S-Equol, a potent ligand for estrogen receptor �, is the exclusiveenantiomeric form of the soy isoflavone metabolite produced byhuman intestinal bacterial flora1–4
Kenneth DR Setchell, Carlo Clerici, Edwin D Lephart, Sidney J Cole, Claire Heenan, Danilo Castellani, Brian E Wolfe,Linda Nechemias-Zimmer, Nadine M Brown, Trent D Lund, Robert J Handa, and James E Heubi
ABSTRACTBackground: The discovery of equol in human urine more than 2decades ago and the finding that it is bacterially derived from daid-zin, an isoflavone abundant in soy foods, led to the current nutritionalinterest in soy foods. Equol, unlike the soy isoflavones daidzein orgenistein, has a chiral center and therefore can occur as 2 distinctdiastereoisomers.Objective: Because it was unclear which enantiomer was present inhumans, our objectives were to characterize the exact structure ofequol, to examine whether the S- and R-equol enantiomers are bio-available, and to ascertain whether the differences in their confor-mational structure translate to significant differences in affinity forestrogen receptors.Design: With the use of chiral-phase HPLC and mass spectrometry,equol was isolated from human urine and plasma, and its enantio-meric structure was defined. Human fecal flora were cultured in vitroand incubated with daidzein to ascertain the stereospecificity of thebacterial production of equol. The pharmaco*kinetics of S- and R-equol were determined in 3 healthy adults after single-bolus oraladministration of both enantiomers, and the affinity of each equolenantiomer for estrogen receptors was measured.Results: Our studies definitively establish S-equol as the exclusiveproduct of human intestinal bacterial synthesis from soy isoflavonesand also show that both enantiomers are bioavailable. S-equol has ahigh affinity for estrogen receptor � (Ki � 0.73 nmol/L), whereasR-equol is relatively inactive.Conclusions: Humans have acquired an ability to exclusively synthe-size S-equol from the precursor soy isoflavone daidzein, and it is sig-nificant that, unlike R-equol, this enantiomer has a relatively high af-finity for estrogen receptor �. Am J Clin Nutr 2005;81:1072–9.
KEY WORDS Equol, soy isoflavones, humans, pharmaco*ki-netics, bacteria
INTRODUCTION
Equol, [7-hydroxy-3-(4'-hydroxyphenyl)-chroman], is a non-steroidal estrogen that was first discovered in the early 1980s inthe urine of adults consuming soy foods (1). It was shown to bea key metabolite of daidzin, one of the main isoflavones presentin most soy foods, and to be formed after intestinal hydrolysis ofthe soy isoflavone glycoside (2) and subsequent colonic bacterialbiotransformation (3) through an intermediate, dihydroequol (4–6). Equol has an infamous history, having been first identified in
the urine of pregnant mares as long ago as 1932 (7) and then in the1940s having been found to be the environmental estrogenicagent that caused a devastating reproductive disease in sheep,referred to as clover disease (8).
Equol is not of plant origin and is exclusively a product ofintestinal bacterial metabolism (9), as evidenced from the findingthat infants fed soy formula up to the age of 4 mo (10, 11) andgerm-free rats fed soy-containing diets (12) do not make equol.When fed soy protein, a common ingredient of most commercialrodent diets (13, 14), rats and mice are prolific equol producers.In contrast, humans are unique among animals in that, for reasonsthat remain unclear, only 20–35% of the adult population makeequol after ingesting soy foods or being challenged with pureisoflavones (3, 15, 16). Several studies have suggested that thosewho are equol producers show more favorable responses to soyisoflavone–containing diets (17–21), which leads to the possi-bility that equol is a more potent isoflavone than genistein (9),which has been so extensively studied in the last decade.
Equol, unlike its precursor daidzein or genistein, is unique inhaving a chiral carbon atom at position C-3 of the furan ring(Figure 1). It therefore can occur as 2 distinct enantiomericforms, S-equol and R-equol, which differ significantly in their
1 From the Division of Pathology (KDRS, BEW, LN-Z, and NMB), theDepartment of Gastroenterology and Nutrition (JEH), and the Department ofPediatrics (SJC and CH), Cincinnati Children’s Hospital Medical Center, andthe Department of Pediatrics, University of Cincinnati College of Medicine,Cincinnati, OH (KDRS and JEH); the Department of Gastroenterology andHepatology, University of Perugia, Perugia, Italy (CC and DC); SanitariumDevelopment and Innovation, Cooranbong, Australia (SJC and CH); theDepartment of Physiology and Developmental Biology and The Neuro-science Center, Brigham Young University, Provo, UT (EDL); and the De-partment of Biomedical Sciences, Colorado State University, Fort Collins,CO (TDL and RJH).
2 These findings were presented at the 4th International Symposium on theRole of Soy in Preventing and Treating Chronic Disease, San Diego, CA,November 4-7, 2001, and at the 5th International Symposium on the Role ofSoy in Preventing and Treating Chronic Disease, Orlando, FL, September21-24, 2003.
3 Supported by grants from the National Institutes of Health (R01CA73328)and the National Center for Research Resources (RR08084).
4 Address reprint requests to KDR Setchell, Clinical Mass Spectrometry,Cincinnati Children’s Hospital Medical Center, 3333 Burnet Avenue, Cin-cinnati, OH 45229. E-mail: [emailprotected].
Received August 14, 2004.Accepted for publication January 13, 2005.
1072 Am J Clin Nutr 2005;81:1072–9. Printed in USA. © 2005 American Society for Clinical Nutrition
Downloaded from https://academic.oup.com/ajcn/article-abstract/81/5/1072/4649521by gueston 01 May 2018
conformational structure. We recently showed that S-equol isunique in that it not only possesses estrogenic properties but alsois a potent antagonist of dihydrotestosterone in vivo, which hassignificant implications for the prevention of prostate cancer andother androgen-related conditions (22). We can find no otherexample of a molecule that is both a selective estrogen and anandrogen antagonist. Establishing the diastereoisomerism ofequol production is therefore important both in view of the pos-sible differences in biological actions of the enantiomers and inaid of the future development of strategies either to use equolpharmacologically or to manipulate equol production in humansto enhance the effectiveness of soy diets.
