Steroids 68 (2003) 1147–1155

Synthesis of hapten and conjugates of coumestroland development of immunoassay

Oldrich Lapcık a,∗, Jan Štursaa, Tereza Kleinováa, Michaela Vıtkováb,Hana Dvorákovác, Borivoj Klejdusd, Jitka Moravcováa

a Department of Chemistry of Natural Compounds, Institute of Chemical Technology, Technická 5, 166 28 Praha 6, Czech Republicb Department of Biochemistry and Microbiology, Institute of Chemical Technology, Technická 5, 166 28 Praha 6, Czech Republic

c Central Laboratories, Institute of Chemical Technology, Technická 5, 166 28 Praha 6, Czech Republicd Department of Chemistry and Biochemistry, Mendel University of Agriculture and Forestry, Brno, Czech Republic

Received 13 January 2003; received in revised form 5 August 2003; accepted 26 August 2003

Abstract

3-O-Carboxymethylcoumestrol was prepared as the hapten for immunoassay by a partial alkylation of coumestrol with ethyl chloroacetatein acetone alkalized with potassium carbonate. 3-O-Ethoxycarbonylmethylcoumestrol was separated by column chromatography and finallywas hydrolyzed with formic acid.1H and13C NMR data (APT, COSY, HMQC, and HMBC) revealed that the reaction was regioselective, as3-O-ethoxycarboxymethylcoumestrol was the only monosubstituted derivative. The hapten was then conjugated to bovine serum albuminand used for immunization of rabbits. A radioimmunoassay (RIA) system was established based on the polyclonal antiserum and a125I-labeled hapten-tyrosine methyl ester conjugate as the radioligand. Parameters of the RIA: sensitivity: 12 pg per tube, 50% intercept:140 pg per tube, working range: 20–4000 pg per tube. The cross-reactivity of a panel isoflavonoid and lignan phytoestrogens was eithernegligible (e.g. formononetin 0.07%; biochanin A 0.06%) or not detectable at all. The major immunoreactive peak in HPLC fractionsfrom an alfalfa extract had the same retention time as coumestrol standard and represented 94.8% of the signal. The remaining 5.2% ofimmunoreactivity was distributed between five minor peaks. We conclude that after the validation for particular matrices, the method willbe a useful tool for analysis of coumestrol, especially in low volume and low concentration samples.© 2003 Elsevier Inc. All rights reserved.

Keywords:Steroid; Phytoestrogen; Coumestrol; Immunoassay; Alfalfa

1. Introduction

Coumestrol (3,9-dihydroxy-6H-benzofuro[3,2-c][1]benz-opyran-6-one) is a coumestan-derived phytoalexin presentin several leguminous genera, e.g. theMedicago, Trifolium,Phaseolus, andGlycinespp.[1–4], present in certain foodsand abundant in forages. Coumestrol interacts with estro-gen receptors (ERs) in vertebrates and acts as an estrogenagonist [5]. It has been reported to display a wide scaleof biological effects on uterus[6,7], bone[7], breast[8],and brain[9], which may be either beneficial or adverseones, depending on the circumstances. It has been hypoth-esized to contribute to the health-promoting effects of soyand other legumes together with the isoflavonoids; on theother hand, the toxicity of high coumestrol doses has beendocumented on cattle and sheep[6]. Coumestrol also has

∗ Corresponding author. Tel.:+420-224354265; fax:+420-224311082.E-mail address:oldrich.lapcik@vscht.cz (O. Lapcık).

been used also as a fluorescent ER ligand thus enablingobservation of ER distribution and trafficking in the cell[10,11]. Due to their expected beneficial effect on humans,isoflavonoid concentrates from different legumes have beenintroduced to the market as dietary supplements. This formhas the potential enables to achieve higher concentrations ofisoflavonoids in body fluids than is possible by diet. Thus,a scale of analytical methods for phytoestrogen estimationin different materials and in different concentration rangesare needed—from food and forage up to the body fluids,from milligrams down to picograms. Analytical methodsfor coumestrol include HPLC, GC, and CZE combined withdifferent suitable types of detection (e.g. UV-spectrometry,fluorescence detection, electrochemical detection, or massspectrometry)[12–14]. Receptor binding assay systemswere also established for analysis of phytoestrogens in-cluding coumestrol after chromatographic pre-separationof individual analytes[15]. Here we present the synthesisof 3-O-carboxymethylcoumestrol and the development of

0039-128X/$ – see front matter © 2003 Elsevier Inc. All rights reserved.doi:10.1016/j.steroids.2003.08.014

1148 O. Lapcık et al. / Steroids 68 (2003) 1147–1155

a competitive radioimmunoassay (RIA) based on the poly-clonal antiserum to the conjugate of this hapten with bovineserum albumin (BSA).

