Nutritional Methodology

Fermentation by Gut Microbiota Cultured in a Simulator of the Human

Intestinal Microbial Ecosystem Is Improved by Supplementing a Soygerm
Powder1

Downloaded from https://academic.oup.com/jn/article-abstract/130/10/2599/4686140 by KERIS National Access user on 20 March 2020

Patrick De Boever, Bart Deplancke* and Willy Verstraete2
Laboratory of Microbial Ecology and Technology, Faculty of Agricultural and Applied Biological Sciences,
University Ghent, B-9000 Ghent, Belgium and *Laboratory of Intestinal Immunobiology, Division of Nutritional
Sciences, University of Illinois at Urbana-Champaign, Urbana, IL 61801

ABSTRACT An in vitro model, designated the Simulator of the Human Intestinal Microbial Ecosystem (SHIME), was
used to study the effect of a soygerm powder rich in ␤-glycosidic phytoestrogenic isoflavones on the fermentation
pattern of the colon microbiota and to determine to what extent the latter metabolize the conjugated phytoestrogens.
Initially, an inoculum prepared from human feces was introduced into the reactor vessels and stabilized over 3 wk using
a culture medium. This stabilization period was followed by a 2-wk control period during which the microbiota were
monitored. The microbiota were then subjected to a 2-wk treatment period by adding 2.5 g/d soygerm powder to the
culture medium. The addition resulted into an overall increase of bacterial marker populations (Enterobacteriaceae,
coliforms, Lactobacillus sp., Staphylococcus sp. and Clostridium sp.), with a significant increase of the Lactobacillus sp.
population. The short-chain fatty acid (SCFA) concentration increased ⬃30% during the supplementation period; this
was due mainly to a significant increase of acetic and propionic acids. Gas analysis revealed that the methane
concentration increased significantly. Ammonium and sulfide concentrations were not influenced by soygerm supple-
mentation. Use of an electronic nose apparatus indicated that odor concentrations decreased significantly during the
treatment period. The ␤-glycosidic bonds of the phytoestrogenic isoflavones were cleaved under the conditions
prevailing in the large intestine. The increased bacterial fermentation after addition of the soygerm powder was
paralleled by substantial metabolism of the free isoflavones (genistein, daidzein and glycitein), resulting in recovery of
only 12–17% of the supplemented isoflavones. J. Nutr. 130: 2599 –2606, 2000.

KEY WORDS: ● soy ● isoflavones ● gut microbiota ● fermentation ● culture system

Soy-based food products have recently gained much atten- the nonestrogenic O-desmethylangolensin, whereas genistein is
tion as functional foods after several epidemiologic studies metabolized to the nonestrogenic p-ethyl phenol (Kurzer and Xu
showed a lower incidence of estrogen-related cancers, cardio- 1997). These data indicate that the gut microbiota play an
vascular diseases and osteoporosis due to a soy-rich diet important role in the generation of biologically active isoflavones
(Messina 1999, Setchell, 1998). Substantial evidence indi- but at the same time in the inactivation of these bioactive
cates that isoflavones present in soy play an important role in compounds after further bacterial fermentation, resulting in a loss
the observed effects. These naturally occurring plant chemi- of their acclaimed beneficial effects.
cals have been shown to compete effectively with endogenous Much of the research concerning the health-promoting
mammalian estrogens in binding to the estrogen receptor of effects of soy products has focused on the putative role of one
mammalian cells, preventing estrogen-stimulated growth of dietary ingredient i.e. the isoflavones. However, soy is an
cancerous cells (Brzezinski and Debi 1999, Molteni et al. important source of many other nutrients including dietary
1995). Other isoflavone effects include inhibition of tyrosine fiber, oligosaccharides, proteins, trace minerals and vitamins,
protein kinases, antioxidative activity and in vitro angiogen- which could influence the host’s well-being (Slavin et al.
esis (Brandi 1997, Kim et al. 1998). 1999). A particular role can be reserved for the oligosaccha-
Isoflavones occur predominantly as biologically inactive ␤-gly- rides, which could meet the standards of a prebiotic. Prebiotics
cosides in soybean (conjugated isoflavones) (Setchell 1998). Af- have been developed to promote the growth and activity of
ter ingestion, the glycosides genistin and daidzin are hydrolyzed beneficial microorganisms in the large intestine. A prebiotic is
by bacteria to release the bioactive phytoestrogens genistein and a nondigestible food ingredient that affects the host benefi-
daidzein (free isoflavones). Daidzein can be metabolized by bac- cially by selectively stimulating the growth and/or activity of
teria in the large intestine to form the more estrogenic equol and one or a limited number of bacteria in the colon that can
improve the health of the host (Gibson and Roberfroid 1995).
1
Soy products contain characteristically high concentrations
Funded by a scholarship from the Flemish Institute for the Improvement of
Scientific-Technological Research in the Industry (IWT). of the ␣-glycosidic galactooligosaccharides, raffinose and
2
To whom correspondence should be addressed. stachyose. These type of sugars are not absorbed in the upper

0022-3166/00 $3.00 © 2000 American Society for Nutritional Sciences.

Manuscript received 30 March 2000. Initial review completed 11 May 2000. Revision accepted 12 July 2000.

2599
2600 DE BOEVER ET AL.

TABLE 1 TABLE 2
Description of the different vessels of the SHIME culture Analyzed microbial groups together with the isolation media
system and their operational parameters1 used and the incubation conditions

Retention Bacterial group Medium Condition Time

Vessel Intestinal segment Volume time pH
h
L h
Staphylococcus sp. Mannitol Salt agar Aerobic 48

Downloaded from https://academic.oup.com/jn/article-abstract/130/10/2599/4686140 by KERIS National Access user on 20 March 2020

1 Stomach 0.2 2 2.0–2.5 Lactobacillus sp. Rogosa agar Anaerobic 72
2 Small intestine 0.3 6 5.0–6.0 Coliforms Violet Red Bile agar Aerobic 24
3 Colon ascendans 0.7 18 5.5–6.0 Enterobacteriaceae Violet Red Bile Glucose Aerobic 24
4 Colon transversum 1.3 36 6.0–6.4 agar
5 Colon descendans 0.8 22 6.6–6.9 Clostridium sp. Tryptose Sulfite Anaerobic 24
Cycloserine agar
1 SHIME, Simulator of the Human Intestinal Microbial Ecosystem.