When equol was originally isolated, it was found to be opti-cally active (7), and subsequently its enantiomeric assignment asR-equol was questioned and redefined (23–26). The few studiesof equol to date have been exclusively performed on the racemic[(�)] mixture because, when equol is chemically synthesized, itis the (�)equol that is usually obtained. Little is known about theenantiomers of equol. For example, it is unclear which form ofequol circulates in human blood or is excreted in human urine,because all of the analytic methods for measuring equol in thesefluids fail to distinguish the diastereoisomers. To our knowledge,before the current studies, little was known about the pharmaco-logic or biological activities of the enantiomers. In this report, wefocus on several key questions related to equol in humans. First,are the R-, S-, or both enantiomeric forms of equol found inhuman urine and blood? Second, are intestinal bacteria stereo-selective in their synthesis of equol? Third, are both enantiomersabsorbed and bioavailable? And fourth, are there differences inthe biological activity of the equol’s enantiomers, specificallywith regard to their binding affinity for estrogen receptors?
SUBJECTS AND METHODS
Human studies
Identity of the enantiomeric form of equol in human urineand plasma
The characterization of equol’s diastereoisomerism was de-termined in plasma and urine samples (n � 10 each) selectedfrom study subjects in previous studies (NIH grant no.R01CA73328) of the pharmaco*kinetics of soy isoflavones inhealthy adults consuming soy foods (18). In addition, plasma andurine samples taken from a group of Seventh-Day Adventistvegetarians (n � 10) after they had consumed soymilk (2 � 240mL/d for 4 d) were analyzed. These samples were collected by
staff members of the Pathology Department of the Sydney Ad-ventist Hospital after informed consent was obtained. Charac-terization of the enantiomeric form of equol was accomplishedby chiral-phase HPLC coupled to electrospray ionization–massspectrometry (ESI-MS) after enzymic hydrolysis with a mixed�-glucuronidase and sulfatase preparation (Helix pomatia), andsolid-phase extraction of equol using a Bond Elut C18 cartridge(Varian, Harbor City, CA). The methanolic extracts of urine andplasma were taken for direct analysis by ESI-MS, or, where gaschromatography–mass spectrometry (GC-MS) was used, a vol-atile tert-butyldimethylsilyl (t-BDMS) ether derivative was pre-pared as described previously (18, 27, 28).
Written informed consent was obtained from all subjects. Thestudies were approved by the Cincinnati Children’s HospitalMedical Center Institutional Review Board.
Studies of S- and R-equol bioavailability
Purified samples of S-equol and R-equol (20 mg) obtained bysemipreparative chiral-phase HPLC chromatographic separationof a racemic mixture were prepared in capsules and randomlyadministered orally to 3 healthy adults (1 female and 2 male) ondifferent occasions separated by a washout period of 1 wk. Thisstudy was performed by the ethical standards of the Departmentof Gastroenterology and Hepatology, University of Perugia, It-aly, in accordance with the Helsinki Declaration of 1975, asrevised in 1983. Each equol enantiomer was taken with a glass ofwater after an overnight fast and before eating breakfast. Bloodsamples (10 mL) were obtained via an indwelling catheter placedin the antecubital vein at timed intervals immediately before andthen 1, 2, 3, 4, 8, 12, and 24 h after the administration of equol.These samples were centrifuged for 10 min at 2200 � g and roomtemperature, and the plasma was removed for the measurementof equol concentrations by stable-isotope dilution GC-MS withselected ion monitoring, as detailed below. The plasma equolconcentration–time profiles for the 3 persons were ascertained byusing a noncompartmental approach. The WINNONLIN com-puter program (version 3.0; Pharsight Corporation, Cary, NC)was used for this analysis. The total area under the plasma con-centration–time curve (AUC 03 �; AUCinf) was computed byusing the following equation:
AUCinf � AUC�0 3 t� � Ct/�z, (1)
where t � the last time point for blood sampling (which in thisstudy was 48 h), Ct � plasma concentration at the last blood-sampling time point, and �z � apparent elimination rate constant.
The �z was determined from the slope of the best-fitting re-gression line of the plasma samples in the terminal phase. At least4 time points were included in the estimation of �z. When re-quired, appropriate weighting schemes (usually 1/y or 1/y2,where y is the observed plasma concentration) were used toimprove the goodness of fit. The choice of the number of pointsincluded in the terminal phase of the plasma concentration-timecurves was based on the weighted residual (difference betweenmodel predicted and observed concentrations) values, dispersionof the residual values, and regression coefficient. In all cases, theregression lines were drawn without exclusion of any timepoints, and the R2 values were �0.91. The terminal half-life wascalculated as ln(2)/�z; the systemic clearance after oral admin-istration was determined as dose/AUCinf, and the apparent vol-ume of distribution after oral administration was determined as
FIGURE 1. Comparison of the chemical structures of the diastereoiso-mers of equol, showing the site position of the chiral carbon center.
EQUOL ENANTIOMERS: FORMATION, FATE, AND PROPERTIES 1073
Downloaded from https://academic.oup.com/ajcn/article-abstract/81/5/1072/4649521by gueston 01 May 2018
dose/(�z.AUCinf). Note that “F” used in the abbreviations of theseterms (CL/F and Vz/F, respectively) refers to the bioavailablefraction after oral administration, which in this case is unknown.
Specificity of intestinal bacterial equol production
Freshly passed stool samples were obtained from 3 recognizedequol producers and 3 equol nonproducers for culture of the fecalbacteria exactly as described previously (3). These samples werecollected by staff of the Pathology Department of the SydneyAdventist Hospital, Australia, after informed consent was ob-tained. The freshly passed fecal sample (1 g) was added to 9 mLsterile distilled water and trypticase broth, to which 0.02 mgdaidzein was added. The broths were incubated anerobically for3 days at 37 °C. A 2-mL aliquot of this incubation mixture wastaken, diluted with distilled water (2 mL), and centrifuged for 5min at 2200 � g and at room temperature. The supernatant waspassed through a Bond Elut C18 solid-phase cartridge to extractthe isoflavones, which were recovered by elution with methanol(5 mL). This sample was subjected to chiral-phase HPLC cou-pled with ESI-MS to separate and identify the S- and R-equoldiastereoisomers as described below. Confirmation of the iden-tity of each enantiomeric form of equol was based on its chro-matographic retention time and mass spectrum as compared withpure standards of S- and R-equol.
Animal study—identification of S-equol in rat urine
A sample of adult Sprague-Dawley rat urine was obtainedfrom animals used in a previously published study of rodents fedcommercial chow containing soy isoflavones that was performedwith the approval of the Cincinnati Children’s Hospital MedicalCenter Animal Use Committee (14).
