FEMS Microbiology Letters 242 (2005) 81–86
www.fems-microbiology.org
Correlation between faecal microbial community structure
and cholesterol-to-coprostanol conversion in the human gut
Patrick Veiga, Catherine Juste, Pascale Lepercq, Katiana Saunier,
Fabienne Béguet, Philippe Gérard *
Unité dÕÉcologie et Physiologie du Système Digestif, Institut National de la Recherche Agronomique, Bâtiment 405, Domaine de Vilvert, 78352
Jouy-en-Josas, France
Received 21 July 2004; received in revised form 19 October 2004; accepted 23 October 2004
First published online 5 November 2004
Edited by M. Schembri
Abstract
Intensity of the cholesterol-to-coprostanol conversion in the intestine, as assessed by the coprostanol-to-cholesterol ratio in faeces, was found highly variable among 15 human volunteers, ranging from absent to almost complete cholesterol conversion. The
number of coprostanoligenic bacteria in the same faecal samples, as estimated by the most probable number method, was found
to be less than 106 cells g1 of fresh stools in the low-to-inefficient converters and at least 108 cells g1 of fresh stools in the highest
converters, indicating that the population level of cultivable faecal coprostanoligenic bacteria correlated with the intensity of cholesterol-to-coprostanol conversion in the human gut. Microbial communities of the samples were profiled by temporal temperature
gradient gel electrophoresis (TTGE) of bacterial 16S rRNA gene amplicons. Dendrogram analysis of the TTGE profiles using the
Pearson product moment correlation coefficient and a unweighted pair group method with arithmetic averages (UPGMA) algorithm
clearly separated banding patterns from low-to-inefficient and high converters in two different clusters suggesting a relationship
between TTGE profiles and coprostanoligenic activity. Principal components analysis further demonstrated that a large subset
of bands rather than some individual bands contributed to this clustering.
2004 Federation of European Microbiological Societies. Published by Elsevier B.V. All rights reserved.
Keywords: Cholesterol; Coprostanol; Temporal temperature gradient gel electrophoresis; Principal component analysis; Faecal microbial community
1. Introduction
Cholesterol escaping absorption in the upper digestive tract can be extensively metabolised by the resident
intestinal microbiota. Bacterial metabolism of cholesterol in the hindgut mainly involves an indirect pathway
with 4-cholesten-3-one and coprostanone as the intermediates and coprostanol as the end-product [1]. The
*
Corresponding author. Tel.: +33 1 34 65 24 28; fax: +33 1 34 65 24
92.
E-mail address: [email protected] (P. Gérard).
organisms responsible for this conversion are unknown
in humans and only a few cholesterol-reducing strains
have been isolated from animal faeces and assigned to
the genus Eubacterium based on morphological and
physiological properties [2,3]. Whereas unmetabolised
cholesterol is subjected to extensive enterohepatic circulation, coprostanol is poorly absorbable and excreted in
the faeces [4]. Therefore, it has been suggested that cholesterol-to-coprostanol conversion by the intestinal
microbiota could facilitate the elimination of cholesterol
from the body [5,6] and therefore decrease the risk of
cardiovascular diseases [6]. However, epidemiological
0378-1097/$22.00 2004 Federation of European Microbiological Societies. Published by Elsevier B.V. All rights reserved.
doi:10.1016/j.femsle.2004.10.042
82
P. Veiga et al. / FEMS Microbiology Letters 242 (2005) 81–86
studies have revealed that high-risk populations for colon cancer have an increased number of total anaerobes
and enhanced metabolism of neutral sterols [7], and
coprostanol is thought to be associated with colorectal
carcinogenesis [8,9].
Several epidemiological studies [10] and nutritional
trials in humans [11] and animals [12] have demonstrated that dietary habits influence the intestinal conversion of cholesterol. Moreover, the distribution of
this conversion among Americans was shown to be
bimodal, with a vast majority of high converters and a
minority of low-to-inefficient converters [13] but differences in the intestinal microbiota of both populations remain unexplored.