ular part of the gastrointestinal microbial ecosystem (Table 1). The

part of the gastrointestinal tract or hydrolyzed by human first two vessels work according to a fill-and-draw system, whereas the
digestive enzymes (Suarez et al. 1999b). Upon delivery to the last three are continuously stirred tank reactors with a total retention
colon, these sugars are fermented by the colonic microbiota time of 76 h. The suspensions are mixed continuously by means of
possessing ␣-galactosidase activity. Administration of soybean magnetic stirrers (Labinco L22, Vel, Leuven, Belgium). pH control-
oligosaccharides to an in vitro culture of human gut micro- lers (pH controller R301, Consort, Turnhout, Belgium) maintain the
biota has been reported to be bifidogenic (Hayakawa et al. pH in the last three vessels between fixed limits by the addition of 0.1
1990). Therefore, Gibson and Roberfroid (1995) suggested a mol/L HCl or 0.1 mol/L NaOH (Table 1). There was no gas exchange
between the different vessels and the headspace of the culture system
potential prebiotic role for soybean oligosaccharides. Studies was flushed twice a day for 15 min with O2/free N2 to ensure
addressing this hypothesis have not been published. anaerobic conditions. A detailed scheme of the reactor set-up is
The aim of our study was to investigate the effects of a provided in Figure 1. At the beginning of the experiment, the last
soygerm powder on the fermentative capacity of the simulated three vessels were inoculated with a pooled fecal sample of five
microbiota of the colon. Furthermore, the capacity of this volunteers (Molly et al. 1993). Aliquots (10 g) of freshly voided fecal
intestinal microbiota to metabolize the phytoestrogens present samples were diluted and homogenized with 100 mL sterilized phos-
in the soygerm powder was evaluated. phate buffer (0.1 mol/L, pH 7.0), containing 1 g/L sodium thiogly-
colate as reducing agent. After removal of the particulate material by
centrifugation (1 min, 500 ⫻ g), the supernatants were pooled and 50
MATERIALS AND METHODS mL was introduced into the last three vessels. The microbial inocu-
Culture system. The Simulator of the Human Intestinal Micro- lum was stabilized over a period of 3 wk by the addition of 200 mL of
bial Ecosystem (SHIME)3 consists of five double-jacketed vessels a carbohydrate-based medium containing arabinogalactan (1 g/L),
maintained at a temperature of 37°C. Each vessel simulates a partic- pectin (2 g/L), xylan (1 g/L), starch (3 g/L), glucose (0.4 g/L) and
mucin (4 g/L) to the first vessel of the culture system three times a day
(Molly et al. 1994). The pH of the feed was gradually decreased to 2,
3
Abbreviations used: GC, gas chromatography; SCFA, short-chain fatty ac-
simulating stomach acidification. The passage of food in the small
ids; SHIME, Simulator of the Human Intestinal Microbial Ecosystem; SOP, sen- intestine was simulated in the next vessel by the addition of 100 mL
soric odor perception. simulated pancreatic and bile liquid [6 g/L oxgall (Difco, Bierbeek,

FIGURE 1 Schematic representation of the Simulator of the Human Intestinal Microbial Ecosystem (SHIME).
 SUPPLEMENTING SOYGERM POWDER IMPROVES GUT FERMENTATION 2601

TABLE 3 Ammonium determination. Liquid samples were frozen at

⫺20°C for subsequent analysis. Using a 1026 Kjeltec Auto Distilla-
Composition of SOYLIFE tion (FOSS Benelux, Amersfoort, The Netherlands), ammonium in
the sample was liberated as ammonia by the addition of an alkali
Compound Concentration Method of analysis (MgO). The released ammonia was distilled from the sample into a
boric acid solution (Bremner and Keeney 1965). The solution was
g/kg back-titrated using a 665 Dosimat (Metrohm, Berchem, Belgium) and
686 Titroprocessor (Metrohm).
Water 50 EEG L 279/81 H2S analysis. The H2S concentration was determined by the
Crude protein 405 AOAC 990.031 addition of 5 mL of liquid samples to 0.4 mL of a solution containing
Crude fat 112 EEG L 15/291 dimethyl-p-phenylenediamine (p-aminodimethyl aniline) and ferric

Downloaded from https://academic.oup.com/jn/article-abstract/130/10/2599/4686140 by KERIS National Access user on 20 March 2020