Analytic methods
Equol concentrations were measured in the urine and plasmaby either ESI liquid chromatography–MS (ESI-LC-MS) orGC-MS techniques, or both, after liquid-solid extraction, hydro-lysis of the conjugates with a mixed �-glucuronidase and sulfa-tase enzyme preparation (Helix pomatia; Sigma Chemicals, StLouis, MO), reextraction, and preparation of a volatile derivativein the case of GC-MS analysis. The methods have been outlinedin previous studies of isoflavones (2, 18, 28, 29). Separate quan-tification of the S- and R-equol was achieved by using a recentlydeveloped chiral-phase HPLC chromatographic method (de-scribed below) that resolves the S- and R- enantiomers of equol.
Determination of S- and R-equol concentrations in plasmaby gas chromatography–mass spectrometry
The concentrations S- and R-equol in plasma (0.5 mL) sam-ples collected after administration of these enantiomers weremeasured by GC-MS after the addition of [13C] stable isotope–labeled analogs of S- and R-equol as internal standards for quan-tification. These internal standards were obtained by chiral-phase HPLC chromatographic separation of (�)([13C]equol thatwas synthesized from [2-13C]daidzein as described elsewhere (30). Isoflavones were extracted on a solid-phaseoctadecylsilane-bonded silica cartridge and hydrolyzed enzy-matically with a mixed �-glucuronidase and sulfatase prepara-tion (H. pomatia). After reextraction and purification, thet-BDMS ether derivatives were prepared, and the derivatized
samples were analyzed by selected ion monitoring GC-MS asdescribed previously (18, 27, 28).
Chiral-phase HPLC and electrospray ionization–massspectrometry separation and identification ofdiastereoisomers of equol
Sample extracts for the characterization of the enantiomericform of equol in urine and plasma were taken up in 100 �L of theHPLC mobile phase, and a 20-�L sample was injected on col-umn. Separation of the enantiomers of equol was performed ona Chiralcel column (Chiral Technologie, Inc, Exton, PA). Themobile phase used to enable baseline resolution of S- andR-equol was a gradient elution beginning with 90% hexane and10% ethanol (by vol) and linearly increasing to a final compo-sition of 10% hexane and 90% ethanol (by vol) over a 15-minperiod at a flow rate of 1 mL/min. Detection of S- and R- equolis possible by ultraviolet absorption at 260 nm, even though equolhas poor ultraviolet absorption characteristics, provided rela-tively high concentrations are injected on column. This wave-length was monitored to establish the conditions using the purestandards of equol. However, for detecting equol enantiomers inhuman plasma and urine, the greater sensitivity of ESI-MS wasnecessary. ESI-MS was performed on a Micromass QuattroLC/MS (Waters Corp, Milford, MA). The HPLC effluent to theESI probe was split 10:1. The desolvation temperature was300 °C, and the source temperature was 100 °C. The samplingcone was held at 50 V and the extractor at 2 V. Data werecollected in the negative ion mode, and the [M-H]� ions moni-tored were a mass-to-charge ratio (m/z) of 241 (equol) and a m/zof 242 ([2-13C]equol). The identity of the enantiomeric form ofequol in the human plasma and urine samples was based on theretention time of the eluting peak in the mass chromatogramcompared with the mass chromatograms obtained for pure stan-dards of S- and R-equol analyzed under identical conditions.
Studies of estrogen receptor binding
Synthesis of hormone receptor proteins
Full-length human estrogen receptor (ER) � [(ER�) pcDNA-ER�; RH Price, University of California, San Francisco] and ratER� (pcDNA-ER�; TA Brown, Pfizer, Groton, CT) expressionvectors were used to synthesize hormone receptors in vitro byusing the transcription and translation–coupled reticulocyte ly-sate system (Promega, Madison, WI) with T7-RNA polymerase,during a 90-min reaction at 30 °C. Translation reaction mixtureswere stored at �80 °C until further use.
Saturation isotherms
To calculate and establish the binding affinity of the R and Sequol enantiomers for ER� and ER�, 100-�L aliquots of reticu-locyte lysate supernatant were incubated at optimal time andtemperature—90 min at room temperature (ER�) and 18 h at4 °C (ER�)—with increasing (0.01–100 nmol/L) concentrationsof [3H]17�-estradiol (E2). These times were determined empir-ically, and they represent optimal binding of receptor with es-trogen. Nonspecific binding was assessed by using a 300-foldexcess of the ER agonist diethylstilbestrol in parallel tubes. Afterincubation, bound and unbound [3H]E2 was separated by passingthe incubation reaction through a 1-mL lipophilic SephadexLH-20 column (Sigma-Aldrich Co, St Louis, MO). Columns
1074 SETCHELL ET AL
Downloaded from https://academic.oup.com/ajcn/article-abstract/81/5/1072/4649521by gueston 01 May 2018
were constructed by packing a disposable pipette tip (1 mL;Labcraft, Curtin Matheson Scientific, Inc, Houston, TX) withTEGMD (10 mmol Tris-Cl/L, 1.5 mmol EDTA/L, 10% glycerol,25 mmol molybdate/L, and 1 mmol dithiothreitol/L; pH 7.4)-saturated Sephadex according to previously published protocols(31, 32). For chromatography, the columns were reequilibratedwith TEGMD (100 �L), and the incubation reactions were addedindividually to each column and allowed to incubate on thecolumn for an additional 30 min. After this incubation, 600 �LTEGMD was added to each column, flow-through was collected,4 mL of scintillation fluid was added, and samples were counted(5 min each) in a 2900 TR Packard scintillation counter (PackardBioscience, Meriden, CT).