Most of the knowledge on bacterial diversity in the
human gastrointestinal tract has been obtained by culture on selective media. However, several limitations
are associated with culture-based approaches and up
to 60–70% of intestinal bacterial species remain uncultivable by known classical methods [14,15]. Recently,
our understanding of complex microbial communities
has been greatly enhanced by the introduction of
molecular techniques [16]. The major advances are
based on the use of 16S rRNA as a phylogenetic marker for analysing community diversity [17]. In particular, denaturing gel electrophoresis (DGE) profiling of
faecal 16S rRNA and rRNA gene amplicons was
shown to be a powerful tool for analysing the dominant species diversity of the faecal bacterial community
[18–20]. These methods separate DNA fragments of the
same length on the basis of differences in base composition. The resulting banding patterns allow the overall
and objective comparison of microbial communities
from different habitats or under different conditions
and avoid conclusions based on an a priori decision
on the type of bacteria to be analysed [14]. Moreover,
a computer-assisted characterisation of the banding
patterns and the subsequent treatment of the data
using a statistical approach can now be applied and
lead to refined results.
The present study addresses the question whether a
relationship exists between the overall structure of the
intestinal microbial community and its ability to metabolise cholesterol in the human gut. For this purpose,
bacterial communities from 15 human faecal samples
were characterised using PCR–TTGE (temporal temperature gradient gel electrophoresis). Cholesterol-to-coprostanol conversion status was assessed by the
coprostanol-to-cholesterol ratio in faeces, and cholesterol-to-coprostanol reducing bacteria were enumerated by
the most probable number (MPN) estimation in the
same faecal samples. Finally, hierarchical cluster analysis (HCA) and principal component analysis (PCA) of
the TTGE profiles were performed and interpreted with
the variations of the cholesterol-to-coprostanol conversion status.
2. Materials and methods
2.1. Faecal samples
Faeces from 15 healthy human subjects between 25 and
45 years of age were collected. Donors were on a Western
European diet. None had any history of digestive pathology nor had received antibiotic, immunosuppressive or
radiological treatments within the last three months. Donors with hypercholesterolemia (cut-off at 2.5 g l1) were
excluded. Faecal samples were collected in sterile plastic
boxes and kept under anaerobic conditions using an
anaerocult A (Merck, Nogent sur Marne, France) and
stored at 4 C for a maximum of 3 h before processing.
2.2. Most probable number enumeration of
coprostanoligenic bacteria and determination of
cholesterol and coprostanol contents in faeces
The MPN enumerations were performed using the
strictly anaerobic technique of Hungate [21], with dilution
solution and growth medium pre-reduced under O2-free
N2. One g of stool was serially diluted 10-fold with dilution solution (casitone 2.0 g l1; yeast extract 2.0 g l1;
NaCl 5.0 g l1; KH2PO4 1.0 g l1, pH 7.0) up to 1012.
One-millilitre aliquots of each dilution were then transferred in triplicate to the MPN culture tubes containing
9 ml of growth medium (brain heart infusion 10 g l1;
yeast extract 10 g l1; L -cysteine 0.5 g l1; 0.1% aqueous
hemin solution 10 ml l1, pH 7.4), enriched with cholesterol (0.2 g l1, Sigma–Aldrich Chimie) solubilised in
1
L -a-phosphatidylcholine (1.0 g l , Type IV-S, Sigma–
Aldrich Chimie) as previously described [22]. After the
MPN cultures had been incubated at 37 C for 7 days,
neutral sterols were extracted from 1 ml of each culture
with 2 ml of n-hexane by magnetic stirring for 3 h [23].
The samples were centrifuged and the sterols in the hexane
supernatant were analysed by gas chromatography (GC)
as their silyl derivatives [24]. Numbers of coprostanoligenic bacteria were obtained from the highest dilutions
showing bacterial growth associated with coprostanol
production. The MPN results were calculated using a micro-computer program [25] and were expressed as cell
number g1 (fresh stools). Neutral sterols were extracted
from faeces as previously described [26]. Briefly, total lipids from 2-g stool aliquots were extracted with ethanol
for 48 h in a Soxhlet apparatus. Neutral sterols were analysed by GC as their silyl derivatives after they had been
saponified and extracted 3 times with petroleum ether.
Coprostanol and cholesterol contents in faeces were expressed in percentages of total neutral animal sterols.
