APPLIED AND ENVIRONMENTAL MICROBIOLOGY, June 2009, p. 4175–4184
Copyright © 2009, American Society for Microbiology. All Rights Reserved.
Vol. 75, No. 12
Diet-Induced Metabolic Improvements in a Hamster Model of
Hypercholesterolemia Are Strongly Linked to Alterations
of the Gut Microbiota?†
Ine ´s Martínez,1‡ Grant Wallace,1‡ Chaomei Zhang,1Ryan Legge,1Andrew K. Benson,1
Timothy P. Carr,2Etsuko N. Moriyama,3and Jens Walter1*
Department of Food Science and Technology1and Department of Nutrition and Health Sciences,2University of Nebraska,
Lincoln, Nebraska 68583-0919, and School of Biological Sciences and Center for Plant Science Innovation,
University of Nebraska, Lincoln, Nebraska 68588-01183
Received 16 February 2009/Accepted 21 April 2009
The mammalian gastrointestinal microbiota exerts a strong influence on host lipid and cholesterol metab-
olism. In this study, we have characterized the interplay among diet, gut microbial ecology, and cholesterol
metabolism in a hamster model of hypercholesterolemia. Previous work in this model had shown that grain
sorghum lipid extract (GSL) included in the diet significantly improved the high-density lipoprotein (HDL)/
non-HDL cholesterol equilibrium (T. P. Carr, C. L. Weller, V. L. Schlegel, S. L. Cuppett, D. M. Guderian, Jr.,
and K. R. Johnson, J. Nutr. 135:2236-2240, 2005). Molecular analysis of the hamsters’ fecal bacterial popu-
lations by pyrosequencing of 16S rRNA tags, PCR-denaturing gradient gel electrophoresis, and Bifidobacte-
rium-specific quantitative real-time PCR revealed that the improvements in cholesterol homeostasis induced
through feeding the hamsters GSL were strongly associated with alterations of the gut microbiota. Bifidobac-
teria, which significantly increased in abundance in hamsters fed GSL, showed a strong positive association
with HDL plasma cholesterol levels (r ? 0.75; P ? 0.001). The proportion of members of the family Coriobac-
teriaceae decreased when the hamsters were fed GSL and showed a high positive association with non-HDL
plasma cholesterol levels (r ? 0.84; P ? 0.0002). These correlations were more significant than those between
daily GSL intake and animal metabolic markers, implying that the dietary effects on host cholesterol metab-
olism are conferred, at least in part, through an effect on the gut microbiota. This study provides evidence that
modulation of the gut microbiota-host metabolic interrelationship by dietary intervention has the potential to
improve mammalian cholesterol homeostasis, which has relevance for cardiovascular health.
The mammalian gut microbiota interacts intimately with its
host, affecting both host metabolic and immunological pheno-
types with important consequences for health (18, 22, 32).
Recent studies have revealed complex linkages between the
gut microbiome and host metabolism, with the microbes ex-
erting effects on the energy balance by influencing glucose and
lipid metabolism (2, 7, 28). This intimate metabolic relation-
ship is most likely the consequence of a long coevolutionary
process that resulted in a mutualistic relationship between the
host and its microbial partners (25). However, life in industri-
alized societies has introduced profound changes into the hu-
man environment (e.g., diet, antibiotics, hospital deliveries,
hygiene, etc.) that are markedly different from the conditions
to which humans have evolved and that are likely to have
occurred too abruptly for the human microbiome to adjust.
Consequently, aberrations of the gut microbiota induced
through lifestyle factors could be relevant to the etiology of
several complex human diseases whose occurrence has mark-
edly increased in developed countries. Interestingly, imbal-
ances in the gut microbiota have been reported for obesity,
type 1 and 2 diabetes, some allergies, and inflammatory bowel
diseases in humans and animal models (7, 24, 43, 45, 48). The
connection between gut bacteria and disease suggests an in-
triguing paradigm on how to view and potentially treat com-
plex diseases. Specific bacterial populations in the intestine
could be pharmaceutical targets to maintain or restore meta-
bolic functions (6, 17).
Coronary heart disease (CHD) continues to be a major
cause of death in developed countries and is another example
of a “western disease” that is less common in underdeveloped
countries but increases in frequency with adoption of western
customs (4). Most risk factors for CHD (obesity, high blood
pressure, type 2 diabetes, heredity, high cholesterol, and diet)
have been linked to the gut microbiota (7, 17, 20, 30, 45), and
gut bacteria have been suggested to play a role in the etiology
of cardiovascular disease (16, 33). Cholesterol metabolism is a
key factor in susceptibility to CHD, and as early as 1959, it has
been shown that germfree rats have higher serum cholesterol
concentrations than their conventional counterparts do (12).
Several mechanisms have been proposed by which gut bacteria
could influence host cholesterol metabolism (13). Bacterial
conversions of bile acids (such as the formation of secondary
bile acids) are likely to play a role, as they affect enterohepatic
circulation, de novo synthesis of bile acids, emulsification, and
cholesterol absorption (10, 28, 30). A further mechanism by
which gut bacteria might influence cholesterol metabolism is
* Corresponding author. Mailing address: Department of Food Sci-
ence and Technology, University of Nebraska, 333 Food Industry
Complex, Lincoln, NE 68583-0919. Phone: (402) 472-2615. Fax: (402)
472-1693. E-mail: firstname.lastname@example.org.
† Supplemental material for this article may be found at http://aem
‡ These authors contributed equally to this work.
?Published ahead of print on 1 May 2009.
through Fiaf (fasting-induced adipocyte factor), which is selec-
tively suppressed in the intestinal epithelium by the gut micro-
biota (1, 2). Fiaf is an important regulator of lipid metabolism
(e.g., through its inhibition of lipoprotein lipase) and has been
shown to increase total cholesterol and high-density lipopro-
tein (HDL) cholesterol levels when overexpressed in trans-
genic mice (26).
There are several reasons why hamsters are an excellent
model for studying the metabolic relationships among diet,
cholesterol metabolism, and gut microbiota in relation to
health. First, hamsters are omnivorous, and their blood lipid
profiles respond to diets in a predictive manner similar to
humans (5). Second, unlike mice and rats which lack choles-
terol ester transfer protein, hamsters exhibit all of the enzy-
matic pathways in lipoprotein and bile metabolism that are
also present in humans. They exhibit limited hepatic synthesis
of cholesterol and bile acids, resulting in more relevant data
when extrapolating to humans (23). Third, hamsters develop
atherosclerosis in a predictive manner in response to dietary
Using the Golden Syrian hamster model, Carr and cowork-
ers have shown that the hexane-extractable lipid fraction of
grain sorghum whole kernels (GSL), when included in the
hamsters’ diet, leads to a significant reduction of plasma non-
HDL and liver cholesterol levels while increasing HDL cho-
lesterol levels (8). We extended this research and performed a
comprehensive molecular characterization of the fecal micro-
biota of the hamsters by pyrosequencing of 16S rRNA tags,
denaturing gradient gel electrophoresis (DGGE), and Bi-
fidobacterium specific quantitative real-time PCR (qRT-PCR)
in order to test whether metabolic effects of GSL were asso-
ciated with specific modifications of the gut microbiota.