RESULTS
Characterization of S-equol in humans and rats bychiral-phase HPLC and mass spectrometry
A typical separation of S- and R-equol enantiomers when aracemic mixture was subjected to chiral-phase chromatographyand the isoflavones were detected by their ultraviolet absorbanceat 260 nm is shown in Figure 2. S-equol eluted from the chiral-phase column with a shorter retention time than that of R-equol,and baseline resolution was achieved. The identity of both en-antiomers was confirmed by isolation of both peaks and mea-surement of optical dichroism. Also shown in Figure 2 is a typicalHPLC-ESI-MS mass chromatogram of the negative ion m/z 241corresponding to the pseudomolecular ion ([M-H]�) of equolobtained from the analysis of hydrolyzed extracts of humanurine. All samples of human urine analyzed were found to con-tain a single equol enantiomer with a retention time (6.75 min)that exactly corresponded to the retention time of the pure stan-dard of S-equol (ie, 6.77 min). There was no evidence for thepresence of R-equol (which elutes from the HPLC column witha significantly longer retention time of 7.51 min) in any sampleof human urine when analyzed under the same chromatographicconditions. Likewise, a sample of urine from a rat, a species thatis exclusively an equol producer (12, 14), was found, on the basisof its retention index, to have exclusively the S-equol enantiomer(Figure 2). Similar findings were obtained from the analysis of a
selection of samples (n � 10) of human plasma collected fromequol producers who had consumed 2 � 240 mL soymilk. Con-sistent with the urinary analysis, the ESI-MS profiles of plasmashowed one major peak in all samples analyzed, and this had aretention time corresponding to that of the S-equol standard.
Pharmaco*kinetics of S- and R-equol enantiomers
Mean (�SEM) appearance and disappearance concentrationcurves for equol in plasma after single-bolus oral administrationof R-equol and S-equol to 3 healthy adults are shown in Figure3. The administration of both enantiomers yielded similar plasmapharmaco*kinetic profiles, which confirmed that the 2 diastereoi-somers are similarly bioavailable. Equol rapidly appeared inplasma and disappeared with a terminal elimination half-life of4.9 � 1.6 and 6.2 � 0.2 h, respectively, for the S- andR-enantiomers, and there was no obvious difference betweenthese values. A comparison of the computed pharmaco*kinetics ofthe diastereoisomers is shown in Table 1. There were no statis-tical differences in maximum plasma concentration, time toreach maximum plasma concentration, terminal eliminationhalf-life, AUCinf, apparent volume of distribution, and systemicclearance between R- and S-equol, and no difference in the ab-sorption rates of the 2 enantiomers.
Because GC-MS was used to quantify equol in these plasmasamples, and thus this technique cannot resolve the individualenantiomers as their t-BDMS ether derivatives, it was essential to
FIGURE 2. Chiral-phase HPLC separation with ultraviolet detection(260 nm) showing resolution of a standard mixture of S- and R-equol (left).These profiles are compared with the superimposed ESI-MS mass chromato-grams of mass-to-charge ratio (m/z) 241 ([M�H]� ion) obtained from the anal-ysis of hydrolyzed extracts of human and rat urine collected after ingestion ofsoy foods (right), which established S-equol as the only enantiomer excreted inhuman urine. ESI-MS, electrospray ionization–mass spectrometry.
FIGURE 3. Mean (�SEM) plasma concentrations of S- and R-equol in3 healthy adults given a single-bolus oral 20-mg dose of each enantiomer onseparate occasions. Data are expressed as linear-linear (left) and log-linear(right) plots. There were no significant differences.
TABLE 1Computed plasma pharmaco*kinetics for S- and R-equol administered to 3healthy adults in a single-bolus oral dose of 20 mg of each diastereoisomerin a randomized crossover design with a 1-wk washout interval1
Pharmaco*kinetic values S-Equol R-Equol
Tmax (h) 2.3 � 0.3 2.7 � 0.7Gmax (ng/mL) 390 � 90 414 � 32t1/2 (h) 4.9 � 1.6 6.2 � 0.2Vz/F (L) 39 � 8.8 50 � 7.7Cl/F (L) 6.4 � 1.6 5.5 � 0.7AUCinf (ng/mL · h) 3595 � 1014 3735 � 448
1 All values are x� � SD. Tmax, time to reach maximum plasma concen-tration; Cmax, maximum plasma concentration; t1/2, terminal eliminationhalf-life; Vz/F, apparent volume of distribution; Cl/F, systemic clearance;AUCinf, total area under the plasma concentration-time curve. There were nosignificant differences between S- and R-equol.
EQUOL ENANTIOMERS: FORMATION, FATE, AND PROPERTIES 1075
Downloaded from https://academic.oup.com/ajcn/article-abstract/81/5/1072/4649521by gueston 01 May 2018
definitively confirm the identity of the R-equol enantiomer inplasma to exclude the possibility of racemization to S-equoloccurring during or after its absorption. Confirmation that theadministered R-equol remained unaltered during absorption oruptake was established by taking the 2-h plasma sample extractand subjecting it to direct chiral-phase HPLC analysis withESI-MS used as the detection system. The mass chromatogramsfor m/z 241 corresponding to the [M-H]� ion of equol for theplasma extract collected 2 h after administration of the R-equolenantiomer are shown in Figure 4. For comparison, the masschromatograms of a pure mixture of S- and R- equol are alsoshown, and, on the basis of the retention time, only R-equol wasfound in the plasma. After administration of S-equol, ESI-MSconfirmed that S-equol appeared unchanged in the 2-h plasmasample (data not shown). These data provided evidence that bothenantiomeric forms of equol were absorbed without change andthat no racemization occurred during or after intestinal uptake.Finding only the S-equol enantiomer in human urine and plasmasuggested that this must be the result of the exclusive bacterialproduction of S-equol in the intestine.
Evidence for enantiomer-specific synthesis of S-equol byhuman intestinal bacteria
In vitro studies were performed on cultured human fecal floracollected from healthy adults who were challenged for 4 d withsoy foods and who were determined from plasma and urinaryisoflavone analysis to be either equol producers or equol non-producers as defined previously (9). Daidzein (20 �g), the pre-cursor isoflavone of equol, was then incubated with culturedfecal flora at 37 °C for 72 h; after extraction of the supernatantfluid by solid-phase chromatography, the extract was analyzedby direct ESI-MS with chiral-phase HPLC separation. Superim-posed mass chromatograms of the negative ion recordings for m/z241 ([M-H]� ion) that were specific for a pure standard ofS-equol and for the equol isolated from the 72-h supernatant fluidfrom one of the equol producers and from one equol nonproducerare shown in Figure 5. The cultured fecal flora from the equolproducers showed a peak corresponding to equol that had a re-tention time identical to that of the authentic pure standard of
S-equol. By contrast, the supernatant fluid from an equol non-producer showed negligible production of S-equol. These resultsestablish conclusively that fecal bacteria are selective in produc-ing only S-equol as the principal metabolite of the soy isoflavone,daidzein.