2.3. DNA isolation, PCR and TTGE analysis
Total DNA was extracted for PCR–TTGE as previously described [27] from 0.2-g stool aliquots frozen at
P. Veiga et al. / FEMS Microbiology Letters 242 (2005) 81–86
2.4. Statistical analyses
Two different and independent approaches were used
to manage the large TTGE data set: (i) HCA, which reduces the number of objects by placing them into
groups, and (ii) PCA, which reduces the dimensionality
of the original data set while retaining those characteristics of the data set that contribute most to its variability
[29]. Gel images were first analysed with the GelCompar
software (version 2.0 Applied Maths, Kortrijk, Belgium). GelCompar software translates TTGE profiles
into densitometric curves whose peak areas are proportional to band intensities. For HCA, the similarity in
shape between two profiles was measured by the Pearson product moment correlation coefficient, which is
based on a whole densitometric curve analysis, taking
into account the number of bands, their position on
the gel, and their intensity. HCA was performed from
similarities values reported in a matrix using the unweighted pair group method with arithmetic averages
(UPGMA). Results are presented as dendrograms,
revealing clustering patterns and relative similarities
among clusters. For PCA, a composite lane of all sample
lanes was generated by the software and band intensities
were reported in a matrix in which rows and columns
contained individuals and all band classes (i.e. points
of migration of bands), respectively. This original data
matrix had 15 rows · 53 columns. To correct for slight
variations in DNA loading between lanes, the intensity
of each band was then divided by the total intensity detected in the considered lane, so that band intensities
were expressed in relative intensity units. PCA was performed from this matrix of relative intensities (where
band classes were considered as active variables) using
the SPAD 4.01 software (DECISIA, Levallois–Perret,
France). Principal components, which altogether represented 100% of the variation in the original data set,
were extracted from the correlation matrix of standardised data. The different extracted PCs were then analysed in terms of correlation with a supplementary
variable representative of the intensity of cholesterolto-coprostanol conversion, i.e. coprostanol-to-cholesterol ratio in fresh stools.
3. Results and discussion
3.1. Relationship between cholesterol-to-coprostanol
conversion status and MPN enumeration of
coprostanoligenic bacteria
The intensity of the cholesterol-to-coprostanol conversion in the human gut was assessed by measuring
the coprostanol-to-cholesterol ratio in the faecal samples. This ratio was found highly variable among the
15 individuals (Fig. 1). As also found in previous studies
[13,30], the conversion patterns in our study were found
to be equally distributed with respect to sex and were
independent of age. Coprostanol was hardly detected
in the faeces of three subjects (C08, C16 and C19) whose
coprostanol-to-cholesterol ratio was less than 0.3. These
results are in accordance with other findings showing
that approximately 20% of subjects, either North Americans consuming a normal mixed Western diet [13], or
omnivorous and vegetarian women [30], or healthy Norwegian subjects [31], have a coprostanol content representing less than one third of their faecal neutral
sterols. In five other subjects (C02, C11, C21, C22, and
C24), the coprostanol-to-cholesterol ratio exceeded 15,
indicating an almost complete cholesterol conversion.
The seven remaining subjects presented incomplete cholesterol conversion patterns with coprostanol-to-cholesterol ratio ranging from 0.9 to 10 (Fig. 1). It has been
suggested that low-to-inefficient conversion could be
due to lack of mucosal receptors for coprostanol-producing bacteria [31] or inhibition of these bacteria by
other members of the gut microbiota [32], but the intestinal microbial communities of both populations have
never been compared. Using the MPN enumeration in
cholesterol-enriched medium, the cultivable coprostanoligenic bacterial community was estimated in the same
faecal samples. This community was found to be present
30
coprostanol-to-cholesterol ratio
70 C. The concentration and integrity of the nucleic
acids were determined visually by electrophoresis on a
1% agarose gel containing ethidium bromide. Primers
GCclamp-U968 (5 0 -GCclamp-GAACGCGAAGAACCTTAC-3 0 ) and L1401 (5 0 -GCGTGTGTACAAGACCC-3 0 ) were used to amplify the V6 to V8 regions of
bacterial 16S rRNA gene [20]. PCR and TTGE were
performed as previously described [22,28]. Gels were
stained in the dark by immersion for 30 min in a solution of SYBR Green I Nucleic Acid Gel Stain (Roche
Diagnostics, GmbH, Mannheim, Germany) and read
on a Storm system (Molecular Dynamics).
