Eitenmiller, R.R. (Eds.), Modern Analytical Methodologies in Fat-and Water-Soluble Vitamins. Wiley-Interscience, New York, pp. 134–158.

Strauss, L., Santti, R., Saarinen, N., Streng, T., Joshi, S., Mäkelä, S., 1998. Dietaryphytoestrogens and their role in hormonally dependent disease. Toxicol. Lett.102–103, 349–354.

Supratim, R., Chandana, S., Kunal, R., 2008. QSAR modeling for lipid peroxidationinhibition potential of flavonoids using topological and structural parameters.Cent. Eur. J. Chem. 6, 267–276.

Tham, D.M., Gardner, C.D., Haskell, W.L., 1998. Potential health benefits of dietaryphytoestrogens: a review of the clinical, epidemiological, and mechanisticevidence. J. Clin. Endocrinol. Metab. 83, 2223–2235.

Van der Woude, H., Alink, G.M., van Rossum, B.E.J., Steeg, K.W., Walle, T., Rietjens,M.C.M., 2005. Formation of transient covalent protein and DNA adducts by cellswith and without oxidative enzyme activity. Chem. Res. Toxicol. 18, 1907–1916.

Vrijsen, R., Michotte, Y., Boeye, A., 1999. Metabolic activation of quercetinmutagenicity. Mutat. Res. 232, 243–248.

Yamaguchi, M., Hamamoto, R., Uchiyama, S., Ishiyama, K., Hashimoto, K., 2006.Anabolic effect of bee pollen Cystus ladaniferus extract on bone components in

554 A. Šaric et al. / Food and Chemical Toxicology 47 (2009) 547–554

the femoral-diaphyseal and -metaphyseal tissues of rats in vitro and in vivo. J.Health Sci. 52, 43–49.

Zubik, L., Meydani, M., 2003. Bioavailability of soybean isoflavones from aglycone andglucoside forms in American women. Am. J. Clin. Nutr. 77, 1459–1465.

Zielinska, M., Kostrzewa, A., Ignatowicz, E., Budzianowski, J., 2001. The flavonoids,quercetin and isorhamnetin 3-0-acylglucosides diminish neutrophiloxidative metabolism and lipid peroxidation. Acta Biochim. Pol. 48, 183–189.