Estrogenic activity of equol enantiomers
Competitive binding studies were used to assess the estrogenicproperties of R- and S-equol. On the basis of the ability of R- andS-equol to compete with [3H]E2 in ER binding, their affinities forERs translated in vitro were shown to be very different. S-equolshowed the greatest affinity for ER� (Ki � 0.73 � 0.2 nmol/L),whereas its affinity for ER� (Ki � 6.41 � 1 nmol/L) was rela-tively poor. In contrast, R-equol possessed only 4.8% and 25.0%as much relative binding affinity, respectively, for ER� (Ki �15.4 � 1.3 nmol/L) and for ER� (Ki � 27.38 � 3.8 nmol/L) asdid S-equol. For comparison, 17�-estradiol binds ER� with a Kd
of 0.13 nmol and ER� with a Kd of 0.15 nmol. S-equol thus showsER selectivity with a high affinity for ER�, whereas R-equol can,at best, be classified as a weak estrogen.
DISCUSSION
In 1932, Marrian and Haslewood, working at the UniversityCollege London, first isolated and elucidated the chemical struc-ture of the isoflavone metabolite equol (7). It was found as aminor metabolite of the urine of pregnant mares and shown to beoptically active, although in subsequent years there was someconfusion as to its enantiomeric assignment; it was first assignedthe R-configuration, and only later was the absolute configura-tion defined as the S-enantiomer (25). The importance of defin-ing the nature of the stereoisomerism in humans relates to themarked differences in the conformational structure of the diaste-reoisomers and the expectation that there would be differences inthe biological activity, primarily related to binding to the ER.When equol was first isolated, it was reported by Marrian andHaslewood to have no estrus-producing activity when injectedinto ovariectomized mice in doses as large as 0.86 mg/animal (7).This observation was inconsistent with the later finding that
FIGURE 5. Definitive evidence for the enantiomer-specific synthesis ofS-equol by cultured human fecal flora. Mass chromatograms obtained byusing chiral-phase HPLC-mass spectrometric analysis of a pure standard ofS-equol (bottom trace) are compared with extracts from in vitro bacterialmetabolism of daidzein by cultured fecal flora from a known equol producer(top trace) and an equol nonproducer (middle trace). ESI-MS, electrosprayionization–mass spectrometry.
FIGURE 4. Chiral-phase HPLC-electrospray ionization–mass spectrom-etry (ESI-MS) analysis of the plasma collected 2 h after administration ofR-equol to a healthy adult (left), which confirmed its presence as unchangedR-equol. For comparison, the ESI-MS mass chromatograms for the ion mass-to-charge ratio (m/z) 241 obtained for S- and R-equol standards are super-imposed (right).
1076 SETCHELL ET AL
Downloaded from https://academic.oup.com/ajcn/article-abstract/81/5/1072/4649521by gueston 01 May 2018
equol was the estrogenic agent responsible for the devastatingreproductive abnormalities, referred to as clover disease, ob-served in sheep in the mid-1940’s (8, 33, 34). Later, in a periodthat predated knowledge of specific ER subtypes (35, 36), equolwas shown in vitro to bind to uterine cytosolic ERs (37, 38).Given the predominance of ER� in the uterus, it is almost certainthat these early reports of binding affinities refer to equol’s bind-ing to ER�, rather than to ER�. Using preparations of recombi-nant steroid receptors, we have shown that only the S enantiomerof equol binds to ERs with sufficient affinity to be of physiologicrelevance based on circulating equol concentrations typicallyfound in humans (9). Furthermore, almost 50% of equol circu-lates unbound to serum protein (39), whereas endogenous estro-gen is �95% protein bound. This protein-binding status of equolis likely to enhance its biological potency because it is only the“free” or unbound fraction that is available for receptor occu-pancy. The relative binding affinity of the R- and S-equol enan-tiomers for ER� were 0.47% and 2.0% with that of 17�-estradiol.However, S-equol is largely ER� selective and has a relativelyhigh affinity for this receptor subtype. S-equol binds ER� �20%with as much affinity as does 17�-estradiol (equol: Ki � 0.7nmol/L; 17�-estradiol: Kd � 0.15 nmol/L), whereas the R en-antiomer bound at �1% of the affinity. These findings are cor-roborated by several studies (40–42) that also show equol tohave selective affinity for ER�, and, therefore, equol can bedefined as a type of selective ER modulator.
As a potent antagonist of dihydrotestosterone (DHT) in vivo,equol is also unique, in that we can find no other example of acompound that has selective estrogen action and yet also has theability to be an antagonist of androgen action (22). It is interestingthat the mechanism of its anti-androgen action differs from thatof the anti-androgen drugs used in clinical practice to block theeffects of DHT. For example, equol has no affinity for the an-drogen receptor (22) and therefore does not function as an an-drogen receptor blocker. It also does not appear to alter thesynthesis of DHT in the way that 5�-reductase inhibitors do, but,rather, it appears to bind directly to DHT (22), and this effect isseen with both R- and S-equol (TD Lund, RJ Handa, ED Lephart,KDR Setchell, unpublished data, 2003).
Given the distinctly different biological actions of the diaste-reoisomers of equol, particularly with regard to their affinitytoward ERs, it is relevant to define the stereoisomerism of equolproduction in humans. Using ESI-MS with selected ion moni-toring, we analyzed urine and plasma samples from healthyadults, and by our ability to separate the 2 diastereoisomers withthe use of chiral-phase column chromatography, we have shownfor the first time that S-equol is the only enantiomer circulatingin human blood and excreted in urine (Figures 2 and 3). This isalso true in the rat, a species that is predominantly an equolproducer (14). The logical explanation for the finding of a singleenantiomer in plasma and urine is that intestinal bacteria arestereoselective in their synthesis, but the possibility that bothenantiomers would be made in the intestine, but only one, theS enantiomer, would be absorbed required addressing. Fur-thermore, racemization of R-equol to S-equol during theformer’s absorption was an alternative possibility that wasfeasible and required investigation.