Crude fiber 39 EEG L 344/351 chloride (Trüper and Schlegel 1964). The methylene blue dye pro-
Ash 46 EEG L 155/201 duced was measured at 670 nm with a UVIKON930 spectrophotom-
Carbohydrates 348 Calculated2 eter (BRS, Brussels, Belgium).
Sugars 133 EEG 155/291 Gas analysis. Before flushing, the headspace of vessels 3, 4 and
Isoflavones 25 Song et al. (1998)
Saponins 41 Ireland et al. (1984)
5 was sampled (1 mL) and analyzed for CO2 and CH4. A 20-mL
Tocopherols 0.496 Ball (1988)
sample was collected to determine the concentration of H2. The
␣-Tocopherols 0.081 Ball (1988) concentrations of CO2 and CH4 were determined with an Intersmat
Lecithin 0.018 NEN 63493 IGC 120MB gas chromatograph (Thermo Separation Products) con-
nected to a Hewlett-Packard 3396A integrator (HP Benelux, Brus-
1 The concentration of all compounds, except carbohydrates, was sels, Belgium). The GC was equipped with a dual-column arrange-
measured analytically. The concentration of carbohydrates was calcu- ment consisting of an in-series connected Porapak (50 – 80 mesh), a
lated by making the total mass balance. molecular sieve (60 – 80 mesh) and a catharometer. The column
2 The directives of the European Community comprising the method temperature was set at 30°C and the flow rate of the carrier gas (He)
of analysis were obtained from the Belgian Institute for Normalization in was fixed at 10 mL/min. The percentage of H2 was determined using
Brussels (Belgium). an Exhaled Hydrogen Monitor (GMI, Renfrew, Scotland) equipped
3 The Dutch norm for analyzing the concentration of lecithin was with a H2-sensitive three-electrode electrochemical cell (Nollet et al.
obtained from the Dutch Institute for Normalization in Delft (The Neth- 1998).
erlands). Further analysis of the gases present in the headspace was done
using the FOX3000 electronic nose (Alpha M.O.S., Toulouse,
France). The electronic nose consists of two measuring chambers
each with six different metal oxide sensors. Each chamber is equipped
Belgium), 0.9 g/L pancreatin (Sigma, Bornem, Belgium) and 12.6 g/L with two extra sensors for measuring temperature and humidity. Each
NaHCO3]. Steady-state conditions were assessed by monitoring the sensor consists of a 50-␮m layer of a metal oxide film deposited on a
concentration of short-chain fatty acids (SCFA). After the initial ceramic film and displays a high sensitivity toward a broad range of
stabilization period of 3 wk, the microbial communities present in the volatile chemical compounds (Table 4). A mass-flow controller reg-
last three vessels of the culture system were fed the carbohydrate- ulates the rate (150 mL/min) with which humidified air flows con-
based medium. During this control period of 14 d, bacterial popula- tinuously over the sensors. The injection time or the sampling time
tions were characterized using plate counts of selected fecal marker was set at 15 s. The headspaces of SHIME vessels (2 mL) were diluted
organisms (Table 2), and the production of fermentation metabolites in 2 L of dry air (Messer, Machelen, Belgium) before injection. The
(SCFA, ammonium, H2S, and gases) was monitored. During the measurement principle of an electronic nose is based on the change
treatment period of 14 d, 2.5 g/d of a soygerm powder (SOYLIFE, in electrical resistance of the sensors when volatile compounds are
Soylife Nederland BV, Giessen, The Netherlands, Table 3) was present. The metal oxide sensors are semiconductors and are there-
added to the carbohydrate-based medium. The effect of this supple- fore gas sensitive. Oxygen present in the air is chemisorbed on
mentation on the simulated microbial communities of the large vacancies in the lattice of the bulk material and removes electrons
intestine was studied by means of the previously mentioned param- from the conducting band as follows:
eters (selected fecal marker organisms and fermentation metabolites).
Participants. Five subjects participated in this study. They were
all healthy males with a mean age of 25 ⫾ 2 y. They all reported to
have no gastrointestinal problems or a recent history of antibiotic TABLE 4
usage. Informed consent was obtained in writing.
Microbiological analyses. The bacterial groups analyzed and the Overview of the sensors in the FOX3000 electronic nose, as
media used are indicated in Table 2. These media have been used well as the chemical compounds to which they are sensitive
successfully by Molly et al. (1994) to select for these bacterial maker
organisms. Liquid samples were withdrawn from the culture system Sensor
and diluted serially in physiologic solution (8.5 g/L NaCl). Three
plates were inoculated with a 0.1-mL sample of three dilutions, and Number Name Sensitive to
incubated at 37°C using conditions given in Table 2. Anaerobic
incubation of plates was performed in jars with a gas atmosphere 1 P30/1 Polar compounds
(84% N2, 8% CO2, and 8% H2) adjusted by the Anoxomat 8000 2 P10/1 Hydrocarbons
system (Mart, Sint-Genesius-Rode, Belgium). 3 P10/2 Methane, propane and aliphatic nonpolar
SCFA determination. Liquid samples were collected and frozen molecules
at ⫺20°C for subsequent analysis. The SCFA were extracted from the 4 P40/1 Chlorinated and fluorinated compounds
samples with diethyl ether and determined with a Di200 gas chro- 5 P40/2 Aldehydes
matograph (GC; Shimadzu, ’s-Hertogenbosch, The Netherlands). 6 PA3 Hydrogen bounds, smoke detection
The GC was equipped with a capillary free fatty acid packed column 7 P70/0 Food aromas, alcohol and aromatic compounds
[EC-1000 Econo-Cap column (Alltech, Laarne, Belgium), 25 m 8 T50/3 Ammonia derivatives and amine-containing
⫻ 0.53 mm; film thickness 1.2 ␮m], a flame ionization detector and compounds
a Delsi Nermag 31 integrator (Thermo Separation Products, Wilrijk, 9 PA2 Low concentrations of hydrogen, H2S and
ammonia
Belgium). Nitrogen was used as the carrier gas at a flow rate of 20
10 T50/1 S-containing compounds
mL/min. The column temperature was set at 130°C and the temper- 11 T40/1 Chlorinated and fluorinated compounds
ature of the injector and detector was set at 195°C (Nollet et al. 12 T70/2 Alcohol vapors and aromatic compounds
1998).
2602 DE BOEVER ET AL.

TABLE 5
Plate counts of selected indicator organisms found in the last three reactor vessels of the SHIME during the control
and soygerm powder treatment periods1,2

Bacterial group Vessel 3 Vessel 4 Vessel 5

log10 CFU/g

Enterobacteriaceae Control 5.79 ⫾ 1.11 6.00 ⫾ 0.59 5.45 ⫾ 0.75

7.43 ⫾ 0.91 6.61 ⫾ 0.61 6.48 ⫾ 0.83

Downloaded from https://academic.oup.com/jn/article-abstract/130/10/2599/4686140 by KERIS National Access user on 20 March 2020

Treatment
Coliforms Control 5.76 ⫾ 1.11 6.32 ⫾ 0.61 5.85 ⫾ 0.60
Treatment 7.76 ⫾ 1.14 6.84 ⫾ 0.68 6.28 ⫾ 0.76
Lactobacillus sp. Control 3.06 ⫾ 0.67 2.75 ⫾ 1.31 3.35 ⫾ 1.14
Treatment 5.79 ⫾ 1.43* 4.56 ⫾ 1.75 4.35 ⫾ 1.30
Clostridium sp. Control 6.10 ⫾ 0.96 6.44 ⫾ 0.62 6.23 ⫾ 1.01
Treatment 7.70 ⫾ 0.64 7.06 ⫾ 0.83 7.13 ⫾ 0.67
Staphylococcus sp. Control 6.60 ⫾ 0.60 5.45 ⫾ 0.30 6.45 ⫾ 1.45
Treatment 7.05 ⫾ 0.40 6.20 ⫾ 0.65 6.30 ⫾ 0.25

1 During the treatment period, 2.5 g/d of soygerm powder was added to the SHIME medium. SHIME, Simulator of the Human Intestinal Microbial
Ecosystem; CFU, colony-forming units.
2 Data are means ⫾ SD, n ⫽ 3.
* Significantly different from the control period, P ⱕ 0.05.

sensor electron ⫹ 1/ 2 O 2 3 O⫺ (s)