2. Experimental

2.1. Material and methods

BSA was purchased from Bioveta (Ivanovice, CzechRepublic), casein from Difco Labs. (Detroit, MI). Di-ethyl ether (stabilized with 0.001% fenidone) was fromSynthesia (Pardubice, Czech Republic); all other organicsolvents were from Merck (Darmstadt, Germany). Dicy-clohexylcarbodiimide (DCC), chloramine-T, and coume-strol were obtained from Fluka (Buchs, Switzerland),N-hydroxysuccinimide (NHS) and charcoal (Norit A) werefrom Serva (Heidelberg, Germany), dextran T-70 from Phar-macia (Uppsala, Sweden), tyrosine methyl ester (TME),tyramine hydrochloride (TYR), complete Freund’s adju-vant, dioctyl sulphosuccinate, daidzein, daidzin, genistein,genistin and biochanin A were from Sigma (St. Louis, MO),3′-hydroxydaidzein, formononetin, and sissotrin were fromIndofine (Somerville, NJ). Isoformononetin and prunetinwere prepared by selective methylation of daidzein andgenistein, respectively, as described elsewhere[16]. Equol,enterolactone, and enterodiol[17] were provided by Prof.Kristiina Wähälä (Department of Chemistry, University ofHelsinki, Finland). 5-Hydroxy-4′,7-dimethoxyisoflavoneand 4′,7-dihydroxy-5-methoxyisoflavone fromEriosematuberosum[18] were generous gifts of Dr. Wei Guang Ma(Faculty of Agriculture, Hokkaido University, Sapporo,Japan). All other substances were of analytical grade fromLachema (Brno, Czech Republic).

Melting points were determined with a Kofler hot blockand are uncorrected. Column chromatography was per-formed on Silica gel 60, particle size 0.063–0.2 mm (Fluka,Germany), and TLC on silica gel according to Stahl(10–40�m, Merck, Germany) with detection by sprayingwith 1% Ce(SO4)2 in 10% H2SO4 and subsequent mineral-ization or DC-Alufolien Kieselgel 60 F 254 plates (Merck,Germany) with UV detection were used. Solutions wereconcentrated under reduced pressure with a bath tempera-ture below 40◦C.

NMR data were extracted from spectra measured inCDCl3 or DMSO-d6 solution (tetramethylsilane as an in-ternal standard) at 25◦C with a Bruker Avance DRX-500instrument (1H, 500 MHz; 13C, 125.7 MHz). Chemicalshifts are given in ppm (δ-scale). Assignments of1H and13C signals are based on APT, COSY, HMQC, and HMBCexperiments. Coupling constants are given in Hz.

The HPLC system consisted of SCL 10-10 AVP Sys-tem Controller, SIL-10 ADVP Auto injector, LC10-ADVTPump, FCV 10 ALVT Gradient Unit, SPD-M10 AVT DiodeArray Detector, FRC-10A Fraction Collector (Shimadzu,Japan). A Purospher Star RP18e 125/4 column with a guard

column Purospher Star RP18e 4/4 (Merck, Germany) wasused.

A gradient elution with mobile phases A (water) and B(methanol) was as followed (all steps linearly): 0 min, A=60%, B= 40%; 5 min, B= 52%; 20 min, B= 70%; 25 min,B = 100%; 25–45 min, B= 100%; then step to A= 60%and reconditioning of the column for 10 min. The flow ratewas 0.8 ml min−1 and the temperature was set at 25◦C. UVat 254 and 350 nm was recorded. Typically 50�l of alfalfaextract diluted in ethanol was applied on the column and0.4 ml fractions were collected. HPLC fractions were evap-orated on a speedvac and reconstituted in the assay bufferfor the immunoanalysis.

2.1.1. HPLC–MSHPLC–MS system was HP 1100 (Hewlett Packard,

Waldbronn, Germany) equipped with vacuum degasser(G1322A), binary pump (G1312A), autosampler (G1313A),column thermostat (G1316A), and diode array detector(model G1315A). The system was coupled on-line tothe mass selective HP MSD detector (G 1946A, HewlettPackard, Palo Alto) and controlled by the ChemStationsoftware (Rev A 07.01). A Zorbax Exlipse XDB C8(150 mm× 4.6 mm; 5�m, Zorbax, Agilent Technologies,USA) analytical column was used. A gradient elutionwith mobile phases A (acetonitrile) and B (0.3% formicacid in water) was as follows (all steps linearly): 0 min,A = 15%, B= 85%; 20 min, B= 65%; 25 min, B= 50%;30 min, B = 45%; 45 min, B= 15%. The flow rate was0.8 ml min−1 and the temperature of the column oven wasset at 40◦C. The volume of sample applied on columnwas 50�l. The effluent from liquid chromatograph wasintroduced directly into the quadrupole mass spectrometeroperating in positive ESI mode. The nebulizer gas pressurewas 60 psi, the drying gas was 12 l min−1 at the tempera-ture 300◦C and capillary voltage was 3500 V. Individualisoflavones and their conjugates were identified by compar-ing their retention times (tR), molecular ions [M+ H]+ andcharacteristic fragments with those of standards[13,19].