MATERIALS AND METHODS
Animal experiments. The fecal samples analyzed here were obtained during a
previous study that determined the effect of GSL included in the diet on the
cholesterol metabolism of hamsters, and the handling of animals, feed compo-
sition, GSL composition, sample collection, and preparation have been described
previously (8). Briefly, groups of seven or eight male F1B Syrian hamsters (Bio
Breeders, Watertown, MA) were housed in cages (each hamster in an individual
cage) and kept at 25°C with a 12-h light–12-h dark cycle. Hamsters were fed a
modified AIN-93 M diet (37) supplemented with 0%, 1% and 5% grain sorghum
lipid extract at the expense of cornstarch. Daily feed intake was determined to
assess individual GSL ingestions. GSL was prepared from whole kernels ob-
tained from a mixture of commercial red grain sorghum hybrids grown in Ne-
braska. Hamsters had free access to food and water throughout the study. After
the hamsters were on their respective diets for 3 weeks, the complete fecal output
for each hamster was collected over 7 days. The fecal samples were ground,
weighed, and stored frozen at ?80°C until the DNA extractions were performed.
DNA extraction. The fecal samples were diluted in ice-cold phosphate-buff-
ered saline (pH 7) in a 1:10 ratio and centrifuged at 8,000 ? g for 5 min. This
washing step was repeated two times. Bacterial cell pellets were resuspended in
750 ?l lysis buffer (200 mM NaCl, 100 mM Tris [pH 8.0], 20 mM EDTA, 20
mg/ml lysozyme) and transferred to a microcentrifuge tube containing 300 mg of
0.1-mm zirconium beads (BioSpec Products). Samples were then incubated at
37°C for 20 min followed by the addition of 85 ?l of 10% sodium dodecyl sulfate
solution and 40 ?l proteinase K (15 mg/ml). After incubation for 15 min at 60°C,
500 ?l of phenol-chloroform-isoamyl alcohol (25:24:1) was added, and the sam-
ples were homogenized in a MiniBeadbeater-8 (BioSpec Products) at maximum
speed for 2 min. Samples were cooled on ice before the layers were separated by
centrifugation at 10,000 ? g for 5 min. The top layer was extracted twice with
phenol-chloroform-isoamyl alcohol (25:24:1) and twice with chloroform-isoamyl
alcohol; DNA was recovered by standard ethanol precipitation. The DNA pellets
were dried for 30 min at room temperature and later resuspended in 100 ?l of
Tris-HCl buffer (10 mM, pH 8.0).
Analysis of the gut microbiota of hamsters by pyrosequencing of 16S rRNA
tags. The V1-V3 region of the 16S rRNA gene was amplified by PCR using
bar-coded universal primers 8F and 518R containing the A and B sequencing
adaptors (454 Life Sciences). The forward primer (A-8FM) was 5?-gcctccctcgcg
ccatcagAGAGTTTGATCMTGGCTCAG-3? where the sequence of the A adap-
tor is shown in lowercase letters. The reverse primer (B-518) was 5?-gccttgccag
cccgctcagNNNNNNNNATTACCGCGGCTGCTGG-3? where the sequence of
the B adaptor is shown in lowercase letters and N represents an eight-base bar
code that is unique for each sample. Prior to sequencing, amplicons from the
individual PCR samples were quantified using the Quant-iT PicoGreen double-
stranded DNA assay (Invitrogen) and quality controlled on an Agilent 2100
bioanalyzer. The amplicons from each reaction mixture were mixed in equal
amounts based on concentration and subjected to emulsion PCR, and amplicon
libraries were generated as recommended by 454 Life Sciences. Sequencing was
performed from the B end using the 454/Roche B sequencing primer kit using a
Roche Genome Sequencer GS-FLX using the standard protocol. Samples were
combined in a single region of the picotiter plate such that approximately 1,000
to 2,000 sequences were obtained from each animal. The data analysis pipeline
removed low-quality sequences (i) that do not perfectly match the PCR primer
at the beginning of a read, (ii) that are shorter than 200 bp in length, (iii) that
contain more than two undetermined nucleotides (N), or (iv) that do not match
a bar code. Sequences (1,000 to 2,000 per animal) were quality controlled and
binned according to bar codes.
Taxonomy-based analyses were performed by assigning taxonomic status to
each sequence using the CLASSIFIER program of the Ribosomal Database
Project (47). To estimate species richness and diversity, taxonomy-independent
methods were used. Sequences were aligned using Infernal Aligner; sequences
from individual animals and then pooled sequences from all animals of a single
treatment group were aligned. Cluster analysis was performed using the com-
plete linkage clustering algorithm available through the Pyrosequencing pipeline
of the Ribosomal Database Project (9). Clustering was done with a 97% cutoff
for inclusion into an operational taxonomic unit (OTU) and was performed on
alignments of sequences from individual animals. The number of species and
species richness were estimated by further sampling-based (rarefaction) analyses
of OTU data and of calculated Shannon diversity indices.
GGGGGGACTCCTACGGGAGGCAGCAG3?) and PRUN518r (5?-ATTACC
GCGGCTGCTGG-3? (34), which amplify the V3 region of the 16S rRNA gene.
DGGE was performed by the method of Walter and coworkers (46) using a
DCode universal mutation detection system (Bio-Rad, Hercules, CA). DNA
bands in the DGGE gel were visualized by standard ethidium bromide staining
and photographed using the InGenius gel documentation system (Syngene, Fre-
derick, MD). DGGE images were analyzed using BioNumerics software version
5.0 (Applied Maths, Kortrijk, Belgium). Bands were manually assigned, and the
normalized banding patterns were used to generate distance matrices by calcu-
lating the Pearson product moment correlation coefficients for all pair-wise
combinations of patterns. This approach compares profiles in a pair-wise manner
based on the entire densitometric curve, therefore accounting for both band
position and intensity. DGGE fingerprints were transformed to peak profiles
using the BioNumerics software, and the intensities of individual bands were
determined as a percentage of the peak surface area relative to the surface area
of the entire molecular fingerprint of the sample. To determine the effects of
feeding hamsters GSL, normalized fragment intensities of all bands in DGGE
fingerprints were determined and compared for the feeding groups.
In order to identify species represented by bands detected by DGGE, bands of
fecal fingerprints from two or three animals were excised from the gel, purified,
and reamplified by the method of ben Omar and Ampe (3), and cloned using the
TOPO TA Cloning kit for sequencing (Invitrogen). Plasmids were isolated from
three transformants per band using the QIAprep spin minprep kit (Qiagen), and
inserts were sequenced by a commercial provider following the manual of the
cloning kit. Closest relatives of the partial 16S rRNA sequences were determined
using the nucleotide blast web tool at the NCBI website (http://blast.ncbi.nlm
.nih.gov/Blast.cgi) and the Seqmatch web tool provided through the Ribosomal
Database Project (http://rdp.cme.msu.edu/seqmatch/seqmatch_intro.jsp). A phy-
logenetic tree was generated from the consensus sequence of the F bands in
three individual animals using the unweighted-pair group method using average
linkages and neighbor-joining algorithms in the MEGA4 software package (42).
There were a total of 178 positions in the final data set, and the evolutionary
distances were computed by using the Kimura two-parameter method and are
reported as the number of base substitutions per site.
Specific quantification of bifidobacteria by qRT-PCR. Quantification of total
bifidobacteria was performed by quantitative real-time PCR using primers Bif-
4176MARTI´NEZ ET AL.APPL. ENVIRON. MICROBIOL.