冘 冉12
i⫽1
R 0,i 冊
R 0,i ⫺ R i
䡠 1000
In the presence of a gas or a fragrant molecule (F), this chemisorbed SOP ⫽
oxygen (O⫺) reacts irreversibly to produce combined molecules (FO) 12 䡠 Volume
as follows:
where R0,i is the baseline resistance of sensor i; Ri is the
F 共 g兲 ⫹ O ⫺ (s) 3 FO (g) ⫹ sensor electron resistance between the gas and sensor i; and Volume is the
The liberated electrons reduce the potential barrier of the oxide volume of gas analyzed.
grains, which decreases the electron mobility. As a consequence, the Isoflavone analysis. The concentration of glycosylated (genistin,
electrical resistance of the sensors drops when a gas sample with daidzin and glycitin) and free isoflavones (genistein, daidzein and
volatile organic compounds passes the sensors. The type of sensor and glycitein) was determined by HPLC (Wang and Murphy 1994). All
the percentage of change of a sensor’s resistance after the injection of samples were analyzed in triplicate.
a sample are an indication of the type and concentration of certain Statistical analysis. Samples were taken on d 7, 10 and 14 of the
organic compounds in the gas sample (Maricou et al. 1998). The odor control and treatment periods. ANOVA was performed on the data with
intensity can be estimated by the sensoric odor perception (SOP) the statistical software SPSS 7.5 (SPSS, Chicago, IL). A P-value ⱕ 0.05
value, which is calculated according to the following formula: was considered to be significant. All values are reported as means ⫾ SD.

TABLE 6
Concentration of different fermentative products (short-chain fatty acids, NH4⫹ and H2S), measured in vessels 3, 4 and 5 of the
SHIME during the control and soygerm powder treatment periods1,2

Parameter Vessel 3 Vessel 4 Vessel 5

␮mol/g suspension

Acetic acid Control 17.91 ⫾ 2.89 17.39 ⫾ 8.13 14.00 ⫾ 5.16

Treatment 24.10 ⫾ 1.68* 24.62 ⫾ 1.99 21.19 ⫾ 1.80
Propionic acid Control 5.39 ⫾ 0.64 6.26 ⫾ 1.74 5.87 ⫾ 1.24
Treatment 7.65 ⫾ 0.74* 6.87 ⫾ 1.25 6.07 ⫾ 0.70
Butyric acid Control 7.73 ⫾ 1.43 7.39 ⫾ 1.54 6.72 ⫾ 1.25
Treatment 8.99 ⫾ 1.34 7.53 ⫾ 1.19 6.49 ⫾ 0.89
Other acids Control 2.08 ⫾ 0.48 2.72 ⫾ 0.91 2.64 ⫾ 0.96
Treatment 2.84 ⫾ 1.10 3.16 ⫾ 1.07 2.89 ⫾ 0.89
Total acids Control 35.19 ⫾ 4.84 36.49 ⫾ 11.51 31.90 ⫾ 6.57
Treatment 46.41 ⫾ 4.55* 45.34 ⫾ 5.24 39.52 ⫾ 4.06
NH4⫹ Control 13.07 ⫾ 0.78 19.64 ⫾ 0.50 18.14 ⫾ 0.93
Treatment 11.50 ⫾ 0.64 18.43 ⫾ 0.64 19.86 ⫾ 2.29
H2S Control 0.32 ⫾ 0.06 0.27 ⫾ 0.05 0.06 ⫾ 0.01
Treatment 0.25 ⫾ 0.02 0.23 ⫾ 0.03 0.09 ⫾ 0.03

1 During the treatment period, 2.5 g/d of soygerm powder was added to the SHIME medium. See Table 1 for abbreviation.
2 Data are means ⫾ SD, n ⫽ 3.
* Significantly different from the control period, P ⱕ 0.05.
 SUPPLEMENTING SOYGERM POWDER IMPROVES GUT FERMENTATION 2603

TABLE 7
Concentration of fermentation gases present in the
headspace of reactor vessels 3, 4 and 5 of the SHIME during
the control and soygerm powder treatment periods1,2

Gas Vessel 3 Vessel 4 Vessel 5

␮mol/L

Downloaded from https://academic.oup.com/jn/article-abstract/130/10/2599/4686140 by KERIS National Access user on 20 March 2020

CO2 Control 9.78 ⫾ 0.70 9.51 ⫾ 0.41 4.92 ⫾ 0.12
Treatment 9.59 ⫾ 0.86 8.94 ⫾ 0.78 5.00 ⫾ 0.49
CH4 Control 0⫾0 0.18 ⫾ 0.09 0.14 ⫾ 0.08
Treatment 0.16 ⫾ 0.21 0.64 ⫾ 0.16* 0.54 ⫾ 0.09*
H2 Control 0.10 ⫾ 0.06 0.15 ⫾ 0.09 2.76 ⫾ 0.51
Treatment 0.16 ⫾ 0.06 0.09 ⫾ 0.07 2.66 ⫾ 0.36

1 During the treatment period, 2.5 g/d of soygerm powder was

added to the SHIME medium. See Table 1 for abbreviation.
2 Data are means ⫾ SD, n ⫽ 3.
* Significantly different from the control period, P ⱕ 0.05.

RESULTS
Microbiology. The SHIME reactor system was subjected to
a control period of 14 d to establish a microbial community after
inoculation of the last three reactor vessels with a pooled fecal
sample. During the treatment period of 14 d, the addition of 2.5
g/d of soygerm powder to the SHIME resulted in a rise in the
concentrations of all analyzed bacterial groups, especially in vessel
3 (Table 5). An increase of ⬎2 log10 units was observed for the
Lactobacillus population in vessel 3 (P ⱕ 0.05). A microbial
ecosystem could not establish itself in vessel 1 or 2 because of the
short retention time and the fill-and-draw principle. Hence, ves-
sels 1 and 2 were not sampled.
Fermentative capacity. The SCFA, acetic, propionic and
butyric acids, comprised ⬎90% of the total fatty acid produc-
tion. The other SCFA were the sum of the concentrations of
isobutyric, isovaleric and valeric acids. The addition of soy-
germ powder led to an overall increase of the fatty acid
concentration. The total concentration in vessel 3 was signif-
icantly higher (P ⱕ 0.05) during the supplementation period
and this was due mainly to a significant increase in the
concentration of acetic and propionic acids (P ⱕ 0.05). The
percentage increase of the total SCFA production amounted
to 32, 24 and 24% for vessel 3, 4 and 5, respectively (Table 6).
Methane production increased significantly (P ⱕ 0.05) in the
last two reactor vessels concomitant with the addition of the
soygerm powder (Table 7). Supplementation of the soygerm