2.2. Synthesis and characterization of the hapten

Coumestrol (1) (80 mg, 0.3 mmol) was suspended in ace-tone (8 ml), K2CO3 (90 mg, 0.66 mmol) was then addedand the mixture was stirred at ambient temperature. Dur-ing 5 min, ethyl chloroacetate (100 mg, 0.8 mmol) wasadded dropwise and stirring was continued under refluxat 80◦C (bath temperature) for 3.5 h until the analysisby TLC (18:1, chloroform/methanol,RF: 1, 0.3; 2, 0.95;3, 0.5) proved the presence of maximal concentration ofethoxycarbonyl derivative3. The reaction mixture was thenconcentrated to dryness in vacuo and a residue (104 mg)was diluted with acetone (1 ml) and passed through a shortsilica gel column (5 g, acetone). Combined fractions con-taining coumestrol and products2 and 3 were collected,solvent was evaporated and a residue was separated on a

O. Lapcık et al. / Steroids 68 (2003) 1147–1155 1149

Table 113C NMR data for compounds2 (in CDCl3) and 3 (in DMSO-d6):chemical shifts (δ, ppm)

O

O

O

OO

O

O

O

O

13

4a

7

9

6a

10a 11a

13

14

18

1716

12

2

4

67a8

10 1a15

19

Nuclei 2 3 Nuclei 2 3

C-1a 106.88 106.47 C-9 158.14 157.88C-1 122.75 122.99 C-10a 157.41 157.73C-2 113.55a 113.73 C-10 98.26 99.15C-3 160.70 160.96 C-11a 160.09 159.33C-4a 154.91 154.70 C-12 66.02b 65.49C-4 102.31 102.90 C-13 168.50 168.61C-6a 103.78 103.53 C-14 61.76c 61.31C-6 156.18 156.57 C-15 14.15d 14.48C-7a 117.61 114.89 C-16 65.42b –C-7 121.85 121.26 C-17 167.85 –C-8 113.32a 114.64 C-18 61.56c –

These signals (with superscript letters a, b, and c) might be interchanged.The value (with superscript letter d) is together with C-19.

silica gel column (18:1, chloroform/methanol). First eluted3,9-diethoxycarbonylmethoxy-6H-benzofuro[3,2-c][1]benz-opyran-6-one (2) was obtained as a light yellow powder,mp 170◦C in the yield of 50% (66 mg).1H NMR (CDCl3),δ (ppm): 7.96 (d, 1H,J 8.5 Hz, H-7), 7.89 (d, 1H,J 8.7 Hz,H-1), 7.18 (s, 1H, H-10), 7.07 (d, 1H,J 8.5 Hz, H-8), 7.02(d, 1H,J 8.7 Hz, H-2), 6.94 (s, 1H, H-4), 4.72 (s, 4H, H-12and H-16), 4.31 (q, 4H,J 7.1 Hz, 2× CH2), 1.33 (t, 3H,J7.1 Hz, CH3) and 1.32 (t, 3H,J 7.1 Hz, CH3). 13C NMRdata are presented inTable 1.

Further elution afforded 3-ethoxycarbonylmethoxy-9-hydroxy-6H-benzofuro[3,2-c][1]benzopyran-6-one (3) asan amorphous solid in the yield of 11% (12 mg).1H NMR(DMSO-d6), δ (ppm): 10.09 (s, OH), 7.93 (d, 1H,J 8.6 Hz,H-1), 7.71 (d, 1H,J 8.3 Hz, H-7), 7.19 (s, 1H, H-4 or H-10),7.18 (s, 1H, H-4 or H-10), 7.10 (d, 1H,J 8.6 Hz, H-2), 6.96(d, 1H, J 8.3 Hz, H-8), 4.96 (s, 2H, H-12), 4.19 (q, 2H,J7.0 Hz, 2× H-14), 1.23 (t, 3H, J 7.0 Hz, 3× H-15). 13CNMR data are presented inTable 1.

Finally, coumestrol was recovered by elution with acetone(27 mg, 34%).

The mixture of ethoxycarbonyl derivative3 (20 mg,0.06 mmol) and aqueous 50% formic acid (4 ml) was stirredat 110◦C (bath temperature) for 20 h until TLC (18:1,chloroform/methanol,RF: 4, 0.25 tailing) indicated thatthe reaction was complete. The mixture was then concen-trated to dryness in vacuo and pure free acid 3-carboxyme-thoxy-9-hydroxy-6H-benzofuro[3,2-c][1]benzopyran-6-one(4) (16 mg, 87%) was obtained.1H NMR (DMSO-d6), δ

(ppm): 8.32 (s, 1H), 7.85 (d, 1H,J 8.6 Hz), 7.60 (d, 1H,J8.4 Hz), 7.06 (s, 1H), 7.00 (d, 1H,J 8.6 Hz), 6.93 (s, 1H),6.88 (d, 1H,J 8.4 Hz), 4.45 (s, 2H).13C NMR (DMSO-d6),

δ (ppm): 170.03, 161.90, 158.95, 158.05, 157.48, 155.91,154.21, 122.03, 120.37, 114.14, 114.04, 113.15, 104.81,102.33, 102.24, 98.55, 67.56.