For (5?-TCGCGTCYGGTGTGAAAG-3? and BifRev (5?-CCACATCCAGCR
TCCAC-3?) (39). PCRs were performed using a Mastercycler Realplex2 (Ep-
pendorf AG, Hamburg, Germany). Each PCR was done in a 25-?l volume. The
reaction mixture comprised 11.25 ?l of the 20? SYBR solution and 2.5? Real-
MasterMix (Eppendorf AG, Hamburg, Germany), 0.5 ?M of each primer, and 1
?l of DNA template. The amplification program consisted of an initial denatur-
ation step of 5 min at 95°C, followed by 35 cycles, where 1 cycle consisted of 15 s
at 95°C (denaturation), 20 s at 58°C (annealing), and 30 s at 68°C (extension),
and fluorescence at each step was measured. To control the specificity of the
amplifications, a melting curve was done consisting of a denaturation step of 15 s
at 95°C, an increase from 58°C to 95°C over a 20-min period, and a final step of
15 s at 95°C. Cultures of B. animalis ATCC 25527Tand B. infantis ATCC 15697T
were used to generate standard curves for absolute quantification of bifidobac-
teria in the fecal samples. Bacterial counts of overnight cultures (12 h) were
determined by plate counting, and a 10-fold dilution series was performed in
phosphate-buffered saline buffer for each strain. DNA was isolated from indi-
vidual samples of the dilution series using the method for fecal samples. Standard
curves were made by plotting the threshold cycle values obtained from DNA of
the dilution series as a linear function of the base 10 logarithm of the number of
bifidobacteria. Two individual qRT-PCR runs of all fecal DNA templates in
duplicate were performed, and means of all four values were used for the
analysis. Despite the use of two different strains of bifidobacteria to generate one
standard curve, its correlation coefficient r2was ?0.96.
To quantify the Bifidobacterium animalis-like phylotype detected by DGGE,
we used a specific primer (Bh1) based on a highly variable region of the sequence
of fragment F in the DGGE gel (5?GGCAGGGGGTTTTCCTC3?). This primer
was used in combination with primer BifRev (39) used for the Bifidobacterium
genus-specific qRT-PCR. PCR was performed as described above. The specific-
ity of the PCR was tested using DNA isolated from fecal samples from 10 human
subjects and DNA from B. animalis ATCC 25527Tand Bifidobacterium infantis
ATCC 15697T. The PCRs all gave negative results with the primer combination
Bh1 and BifRev and positive results with primers BifFor and BifRev (data not
shown). As we had no cultural representative of the phylotype represented by
band F, we used a standard curve generated as described above with B. animalis
ATCC 25527Tand B. infantis ATCC 15697Tand primers BifFor and BifRev.
Although the standard curve was generated with a different forward primer, it
can be assumed that no significant bias is introduced, as the efficiencies of the two
PCR systems were virtually identical (0.51 for primers BifFor and BifRev and
0.56 for primers Bh1 and BifRev).
Correlation analysis of gut microbiota-host metabolic functional relation-
ships. Correlation analysis between metabolic host parameters and bacterial
populations at different taxonomic levels was performed by the method of Cani
and coworkers (7). Metabolic parameters included in the correlation analysis
were the levels of cholesterol absorption, fecal cholesterol, plasma total choles-
terol, plasma HDL cholesterol, plasma non-HDL cholesterol, liver total choles-
terol, liver-free cholesterol, liver triglycerides, liver phospholipids, and liver-
esterified cholesterol. The determination of these metabolic phenotypes and the
methods applied were reported previously (8).
Genome comparisons. The web-based Integrated Genomics Platform of the
Joint Genome Institute (JGI) was used to identify functions enriched in bi-
fidobacteria (27). The Abundance Profile Search was used to identify clusters of
orthologous groups of proteins (COGs) that were more abundant in individual
bifidobacterial genomes (Bifidobacterium adolescentis ATCC 15703, Bifidobacte-
rium adolescentis L2-32, Bifidobacterium animalis subsp. lactis HN019, Bifidobac-
terium dentium ATCC 27678, Bifidobacterium longum DJO10A, and Bifidobac-
terium longum NCC2705) compared to the genomes of a selection of bacteria
commonly present in the mammalian gastrointestinal tract (Bacteroides caccae
ATCC 43185, Bacteroides capillosus ATCC 29799, Bacteroides fragilis NCTC
9343, Bacteroides fragilis YCH46, Bacteroides ovatus ATCC 8483, Bacteroides
stercoris ATCC 43183, Bacteroides thetaiotaomicron VPI-5482, Bacteroides uni-
formis ATCC 8492, Bacteroides vulgatus ATCC 8482, Clostridium acetobutylicum
ATCC 824, Clostridium bartlettii DSM 16795, Clostridium bolteae ATCC BAA-
613, Clostridium leptum DSM 753, Clostridium ramosum DSM 1402, Clostridium
sp. strain L2-50, Clostridium sp. strain SS2/1, Clostridium thermocellum ATCC
27405, Collinsella aerofaciens ATCC 25986, Coprococcus eutactus ATCC 27759,
Dorea formicigenerans ATCC 27755, Dorea longicatena DSM 13814, Enterobacter
sp. strain 638, Enterococcus faecalis V583, Enterococcus faecium DO, Escherichia
coli K-12, Eubacterium dolichum DSM 3991, Eubacterium siraeum DSM 15702,
Eubacterium ventriosum ATCC 27560, Faecalibacterium prausnitzii M21/2, Lac-
tobacillus reuteri 100-23, Lactobacillus reuteri F275, Lactobacillus salivarius subsp.
salivarius UCC118, Methanobrevibacter smithii ATCC 35061, Parabacteroides dis-
tasonis ATCC 8503, Parabacteroides merdae ATCC 43184, Peptostreptococcus
micros ATCC 33270, Providencia stuartii ATCC 25827, Ruminococcus gnavus
ATCC 29149, Ruminococcus obeum ATCC 29174, Ruminococcus torques ATCC
27756, and Salmonella enterica serovar Typhimurium LT2). Functions associated
with lipid metabolism were specifically selected from the enriched COGs and
added to the function list. A function profile of these COGs was then generated
for all of the gut species. Please refer to the IMG web page for details (http:
Statistical analysis. Results are presented as means ? standard deviations
(SDs). Statistical tests for treatment effects of the GSL on the abundance of
individual taxonomic ranks or DGGE band intensities were performed by one-
way analysis of variance (ANOVA) analysis followed by Tukey’s posthoc multi-
ple comparison tests. The Mann-Whitney test was used to compare Shannon
diversity indices of gut populations. Correlations between metabolic parameters
and bacterial populations were assessed by Pearson’s correlation test using
GraphPad Prism version 5.00 (GraphPad Software, San Diego, CA).
Characterization of the hamster gut microbiota by pyrose-
quencing of 16S rRNA tags. To determine whether propor-
tional changes of the gut microbiota were associated with the
effects of GSL on cholesterol metabolism in hamsters, we an-
alyzed the fecal microbiota of hamsters fed 0% (n ? 7), 1%
(n ? 7), and 5% (n ? 7) GSL by pyrosequencing of the V3
region of the 16S rRNA gene. A total of 34,424 sequences were
studied; the average sequence length was around 250 bp, and
an average of 1,639 sequences per animal were studied. Tax-
onomy-based analysis showed that the composition of the ham-
ster gut microbiota at the phylum level is similar to that of
humans and mice, being dominated by Firmicutes and Bacte-
roidetes. An unusual feature of the hamster gut microbiota,
however, was that Firmicutes comprised the vast majority of the
taxa (94%) with Bacteroidetes making up only 4% of the pop-
ulation. The remaining bacteria belonged to the phyla Verru-
comicrobia and Actinobacteria (each representing around 1%
of the total sequence tags) and Proteobacteria and candidate
division TM7 (0.07% and 0.024% of sequences, respectively).
