RES E AR C H Open Access
Gut microbiota dysbiosis contributes to the
development of hypertension
Jing Li
1,2,3†
, Fangqing Zhao
4†
, Yidan Wang
1†
, Junru Chen
5†
, Jie Tao
6†
, Gang Tian
7
, Shouling Wu
8
, Wenbin Liu
5
,
Qinghua Cui
9
, Bin Geng
1
, Weili Zhang
1
, Ryan Weldon
10
, Kelda Auguste
10
, Lei Yang
11
, Xiaoyan Liu
11
, Li Chen
10,12,13
,
Xinchun Yang
2,3*
, Baoli Zhu
14,15*
and Jun Cai
1*
Abstract
Background: Recently, the potential role of gut microbiome in metabolic diseases has been revealed, especially in
cardiovascular diseases. Hypertension is one of the most prevalent cardiovascular diseases worldwide, yet whether
gut microbiota dysbiosis participates in the development of hypertension remains largely unknown. To investigate
this issue, we carried out comprehensive metagenomic and metabolomic analyses in a cohort of 41 healthy
controls, 56 subjects with pre-hypertension, 99 individuals with primary hypertension, and performed fecal
microbiota transplantation from patients to germ-free mice.
Results: Compared to the health y control s, we found dramatically decreased microbial richness and diversity, Prevotella-
dominated gut enterotype, distinct metagenomic composition with reduced bacteria associated with healthy status and
overgrowth of bacteria such as Prevotella and Klebsiella, and disease-linked microbial function in both pre-hypertensive and
hypertensive populations. Unexpectedly, the microbiome characteristic in pre-hypertension group was quite similar to that
in hypertension. The metabolism changes of host with pre-hypertension or hypertension were identified to be closely
linked to gut microbiome dysbiosis. And a disease classifier based on microbiota and metabolites was constructed to
discriminate pre-hypertensive and hypertensive individuals from controls accurately. Furthermore, by fecal transplantation
from hypertensive human donors to germ-free mice, elevated blood pressure was observed to be transferrable through
microbiota, and the direct influence of gut microbiota on blood pressure of the host was demonstrated.
Conclusions: Overall, our results describe a novel causal role of aberrant gut microbiota in contributing to the
pathogenesis of hypertensio n. And the significance of early intervention for pre-hypertension was emphasized.
Keywords: Hypertension, Pre-hypertension, Gut microbiota, Metabolism, Fecal transplant
Background
In recent decades, the potential role of the gut mi-
crobiome in altering health status of the hosts ha s drawn
considerable attention. Emerging evidence suggests a
link between gut microbiome and various diseases,
including colorectal cancer, liver cirrhosis, arthritis, type
2 diabetes, and atherosclerosis [1–5]. A number of mi-
crobial biomarkers specific to these diseases have been
discovered, and fecal microbiome-targeted strategies are
recommended to be a powerful tool for early diagnosis
and treatment of different diseases.
More importantly, by fecal transfer experiment and
gut microbiota (GM) remodeling, intestinal microbiome
has been further indicated to conduce to the pathogen-
esis of multiple diseases such as obesity, depressive dis-
order, chronic ileal inflammation, liver diseases, and
atherosclerosis [6–12]. Specific mechanisms underlying
the causal function of GM have been revealed. For ex-
ample, the metabolism by intestinal microbiota of dietary
L-carnitine, a nutrient in red meat, was demonstrated to
* Correspondence: yxc6229@sina.com; zhubaoli@im.ac.cn;
caijun@fuwaihospital.org
†
Equal contributors
2
Department of Cardiology, Beijing ChaoYang Hospital, Capital Medical
University, Beijing 100020, China
14
CAS Key Laboratory of Pathogenic Microbiology and Immunology, Institute
of Microbiology, Chinese Academy of Sciences, Beijing 100101, China
1
Hypertension Center, Fuwai Hospital, State Key Laboratory of Cardiovascular
Disease of China, National Center for Cardiovascular Diseases of China,
Chinese Academy of Medical Sciences and Peking Union Medical College,
Beijing 100037, China
Full list of author information is available at the end of the article
© The Author(s). 2017 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0
International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and
reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to
the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver
(http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
Li et al. Microbiome (2017) 5:14
DOI 10.1186/s40168-016-0222-x
promote atherosclerosis and lead to cardiovascular disease
risk via producing trimethylamine and trimethylamine-N-
oxide [12]. Targeting gut microbial production of trimethy-
lamine specifically and non-lethal microbial inhibitors were
confirmed to relieve diet -induced atherosclerotic lesion
development [13]. Thus GM may serve as a potential thera-
peutic approach for the treatment of cardiovascular and
metabolic diseases.