2.3. Immunogen synthesis and immunization

The immunogen was synthesized in a reversed micel-lar system, developed originally for steroid immunogensynthesis by Yatsimirskaya et al.[20], with minor modifica-tions. In brief, 4.0 mg of the 3-O-carboxymethylcoumestrol(4) was left to react with DCC and NHS (molar ra-tio 3:4:5) in 200�l of anhydrous mixture dimethylfor-mamide/dimethylsulfoxide (1:1). After 4 h, the reactionmixture was centrifuged to sediment the crystals of dicy-clohexylurea and the supernatant used for conjugation withBSA in a reversed micellar system. The starting molar ratio4:BSA was 50:1. BSA (12 mg) was dissolved in 0.75 ml0.02 M bicarbonate buffer, pH 8.5. This solution was addeddropwise to 5 ml of 0.3 M dioctyl sulphosuccinate in octaneunder continuous stirring. After the mixture became clear,the dimethylformamide/dimethylsulfoxide solution of theactive coumestrol intermediate was added. The mixture wasstirred an additional 24 h at ambient temperature. Coume-strol 4–BSA conjugate was isolated from the mixture byprecipitation with three volumes of cold acetone at−20◦Cfollowed by centrifugation. The supernatant was removedand the sediment was washed twice with 2 ml of cold ace-tone, dissolved in 1 ml of distilled water, filtered through a0.22�m Millipore filter and lyophilized. The hapten/carrierprotein ratio of the conjugate was estimated by UV spec-trometry at 350 nm yielding 11 molecules of the hapten4per one molecule of BSA.

Rabbits were immunized and antisera collected by usinga standard procedure[21]. Briefly, the immunogen dissolvedin sterile isotonic saline and emulsified with equal volumeof complete Freund’s adjuvans was applied subcutaneouslyinto three to four sites on the rabbits’ backs and legs. Im-munization was repeated four times in 4-week periods; oneboost represented 0.3 mg of the immunogen in 0.3 ml of theemulsion. The final serum harvest was performed 10 daysafter the last immunization by a cordial puncture under com-plete anesthesia (ketamine/xylasine).

2.4. Synthesis of125I-labeled tracers

Two different approaches to radioligand synthesis wereevaluated. The first one consists of iodination of tyrosinemethylester (TME) and its subsequent conjugation of the io-dinated molecule to the 3-O-carboxymethylcoumestrol (4),as described elsewhere for daidzein and genistein[22,23].

The second approach is based on iodination of previouslysynthesized conjugate of the hapten4 with TME or withTYR. The conjugates of 3-O-carboxymethylcoumestrolwith TME and with TYR were prepared as describedfor steroid-TME radioligands[24]. Briefly, 0.5 mg of the3-O-carboxymethylcoumestrol (4) was dissolved in 10�l

1150 O. Lapcık et al. / Steroids 68 (2003) 1147–1155

of dimethylsulfoxide/dimethyl formamide mixture (1:1)and 0.4 mg DCC and 0.25 mg NHS were added (both dis-solved in 10�l of dimethylformamide). After 4 h, 0.4 mgof TME or TYR in 10�l of dimethylformamide wereadded and reacted overnight. Resulting conjugates werepurified by TLC (plate: Merck Art. 5583, mobile phase:dichloromethane/isopropanol 95:5) and approximately 5�gwere iodinated by the chloramine-T method[20].

2.5. RIA system

Assay buffer was prepared by dissolving 20 mM sodiumphosphate in saline, containing sodium azide and casein,1.0 g/l each. The radioligand (20,000–30,000 cpm) andthe antibody (working dilution 1:30,000) dissolved in theassay buffer (100�l of each) were added to the tubes con-taining diluted samples and standards. Final volume inthe tube was 300�l. After vortex-mixing and incubationovernight at 4◦C, bound and free portions were separatedby dextran-charcoal adsorption (0.5 ml of the suspensionof 0.25 g charcoal and 0.025 g dextran T-70 in 100 mlassay buffer) and after additional incubation at 4◦C for15 min. the tubes were centrifuged at 4◦C for 10 min. Ra-dioactivity in the supernatants was measured in BertholdTwelve-Channel Gamma Counter (Wilbad, Germany) andthe integral built-in software was used for evaluation. Eachsample was assayed in duplicate.

2.6. Preparation of alfalfa samples

2.6.1. SeedlingsAlfalfa seeds were purchased from the “Country Life”

food store, which specializes in healthy nutrition products.Seeds were germinated 8 days according to the recommen-dations of the producer, using tap water for their wettingtwice daily. Samples were taken every day and kept deepfrozen until lyophilized. The dry material was disintegratedin a grinder and extracted 4 h with 80% ethanol. The extractswere filtered and adjusted on final volume 25 ml of extractper 1.0 g dry matter. Before the analysis, the extracts werediluted 100 times with the assay buffer.

O

O

O

OH

OH O

O

O

OCH2COOEt

EtOOCCH2O

O

O

O

OCH2COOH

OH

O

O

O

OCH2COOEt

OH

acetone

K2CO

3

ClCH2COOEt

1 2

4

50% HCOOH

+

3

Fig. 1. The coumestrol hapten4 preparation.