83
25
C22
C24
20
C02
C11
C21
15
10
C04
5
C27
C12
C18
C14
C08 C19C16
0
4
5
C15
C25
6
7
log(MPN)
8
9
Fig. 1. Relationship between the MPN estimates of cultivable coprostanoligenic bacteria and the coprostanol-to-cholesterol ratios in fresh
stools of 15 individuals.
84
P. Veiga et al. / FEMS Microbiology Letters 242 (2005) 81–86
at low levels (<106 bacteria g1 (fresh stools)) in the
three low-to-inefficient converters whereas it was part
of the dominant faecal microbiota (P108 bacteria g1
(fresh stools)) in the five highest converters (Fig. 1). Subjects presenting an incomplete cholesterol conversion
pattern harboured intermediate levels of coprostanoligenic bacteria (comprised between 106 and 108 cells g1
(fresh stools)) (Fig. 1). In a previous study, we used
MPN enumeration to estimate for the first time the cultivable coprostanoligenic bacterial community in stools
from a high and a low cholesterol converter [22]. It
was estimated at 9.2 · 108 and 9.2 · 103 cells g1 (fresh
stools), respectively. In the present study, we confirmed
that the rate of cholesterol-to-coprostanol conversion in
the human gut mainly results from the abundance of
coprostanoligenic bacteria in stools. Moreover, our results suggest that the level of coprostanoligenic bacteria
must be at least 106 cells g1 (fresh stools) to efficiently
convert cholesterol in the human gut whereas a population level higher than 108 cells g1 (fresh stools) is
needed to reach a nearly complete conversion.
3.2. Correlation between composition of the dominant
faecal microbiota and coprostanoligenic activity
Faecal microbial communities of the 15 individuals
were profiled by TTGE of amplified 16S rRNA gene
fragments (Fig. 2). Denaturing gel electrophoresis of
the amplified sequences of 16S rRNA or rRNA gene
have been used successfully to characterise or monitor
gut microbial communities in humans [18–20] and have
shown that the dominant bacterial community in human
faeces is stable over time and host specific [20]. To compare samples, the entire densitometric curves for each
lane were numerically compared using the Pearson
product moment correlation coefficient. This coefficient
is robust and objective, since whole curves are compared
thus avoiding subjective band-scoring inherent to bandbased methods like the Dice or Jaccard coefficients [29].
A value of 100 indicates that distribution of the dominant species is identical in the samples. In the present
study, the Pearson coefficients (i.e. similarity values)
for comparison between patterns ranged between 2.8%
and 82.0%, with an average value of 45.1%, confirming
that each individual harbours his own complex faecal
microbiota although some common bands could be observed across different subjects. From the similarity matrix, a cluster analysis using GelCompar software was
applied. It separated TTGE profiles into two main clusters (Fig. 2). Interestingly, the five lowest cholesterol
converters fell into cluster A whereas the eight highest
converters fell into cluster B. The mean faecal coprostanol-to-cholesterol ratio of individuals belonging to each
cluster was then calculated and found to be significantly
higher (P = 0.005) in cluster B (13.94 ± 7.97) than in
cluster A (1.03 ± 1.13). This finding suggested that a
relationship could exist between the overall structure
of the dominant faecal microbiota, as inferred by the
TTGE profiles, and the coprostanoligenic activity.
PCA was then performed for extracting the main
sources of variation in the 15 TTGE profiles, and for
tentatively highlighting those bands which contributed
most to this variation, in relationship with coprostanoligenic activity. PCA is a method that is applicable to a
single treatment, thus allowing statistical analysis of
unreplicated ecosystems and rigorous testing of hypotheses [33]. It has often been used for interpreting DGE
community fingerprinting analysis [29]. The original
TTGE data set (53 different band classes) was compressed to fourteen PCs (Table 1). Because we were
interested in that part of the total variation which was
related to coprostanoligenic activity, the different extracted PCs were then analysed in terms of correlation
with coprostanol-to-cholesterol ratio in fresh stools.