Hypertension (HTN) has become a global public health
concern and a major risk factor for cardiovascular, cere-
brovascular, and kidney diseases [14, 15]. It is believed that
the etiology of HTN depends on the complex interplay of
both genetic and environmental factors [16, 17], and the
precise cause of this morbidity has not been elucidated to
date. It has been suggested that the germ-free (GF) mice,
in which the intestinal bacteria is completely absent,
present relatively lower blood pressure (BP) when com-
pared to conventional mice [18]. And therefore we sus-
pected that GM might have the potential to regulate BP.
Most recently, many lines of seminal evidence, which
for the first time demonstrate that aberrant gut micro-
bial community are linked to BP changes of the host,
support this hypothesis. For example, disordered GM as
a result of decreased microbial richness, diversity, even-
ness, and increased Firmicutes/Bacteroidetes ratio was
reported in hypertensive animals and seven HTN pa-
tients, as sequenced by 16S ribosomal RNA [19]. In Dahl
rats, distinct metagenomic composition have been re-
vealed between salt-sensitive and salt-resistant strains,
and the GM of salt-sensitive rats was suggested to be in
a symbiotic relationship with the host [20]. In addition,
by rat models of HTN and meta-analyses in randomized
human clinical trials, investigators have revealed that ad-
ministration of probiotics can reduce BP [21, 22]. This
drove us to speculate that the alteration in GM by pro-
biotic use may lead to BP changes. Furthermore, it has
been proved that transplantation of cecal contents from
hypertensive obstructive sleep apnea rats on high-fat diet
into recipient rats on normal chow diet lead to higher
BP levels, and a major contributor to the gut dysbiosis
of obstructive sleep apnea-induced HTN is high-fat diet
[23]. These studies have emphasized a strong correlation
between gut dysbiosis and HTN, and further implied the
significance of GM in BP regulation, yet animal models
could not perfectly substitute human disease, and the
sample size of human participants for microbial analysis
was quite limited.
In consideration of the BP levels being classified into
optimal, pre-hypertension (pHTN), and HTN according
to the most recent clinical guidelines [24], it remains ob-
scure how exactly the composition of gut microbes and
the products of microbial fermentation change in human
patients with HTN, espe cially in pHTN populations. In
addition, decisive evidence is still needed to determine
whether gut dysbiosis is a consequence or an important
causal factor for the pathogenesis of HTN. Fecal trans-
plantation from human samples into GF mice is re-
quired to uncover the involvement of GM dysbiosis in
pathophysiology of HTN. Collectively, these key issues
are the major goal of the present study.
To addr ess the questions above, we performed deep
metagenomic sequencing of stool samples from 196 par-
ticipants of healthy control, pHTN, and HTN; took
metabolomic analyses of their metabolic profiles, further
constructed a disea se classifier for pHTN and HTN
based on GM and metabolites; and demonstrated the
crucial role of disordered GM in triggering thigh BP by
human fecal microbiota transplantation into GF mice.
Results
GM diversity and enterotype in pHTN and HTN
To identify whether gut microbial changes are associated
with HTN, we performed shotgun metagenomic sequen-
cing of fecal samples from a cohort of 196 Chinese indi-
viduals. The cohort consisted of 41 healthy controls, 56
subjects with pHTN, and 99 patients with primary HTN.