2.6.2. Alfalfa cutAlfalfa herbage was obtained from the Research Insti-

tute of Animal Production (Prague, Czech Republic). Pow-dered alfalfa hay (0.5 g) was refluxed with 66% acetonitrilein water (20 ml) containing hydrochloric acid (3.5 M, 5 ml)at 95◦C for 45 min. The mixture was filtered and the vol-ume of filtrate adjusted to 25 ml with 66% acetonitrile. Be-fore the analysis the extracts were diluted with water andre-extracted with diethyl ether. The ether was evaporatedand the residue reconstituted in the solvent suitable for thesubsequent procedure (i.e. 80% ethanol for HPLC and theassay buffer for RIA, respectively).

3. Results

3.1. Characterization of the hapten4

Coumestrol was alkylated according to the procedure de-scribed for isoflavones previously[25] with ethyl chloroac-etate being used instead of ethyl bromoacetate (Fig. 1). Thereaction was very fast and the maximum of concentrationof the desired product3 was achieved in 3.5 h. The C-3hydroxyl group was alkylated regioselectively according toNMR measurements (seeSection 2and Table 1). Finally,saponification of ethyl ester3 afforded the correspondingacid 4 and its structure was confirmed by both1H and13CNMR data.

3.2. Characterization of the radioligands

Radioligands were prepared by conjugation of iodinatedTME to hapten4 as well as by iodination of previouslyprepared conjugates of4 with TME or with TYR. Bothapproaches gave mixtures of radioactive products, whichwere then separated by TLC. The immunoreactive com-pounds represented minor portions of total radioactivity;their TLC and immunochemical characteristics are sum-marized inTable 2. TLC mobility of the iodinated hapten4–TME conjugate was equal to that of conjugate of hapten4with iodinated TME; however, the radiochemical yield and

O. Lapcık et al. / Steroids 68 (2003) 1147–1155 1151

Table 2Characterization of hapten4/3-TME and hapten4/3-TYR radioligands

Radioligand TLCcharacteristicsa (RF)

Immunochemicalcharacteristics

S1 S2 Titerb 50% interceptc

(pg per tube)

4-TYR 0.39 0.38 – –4-TYR iodinated 0.43 0.49 25000 1774-TME 0.47 0.52 – –4-TME iodinated 0.52 0.66 30000 1404-conjugated

TME-[125I]0.52 0.66 8000 807

a TLC plate: Merck SiO2 Art. 5583; S1: dichlormethane/2-propanol(95:5); S2: dichlormethane/2-propanol/acetic acid (96:4:0.5), developedtwice.

b Working dilution of the antiserum no. 61 at which 50% of theradioactivity present in the system was specifically bound.

c Derived from the calibration curve in a log-logit plot.

the immunochemical data favored the first approach. Theiodinated coumestrol4–TME conjugate was selected as theligand for further studies.

3.3. RIA system

The sensitivity expressed as a minimal amount of coume-strol distinguishable from the zero with 95% probability was12 pg per tube. The 50% intercept of the calibration curvewas 140 pg per tube; working range was 20–4000 pg per tube(Fig. 2). Samples of alfalfa extracts and of HPLC fractionsof alfalfa extracts were analyzed six times in one series aswell as on 6 separate days, for the determination of intra- andinter-assay coefficient of variation (Table 3). The indepen-dence of results on dilution of the sample was demonstratedon two alfalfa extracts diluted one to four times (Fig. 3).

S1: y = 4653x - 97; r2 = 0.9936

S2: y = 4278x - 13; r2 = 0.9755

0

1000

2000

3000

4000

5000

0.00 0.20 0.40 0.60 0.80 1.00

Reciprocal dilution

Con

cent

rati

on (

pg/t

ube)

Fig. 3. Dependence of immunoreactivity on dilution of two alfalfa extracts diluted 1.5, 2.0, 2.5, 3, and 4 times.

0

20

40

60

80

100

10 1000 10000

Concentration of coumestrol (pg/tube)

B/B

0 (%

)

100

Fig. 2. Standard curve for coumestrol.

Specificity of the antiserum was tested by a panel ofisoflavonoids and lignans in a single concentration 20,000 pgper tube and the signal was compared to that of coume-strol. Following substances were tested: daidzein, daidzin,

Table 3Intra- and inter-assay coefficients of variation

Samplea Coumestrol(pg per tube)

Intra-assaycoefficient (%)

Inter-assaycoefficient (%)

HPLC fraction 35 939.8 – 9.07HPLC fraction 36 1907.5 – 3.85HPLC fraction 37 240.3 – 6.88Seeds day 0 135.5 7.34 –Sprouts day 2 279.1 1.69 –Sprouts day 4 439.9 1.51 –Sprouts day 6 727.7 5.35 –

a Each sample was processed six times.

1152 O. Lapcık et al. / Steroids 68 (2003) 1147–1155

0

200

400

600

800

1000

1200

1400

1600

1800

2000

0 5 10 15 20 25 30 35 40 45 50 55 60 65

HPLC fraction

Imm

unor

eact

ivit

y (p

g/tu

be)

Fig. 4. Immunoreactivity of individual HPLC fractions of an extract from alfalfa hay.

formononetin, isoformononetin, 3′-hydroxydaidzein, genis-tein, genistin, biochanin A, prunetin, 4′,7-dihydroxy-5-methoxyisoflavone, 5-hydroxy-4′,7-dimethoxyisoflavone,sissotrin, equol, enterolactone, enterodiol, and severalsteroids. The only detectable cross-reactivities were ob-served for formononetin (0.07%) and biochanin A (0.06%);the other compounds did not cross-react at all.