Only PC1 was found to correlate highly with this ratio
(0.79, Table 1). The original data set was therefore
projected onto PC1 and PC2. This is illustrated on
Fig. 3, where diameters of markers are proportional to
coprostanol-to-cholesterol ratio in fresh stools. As expected (because PC1 was highly correlated to coprostanoligenic activity), distribution of individuals along the
PC1 axis coincided with cholesterol-to-coprostanol con-
Fig. 2. Clustering of TTGE profiles obtained with universal primers V6–V8 from 15 faecal samples using Pearson product moment correlation
coefficient and the unweighted pair group clustering method with arithmetic averages (UPGMA). A and B represents the two distinct clusters.
P. Veiga et al. / FEMS Microbiology Letters 242 (2005) 81–86
85
Table 1
Contributions of principal components (PCs) to the overall variation and correlations with coprostanol-to-cholesterol ratio in fresh stools
PC
Individual contribution (%)
Cumulative contribution (%)
Correlation with coprostanol-to-cholesterol ratio
1
2
3
4
5
6
7
8
9
10
11
12
13
14
18.3
13.4
11.3
8.3
7.7
6.6
6.4
5.8
5.0
4.6
3.9
3.7
2.9
2.1
18.3
31.7
43.0
51.4
59.0
65.7
72.0
77.8
82.8
87.4
91.3
95.0
97.9
100.0
0.79
0.22
0.23
0.11
0.10
0.22
0.10
0.00
0.19
0.24
0.08
0.29
0.04
0.09
Therefore, two different and independent ways of
managing (HCA and PCA) the large TTGE data set
converged to demonstrate a relationship between faecal
microbial community structure and cholesterol-to-coprostanol conversion in the human gut. PCA extracted
one PC which was highly correlated with coprostanoligenic activity and further demonstrated that a large subset of bands equally contributed to this PC, indicating
that the correlation observed between coprostanoligenic
activity and TTGE patterns was not due to some
remarkable bacterial species but to the overall structure
of the intestinal microbial community. Although they
have failed so far, classical culture methods will therefore be needed to identify and isolate coprostanoligenic
bacterial species from the human gut. This work is in
progress in our laboratory.
In conclusion, this work constitutes the first application of a multivariate data analysis, integrating structural and metabolic variable data sets, to the gut
microbial community, and revealing a relationship between microbial community structure and function.
We believe that coupling DGE or other fingerprinting
methods with such a statistical approach will make it
possible to define new hypotheses concerning microbial
percent contribution to PC1
version intensity. Indeed, the seven subjects exhibiting a
faecal coprostanol-to-cholesterol ratio higher than five
were grouped on the left side of the PC1 axis whereas
other individuals, including C14 and C27, were on the
opposite side. This result confirmed that there was a
relationship between structure of faecal microbiota
and cholesterol-to-coprostanol conversion. PCA further
returned the contributions of the original active variables (band classes in our case) to PC1, i.e. to that part
of the variation in TTGE profiles which was associated
with coprostanoligenic activity. Interestingly, none of
the 53 band classes preponderantly*** contributed to
PC1 (maximum percent contribution 7.3%), and a large
subset of band classes (20 out of a total of 53) had close
contributions above the mean contribution value (Fig.
4). We therefore concluded that distribution of samples
along the PC1 axis, which is linked to the intensity of
cholesterol-to-coprostanol conversion, could not be correlated with any particular bacterial species (i.e. TTGE
bands), but rather resulted from small and equal contributions of a large subset of band classes.
8
7
6
5
4
3
2
1
0
mean percent contribution = 1.89
1
Fig. 3. Projection of the original TTGE data set onto the first two
principal components PC1 and PC2. Correlation of PC1 with
coprostanol-to-cholesterol ratio was 0.79. Diameters of filled black
circles () are proportional to coprostanol-to-cholesterol ratio.
5
9
13 17 21 25 29 33 37 41 45 49 53
band classes
Fig. 4. Contributions of the 53 original band classes to PC1. PC1 was
the only principal component which correlated highly (0.79) with
coprostanoligenic activity.
86
P. Veiga et al. / FEMS Microbiology Letters 242 (2005) 81–86
community structure and function as fingerprinting patterns can be tested for correlation analysis against any
metabolic or environmental data set.
Acknowledgements
We thank Jérémy Mazet for his advice on PCA interpretation, Christine Young for the English correction,
and Lionel Rigottier-Gois for comments on the
manuscript.