At the family level, the predominant groups in hamsters of
the control group were the Erysipelotrichaceae, Eubacteriaceae,
Ruminococcaceae, and Lactobacillaceae, represented by an av-
erage of 59%, 19%, 13% and 5% of the total fecal microbiota,
respectively (Fig. 1A). Of the bacterial groups on the genus
level, the most dominant were unclassified Erysipelotrichaceae,
Allobaculum, unclassified Eubacteriaceae, Ruminococcus, and
Lactobacillus, comprising 31%, 28%, 19%, 11%, and 5% of the
total sequence pool on average in control animals, respectively
(Fig. 1B). With the exception of Allobaculum, these genera are
also shared with the gut microbiota reported for mice, humans,
and primates (15, 24, 29, 44). As shown in Fig. S1 in the
supplemental material, pyrosequencing revealed high animal-
to-animal variability on both the family and genus levels.
Effects of GSL on specific taxa of the hamster gut micro-
biota. Sequence proportions determined by pyrosequencing
were used to establish the effects of the GSL on the gut mi-
crobiota composition. To identify specific taxa that were af-
fected by the feeding treatments, the proportions of taxa in
each rank of each animal were tested for treatment effects. As
shown in Table 1, ANOVA identified one family, the Coriobac-
teriaceae (P ? 0.042), and two bacterial groups at the genus
level, unclassified members of the family Erysipelotrichaceae
(P ? 0.0016) and genus Pseudoramibacter (P ? 0.017), as being
significantly affected by the inclusion of GSL to the hamsters’
diet. Moreover, values for the genus Allobaculum (P ? 0.096)
VOL. 75, 2009EFFECT OF DIET ON GUT MICROBIOTA IN HAMSTERS4177
and unclassified members of the family Coriobacteriaceae (P ?
0.064) approached statistical significance.
Taxonomy-independent analysis of the hamster gut micro-
biota from individual animals showed that with a conservative
level of 97% identity as a cutoff for OTUs, nearly 200 OTUs
were observed in the average of 1,600 sequences from each
animal (Fig. 1C). Individual animals in the 5% GSL feeding
group showed a trend toward fewer OTUs in the rarefaction
curves. The Shannon diversity indices from individual animals
also showed a trend of fewer OTUs in animals fed 5% GSL
(blue lines) compared to animals fed 0% GSL (Fig. 1D).
Grouping of the samples by GSL showed significant differences
between 0% and 5% GSL (P ? 0.0001, Mann-Whitney test).
Thus, 5% GSL had the effect of reducing the diversity of the
DGGE analysis of fecal microbiota of hamsters fed GSL. To
validate the findings obtained with pyrosequencing, fecal bac-
terial populations of the hamsters were also analyzed by PCR-
DGGE. The DGGE gel is shown in Fig. 2A, and the results of
analysis of the gel are presented in Table 2. Feeding the ham-
FIG. 1. Characterization of the gut microbiota composition of hamsters fed different amounts of GSL as determined by pyrosequencing of 16S
rRNA tags (V3 region). Composition of the gut microbiota of hamsters fed 0%, 1%, and 5% GSL (n ? 7 per group) at the family level (A) and
the genus level (B). (C) Rarefaction curves of OTUs from sequences of fecal samples from individual hamsters fed 0% GSL (red), 1% GSL
(green), and 5% GSL (blue). (D) Shannon diversity indices of the gut microbiota of individual hamsters fed 0% GSL (red), 1% GSL (green), and
5% GSL (blue). OTUs were identified using 97% cutoffs for rarefaction and Shannon diversity indices.
4178MARTI´NEZ ET AL.APPL. ENVIRON. MICROBIOL.
sters 5% GSL significantly increased the staining intensity of
band C (P ? 0.037) and band F (P ? 0.011). Band A showed
a high animal-to-animal variation, and no consistent impact of
GSL was detected. Sequence analysis of amplicons from these
bands revealed that they represent bacteria related to Rumi-
nococcus bromii (band A), Allobaculum stercoricanis (band C),
and Bifidobacterium animalis (band F). Phylogenetic compar-
ison of the sequence from band F with the closest hits in the
RDP database revealed this sequence to be most similar to
Bifidobacterium animalis, and it is referred to as the Bifidobac-
terium animalis-like phylotype in this article (a phylogenetic
tree is shown in Fig. 2B).
Although providing less depth, DGGE analysis showed good
agreement with the results from pyrosequencing. Both meth-
ods detected the increase of bacteria related to Allobaculum
and the high animal-to-animal variability of bacteria related to
Ruminococcus. Relative species quantification obtained with
pyrosequencing showed high correlations with staining inten-
sities of DGGE bands representing the same bacterial groups
(for Ruminococcus, r ? 0.94; for Allobaculum, r ? 0.81; both
P ? 0.0001) (see Fig. S2 in the supplemental material). This is
remarkable, as both methods are only semiquantitative and
entail multitemplate PCR susceptible to PCR bias. The main
difference between the findings by pyrosequencing and DGGE
was in the proportions of bifidobacteria, where a significant
proportion could be detected only by DGGE, while only 0.03%
of the total sequences obtained by pyrosequencing corre-
sponded to bifidobacteria. The difference can be explained by
the use of primer 8F in pyrosequencing, which shows three
mismatches with the 16S rRNA genes from five Bifidobacte-
rium species for which whole-genome sequences were available
(data not shown). Accordingly, many studies employing direct
analysis of 16S rRNA genes to study the human gut micro-
biota and using the primer 8F, which is one of the most
commonly used primers for such approaches, resulted in a
significant underrepresentation of Bifidobacterium species
(15, 35, 41, 50).
Quantification of bifidobacteria using qRT-PCR. Since
primer 8F resulted in an underrepresentation of bifidobacteria
in pyrosequencing and to confirm and quantify the bifidogenic
effect of the GSL detected by DGGE analysis, a specific qRT-
PCR procedure was used to determine the numbers of total
bifidobacteria and the Bifidobacterium animalis-like phylotype.
As shown in Fig. 2C and D, qRT-PCR analysis showed a
significant increase in cell numbers of total bifidobacteria (P ?
0.012) and the Bifidobacterium animalis-like phylotype (P ?
0.019). As shown in Fig. 2E, the numbers of bifidobacteria
from individual hamsters were highly variable, but a significant
correlation between cell numbers and daily GSL intake was
Bifidobacteria and Coriobacteriaceae showed high correla-
tions with important markers of host cholesterol metabolism.