3.4. Coumestrol immunoreactivity in alfalfa

Analysis of HPLC fractions from alfalfa extracts showedone major immunoreactive peak with identical retention timeas the coumestrol standard (tR: 19.5 min;Fig. 4). This peakrepresented 94.8% of whole immunoreactivity on the chro-matogram. The rest of the signal was divided between five

0.0

1.0

2.0

3.0

4.0

5.0

6.0

7.0

8.0

0 1 2 3 4 5 6 7 8

Day

Cou

mes

trol

(m

g/kg

; dr

y w

eigh

t)

Light

Dark

Fig. 5. Coumestrol immunoreactivity in alfalfa seedlings incubated under a glass lid (light) or a non-transparent lid (dark).

small peaks. One of them (tR: c14.5 min) was more polarand the others were less polar (tR: 23.0, 27.5, 29.0, and30.5 min, respectively) than coumestrol.

Coumestrol content in alfalfa sprouts increased during thegermination from 1.2 mg/kg on day 0 up to 7.3 mg/kg (dryweight) on day 8 (Fig. 5).

3.5. Correlation with HPLC–MS

Eighteen extracts of alfalfa hay with coumestrol con-tent from 1.26 to 72.2 mg/kg (dry weight) were analyzedwith both immunoassay and HPLC–MS. For RIA, dupli-cate determinations were used and for HPLC–MS the valueswere based on single determinations. The correlation be-tween RIA and HPLC–MS values was 0.8989 (RIA result=

O. Lapcık et al. / Steroids 68 (2003) 1147–1155 1153

y = 1.179x

r = 0.8989

0

10

20

30

40

50

60

70

0 10 20 30 40 50 60 70

HPLC-MS (mg/kg)

RIA

(m

g/kg

)

Fig. 6. Correlation between coumestrol RIA and HPLC–MS in alfalfahay extracts.

1.18×HPLC-MS result) over the entire range of concentra-tions (Fig. 6).

4. Discussion

Immunoassay is a suitable complement to the instrumentalmethods (i.e. HPLC, GC, and CZE combined with differenttypes of detection), especially for screening purposes, forsamples with low concentration of the analyte and for smallvolume samples such as the body fluids, cell, and tissueculture supernatants, etc.

First step in developing immunoassay for a smallmolecule is to design and synthesize a suitable hapten. Hy-droxylated analytes offer the opportunity to use the hydroxygroup for the introduction of anO-carboxyalkyl spacer arm.Coumestrol has two hydroxyl groups available for suchconstruction and it was necessary to know exactly whichof them was modified. The location of the ethoxycarbonyllinkage in3 was deduced from its HMBC spectra. The as-signment of the C-3 and C-9 carbon resonance was the keyto the determination of the linkage position in3, but dif-ferentiation between these carbons was difficult to achieve.Thus, carbon C-6a was chosen as a starting point because itcorrelates through a three-bond interaction with only protonH-7 (Fig. 7). Moreover, the chemical shift of the C-6a atom(103.53 ppm) was quite different from the ones of carbonsC-3, C-4a, C-9, C-10a, and C-11 which are linked to an oxy-gen atom (Table 1), so it can be easily assigned. The protonH-7 (7.71 ppm) gave another three-bond interaction withboth C-9 and C-10a atoms (Fig. 5) and a two-bond correla-tion of carbon C-9 with the proton H-8 (6.96 ppm) allowedthe distinction between them. Thus, the signal located at157.88 ppm was attributed to the carbon C-9. A similarapproach was applied for the elucidation of the carbon C-3location. The presence of both the three-bond interactionwith H-1 (7.93 ppm) and the two-bond interaction with

H-2 (7.10 ppm) indicated that the13C signal at 160.96 ppmcorresponded to the C-3 atom. After both C-9 and C-3 car-bon signals were assigned, we were able to see which ofthem correlated with the H-12 protons (Fig. 5). It was justthe C-3 carbon, which displayed cross-peak with the H-12protons (4.96 ppm) and the integrity of the side chain wasfurther demonstrated by the correlation of the H-12 protonswith the C-13 carbon and also by the correlation of theH-14 protons with the carbon C-13. Based on the analysisof the HMBC spectra, all proton (seeSection 3) and car-bon (Table 1) chemical shifts were assigned, allowing us toestablish that the synthesized monoester3 was substitutedat the position 3 and not at the position 9. The reaction ofcoumestrol with ethyl chloroacetate was strictly regioselec-tive giving only 3-O-carboxymethylcoumestrol3 and fullysubstituted product2. The hydroxyl at position 3 of coume-strol is analogous to the 7-hydroxy group of isoflavones,the coumestrol biosynthetic precursors. In daidzein andgenistein, this hydroxyl is by two orders of magnitude moreacidic than that at position 4′ (the analog of coumestrolhydroxyl 9) and is accessible to selective methylation[16]or carboxymethylation[26] in relatively mild conditions.