All the participants were from a cohort study among
employees of the Kailuan Group Corporatio n. The Kai-
luan study is a prospective cohor t study focusing on the
Kailuan community in Tangshan, a large modern city in
northern China. All the subjects in the hypertension
group were newly diagnosed hypertensive patients prior
to antihypertensive treatment. Patients suffering from
cancer, heart failure, renal failure, smoking, stroke, per-
ipheral artery disease, and chronic inflammatory disease
were all excluded. Drugs including statins, aspirin, insu-
lin, metformin, nifedipine, and metoprolol were not used
on the patients, and other drug consumption was not
compared because the sample size was quite small. Indi-
viduals were also excluded if they had received antibi-
otics or probiotics within the last 8 weeks. Other than
SBP and DBP, there was no significant difference in
other clinical parameters among groups, except for fast-
ing blood glucose level (FBG) (P = 0.026, C vs H;
Kruskal-Wallis test, Additional file 1: Table S1). Bacterial
DNA was extracted from stool samples, sequenced on
the Illumina platform, and a total of 1211 Gb 125-bp
paired-end reads were generated, with an average of
6.18 ± 1.43 (s.d.) million reads per sample (Add itional file
2: Table S2). For each sample, a majority of high-quality
sequencing reads (83.74–97.24%) were de novo a ssem-
bled into long contigs or scaffolds, which were used for
gene prediction, taxonomic classification, and functional
annotation.
To characterize the bacterial richness, rarefaction ana-
lysis was performed by randomly sampli ng 100 times
with replacement and estimating the total number of
genes that could be identified from these samples. The
Li et al. Microbiome (2017) 5:14 Page 2 of 19
curve in each group was near saturation, which sug-
gested the sequenci ng data were great enough with very
few new genes undetected. The rate of acquisition of
new genes in control samples rapidly outpaced new gene
acquisition in disease samples, suggesting lower levels of
gene richness in the pHTN and HTN groups (Fig. 1a).
The number of genes in both pHTN and HTN groups
were significantly decreased as compared to the controls
(P = 0.024, C vs P; P = 0.04, C vs H; Kruskal-Wallis test,
Fig. 1b). Shannon index based on the genera profile was
calculated to estimate the within-sample (α) diversity.
Consistently, the α diversity at the genus level was mu ch
lower in pHTN and HTN groups (P = 0.023, C vs P; P =
0.016, C vs H; Kruskal-Wallis test, Fig. 1c). The reduced
richness of genes and genera in the GM of pHTN and
HTN groups is consistent with previous findings [19],
suggesting possible deficiency of healthy microflora in
hypertensive patients.
To explore the difference between the microbial commu-
nities at different stages of HTN, enterotypes were identi-
fied based on the abundance of genera using Partitioning
Around Medoid (PAM) clustering method. The optimal
number of enterotypes was two as indicated by Calinski-
Harabasz (CH) index (Additional file 3: Figure S1). Then
Principal Coordinate Analysis (PCoA) using Jensen-
Shannon distance was performed to cluster the 196 sam-
ples into two distinct enterotypes (Fig. 1d). Prevotella was
the most enriched genus in enterotype 1; Bacteroides was
a
def
bc
Fig. 1 Decreased diversity and shift of gut enterotypes in human adults with pHTN and HTN. a Rarefaction curves for gene number in control (n =41),
pHTN (n = 56), and HTN (n = 99) after 100 random sampling. The curve in each group is near smooth when the sequencing data are great enough
with few new genes undetected. b, c Comparison of the microbial gene count and α diversity (as accessed by Shannon index) based on the genera
profile in the three groups. C, control; P, pHTN; H, HTN. P =0.024, C vs P; P = 0.04, C vs H; for gene c ount. P = 0.023, C vs P; P = 0.016, C
vs H; for α diversity. P values are from Kruskal -Wallis test. d A total of 196 samples are c lustered into ent erotype 1 (blue) and enterotype
2(red) by PCA of Jensen-Shannon divergence values at the genus level. The major contributor in the two enterotypes is Prevotella and Bacter-
oides, respectively. e Relative abundances of the top genera (Prevotella and Bacteroides) in each enterotype. P =6.31e−31 and P =2.09e−15, respectively;
Wilcoxon rank sum test. f The percentage of control, pHTN and HTN samples distributed in two enterotypes. 26.83% normotensive controls, 48.21%
pHTN, and 45.45% HTN are found in enterotype 1. P = 0.02, C vs P; P =0.03,CvsH;Fisher’s exact test. Boxes represent the inter quartile ranges, the
inside line or points represent the median, and circles are outliers
Li et al. Microbiome (2017) 5:14 Page 3 of 19
the most enriched genus in enterotype 2 (P = 6.31e−31 and
P = 2.09e−15, respectively; Wilcoxon rank sum test, Fig. 1e).