In a previous study using the animals studied here, dietary GSL
reduced cholesterol absorption, plasma non-HDL cholesterol
concentrations, and liver esterified cholesterol levels, while
raising plasma HDL cholesterol levels (8). To determine
whether alterations of the gut microbiota in hamsters fed GSL
were associated with an improvement in cholesterol metabo-
lism, a correlation analysis was used to determine correlations
between all bacterial taxa at different taxonomic levels and host
metabolic phenotypes. The analysis revealed highly positive
correlations between HDL plasma concentrations and total
bifidobacteria (r ? 0.75; P ? 0.0011), between HDL plasma
concentrations and Bifidobacterium animalis-like phylotype
(r ? 0.77; P ? 0.0009), among total Coriobacteriaceae and
non-HDL plasma concentrations (r ? 0.84; P ? 0.0002), and
between unclassified Coriobacteriaceae and both non-HDL
plasma concentration (r ? 0.82; P ? 0.0004) and cholesterol
absorption (r ? 0.71; P ? 0.0042). These high correlations
were observed only in animals fed 1% and 5% GSL, and
inclusion of the values from control animals significantly
reduced correlations (Table 3). Graphs showing the highest
correlations between bacterial taxa and metabolic pheno-
types are shown in Fig. 3, and a metabolic network diagram
linking GSL, bacterial phylotypes, and host cholesterol me-
tabolism is shown in Fig. 4. Interestingly, the correlations
between bifidobacteria and HDL cholesterol concentration
and between Coriobacteriaceae and non-HDL concentration
showed higher significance than correlations achieved be-
tween GSL intake and the respective host metabolic pheno-
Genome comparisons of bifidobacteria and other gut organ-
categories in 47 genomes of gut bacteria, we observed that pro-
teins belonging to the COG clusters COG0400 (predicted ester-
ase), COG0657 (esterase/lipase), and COG2272 (carboxylester-
ase type B) are enriched in six Bifidobacterium genomes (see
Table S1 in the supplemental material). Carboxylesterases repre-
sented by COG0400 and COG2272 belong to enzymes that hy-
drolyze a wide variety of substrates, ranging from methylcaplyrate
to p-nitrobenzyl (21, 49).
TABLE 1. Abundance of bacterial groups in the fecal microbiota of
hamsters that changed by including GSL in the diet as determined
by pyrosequencing of 16S rRNA tagsa
Abundance of bacterial groupb(% of total
sequences obtained with sample
?mean ? SD?) in hamsters fed:
0% GSL1% GSL5% GSL
1.22 ? 0.79 0.79 ? 0.63
0.31 ? 0.33c
27.82 ? 15.9
0.11 ? 0.09
30.63 ? 15.8
0.48 ? 0.3c
43.55 ? 7.0e
0.35 ? 0.21c
of the following
1.0 ? 0.7
31.0 ? 7.4
0.69 ? 0.62
32.7 ? 14.0
0.23 ? 0.31e
12.39 ? 6.0d
aThere were seven hamsters in each group.
bValues that were significantly different or approaching statistical significance
are shown in boldface type.
cStatistically significantly different from the value for hamsters fed 0% GSL
(P ? 0.05) by ANOVA.
dThis value was statistically significantly different from the value for hamsters
fed 0% GSL (P ? 0.01) and from the value for hamsters fed 1% GSL (P ? 0.01)
eApproaching statistical significance compared to the value for hamsters fed
0% GSL (P ? 0.1) by ANOVA.
VOL. 75, 2009 EFFECT OF DIET ON GUT MICROBIOTA IN HAMSTERS4179
In humans, CHD is associated with high levels of low-density
lipoprotein and low levels of high-density lipoprotein. The
characterization of the gut microbiota in a hamster model of
hypercholesterolemia showed that dietary intervention with
GSL had a major impact on the composition of the gut micro-
biota and that these modulations were highly associated with
improvements in the HDL and non-HDL cholesterol equilib-
rium. With consumption of GSL, population levels of bi-
fidobacteria increased and showed a strong positive association
with increases in HDL cholesterol levels. In contrast, relative
abundance of members of the family Coriobacteriaceae de-
creased with feeding the hamsters GSL, and a high positive
correlation with non-HDL cholesterol and cholesterol absorp-
tion was discovered. The findings indicate that GSL intake
influences the HDL/non-HDL equilibrium, at least in part,
through an alteration of the gut microbiota. We infer this
because correlation coefficients between bifidobacterial and
Coriobacteriaceae populations and plasma cholesterol concen-
trations were higher than associations among GSL intake, host
phenotypes, and cholesterol absorption (Fig. 4). In addition, if
bacterial phylotype/host phenotype correlations were merely a
result of GSL affecting both bacterial taxa and cholesterol
metabolism independently, one would assume that all bacterial
taxa whose abundance correlated with GSL intake would show
an association with host phenotypes. However, much lower
correlation coefficients with non-HDL and HDL plasma con-
centrations were observed between relative abundance of un-
classified members of the family Erysipelotrichaceae and the
genus Allobaculum, although these taxa showed significant as-
FIG. 2. Impact of GSL on the gut microbiota composition of hamsters fed 0% GSL (n ? 7), 1% GSL (n ? 7), and 5% GSL (n ? 8) as
determined by DGGE and qRT-PCR. (A) DGGE showing fingerprints of DNA isolated from the fecal samples of hamsters. Lanes 1 to 32 contain
DNA from individual hamsters. Lane M contains markers from reference strains. Bands C and F showed significant increases in staining intensity
in fecal fingerprints of hamsters fed 5% GSL. The bands A, C, and F marked by an arrow were excised, purified, and sequenced (Table 2).
(B) Phylogenetic tree of DGGE band F with sequences that revealed highest similarities in GenBank. The tree was inferred using the
unweighted-pair group method using average linkages, and the percentage of replicate trees in which the associated taxa clustered together in the
bootstrap test (1,000 replicates) are shown next to the branches. A neighbor-joining tree resulted in essentially the same phylogeny (data not
shown). (C) Cell numbers of total bifidobacteria in hamster fecal samples as determined by qRT-PCR. (D) Quantification of the Bifidobacterium
animalis-like phenotype detected by DGGE in hamster fecal samples by qRT-PCR. (E) Correlation of cell numbers of bifidobacteria with daily
4180MARTI´NEZ ET AL.APPL. ENVIRON. MICROBIOL.
sociations with GSL intake (Table 3). However, it should be
considered that bifidobacteria and Coriobactericeae are just
two of hundreds of groups, and other bacteria, independent of
GSL administration, are likely to interact with host cholesterol
Changes in the composition of the hamsters’ gut microbiota
induced by GSL consumption were limited to a relatively small
number of bacterial groups (Table 1 and 2). These composi-
tional adjustments had the net effect of reducing the overall
species richness (number of individual species per unit popu-
lation). However, the overall composition of the microbiota at
the phylum level was not affected by GSL. The reason for this
finding was that increases of dominant bacterial taxa were
“balanced” by a reduction of related bacteria, leaving the rel-
ative proportions of higher taxonomic taxa unaffected. Al-
lobaculum belongs to the family Erysipelotrichaceae, and un-
characterized bacteria of this family declined as Allobaculum
increased with feeding the hamsters GSL. Thus, the overall
proportions within the family Erysipelotrichaceae and the phy-
lum Firmicutes were not changed. Similar findings were ob-
tained for the phylum Actinobacteria, where numbers of bi-
fidobacteria increased while the abundance of members of the
family Coriobacteriaceae declined. As shown in Fig. S3 in the
supplemental material, significant inverse correlations were
obtained between these related bacterial groups in individual
animals. Similar diet-induced compositional adjustments of the
gut microbiota that maintain the overall composition at higher
taxonomic levels have also been observed in humans. For ex-
ample, the decline of bacteria belonging to the Roseburia and
Eubacterium rectale groups induced through reduced carbohy-
drate intake was balanced by an increase in the number of
related members of the Clostridium coccoides cluster in human
fecal samples (14). Furthermore, the diet of human infants
appears to influence the Bifidobacterium/Coriobacteriaceae ra-
tio, with higher numbers of bifidobacteria in breast-fed infants
while there were higher numbers of coriobacteria when the
infants were fed formula (19). Collectively, these findings in-
dicate that homeostatic reactions that restore the overall equi-
TABLE 2. Ratio of staining intensities of major bands as a proportion of total fingerprint intensity and results of sequence analysis of
Mean band intensitya(SD) in DNA from
Closest GenBank hitd
Closest type straine
0% GSL1% GSL 5% GSL
A 20.0 (17.5)
14.2 (14.3) AM265444, uncultured bacteria, clone
EU777003, uncultured bacterium
clone molerat_aai70g11 (92.4–93.0)
AB186296, Bifidobacterium animalis
strain DBF 1307 (96.8)
Ruminococcus bromii ATCC 27255T
Allobaculum stercoricanis DSM 13633T
Bifidobacterium animalis subsp. animalis
DSM 10140T?X89513? (96.8)
aRatio of staining intensities of major bands as a proportion of total fingerprint intensity (shown as a percentage). Values that were significantly different are shown
in boldface type.