TABLE 8
Sensoric odor perception of the headspaces of reactor
vessels 3, 4 and 5 of the SHIME during the control and
soygerm powder treatment periods1,2 FIGURE 2 Radar plots of the odor profiles of the headspace of
reactor vessel 3 (a), vessel 4 (b) and vessel 5 of (c) control and
Reactor vessel Control Treatment soygerm powder treatment samples. The 12 angular data corre-
spond to the 12 different sensors of the FOX3000 electronic nose.
% change in electrical resistance/L The radial data indicate the change in electrical resistance of each
sensor due to the injection of gas samples. Headspaces were sam-
3 561 ⫾ 55 450 ⫾ 15* pled five times during the control period (F) and treatment period (f).
4 546 ⫾ 48 427 ⫾ 30* During the treatment period, 2.5 g/d of soygerm powder was added
5 318 ⫾ 39 127 ⫾ 75*
to the SHIME medium. SD are not represented to improve interpret-
1 During the treatment period, 2.5 g/d of soygerm powder was ability of the plots. Significant differences between the control period
and the treatment period are indicated next to the angular data:
added to the SHIME medium. See Table 1 for abbreviation.
2 Data are means ⫾ SD, n ⫽ 5. *P ⱕ 0.05.
* Significantly different from the control period, P ⱕ 0.05.
2604 DE BOEVER ET AL.

TABLE 9
Concentration of isoflavones measured in the vessels of the SHIME during the supplementation of 2.5 g/d of soygerm powder
containing ⬃90 ␮mol conjugated isoflavones/g1

7d 14 d

Compound Vessel 1 Vessel 2 Vessel 3 Vessel 4 Vessel 5 Vessel 1 Vessel 2 Vessel 3 Vessel 4 Vessel 5

mmol/L

Downloaded from https://academic.oup.com/jn/article-abstract/130/10/2599/4686140 by KERIS National Access user on 20 March 2020

Conjugated
isoflavones 255.6 256.0 25.9 25.9 14.8 254.5 253.8 —2 18.5 14.8
Free
isoflavones ⬍3 ⬍3 11.1 11.1 14.8 ⬍3 ⬍3 —2 11.1 14.8

1 See Table 1 for abbreviation.

2 No data available.

powder did not influence ammonium and sulfide production. (Fooks et al. 1999, Fuller and Gibson 1997, Roberfroid 1998).
Supplementation of the soygerm powder had no significant Criteria that allow classifying a food ingredient as a prebiotic
effect on the concentration of H2 and CO2 (Table 7). include the following (Gibson 1999): it must neither be hy-
The addition of the soygerm powder significantly (P drolyzed nor absorbed in the upper part of the gastrointestinal
ⱕ 0.05) reduced the SOP value in the headspace of all three tract; it must be selectively fermented by one or a limited
reactor vessels as measured by the FOX3000 electronic nose number of potentially beneficial bacteria in the colon; it must
(Table 8). The percentage of change of electrical resistance alter the composition of the colonic microbiota toward a
for the 12 FOX3000 sensors was plotted in a radar plot to healthier composition; and it must preferably induce effects
distinguish qualitatively the odor profiles of the headspace in that are beneficial to the host’s health.
the reactor vessels before and during the addition of the The oligosaccharides present in the soygerm powder consist
soygerm powder. During the supplementation period, signifi- mainly of the nondigestible raffinose (13 g/kg) and stachyose
cant decreases (P ⱕ 0.05) of the electrical resistance were (87 g/kg) oligosaccharides, which therefore serve as substrates
observed for sensors 1, 3, 4 and 6, and sensors 4 and 6 for for bacterial fermentation (Suarez et al. 1999b). Moreover, use
vessels 3 and 4, respectively. When the headspace of vessel 5 of the SHIME in vitro model, which lacks resorption of the
was analyzed, the resistance dropped significantly for all sen- small intestine, implied that all nutrients present in the soy-
sors, except 5 and 10, during the treatment (Fig. 2). germ powder were considered as colonic foods. During the
The concentrations of conjugated (genistin, daidzin and treatment period, the concentration of all bacterial groups
glycitin) and free isoflavones (genistein, daidzein and gly- increased and the strongest increase was observed for Lactoba-
citein) were determined in the reactor vessels of the SHIME 1 cillus sp. An increase in lactic acid bacteria is often driven by
and 2 wk after the initiation of the soygerm powder supple- a decrease in luminal pH because the capacity to thrive in low
mentation. Conjugated isoflavones were not hydrolyzed due to pH environments gives these bacteria a selective advantage
the acid environment in vessel 1 or the subsequent enzymatic over other intestinal bacteria (Sghir et al. 1998). Gibson
digestion, which was taking place in vessel 2 of the SHIME. In (1999) stated that lowering of the gut pH by metabolic prod-
the last three reactor vessels, similar isoflavone concentrations ucts may be one of the most important mechanisms by which
were recovered after 7 and 14 d. The concentration of conju- a prebiotic can improve colonic health. The pH drop in a
gated isoflavones was somewhat higher than that of the free microniche could select for lactic acid-producing bacteria in
isoflavones (Table 9). The free isoflavones comprised solely this environment, possibly excluding long-term colonization
daidzein. by invasive pathogens (Gibson and Wang 1994). In our in
vitro model, however, the colon vessels are pH controlled,
DISCUSSION allowing nonacid-resistant microorganisms to develop. The
We investigated the potential prebiotic effects of a soygerm pH in the colon compartments of our system was set at the
powder. This product is currently being marketed mainly be- lower limit of the pH interval, and the amount of base dosed
cause it contains a large amount of isoflavones, through which to maintain this pH (data not shown) increased during sup-
the product could relieve menopausal problems, reduce osteo- plementation of the soygerm powder, giving the lactic acid–
porosis, improve blood cholesterol levels and lower the risk of producing bacteria a slight growth advantage over the other
certain types of cancers (Watanabe et al. 1998, Zhang et al. microorganisms. Furthermore, it is now recognized that the
1999). During a 2-wk period, the soygerm powder was supple- classical culture-based methods give a biased view of the gut
mented to the SHIME, an in vitro model simulating the microbiota composition. Use of molecular approaches has in-
colonic microbial ecosystem of healthy adults, to determine dicated that 60 – 80% of the gut microbiota have not been
the effect of the soygerm powder on the microbial ecosystem. cultivated and that the contribution of some species may not
A second aim was to determine to what extent the microbial have been estimated correctly (Langendijk et al. 1995, Suau et
ecosystem metabolizes the isoflavones present in this soygerm al. 1999). Moreover, molecular techniques are more sensitive
powder. than culture-based techniques, making it plausible that we
A growing body of scientific publications indicates that missed changes in bacterial populations. However, we contend
dietary modulation, by ingestion of prebiotics, may shift the that the plate counts obtained remain indicative of relative
microbial ecosystem toward higher concentrations of lactic major trends in microbial changes, although they do not
acid–producing bacteria, which may be beneficial for the host necessarily represent absolute changes in the complex micro-
 SUPPLEMENTING SOYGERM POWDER IMPROVES GUT FERMENTATION 2605