References
[1] Ren, D., Li, L., Schwabacher, A.W., Young, J.W. and Beitz, D.C.
(1996) Mechanism of cholesterol reduction to coprostanol by
Eubacterium coprostanoligenes ATCC 51222. Steroids 61, 33–40.
[2] Brinkley, A.W., Gottesman, A.R. and Mott, G.E. (1982) Isolation
and characterization of new strains of cholesterol-reducing
bacteria from baboons. Appl. Environ. Microbiol. 43, 86–89.
[3] Freier, T.A., Beitz, D.C., Li, L. and Hartman, P.A. (1994)
Characterization of Eubacterium coprostanoligenes sp. nov., a
cholesterol-reducing anaerobe. Int. J. Syst. Bacteriol. 44, 137–142.
[4] Lichtenstein, A.H. (1990) Intestinal cholesterol metabolism. Ann.
Med. 22, 49–52.
[5] Illman, R.J., Storer, G.B. and Topping, D.L. (1993) White wheat
flour lowers plasma cholesterol and increases fecal steroids
relative to whole wheat flour, wheat bran and wheat pollard in
rats. J. Nutr. 123, 1094–1100.
[6] Navder, K.P., Fryer, E.B. and Fryer, H.C. (1992) Neutral sterol
excretion in rats fed skim milk and skim milk yogurt diets. Indian
J. Med. Sci. 46, 275–280.
[7] Reddy, B.S., Engle, A., Simi, B. and Goldman, M. (1992) Effect of
dietary fiber on colonic bacterial enzymes and bile acids in
relation to colon cancer. Gastroenterology 102, 1475–1482.
[8] Panda, S.K., Chattoraj, S.C. and Broitman, S.A. (1999) Correlation of neomycin, faecal neutral and acid sterols with colon
carcinogenesis in rats. Brit. J. Canc. 80, 1132–1136.
[9] Peuchant, E., Salles, C. and Jensen, R. (1987) Relationship
between fecal neutral steroid concentrations and malignancy in
colon cells. Cancer 60, 994–999.
[10] Reddy, S., Sanders, T.A., Owen, R.W. and Thompson, M.H.
(1998) Faecal pH, bile acid and sterol concentrations in premenopausal Indian and white vegetarians compared with white
omnivores. Brit. J. Nutr. 79, 495–500.
[11] van Faassen, A., Bol, J., van Dokkum, W., Pikaar, N.A.,
Ockhuizen, T. and Hermus, R.J. (1987) Bile acids, neutral
steroids, and bacteria in feces as affected by a mixed, a lactoovovegetarian, and a vegan diet. Am. J. Clin. Nutr. 46, 962–967.
[12] Vanhoof, K. and De Schrijver, R. (1998) The influence of enzymeresistant starch on cholesterol metabolism in rats fed on a
conventional diet. Brit. J. Nutr. 80, 193–198.
[13] Wilkins, T.D. and Hackman, A.S. (1974) Two patterns of neutral
steroid conversion in the feces of normal North Americans.
Cancer Res. 34, 2250–2254.
[14] Tannock, G.W. (2001) Molecular assessment of intestinal microflora. Am. J. Clin. Nutr. 73 (2 Suppl), 410S–414S.
[15] Hayashi, H., Sakamoto, M. and Benno, Y. (2002) Phylogenetic
analysis of the human gut microbiota using 16S rDNA clone
libraries and strictly anaerobic culture-based methods. Microbiol.
Immunol. 46, 535–548.
[16] Pace, N.R. (1997) A molecular view of microbial diversity and the
biosphere. Science 276, 734–740.
[17] Amann, R.I., Ludwig, W. and Schleifer, K.H. (1995) Phylogenetic
identification and in situ detection of individual microbial cells
without cultivation. Microbiol. Rev. 59, 143–169.
[18] Tannock, G.W., Munro, K., Harmsen, H.J.M., Welling, G.W.,
Smart, J. and Gopal, P.K. (2000) Analysis of the fecal microflora of
human subjects consuming a probiotic product containing Lactobacillus rhamnosus DR20. Appl. Environ. Microbiol. 66, 2578–2588.