bStatistically significantly different from the value for hamsters fed 0% GSL (P ? 0.05) by ANOVA.
cStatistically significantly different from the value for hamsters fed 1% GSL (P ? 0.05) by ANOVA.
dThe GenBank accession number and species or clone is shown. The values in parentheses are the percentages of similarity.
eThe closest type strain is shown first. The GenBank accession number is shown in brackets. The values in parentheses are the percentages of similarity.
fND, not determined.
TABLE 3. Correlations between abundance of bacterial taxa and markers of cholesterol metabolisma
Correlationb(r value) between abundance of bacterial taxa and the following marker of cholesterol metabolism:
GSL intakeNon-HDL level HDL level
Family level Coriobacteriaceae
?0.56 (?0.18)0.68 (0.50)
Unclassified members of the
aValues for animals fed 1% and 5% GSL are presented.
bValues for all animals, including control animals, are presented in parentheses. Correlation coefficients with an r of ?0.7 are shown in boldface type.
VOL. 75, 2009 EFFECT OF DIET ON GUT MICROBIOTA IN HAMSTERS4181
librium of the gut microbiota are often a natural consequence
of compositional changes induced through diet. Nevertheless,
as indicated by the correlation analysis in our study, an alter-
ation of the gut microbiota at lower taxonomic levels is still
likely to have important functional consequences for the host.
The mechanisms by which bifidobacteria and coriobacteria
affect cholesterol metabolism remain an important field of
future research. Including GSL in the diet reduced cholesterol
absorption efficiency, which was directly correlated with non-
HDL cholesterol concentration (8). The high correlations of
unclassified members of the family Coriobacteriaceae with both
non-HDL cholesterol and cholesterol absorption suggest that
these bacteria could have a negative impact on cholesterol
homeostasis through increasing cholesterol absorption. Bi-
fidobacteria, on the other hand, showed high positive correla-
tion with HDL cholesterol levels and no association with cho-
FIG. 3. Specific bacterial populations in the guts of hamsters show high associations with both cholesterol metabolic phenotypes and GSL
intake. (A and B) Correlations between cell numbers of total bifidobacteria (A) and the Bifidobacterium animalis-like phenotype (B) with HDL
cholesterol. (C) Correlation between proportion of Coriobacteriaceae and non-HDL cholesterol. (D) Correlation between unclassified members
of the family Coriobacteriaceae and cholesterol absorption. Data from control animals (0% GSL) were excluded from the analysis.
p = 0.0015
(mainly a phylotype
related to B. animalis)
p = 0.0026
p = 0.12
Plasma HDL cholesterol
p = 0.0078
p = 0.054
r = 0.75
p = 0.0011
p = 0.0065
p < 0.0001
p = 0.0039
r = -0.56
p = 0.037
p = 0.0038
p = 0.016
p = 0.0003
r = 0.84
p = 0.0002
p = 0.011
FIG. 4. Metabolic network showing the associations between daily GSL intake, gut microbiota composition, and host cholesterol metabolism
in hamsters fed 0%, 1%, and 5% GSL. Results of the correlations of cell numbers of bifidobacteria and proportions of Coriobacteriaceae and
phenotypic markers were obtained with data from animals fed 1% and 5% GSL. Red connections indicate a positive correlation, while blue
connections show correlations that are inverse. Green connections show associations with no statistical significance. Metabolic data were obtained
by Carr and coworkers in a previous study (8).
4182MARTI´NEZ ET AL.APPL. ENVIRON. MICROBIOL.
lesterol absorption. Bifidobacteria have been shown to affect
cholesterol and lipid metabolism in animal models when ad-
ministered as probiotics or when stimulated by prebiotics (11,
13). The mechanism by which bifidobacteria achieve these ef-
fects remain speculative, but they might impact cholesterol
metabolism indirectly by suppressing numbers of Coriobacteri-
aceae. For both bacterial groups, the capability to transform
bile acids has been reported (38), and this phenotypic trait
might influence host cholesterol metabolism through an im-
pact on enterohepatic circulation. The strong correlations be-
tween bacterial taxa and cholesterol metabolism were observed
only in animals fed GSL and not in control animals, suggesting
that GSL influences the relative abundance of these organisms
as well as metabolic characteristics.
The consumption of lipids has not yet been associated with
increases in numbers of intestinal bifidobacteria. In contrast,
Cani and coworkers (7) showed that a high-fat diet significantly
lowered the number of bifidobacteria in mice. The composition
of the GSL administered to the hamsters in our study con-
tained not only mono-, di-, and triglycerides but also esters,
alcohols, and other lipophilic compounds, such as waxes, ste-
rols, and polycosanols (8). Carbohydrates or fiber are an un-
likely explanation for the bifidogenic effect of GSL, as the
amounts of fiber in hexane extracts of grains are negligible (8).
Interestingly, genome comparisons revealed that bifidobacteria
possess metabolic capacities that could allow them to utilize
complex lipids, including lipids that may not be utilized by
other members of the gut microbiota or the host. Schell and
coworkers (40) detected four genes encoding long-chain fatty
acyl-coenzyme A synthetases in the genome of Bifidobacterium
longum, more than any other prokaryote genome available at
that time, except for Streptomyces coelicolor and another gas-
trointestinal tract inhabitant, Bacteroides fragilis. These find-
ings together with the enrichment of putative esterases in bi-
fidobacterial genomes detected above indicate that bifidobacteria
are likely to utilize specific components of GSL leading to the
increase in numbers in the gut.
bacteriaceae equilibrium to be important for the plasma cho-
lesterol levels in hamsters, with bifidobacteria being beneficial
and coriobacteria being detrimental. While extrapolation of
our observations to humans is still speculative, our findings
suggest that bifidobacteria and coriobacteria could be potential
targets for the prevention of metabolic aberrations that play a
role in CHD. Clearly, it will be essential to first identify the
exact bacterial taxa within the human gut microbiota that have
strong correlations to cholesterol metabolism, which in itself is
a challenging task. Unlike the inbred population of hamsters
used in our study, human subjects have significant genetic
diversity, and genetic factors that affect cholesterol metabolism
play a more important role in humans than in the animal
model. Furthermore, human subjects follow individual life-
styles and consume different diets, and they harbor variable
and individual communities of the gut bacteria. All these fac-
tors will hamper the identification of bacterial contributors to
human cholesterol metabolism. Nevertheless, it is tempting to
speculate that the positive impact of breast-feeding on the
Bifidobacterium/Coriobacteriaceae ratio in human infants (19)
could be responsible for the higher HDL cholesterol levels
observed in adults that were breast-fed in infancy (36).