Although the nonradioisotopic methods should be pre-ferred for routine work, non-radioactive labeling of smallcompounds often brings additional difficulties and thus it isnot unusual that radioisotopic methods are developed first topave the way. In this particular case, the initial experimentsto develop an indirect ELISA for coumestrol using the sameantiserum and the immunogen-coated microtitration plateslead to a far less sensitive system (data not shown). Now wetry to overcome these difficulties by synthesis of alternativehapten–carrier conjugates for the plates coating.

In developing the RIA system, we have exploited our ex-perience with numerous RIAs for steroids and isoflavonoids.Thus, the reversed micellar system, designed originally forpreparation of steroid–BSA conjugates[20] was success-fully applied in preparation of immunogens for isoflavonoids[22,23], a lignan [27], a chalcone[28] and now for acoumestan immunogen. Both approaches to synthesis of theradioligand were also originally used in the steroid field.Iodination of a hapten-TME or a hapten–TYR conjugateis a convenient method for non-phenolic analytes. How-ever, when conjugates of phenolic haptens with TME areiodinated, competition may occur and the hapten may be la-beled instead of TME. The resulting product then possessesa weaker or even no immunoreactivity. Thus, we have suc-cessfully prepared radioligands for four isoflavonoids thisway, but it completely failed in the case of genistein-4′–TMEconjugate[29]. In such circumstances, the method of choiceis to conjugate the hapten to a suitable molecule iodinatedbeforehand. In this study, iodination of coumestrol-3–TMEconjugate resulted in a mixture of radioactive products. Oneof them was a useful radioligand for RIA. Conjugation ofpreviously iodinated TME with coumestrol gave a radioli-gand with the same TLC mobility, but the radiochemicalyield was low. Also its immunochemical properties were

1154 O. Lapcık et al. / Steroids 68 (2003) 1147–1155

Fig. 7. Important1H–13C HMBC correlations observed in monoester3.

worse, probably due to a less efficient separation of theradioligand from the more complex reaction mixture.

The RIA method is highly specific for coumestrol, nocross-reactivity was recorded with isoflavonoids and lig-nans tested. We admit that none of these compounds wasstructurally really close to coumestrol, as the necessarystandards were not commercially available. The coumestrolpeak represented 95% of total immunoreactivity on theHPLC chromatogram of combined alfalfa extracts (Fig. 4),indicating that the method is suitable for analysis of thismaterial. The small peaks, which altogether account forthe remaining 5% of immunoreactivity, were found re-peatedly at the same positions on the chromatogram, sowe conclude that they represent true cross-reactants ratherthan noise. At this time, we have no experimental proofof the substantiality of these entities. If we tried to spec-ulate what compounds could cross-react with this method,we would focus particularly on coumestrol derivatives dif-

fering at the bridge position (i.e. position 3) or bringinga small substituent on one of vicinal carbons (e.g. lucer-nol and sativol, i.e. the 2,3,9-trihydroxycoumestan and4,9-dihydroxy-3-methoxycoumestan, respectively), as theimmunoassays usually do not discriminate such deriva-tives very strongly. As expected, the coumestrol levelsmeasured by RIA were slightly overestimated when com-pared with HPLC–MS (Fig. 6), but the correlation betweenboth methods was good (r = 0.8989). Further improve-ment of the specificity may be achieved by pre-separationof the samples, if needed in future applications of themethod.

To our best knowledge, this is the first report on im-munoassay for coumestrol. We conclude that this methodwill be a useful tool for screening purposes and for serialanalysis in plant materials as well as in human and animaltissues and body fluids. However, validation for particularmatrices will be necessary.

O. Lapcık et al. / Steroids 68 (2003) 1147–1155 1155

Acknowledgements

This study was supported by Grant 523/00/0567 from theGrant Agency of Czech Republic; HPLC–MS analysis ofalfalfa extracts was supported by the Grant MSM 432100001from the Grant Agency of the Ministry of Education, Youthand Sports of the Czech Republic. We thank Ms. TatianaVrbska for skillful technical assistance.

References

[1] Oldfield JE, Fox CW, Bahn AV, Bickoff EM, Kohler GO. Coumestrolin alfalfa as a factor in growth and carcass quality in lambs. J AnimSci 1966;25:167–74.

[2] Dewick PM, Barz W, Griesbach H. Biosynthesis of coumestrol inPhaseolus aureus. Phytochemistry 1970;9:775–83.

[3] Wong E, Latch GCM. Coumestans in diseased white clover.Phytochemistry 1971;10:466–8.

[4] O’Neil M, Adesnaya SA, Roberts MF, Pantry I. Inducibleisoflavonoids from the Lima beanPhaseolus lunatus. Phytochemistry1986;25:1315–32.

[5] Gutendorf B, Westendorf J. Comparison of an array of in vitroassays for the assessment of the estrogenic potential of natural andsynthetic estrogens, phytoestrogens and xenoestrogens. Toxicology2001;166:79–89.

[6] Bickoff EM, Livingston AL, Hendrickson AP, Booth AN. Relativepotencies of several estrogen-like compounds found in forages. JAgric Food Chem 1962;10:410–2.