bial ecosystem of the colon. The knowledge gained via the initiation of cancer, malabsorption and anemia (Macfarlane
application of molecular techniques and the limited informa- and Macfarlane 1997). During our study, we investigated the
tion obtained with plate counts emphasize that the use of odor intensity and pattern generated by the microbial com-
accurate methods for monitoring microbial ecosystems is es- munity by means of an electronic nose. This apparatus has
sential to understand the true effect of a dietary strategy that been used in the past to measure the concentration of volatile
aims at modulating the gut microbiota. organic compounds (Maricou et al. 1998). Our data revealed
During the supplementation period, the fermentative ca- that during the addition of the soygerm powder, the amount of
pacity of the microbial community was increased and this led odor decreased significantly. In the headspace of vessels 3 and
to a general rise in SCFA. The strongest increase was observed 4, the decreased amount of odor was due to a significantly
in vessel 3, indicating high carbohydrate fermentation. A rise decreased response of the electrical sensors 1, 3, 4 and 6.

Downloaded from https://academic.oup.com/jn/article-abstract/130/10/2599/4686140 by KERIS National Access user on 20 March 2020

of the SCFA concentration is a positive property because these Nearly all sensors showed a low response when the headspace
acids, especially butyric acid, are the main energy source for of vessel 5 was analyzed. Low electrical response of the sensors
colonocytes and influence colonic function by stimulating 1, 3, 4 and 6 was caused by a concentration decrease of certain
water and sodium absorption and modulating motility (Cher- gaseous compounds. To determine which classes of compounds
but et al. 1997). Furthermore, butyric acid induces differenti- were reduced in concentration during the soygerm powder
ation, stimulates apoptosis of cancerous cells in vitro and thus treatment, the FOX pattern of a range of pure volatile organic
arrests the development of cancer (Scheppach et al. 1995). compounds was determined. Sensors 1, 3, 4 and 6 were very
During the supplementation of the soygerm powder, no effect sensitive to skatol, indol and phenol (data not shown). These
on ammonium and hydrogen sulfide concentrations was ob- compounds can be recovered in a large concentration in the
served. Ammonia can alter the morphology and intermediary distal part of the colon and originate from bacterial metabo-
metabolism of intestinal cells, increase DNA synthesis and lism of aromatic amino acids (Macfarlane and Macfarlane
promote tumorigenesis (Ichikawa and Sakata 1998). Hydrogen 1997). To our knowledge, this is the first study in which the
sulfide selectively inhibits butyric acid oxidation in colono- odor produced by the gut microbiota was monitored. Taking
cytes and this may play a pathogenic role in inflammatory into account that compounds produced by intestinal microor-
bowel diseases such as ulcerative colitis [see reviews by Pitcher ganisms can be responsible for malodorous breath and various
and Cummings (1996) and Roediger et al. (1997)]. Hence, diseases, the capacity of the soygerm powder to reduce the
increases in the concentration of these compounds are con- concentration of odoriferous compounds is an interesting fea-
sidered to be potentially harmful for the host. Quantitative gas ture.
analyses revealed that the methane concentration increased The soygerm powder used in this study is being marketed as
significantly in the last two colon compartments during the a product containing a naturally high level of ␤-glycosidic
supplementation period. Production of methane by the co- isoflavones. Once the isoflavone moiety is hydrolyzed from the
lonic microbiota is important because it permits a more com- sugar residue by bacterial enzymes, the free isoflavone can
plete fermentation in that the removal of hydrogen is ener- exert estrogenic effects locally (on colonocytes) or on other
getically more favorable for fermentation (Christl et al. 1992). tissues (after resorption in the bloodstream). Gut microbiota
Although clinical relevance is not clear, the prevalence of play important roles in further isoflavone metabolism and
methanogenesis is inversely related to the prevalence of bowel bioavailability (Xu et al. 1995). In our study, we did not aim
cancer. Long-term methanogenic conditions in the intestine to identify the metabolites generated by the microorganisms,
can eventually result in out-competition of the potentially but rather to confirm that the bacteria were able to hydrolyze
harmful sulfate-reducing bacteria (Gibson et al. 1988). Sec- the conjugated isoflavones. Therefore, the concentration of
ond, there is some evidence that methanogenic persons suffer isoflavones was determined 1 and 2 wk after the initiation of
less from gas symptoms. Excessive gas production, primarily the treatment. The isoflavones were released in neither vessel
hydrogen gas, is an important cause for irritable bowel syn- 1 or 2, simulating the stomach and small intestine, respec-
drome. This common gastrointestinal disorder causes abdom- tively. The concentration of isoflavones (conjugated and free)
inal pain, distention, flatulence and borborygmus. These symp- in the three last vessels of the SHIME ranged between 12 and
toms may be reduced in methanogenic persons because the 17% of the concentration that could be expected, on the basis
oxidation of hydrogen gas to methane gas reduces the volume of the amount of soygerm powder added daily to the reactor
of gas to 25% (Florin and Jabbar 1994). Moreover, pulmonary and the concentration of conjugated isoflavones present in the
excretion of methane may also be less limited by intestinal powder (Table 3). The data are consistent with the literature
blood flow because methane is more soluble than hydrogen and indicate that isoflavones are not only hydrolyzed but also
(Florin and Woods 1995). extensively metabolized (Chang and Nair 1995, Xu et al.
Many of the putrefactive compounds (such as sulfur-con- 1995). Although Zhang et al. (1999) indicated that genistein
taining compounds, indoles, aliphatic amines and phenols) and glycitein are more resistant to bacterial breakdown than
generated by bacterial fermentation in the colon are respon- daidzein, we recovered only daidzein. It is known that there is
sible for the malodor of flatus and feces (Hussein et al. 1999). considerable interindividual variation of isoflavone metabo-
Because of its offensive odor, rectal gas has been a topic of lism. Zhang et al. (1999) hypothesized that isoflavone metab-
scientific interest for many years. Furthermore, there have olism may be influenced by the dietary pattern because diet has
been allusions to the possibility that malodorous breath gases an effect on the gut microbial population and its metabolizing
may originate from the gut. In a recent study, Suarez et al. capacity. The extent to which the host can benefit from the
(1999a) showed that the predominant sulfur gas causing the positive effects of free isoflavones is determined by the com-
persistent malodor of breath after garlic consumption origi- petition between resorption of free isoflavones by the host and
nated from the gut. The authors suggested that if these intes- further metabolism and inactivation of these compounds by
tinal gases were the major cause of malodorous breath, a the intestinal microbiota. The kinetics of isoflavone metabo-
reduction of the breath concentration of these gases would lism by the microbial ecosystem is a point that merits more
presumably require manipulation of the diet and/or gut micro- thorough research.
biota. Furthermore, phenolic and indolic compounds have This study indicates that the consumption of soygerm pow-
been linked to a variety of disease states in humans, including der may influence the gut microbiota in a beneficial way. The
2606 DE BOEVER ET AL.