This study provides new and important perspectives on di-
etary modulation of the mammalian gut microbiota and its
effects on the host. The findings indicate that a complex mix-
ture of lipids can exert a “prebiotic” effect that leads to im-
provements in host cholesterol metabolism. In conclusion, this
study provided evidence that modulation of bacterial popula-
tions in the gut has the potential to improve mammalian
cholesterol homeostasis, which has relevance in the preven-
tion of CHD.
We thank the members of the University of Nebraska—Lincoln
Nutraceutical Team and especially Curtis Weller, Vicki Schlegel, and
Susan Cuppett for their contributions to the hamster feeding trial. We
thank Ty Nguyen for programming the pyrosequencing data analysis
Grant Wallace was supported by the UCARE Program of the Uni-
versity of Nebraska. This study was funded in part by the Nebraska
Grain Sorghum Board.
1. Backhed, F., H. Ding, T. Wang, L. V. Hooper, G. Y. Koh, A. Nagy, C. F.
Semenkovich, and J. I. Gordon. 2004. The gut microbiota as an environmen-
tal factor that regulates fat storage. Proc. Natl. Acad. Sci. USA 101:15718–
2. Backhed, F., J. K. Manchester, C. F. Semenkovich, and J. I. Gordon. 2007.
Mechanisms underlying the resistance to diet-induced obesity in germ-free
mice. Proc. Natl. Acad. Sci. USA 104:979–984.
3. ben Omar, N., and F. Ampe. 2000. Microbial community dynamics during
production of the Mexican fermented maize dough pozol. Appl. Environ.
4. Bickler, S. W., and A. DeMaio. 2008. Western diseases: current concepts and
implications for pediatric surgery research and practice. Pediatr. Surg. Int.
5. Bravo, E., A. Cantafora, and G. Ortu. 1994. Why prefer the Golden Syrian
hamster (Mesocritus auratus) to the Wistar rat in experimental studies on
plasma lipoprotein metabolism? Comp. Biochem. Physiol. 107B:347–355.
6. Cani, P. D., and N. M. Delzenne. 2007. Gut microflora as a target for energy
and metabolic homeostasis. Curr. Opin. Clin. Nutr. Metab. Care 10:729–734.
7. Cani, P. D., A. M. Neyrinck, F. Fava, C. Knauf, R. G. Burcelin, K. M. Tuohy,
G. R. Gibson, and N. M. Delzenne. 2007. Selective increases of bifidobacteria
in gut microflora improve high-fat-diet-induced diabetes in mice through a
mechanism associated with endotoxaemia. Diabetologia 50:2374–2383.
8. Carr, T. P., C. L. Weller, V. L. Schlegel, S. L. Cuppett, D. M. Guderian, Jr.,
and K. R. Johnson. 2005. Grain sorghum lipid extract reduces cholesterol
absorption and plasma non-HDL cholesterol concentration in hamsters. J.
9. Cole, J. R., Q. Wang, E. Cardenas, J. Fish, B. Chai, R. J. Farris, A. S.
Kulam-Syed-Mohideen, D. M. McGarrell, T. Marsh, G. M. Garrity, and
J. M. Tiedje. 2009. The Ribosomal Database Project: improved alignments
and new tools for rRNA analysis. Nucleic Acids Res. 37:D141–D145.
10. Cowles, R. L., J. Y. Lee, D. D. Gallaher, C. L. Stuefer-Powell, and T. P. Carr.
2002. Dietary stearic acid alters gallbladder bile acid composition in ham-
sters fed cereal-based diets. J. Nutr. 132:3119–3122.
11. Crittenden, R., and M. J. Playne. 2006. Modifying the human intestinal
microbiota with prebiotics, p. 285–314. In A. C. Ouwehand and E. E.
Vaughan (ed.), Gastrointestinal microbiology. Taylor & Francis, New
12. Danielsson, H., and B. Gustafsson. 1959. On serum-cholesterol levels and
neutral fecal sterols in germ-free rats: bile acids and steroids 59. Arch.
Biochem. Biophys. 83:482–485.
13. Delzenne, N. M., P. D. Cani, and A. M. Neyrinck. 2008. Prebiotics and lipid
metabolism. In J. Versalovic and M. Wilson (ed.), Therapeutic microbiology:
probiotics and related strategies. ASM Press, Washington, DC.
14. Duncan, S. H., G. E. Lobley, G. Holtrop, J. Ince, A. M. Johnstone, P. Louis,
and H. J. Flint. 2008. Human colonic microbiota associated with diet, obesity
and weight loss. Int. J. Obes. (London) 32:1720–1724.
15. Eckburg, P. B., E. M. Bik, C. N. Bernstein, E. Purdom, L. Dethlefsen, M.
Sargent, S. R. Gill, K. E. Nelson, and D. A. Relman. 2005. Diversity of the
human intestinal microbial flora. Science 308:1635–1638.
16. Fava, F., J. A. Lovegrove, R. Gitau, K. G. Jackson, and K. M. Tuohy. 2006.
The gut microbiota and lipid metabolism: implications for human health and
coronary heart disease. Curr. Med. Chem. 13:3005–3021.
17. Flint, H. J., S. H. Duncan, K. P. Scott, and P. Louis. 2007. Interactions and
competition within the microbial community of the human colon: links
between diet and health. Environ. Microbiol. 9:1101–1111.
VOL. 75, 2009 EFFECT OF DIET ON GUT MICROBIOTA IN HAMSTERS4183
18. Gordon, H. A., and L. Pesti. 1971. The gnotobiotic animal as a tool in the
study of host microbial relationships. Bacteriol. Rev. 35:390–429.
19. Harmsen, H. J., A. C. Wildeboer-Veloo, J. Grijpstra, J. Knol, J. E. Degener,
and G. W. Welling. 2000. Development of 16S rRNA-based probes for the
Coriobacterium group and the Atopobium cluster and their application for
enumeration of Coriobacteriaceae in human feces from volunteers of differ-
ent age groups. Appl. Environ. Microbiol. 66:4523–4527.
20. Holmes, E., R. L. Loo, J. Stamler, M. Bictash, I. K. Yap, Q. Chan, T. Ebbels,
M. De Iorio, I. J. Brown, K. A. Veselkov, M. L. Daviglus, H. Kesteloot, H.
Ueshima, L. Zhao, J. K. Nicholson, and P. Elliott. 2008. Human metabolic
phenotype diversity and its association with diet and blood pressure. Nature
21. Hong, K. H., W. H. Jang, K. D. Choi, and O. J. Yoo. 1991. Characterization
of Pseudomonas fluorescens carboxylesterase: cloning and expression of the
esterase gene in Escherichia coli. Agric. Biol. Chem. 55:2839–2845.
22. Hooper, L. V., and J. I. Gordon. 2001. Commensal host-bacterial relation-
ships in the gut. Science 292:1115–1118.
23. Horton, J. D., J. A. Cuthbert, and D. K. Spady. 1995. Regulation of hepatic
7 alpha-hydroxylase expression and response to dietary cholesterol in the rat
and hamster. J. Biol. Chem. 270:5381–5387.
24. Ley, R. E., F. Backhed, P. Turnbaugh, C. A. Lozupone, R. D. Knight, and J. I.
Gordon. 2005. Obesity alters gut microbial ecology. Proc. Natl. Acad. Sci.