[7] Dodge JA, Glaesbbrook AL, Magee DE, Philips DL, Sato M, ShortL, et al. Environmental estrogens: effect of cholesterol loweringand bone in the ovariectomized rat. J Steroid Biochem Mol Biol1996;59:155–61.

[8] Martin PM, Horwitz KB, Ryan DS, McGuire WL. Phytoestrogeninteraction with estrogen receptors in human breast cancer cells.Endocrinology 1978;1860–1867:1978.

[9] Whitten PL, Patisaul HB, Young LJ. Neurobehavioral actions ofcoumestrol and related is of lavonoids in rodents. NeurotoxicolTeratol 2002;24:47–54.

[10] Lee YJ, Notides AC, Tsay YG, Kende AS. Coumestrol, NBD-norhexstrol, and dansyl-norhextrol, fluorescent probes of estrogen-binding proteins. Biochemistry 1977;16:2896–901.

[11] van der Oord CJ, Jones GR, Shaw DA, Munro IH, LevineYK, Gerritsen HC. High-resolution conofocal microscopy usingsynchrotron radiation. J Microsc 1996;182:217–24.

[12] Franke AA, Custer LJ, Cerna CM, Narala K. Rapid HPLC analysisof dietary phytoestrogens from legumes and from human urine. ProcSoc Exp Biol Med 1995;208:18–26.

[13] Hutabarad LS, Greenfield H, Mullholand M. Quantitativedetermination of isoflavones and coumestrol in soybean by columnliquid chromatography. J Chromatogr A 2000;886:55–63.

[14] Moravcová J, Kleinová T, Loucka R. The determination of coumestrolin alfalfa (Medicago sativa) by capillary electrophoresis. RostlinnáVýroba 2002;48:224–9.

[15] Wong MSF, Cox RI. Microanalysis of phyto-oestrogens. J ReprodFertil 1972;28:S158–159.

[16] Lapcık O, Hill M, Cerný I, Lachman J, Al-Maharik N, AdlercreutzH, et al. Immunoanalysis of isoflavonoids inPisum sativumandVigna radiata. Plant Sci 1999;148:111–9.

[17] Raffaelli B, Hoikkala A, Leppala E, Wähälä K. Enterolignans. JChromatogr B 2002;777:29–43.

[18] Ma WG, Fukushi Y, Hostettmann K, Tahara S. Isoflavonoidglycosides from Eriosema tuberosum. Phytochemistry 1998;49:251–4.

[19] Klejdus B, Vitamvásová-Šterbová D, Kubán V. Identification ofisoflavone conjugates in red clover (Trifolium pratense) by liquidchromatography-mass spectrometry after two-dimensional solid–phase extraction. Anal Chim Acta 2001;450:81–97.

[20] Yatsimirskaya EA, Gavrilova EM, Egorov AM, Levashov A.Preparation of conjugates of progesterone with bovine serumalbumin in the reversed micellar medium. Steroids 1993;58:547–50.

[21] Cook B, Beastall GH. Measurement of steroid hormoneconcentrations in blood, urine and tissues. In: Green B, Leake RE,editors. Steroid hormones, a practical approach. Oxford: IRL PressLimited; 1987. p. 1–65.

[22] Lapcık O, Hampl R, Al-Maharik N, Salakka A, Wähälä K,Adlercreutz H. A novel radioimmunoassay for daidzein. Steroids1997;62:315–20.

[23] Lapcık O, Hill M, Hampl R, Wähälä K, Al-Maharik N, AdlercreutzH. Radioimmunoassay of free genistein in human serum. J SteroidBiochem Mol Biol 1998;64:261–8.

[24] Lapcık O, Hampl R, Hill M, Stárka L, Kasal A, PouzarV, et al. Radioimmunological and chromatographic propertiesof tyrosine methyl ester conjugates with stereoisomeric steroidcarboxy derivatives. Collect Czech Chem Commun 1996;61:799–807.

[25] Bennetau-Pellisero C, Le Houérou C, Lamothe V, LeMenn F, BabinP, Bennetau B. Synthesis of haptens and conjugates for ELISAs ofphytoestrogens. Development of the immunological tests. J AgricFood Chem 2000;48:305–11.

[26] Al-Maharik N, Kaltia SA, Wähälä K. Regioselective mono-O-carboxymethylation of polyhydroxyisoflavones. Mol Online1999;3:20–4.

[27] Adlercreutz H, Wang GJ, Lapcık O, Hampl R, Wähälä K, MäkeläT, et al. Time resolved fluoroimmunoassay for plasma enterolactone.Anal Biochem 1998;265:208–15.

[28] L’homme R, Brouwers E, Al-Maharik N, Lapcık O, Hampl R, MikolaH, et al. Time-resolved fluoroimmunoassay of plasma and urineO-desmethylangolensin. J Steroid Biochem Mol Biol 2002;81:353–61.

[29] Lapcık O. Radioimmunoassay of isoflavonoid phytoestrogens.Ph.D. Thesis. Faculty of Science, Charles University, Prague;1998.