The separate oral administration of pure S- and R-equol to 3healthy adults clearly showed that both enantiomers, when
present in the intestine, are efficiently absorbed and appear rap-idly in plasma. There were no differences in the pharmaco*kinet-ics of the 2 enantiomers. The bioavailability of equol as measuredby the dose-adjusted AUCinf is relatively high when comparedwith the bioavailability of genistein and daidzein reported inprevious studies (18, 27, 28). The clearance rate of equol was alsomuch slower than that of the soy isoflavones, which contributesto the maintenance of high circulating equol concentrations ob-served in rodents (14). ESI-MS established that, after its admin-istration, R-equol appeared in plasma unchanged, and thereforethe possibility of bacterial production of R-equol in the intestinewith racemization to S-equol during absorption can be confi-dently excluded. Thus, these data taken together are indicative ofthe enantiomer-specific production of S-equol by intestinal bac-teria. This is now confirmed by in vitro experiments in whichhuman fecal flora from equol producers were cultured underanerobic conditions and incubated with daidzein or soy isofla-vones. After 72 h in culture, S-equol was the sole enantiomeridentified in the supernatant. Thus, given that humans, rats, andsheep all produce S-equol—and it is likely that macaque mon-keys (43), chimpanzees (44), dogs (45), domestic fowl (46), cows(47), and mice (14) also do so—it is evident that the bacteriaresponsible for equol production are all highly selective in per-forming an asymmetric synthesis with production of the oneenantiomer that shows the highest ligand affinity for ER�.
The formation of equol from its precursor daidzein proceedsthrough an intermediate, dihydrodaidzein. Our pharmaco*kineticstudies show that equol is rapidly absorbed from the intestine, butit* formation after the initial intake of daidzein or of soy foodscontaining daidzin or daidzein is a time-dependent process. Itgenerally takes �12 h for equol to appear in the plasma, and, insome adults, it may not appear for 36 h, which indicates that thecolon is the site of its formation (18, 28, 48). Identification of thebacteria responsible for equol production has thus far been elu-sive. It is apparent that there is more than one bacterium involvedbecause we have observed cases in which dihydrodaidzein ispresent in urine in the absence of equol, which is consistent withpartial conversion of daidzein to equol (KDR Setchell, unpub-lished observations, 1995). Furthermore, in vitro incubation offecal hom*ogenates from some adults was shown to produce di-hydrodaidzein and O-desmethylangolensin but not equol,whereas recently it was shown that some antibiotics, such asrifampicin and kanamycin, may inhibit the production of equolbut not of dihydrodaidzein (6). In contrast, kanamycin virtuallyeliminated equol production in the plasma of cynomologus mon-keys (49), which highlights the complexity of the bacterial pro-duction of equol. Attempts to identify the species of bacteriainvolved in equol production have yielded some informationregarding strains that are capable of hydrolyzing the �-glucosideof daidzin (50, 51) or of converting daidzein to dihydrodaidzein(52), and one report claimed that Streptococcus intermedius spp,Ruminococcus productus spp, and Bacteroides ovatus in vitroperform the required conversion (53).
In view of the apparent advantages of being an equol producer(9, 17, 19–21, 54), the question of whether it is possible tomanipulate the intestinal milieu in favor of equol productionwhen soy foods are given is relevant. Early studies by Setchelland Cassidy (55) using an in vitro model of human colonicfermentation showed that, with a background of a high nonstarchpolysaccharide environment, which affords increased colonicfermentation, the conversion of daidzein to equol is complete, but
EQUOL ENANTIOMERS: FORMATION, FATE, AND PROPERTIES 1077
Downloaded from https://academic.oup.com/ajcn/article-abstract/81/5/1072/4649521by gueston 01 May 2018
no conversion occurs under low-carbohydrate conditions. This invitro observation is supported by data from a study of 24 healthyadults showing that good equol producers were associated witha diet that was lower in saturated fat and higher in total carbo-hydrate (16). The role of carbohydrate in equol production, alsoreported by Lampe et al (15), suggests that there are importantprebiotic or probiotic factors that may influence equol formationin adults. The addition of fructooligosaccharides to the diet ofmice has been shown to yield higher equol concentrations (56),as did feeding potato starch (57), and, in the former study, higherequol concentrations were associated with a greater effect onbone density in this animal model. This has also been shown tobe the case in a 2-y clinical study of the effectiveness of soy foodsin preventing bone loss in postmenopausal women (20), in whichequol producers showed a significant increase in lumbar spinebone mineral density. Whether probiotic or prebiotic diets caninfluence the metabolism of soy isoflavones to enhance equolproduction in humans is unclear (58–60), but, if not, the potentialbenefits of equol, with its selective ER modulator properties andits antiandrogen actions, could be optimized by the use of pureenantiomeric forms of equol as a supplement or nutraceutical.
In conclusion, our studies described here definitively establishS-equol as the only enantiomer found in human plasma and urine,and we show for the first time that this exclusivity is not due todifferences in the bioavailability or metabolism of S- andR-equol but rather to the fact that intestinal microflora showenantiomeric specificity in their production of equol. These find-ings are of immense clinical relevance because we also found thatS-equol, but not R-equol, has a relatively high affinity for ER�and is in fact a more potent estrogen than is estradiol; thesefindings have been corroborated by others (40–42). The signif-icance of these findings is that humans have acquired intestinalmicroflora that perform an asymmetric synthesis of the onlydiastereoisomer of equol that has affinity for the ER and thus hasthe greatest potential for physiologic effects.
KDRS was the principal investigator; conceived, designed, and directedthese studies; and prepared the manuscript. CC and DC were responsible forconducting the pharmaco*kinetic studies of R- and S-equol in healthy sub-jects. SC and CH performed the studies on human fecal flora and collectedblood and urine samples from healthy vegetarian Seventh Day Adventistvolunteers in Sydney. EDL, TDL, and RJH conducted the estrogen-bindingstudies of equol. NMB was the clinical coordinator for studies conducted onhealthy subjects in Cincinnati, and JEH was the physician responsible for themonitoring of these studies conducted on the General Clinical ResearchCenter. BEW and LN-Z, the research assistants, performed the laboratoryanalysis of equol by mass spectrometry. All authors provided input to themanuscript, and none of the authors had any conflict of interest.
REFERENCES1. Axelson M, Kirk DN, Farrant RD, Cooley G, Lawson AM, Setchell
KDR. The identification of the weak oestrogen equol [7-hydroxy-3-(4'-hydroxyphenyl)chroman] in human urine. Biochem J 1982;201:353–7.
2. Setchell KDR, Brown NB, Zimmer-Nechemias L, et al. Evidence forlack of absorption of soy isoflavone glycosides in humans, supportingthe crucial role of intestinal metabolism for bioavailability. Am J ClinNutr 2002;76:447–53.
3. Setchell KDR, Borriello SP, Hulme P, Kirk DN, Axelson M. Nonste-roidal estrogens of dietary origin: possible roles in hormone-dependentdisease. Am J Clin Nutr 1984;40:569–78.