[19] Vanhoutte, T., Huys, G., De Brandt, E. and Swings, J. (2004)
Temporal stability analysis of the microbiota in human feces by
denaturing gradient gel electrophoresis using universal and groupspecific 16S rRNA gene primers. FEMS Microbiol. Ecol. 48, 437–
446.
[20] Zoetendal, E.G., Akkermans, A.D.L. and de Vos, W.M. (1998)
Temperature gradient gel electrophoresis analysis of 16S rRNA
from human fecal samples reveals stable and host-specific
communities of active bacteria. Appl. Environ. Microbiol. 64,
3854–3859.
[21] Hungate, R.E. (1969) A roll tube method for cultivation of strict
anaerobes (Norris, J.R. and Ribbons, D.W., Eds.), Methods in
Microbiology, vol. 3B, pp. 117–132. Academic Press, Inc., New
York.
[22] Gérard, P., Béguet, F., Lepercq, P., Rigottier-Gois, L., Rochet,
V., Andrieux, C. and Juste, C. (2004) Gnotobiotic rats harboring
human intestinal microbiota as a model for studying cholesterolto-coprostanol conversion. FEMS Microbiol. Ecol. 47, 337–343.
[23] Hoskin, G.P. and Bandler, R. (1987) Identification of mammalian
feces by coprostanol thin layer chromatography: method development. J. Assoc. Off. Anal. Chem. 70, 496–498.
[24] Catala, I., Juste, C., Boehler, N., Férézou, J., André, M., Riottot,
M., Lutton, C., Lafont, H., Bornet, F. and Corring, T. (2000)
Cholesterol crystallization in gall-bladder bile of pigs given
cholesterol-b-cyclodextrin-enriched diets with either casein or
soyabean concentrate as protein sources. Brit. J. Nutr. 83, 411–420.
[25] Clarke, K.R. and Owens, N.J.P. (1983) A simple and versatile
micro-computer program for the determination of Ômost probable
numberÕ. J. Microbiol. Meth. 1, 133–137.
[26] Riottot, M., Olivier, P., Huet, A., Caboche, J.J., Parquet, M.,
Khallou, J. and Lutton, C. (1993) Hypolipidemic effects of bcyclodextrin in the hamster and in the genetically hypercholesterolemic Rico rat. Lipids 28, 181–188.
[27] Godon, J.-J., Zumstein, E., Dabert, P., Habouzit, F. and Moletta,
R. (1997) Molecular microbial diversity of an anaerobic digestor
as determined by small-subunit rDNA sequence analysis. Appl.
Environ. Microbiol. 63, 2802–2813.
[28] Seksik, P., Rigottier-Gois, L., Gramet, G., Sutren, M., Pochart,
P., Marteau, P., Jian, R. and Doré, J. (2003) Alterations of the
dominant faecal bacterial groups in patients with CrohnÕs disease
of the colon. Gut 52, 237–242.
[29] Fromin, N., Hamelin, J., Tarnawski, S., Roesti, D., JourdainMiserez, K., Forestier, N., Teyssier-Cuvelle, S., Gillet, F.,
Aragno, M. and Rossi, P. (2002) Statistical analysis of denaturing
gel electrophoresis (DGE) fingerprinting patterns. Environ.
Microbiol. 4, 634–643.
[30] Korpela, J.T. and Adlercreutz, H. (1985) Fecal neutral sterols in
omnivorous and vegetarian women. Scand. J. Gastroenterol. 20,
1180–1184.
[31] Midtvedt, T., Lingaas, E., Carlstedt-Duke, B., Höverstad, T.,
Midtvedt, A.-C., Saxerholt, H., Steinbakk, M. and Norin, K.E.
(1990) Intestinal microbial conversion of cholesterol to coprostanol in man. Influence of antibiotics. APMIS 98, 839–844.
[32] Sadzikowski, M.R., Sperry, J.F. and Wilkins, T.D. (1977)
Cholesterol-reducing bacterium from human feces. Appl. Environ. Microbiol. 34, 355–362.
[33] Dollhopf, S.L., Hashsham, S.A. and Tiedje, J.M. (2001) Interpreting 16S rDNA T-RFLP data: application of self-organizing
maps and principal component analysis to describe community
dynamics and convergence. Microb. Ecol. 42, 495–505.