concentration of SCFA increased, whereas concentrations of Kurzer, M. & Xu, X. (1997) Dietary phytoestrogens. Annu. Rev. Nutr. 17:
the putative harmful metabolites (NH4⫹ and H2S) were not 353–381.
Langendijk, P. S., Schut, F., Jansen, G. J., Raangs, G., Kamphuis, G. R., Wilkin-
affected. Gas analyses revealed that methanogenesis was stim- son, M. H. F. & Welling, G. W. (1995) Quantitative fluorescence in situ
ulated significantly and that the odor produced by the micro- hybridisation of Bifidobacterium spp. with genus-specific 16S rRNA-targeted
biota decreased. The microbial community of the large intes- probes and its application in fecal samples. Appl. Environ. Microbiol. 61:
3069 –3075.
tine, as simulated by the SHIME, was able to hydrolyze Macfarlane, G. T. & Macfarlane, S. (1997) Human colonic microbiota: ecology,
conjugated isoflavones and to transform the free isoflavones. physiology and metabolic potential of intestinal bacteria. Scand. J. Gastro-
Because of the limited experimental set-up, we cannot draw enterol. 32: 3–9.
firm conclusions concerning the effect of soygerm powder on Maricou, H., Pereira, D., Verschuere, L., Philips, S. & Verstraete, W. (1998)
Measurement of some volatile compounds by means of the electronic nose.
the host health. This is the first study indicating the ability of

Downloaded from https://academic.oup.com/jn/article-abstract/130/10/2599/4686140 by KERIS National Access user on 20 March 2020