4. Kelly GE, Nelson C, Waring MA, Joannou GE, Reeder AY. Metabolitesof dietary (soya) isoflavones in human urine. Clin Chim Acta 1993;223:9–22.
5. Heinonen S, Wahala K, Adlercreutz H. Identification of isoflavone me-tabolites dihydrodaidzein, dihydrogenistein, 6'-OH-O-dma, and cis-4-OH-equol in human urine by gas chromatography-mass spectroscopyusing authentic reference compounds. Anal Biochem 1999;274:211–9.
6. Atkinson C, Berman S, Humbert O, Lampe JW. In vitro incubation ofhuman feces with daidzein and antibiotics suggests interindividual dif-ferences in the bacteria responsible for equol production. J Nutr 2004;134:596–9.
7. Marrian G, Haslewood G. Equol, a new inactive phenol isolated fromthe ketohydroxyoestrin fraction of mare’s urine. Biochem J 1932;26:1227–32.
8. Bennetts HW, Underwood EJ, Shier FL. A specific breeding problem ofsheep on subterranean clover pastures in Western Australia. Aust J AgricRes 1946;22:131–8.
9. Setchell KDR, Brown NM, Lydeking-Olsen E. The clinical importanceof the metabolite Equol—a clue to the effectiveness of soy and itsisoflavones. J Nutr 2002;132:3577–84.
10. Setchell KDR, Zimmer-Nechemias L, Cai J, Heubi JE. Isoflavone con-tent of infant formulas and the metabolic fate of these phytoestrogens inearly life. Am J Clin Nutr 1998;68(suppl):1453S–61S.
11. Setchell KDR, Zimmer-Nechemias L, Cai J, Heubi JE. Exposure ofinfants to phyto-oestrogens from soy-based infant formula. Lancet 1997;350:23–7.
12. Axelson M, Setchell KDR. The excretion of lignans in rats—evidencefor an intestinal bacterial source for this new group of compounds. FEBSLett 1981;123:337–42.
13. Thigpen JE, Setchell KDR, Ahlmark KB, et al. Phytoestrogen content ofpurified, open- and closed-formula laboratory animal diets. Lab AnimSci 1999;49:530–6.
14. Brown NM, Setchell KDR. Animal models impacted by phytoestrogensin commercial chow: implications for pathways influenced by hor-mones. Lab Invest 2001;81:735–47.
15. Lampe JW, Karr SC, Hutchins AM, Slavin JL. Urinary equol excretionwith a soy challenge: influence of habitual diet. Proc Soc Exp Biol Med1998;217:335–9.
16. Rowland IR, Wiseman H, Sanders TA, Adlercreutz H, Bowey EA.Interindividual variation in metabolism of soy isoflavones and lignans:influence of habitual diet on equol production by the gut flora. J Nutr2000;36:27–32.
17. Duncan AM, Merz-Demlow B, Xu X, Phipps WR, Kurzer MS. Pre-menopausal equol excretors show plasma hormone profiles associatedwith lowered risk of breast cancer. Cancer Epidemiol Biomarkers Prev2000;9:581–6.
18. Setchell KDR, Brown NM, Desai PB, et al. Bioavailability, disposition, anddose-responseeffectsof soy isoflavoneswhenconsumedbyhealthywomenat physiologically typical dietary intakes. J Nutr 2003;133:1027–35.
19. Uchiyama S, Ueno T, Shirota T. The relationship between soy isofla-vones and the menopausal symptoms in Japanese perimenopausalwomen. Ann Nutr Metab 2001;45:113 (abstr).
20. Lydeking-Olsen E, Beck-Jensen JE, Setchell KDR, Holm-Jensen T.Soymilk or progesterone for prevention of bone loss–a 2 year random-ized, placebo-controlled trial. Eur J Nutr 2004;43:246–57.
21. Akaza H, Miyanaga N, Takahima N, et al. Comparisons of percent equolproducers between prostate cancer patients and controls: case-controlledstudies of isoflavones in Japanese, Korean and American residents. JpnJ Clin Oncol 2004;34:86–9.
22. Lund TD, Munson DJ, Haldy ME, Setchell KDR, Lephart ED, Handa RJ.The phytoestrogen equol acts as an anti-androgen to inhibit prostategrowth and hormone feedback. Biol Reprod 2004;70:1188–95.
23. Suginome H. Absolute and relative stereochemical correlation ofisoflvanone sophorol and naturally occurring isoflavan derivatives. BullChem Soc Jpn 1966;39:1544–7.
24. Clark-Lewis JW. Configurations of optically active flavonoid com-pounds. Rev Pure Appl Chem 1962;12:96–116.
25. Clark-Lewis JW, Dainis I, Ramsay GC. Flavan derivatives XIV. Theabsolute configurations of some 1,2-diarylpropane derivatives and someisoflavans. Austral J Chem 1965;18:1035–48.
26. Kurosawa K, Ollis WD, Redman BT, Sutherland OI, Gottlieb OR, AlvesHM. The absolute configuration of the animal metabolite equol, threenaturally occurring isoflavans and one natural isoflavanquinone. ChemCommun 1968;1265–7.
27. Setchell KDR, Brown NM, Desai P, et al. Bioavailability of pure isofla-vones in healthy humans and analysis of commercial soy isoflavonesupplements. J Nutr 2001;131:1362S–75S.
1078 SETCHELL ET AL
Downloaded from https://academic.oup.com/ajcn/article-abstract/81/5/1072/4649521by gueston 01 May 2018
28. Setchell KDR, Faughnan MS, Avades T, et al. Comparing the pharma-co*kinetics of daidzein and genistein with the use of 13C-labeled tracersin premenopausal women. Am J Clin Nutr 2003;77:411–9.
29. Setchell KDR, Cole SJ. Variations in isoflavone levels in soy foods andsoy protein isolates and issues related to isoflavone databases and foodlabeling. J Agric Food Chem 2003;51:4146–55.
30. Baraldi PG, Spalluto G, Cacciari B, Romangnoli R, Setchell KDR.Chemical synthesis of [13C]daidzein. J Med Food 1999;2:99–102.
31. Handa RJ, Reid DL, Resko JA. Androgen receptor in brain and pituitaryof female rats: cyclic changes and comparisons with the male. BiolReprod 1986;34:293–303.
32. O’Keefe JA, Handa RJ. Transient elevation of estrogen receptors in theneonatal rat hippocampus. Dev Brain Res 1990;57:119–27.