25. Ley, R. E., D. A. Peterson, and J. I. Gordon. 2006. Ecological and evolution-
ary forces shaping microbial diversity in the human intestine. Cell 124:837–
26. Mandard, S., F. Zandbergen, E. van Straten, W. Wahli, F. Kuipers, M.
Muller, and S. Kersten. 2006. The fasting-induced adipose factor/angiopoi-
etin-like protein 4 is physically associated with lipoproteins and governs
plasma lipid levels and adiposity. J. Biol. Chem. 281:934–944.
27. Markowitz, V. M., E. Szeto, K. Palaniappan, Y. Grechkin, K. Chu, I. M.
Chen, I. Dubchak, I. Anderson, A. Lykidis, K. Mavromatis, N. N. Ivanova,
and N. C. Kyrpides. 2008. The integrated microbial genomes (IMG) system
in 2007: data content and analysis tool extensions. Nucleic Acids Res. 36:
28. Martin, F. P., M. E. Dumas, Y. Wang, C. Legido-Quigley, I. K. Yap, H. Tang,
S. Zirah, G. M. Murphy, O. Cloarec, J. C. Lindon, N. Sprenger, L. B. Fay,
S. Kochhar, P. van Bladeren, E. Holmes, and J. K. Nicholson. 2007. A
top-down systems biology view of microbiome-mammalian metabolic inter-
actions in a mouse model. Mol. Syst. Biol. 3:112.
29. McKenna, P., C. Hoffmann, N. Minkah, P. P. Aye, A. Lackner, Z. Liu, C. A.
Lozupone, M. Hamady, R. Knight, and F. D. Bushman. 2008. The macaque
gut microbiome in health, lentiviral infection, and chronic enterocolitis.
PLoS Pathog. 4:e20.
30. Midtvedt, T. 1999. Microbial functional activities, p. 79–96. In L. A. Hanson
and R. H. Yolken (ed.), Probiotics, other nutritional factors, and intestinal
microflora, vol. 42. Lippincott-Raven Publishers, Philadelphia, PA.
31. Mitchell, P. L., and R. S. McLeod. 2008. Conjugated linoleic acid and
atherosclerosis: studies in animal models. Biochem. Cell Biol. 86:293–301.
32. Nicholson, J. K., E. Holmes, and I. D. Wilson. 2005. Gut microorganisms,
mammalian metabolism and personalized health care. Nat. Rev. Microbiol.
33. Ordovas, J. M., and V. Mooser. 2006. Metagenomics: the role of the micro-
biome in cardiovascular diseases. Curr. Opin. Lipidol. 17:157–161.
34. Øvreås, L., L. Forney, F. L. Daae, and V. Torsvik. 1997. Distribution of
bacterioplankton in meromictic Lake Saelenvannet, as determined by dena-
turing gradient gel electrophoresis of PCR-amplified gene fragments coding
for 16S rRNA. Appl. Environ. Microbiol. 63:3367–3373.
35. Palmer, C., E. M. Bik, D. B. Digiulio, D. A. Relman, and P. O. Brown. 2007.
Development of the human infant intestinal microbiota. PLoS Biol. 5:e177.
36. Parikh, N. I., S.-J. Hwang, E. Ingelsson, E. J. Benjamin, C. S. Fox, R. S.
Vasan, and J. M. Murabito. 2007. Abstract 3498: the association of breast-
feeding in infancy and adult cardiovascular disease risk factors: the Fram-
ingham Third Generation Cohort. Circulation 116(Suppl.):II_792.
37. Reeves, P. G., F. H. Nielsen, and G. C. Fahey, Jr. 1993. AIN-93 purified diets
for laboratory rodents: final report of the American Institute of Nutrition ad
hoc writing committee on the reformulation of the AIN-76A rodent diet. J.
38. Ridlon, J. M., D. J. Kang, and P. B. Hylemon. 2006. Bile salt biotransfor-
mations by human intestinal bacteria. J. Lipid Res. 47:241–259.
39. Rinttila, T., A. Kassinen, E. Malinen, L. Krogius, and A. Palva. 2004.
Development of an extensive set of 16S rDNA-targeted primers for quanti-
fication of pathogenic and indigenous bacteria in faecal samples by real-time
PCR. J. Appl. Microbiol. 97:1166–1177.
40. Schell, M. A., M. Karmirantzou, B. Snel, D. Vilanova, B. Berger, G. Pessi,
M. C. Zwahlen, F. Desiere, P. Bork, M. Delley, R. D. Pridmore, and F.
Arigoni. 2002. The genome sequence of Bifidobacterium longum reflects its
adaptation to the human gastrointestinal tract. Proc. Natl. Acad. Sci. USA
41. Suau, A., R. Bonnet, M. Sutren, J. J. Godon, G. R. Gibson, M. D. Collins,
and J. Dore. 1999. Direct analysis of genes encoding 16S rRNA from com-
plex communities reveals many novel molecular species within the human
gut. Appl. Environ. Microbiol. 65:4799–4807.
42. Tamura, K., J. Dudley, M. Nei, and S. Kumar. 2007. MEGA4: Molecular
Evolutionary Genetics Analysis (MEGA) software version 4.0. Mol. Biol.
43. Tannock, G. W. 2008. The search for disease-associated compositional shifts
in bowel bacterial communities of humans. Trends Microbiol. 16:488–495.
44. Turnbaugh, P. J., F. Backhed, L. Fulton, and J. I. Gordon. 2008. Diet-
induced obesity is linked to marked but reversible alterations in the mouse
distal gut microbiome. Cell Host Microbe 3:213–223.
45. Turnbaugh, P. J., R. E. Ley, M. A. Mahowald, V. Magrini, E. R. Mardis, and
J. I. Gordon. 2006. An obesity-associated gut microbiome with increased
capacity for energy harvest. Nature 444:1027–1031.
46. Walter, J., G. W. Tannock, A. Tilsala-Timisjarvi, S. Rodtong, D. M. Loach,
K. Munro, and T. Alatossava. 2000. Detection and identification of gastro-
intestinal Lactobacillus species by using denaturing gradient gel electro-
phoresis and species-specific PCR primers. Appl. Environ. Microbiol. 66:
47. Wang, Q., G. M. Garrity, J. M. Tiedje, and J. R. Cole. 2007. Naïve Bayesian
classifier for rapid assignment of rRNA sequences into the new bacterial
taxonomy. Appl. Environ. Microbiol. 73:5261–5267.
48. Wen, L., R. E. Ley, P. Y. Volchkov, P. B. Stranges, L. Avanesyan, A. C.
Stonebraker, C. Hu, F. S. Wong, G. L. Szot, J. A. Bluestone, J. I. Gordon, and
A. V. Chervonsky. 2008. Innate immunity and intestinal microbiota in the
development of type 1 diabetes. Nature 455:1109–1113.
49. Zock, J., C. Cantwell, J. Swartling, R. Hodges, T. Pohl, K. Sutton, P. Ros-
teck, Jr., D. McGilvray, and S. Queener. 1994. The Bacillus subtilis pnbA
gene encoding p-nitrobenzyl esterase: cloning, sequence and high-level ex-
pression in Escherichia coli. Gene 151:37–43.
50. Zoetendal, E. G., M. Rajilic-Stojanovic, and W. M. de Vos. 2008. High-
throughput diversity and functionality analysis of the gastrointestinal tract
microbiota. Gut 57:1605–1615.
4184 MARTI´NEZ ET AL.APPL. ENVIRON. MICROBIOL.