Water Air Soil Pollut. 107: 423– 442.
soygerm powder to have a positive influence on the fermen- Messina, M. J. (1999) Legumes and soybeans: overview of their nutritional
tation balance in the large intestine, but these results are to be profiles and health effects. Am. J. Clin. Nutr. 70: 439S– 450S.
confirmed in larger and better-designed in vitro experiments. Minekus, M., Smeets-Peeters, M., Bernalier, A., Marol-Bonnin, S., Havenaar, R.,
Marteau, P., Alric, M., Fonty, G. & Huis in ’t Veld, J. H. (1999) A computer-
The limitations associated with in vitro systems (Minekus et controlled system to simulate conditions of the large intestine with peristaltic
al. 1999) also should trigger further research to study the mixing, water absorption and absorption of fermentation products. Appl.
beneficial effects of soygerm powder consumption in vivo. Microbiol. Biotechnol. 53: 108 –114.
Molly, K., Vande Woestyne, M., De Smet, I. & Verstraete, W. (1994) Validation
of the simulator of the human intestinal microbial ecosystem (SHIME) reactor
ACKNOWLEDGMENTS using microorganism-associated activities. Microb. Ecol. Health Dis. 7: 191–
200.
The authors are indebted to Soylife Nederland BV (Giessen, The Molly, K., Vande Woestyne, M. & Verstraete, W. (1993) Development of a
Netherlands) for supplying SOYLIFE and Nutrilab S/G (Giessen) for 5-step multi-chamber reactor as a simulation of the human intestinal microbial
the isoflavone analyses. K. Van Hege, J. Kielemoes, R. Wouters, G. ecosystem. Appl. Microbiol. Biotechnol. 39: 254 –258.
Rombaut, E. Top and H.R. Gaskins are acknowledged for their Molteni, A., Brizio-Molteni, L. & Persky, V. (1995) In vitro hormonal effects of
critical reading of the manuscript. soybean isoflavones. J. Nutr. 125: 751S–756S.
Nollet, L., Mbanzamihigo, L., Demeyer, D. & Verstraete, W. (1998) Effect of the
addition of Peptostreptococcus productus ATCC 35244 on reductive aceto-
LITERATURE CITED genesis in the ruminal ecosystem after inhibition of methanogenesis by cell-
free supernatant of Lactobacillus plantarum 80. Anim. Feed Sci. Technol. 71:
Ball, G.F.M. (1988) Fat Soluble Vitamin Assays in Food Analyses, pp. 64 –75 49 – 66.
and 234 –258. Elsevier Applied Science, London, UK. Pitcher, M. C. & Cummings, J. H. (1996) Hydrogen sulphide: a bacterial toxin
Brandi, M. L. (1997) Natural and synthetic isoflavones in the prevention and in ulcerative colitis? Gut 39: 1– 4.
treatment of chronic diseases. Calcif. Tissue Int. 61: S5–S8. Roberfroid, M. B. (1998) Prebiotics and synbiotics: concepts and nutritional
Bremner, J. M. & Keeney, R. D. (1965) Steam distillation methods for deter- properties. Br. J. Nutr. 80: S197–S202.
mination of ammonium, nitrate and glycine. Anal. Chem. Acta 32: 485– 495. Roediger, W. E., Moore, J. & Babidge, W. (1997) Colonic sulfide in pathogen-
Brzezinski, A. & Debi, A. (1999) Phytoestrogens: the “natural” selective estro-
esis and treatment of ulcerative colitis. Dig. Dis. Sci. 42: 1571–1579.
gen receptor modulators? Eur. J. Obstet. Gynaecol 85: 47–51.
Scheppach, W., Bartram, H. P. & Richter, F. (1995) Role of short-chain fatty
Chang, Y. C. & Nair, M. (1995) Metabolism of daidzein and genistein by
acids in the prevention of colorectal cancer. Eur. J. Cancer: 1077–1080.
intestinal bacteria. J. Nat. Prod. 58: 1892–1896.
Cherbut, C., Aubé, A. C., Blottière, H. M. & Galmiche, J. P. (1997) Effects of Setchell, K.D.R. (1998) Phytoestrogens: the biochemistry, physiology, and
short-chain fatty acids on gastrointestinal motility. Scand. J. Gastroenterol. implications for human health of soy isoflavones. Am. J. Clin. Nutr. 68:
32: 52–57. 1333S–1346S.
Christl, S. U., Murgatroyd, P. R., Gibson, G. R. & Cummings, J. H. (1992) Sghir, A., Chow, J. M. & Mackie, R. I. (1998) Continuous culture selection of
Production, metabolism and excretion of hydrogen in the large intestine. bifidobacteria and lactobacilli from human faecal samples using fructooligo-
Gastroenterology 102: 1269 –1277. saccharide as selective substrate. J. Appl. Microbiol. 85: 769 –777.
Florin, T.H.J. & Jabbar, I. A. (1994) A possible role for bile acid in the control Slavin, J. L., Martini, M. C., Jacobs, D. R. & Marquart, L. (1999) Plausible
of methanogenesis and the accumulation of hydrogen gas in the colon. J. mechanisms for the protectiveness of whole grains. Am. J. Clin. Nutr. 70:
Gastroenterol. Hepatol. 9: 112–117. 459S– 463S.
Florin, T.H.J. & Woods, H. J. (1995) Inhibition of methanogenesis by human Song, T., Barua, K., Buseman, G. & Murphy, P. A. (1998) Soy isoflavones
bile. Gut 37: 418 – 421. analysis: quality control and a new internal standard. Am. J. Clin. Nutr. 68:
Fooks, L. J., Fuller, R. & Gibson, G. R. (1999) Prebiotics, probiotics and human 1474S–1479S.
gut microbiology. Int. Dairy J. 9: 53– 61. Suarez, F., Springfield, J., Furne, J. & Levitt, M. (1999a) Differentiation of
Fuller, R. & Gibson, G. R. (1997) Modification of the intestinal microflora using mouth versus gut as site of origin of odoriferous breath gases after garlic
probiotics and prebiotics. Scand. J. Gastroenterol. 32: 28 –31. ingestion. Am. J. Physiol. 39: G425–G430.
Gibson, G. R. (1999) Dietary modulation of the human gut microflora using the Suarez, F. L., Springfield, J., Furne, J. K., Lohrmann, T. T., Kerr, P. S. & Levitt,
prebiotics oligofructose and inulin. J. Nutr. 129: 1438S–1441S. M. D. (1999b) Gas production in humans ingesting a soybean flour derived
Gibson, G. R., Macfarlane, G. T. & Cummings, J. H. (1988) Occurrence of from beans naturally low in oligosaccharides. Am. J. Clin. Nutr. 69: 135–139.
sulphate-reducing bacteria in human faeces and the relationship of dissimi- Suau, A., Bonnet, R., Sutren, M., Godon, J. J., Gibson, G. R., Collins, M. D. &
latory sulphate reduction to methanogenesis. J. Appl. Bacteriol. 65: 103–111. Dore, J. (1999) Direct analysis of genes encoding 16S rRNA from complex
Gibson, G. R. & Roberfroid, M. B. (1995) Dietary modulation of the human communities reveals many novel molecular species within the human gut.
colonic microbiota: introducing the concept of prebiotics. J. Nutr. 125: 1401– Appl. Environ. Microbiol. 65: 4799 – 4807.
1412. Trüper, H. G. & Schlegel, H. G. (1964) Sulphur metabolism in Thiorhodaceae
Gibson, G. R. & Wang, X. (1994) Regulatory effects of bifidobacteria on the
I. Quantitiative measurements on growing cells of Chromation okenii. Antonie
growth of other colonic bacteria. J. Appl. Bacteriol. 77: 412– 420.
Leeuwenhoek 30: 225–238.
Hayakawa, K., Mizutani, J., Wada, K., Masai, T., Yoshihara, I. & Mitsuoka, T.
Wang, H. J. & Murphy, P. A. (1994) Isoflavone content in commercial soybean
(1990) Effect of soybean oligosaccharides on human fecal flora. Microb.
Ecol. Health Dis. 3: 293–303. foods. J. Agric. Food Chem. 42: 1666 –1673.
Hussein, H. S., Flickinger, E. A. & Fahey, G. C. (1999) Petfood applications of Watanabe, S., Yamaguchi, M., Sobue, T., Takahashi, T., Miura, T., Arai, Y., Mazur,
inulin and oligofructose. J. Nutr. 129: 1454S–1456S. W., Wahala, K. & Adlercreutz, H. (1998) Pharmacokinetics of soybean
Ichikawa, H. & Sakata, T. (1998) Stimulation of epithelial cell proliferation of isoflavones in plasma, urine and feces of men after ingestion of 60 g baked
isolated distal colon of rats by continuous colonic infusion of ammonia or soybean powder (kinako). J. Nutr. 128: 1710 –1715.
short-chain fatty acids is nonadditive. J. Nutr. 128: 843– 847. Xu, X., Harris, K. S., Wang, H. J., Murphy, P. A. & Hendrich, S. (1995) Bio-
Ireland, P. A., Dziedzic, S. Z. & Kearsley, M. W. (1984) Saponin content of soya availability of soybean isoflavones depends upon gut microflora in women. J.
and some commercial soya products by means of high-performance liquid- Nutr. 125: 2307–2315.
chromatography of the sapogenins. J. Sci. Food Agric. 37: 694 – 698. Zhang, Y., Wang, G. J., Song, T. T., Murphy, P. A. & Hendrich, S. (1999)
Kim, H. K., Peterson, T. G. & Barnes, S. (1998) Mechanism of action of the soy Urinary disposition of the soybean isoflavones daidzein, genistein and gly-
isoflavone genistein: emerging role of its effects via transforming growth citein differs among humans with moderate isoflavone degradation activity. J.
factor ␤ signaling pathways. Am. J. Clin. Nutr. 68: 1418S–1425S. Nutr. 129: 957–962.