33. Kaziro R, Kennedy JP, Cole ER, Southwell-Keely PT. The oestrogenic-ity of equol in sheep. J Endocrinol 1984;103:395–9.
34. Shutt D, Braden A. The signficance of equol in relation to the oestrogenicresponses in sheep ingesting clover with a high formononetin content.Aust J Agric Res 1968;19:545–53.
35. Kuiper GG, Enmark E, Pelto-Huikko M, Nilsson S, Gustafsson JA.Cloning of a novel receptor expressed in rat prostate and ovary. Proc NatlAcad Sci U S A 1996;93:5925–30.
36. Kuiper GG, Carlsson B, Grandien K, et al. Comparison of the ligandbinding specificity and transcript tissue distribution of estrogen recep-tors alpha and beta. Endocrinology 1997;138:863–70.
37. Shutt DA, Cox RI. Steroid and phyto-oestrogen binding to sheep uterinereceptors in vitro. J Endocrinol 1972;52:299–310.
38. Tang BY, Adams NR. Effect of equol on oestrogen receptors and onsynthesis of DNA and protein in the immature rat uterus. J Endocrinol1980;85:291–7.
39. Nagel SC, vom Saal FS, Welshons WV. The effective free fraction ofestradiol and xenoestrogens in human serum measured by whole celluptake assays: physiology of delivery modifies estrogenic activity. ProcSoc Exp Biol Med 1998;217:300–9.
40. Morito K, Hirose T, Kinjo J, et al. Interaction of phytoestrogens withestrogen receptors � and �. Biol Pharm Bull 2001;24:351–6.
41. Kostelac D, Rechkemmer G, Briviba K. Phytoestrogens modulate bind-ing response of estrogen receptors � and � to the estrogen responseelement. J Agric Food Chem 2003;51:7632–5.
42. Muthyala RS, Ju YH, Sheng S, et al. Equol, a natural estrogenic metab-olite from soy isoflavones: convenient preparation and resolution ofR- and S-equols and their differing binding and biological activitythrough estrogen receptors alpha and beta. Bioorgan Med Chem2004;12:1559 – 67.
43. Monfort SL, Thompson MA, Czekala NM, Kasman LH, ShackletonCHL, Lasley BL. Identification of a non-steroidal estrogen, equol, in theurine of pregnant macaques: correlation with steroidal estrogen excre-tion. J Steroid Biochem 1984;20:869–76.
44. Adlercreutz H, Musey PI, Fotsis T, et al. Identification of lignans andphytoestrogens in urine of chimpanzees. Clin Chim Acta 1986;158:147–54.
45. Juniewicz PE, Pallante Morell S, Moser A, Ewing LL. Identification ofphytoestrogens in the urine of male dogs. J Steroid Biochem 1988;31:987–94.
46. Common R, Aimsworth L. Identification of equol in the urine of thedomestic fowl. Biochim Biophys Acta 1961;53:403–4.
47. Klyne W, Wright S. Steroids and other lipids of pregnant cows’ urine. JEndocrinol 1959;18:32–45.
48. Watanabe S, Yamaguchi M, Sobue T, et al. Pharmaco*kinetics of soybeanisoflavones in plasma, urine and feces of men after ingestion of 60 gbaked soybean powder (kinako). J Nutr 1998;128:1710–5.
49. Blair RM, Appt SE, Clarkson TB. Treatment with antibiotics reducesplasma equol concentration in cynomolgus monkeys (Macaca fascicu-laris). J Nutr 2003;133:2262–7.
50. Tsangalis D, Ashton JE, McGill AE, Shah NP. Enzymic transformationof isoflavone phytoestrogens in soymilk by �-glucosidase-producingBifidobacteria. J Food Sci 2002;67:3104–33.
51. Tsangalis D, Ashton JE, McGill AE, Shah NP. Biotransformation ofisoflavones by Bifidobacteria in fermented soymilk supplemented withD-glucose and L-cysteine. J Food Sci 2003;68:623–31.
52. Hur HG, Lay JO, Beger RD, Freeman JP, Rafii F. Isolation of humanintestinal bacteria metabolizing the natural isoflavone glycosides daid-zin and genistin. Arch Microbiol 2000;174:422–8.
53. Ueno T, Uchiyama S. Identification of the specific intestinal bacteriacapable of metabolising soy isoflavone to equol. Ann Nutr Metab 2001;45:114 (abstr).
54. Atkinson C, Skor H, Fitzgivvons E, Scholes D, Chen C, Wahala K.Urinary equol excretion in relation to 2-hydroxyestrone and 16a-hydroxyestrone concentrations: an observational study of young tomiddle-aged women. J Steroid Biochem Mol Biol 2003;86:71–7.
55. Setchell KDR, Cassidy A. Dietary isoflavones: biological effects andrelevance to human health. J Nutr 1999;129:758S–67S.
56. Ohta A, Uehara M, Sakai K, et al. A combination of dietary fructooli-gosaccharides and isoflavone conjugates increases femoral bone mineraldensity and equol production in ovariectomized mice. J Nutr 2002;132:2048–54.
57. Tamura M, Hirayama K, Itoh K, Suzuki H, Shinohara K. Effects of ricestarch-isoflavone deit or potato starch-isoflavone diet on plasma isofla-vone, plasma lipids, cecal enzyme activity, and composition of fecalmicroflora in adult mice. J Nutr Sci Vitam 2002;48:225–9.
58. Steer TE, Johnson IT, Gee JM, Gibson GR. Metabolism of the soybeanisoflavone glycoside genistin in vitro by human gut bacteria and theeffect of prebiotics. Br J Nutr 2003;90:635–42.
59. Nettleton JA, Greany KA, Thomas W, Wangen KE, Adlercreutz H,Kurzer MS. Plasma phytoestrogens are not altered by probiotic con-sumption in postmenopausal women with and without a history of breastcancer. J Nutr 2004;134:1998–2003.
60. Bonorden MJ, Greany KA, Wangen KE, et al. Consumption of Lacto-bacillus acidophilus and Bifidobacterium longum do not alter urinaryequol excretion and plasma reproductive hormones in premenopausalwomen. Eur J Clin Nutr 2004;58:1635–42.
EQUOL ENANTIOMERS: FORMATION, FATE, AND PROPERTIES 1079
Downloaded from https://academic.oup.com/ajcn/article-abstract/81/5/1072/4649521by gueston 01 May 2018