Both contributors in the two enterotypes have been re-
ported in European and Chinese populations before [2, 3].
There was a higher percentage of pre-hypertensive and
hypertensive patients distributed in enterotype 1 (48.21%
for pHTN, and 45.45% for HTN), while more healthy con-
trols (73.17%) were found in enterotype 2 (P = 0.02, C vs P;
P = 0.03, C vs H; Fisher’sexacttest;Fig.1f).Thesefindings
suggest that enterotype 2 may represent a GM community
structure associated with healthy control, while enterotype
1 may be associated with pHTN and HTN.
Considering the higher percentage of HTN patients in
enterotype 1, we clustered the genera in this enterotype
and further explored the microbial co-occurrence net-
work by Spearman’s correlation. There was a positively
interacted network constituted by 12 genera, which were
negatively correlated with Prevotella, the core genus in
this enterotype (Additional file 4: Figure S2a). All these
genera were decreased in enterotyp e 1 as compared with
enterotype 2 (Additional file 4: Figure S2b). Eight out of
them were directly linked to Prevotella, while the other
four, including Oscillibacter, Faecalibacterium, Butyrivibrio,
and Roseburia, were indirectly linked to Prevotella.These
findings highlighted the possibility of Prevotella as a key
genus associated with pHTN and HTN. The difference in
gut enterotype distribution revealed profound changes of
the intestinal microbiome structure in both pHTN and
HTN, implying the significance of gut microbes in the de-
velopment of HTN.
pHTN and HTN-associated genera in GM
Genes were aligned to the NR database and annotated
to taxonomic groups. The relative abundance of gut mi-
crobes was calculated by summing the abundance of
genes as liste d in Additional file 2: Table S3–S4. P values
were tested by Wilcoxon rank sum test and corrected
for multiple testing with Benjamin & Hochberg meth od
as others previously did [4, 25]. It is worth mentioning
that 44 gener a were differentially enriched in control,
pHTN, and HTN (P < 0.1, Wilcoxon rank sum test ,
Fig. 2a and Additional file 2: Table S5). Fifteen of them
were further shown in Fig. 2b. Genera such as Prevotella
and Klebsiella were overrepresented in individuals with
pHTN or HTN (Fig. 2b). Prevotella, originated from
mouth and vagina, was abundant in the microbiome of
our study cohort. The pathogenesis of periodontal
diseases and rheumatoid arthritis are thought to be
attributed to Prevotella [3, 26]. A wide range of infec-
tious diseases are known to be attributed to Klebsiella
[27, 28]. Porphyromonas and Actinomyces, which were
also elevated in the HTN group, are morbific oral bac-
teria that cause infections and periodontal diseases [29].
By contrast, Faecalibacterium, Oscillibacter, Roseburia,
Bifidobacterium, Coprococcus, and Butyrivibrio, which
were enriched in healthy controls, declined in pHTN
and HTN patients (Fig. 2b). Our observations were con-
sistent with the genera negat ively correlated with Prevo-
tella in the network of enterotype 1 (Additional file 4:
Figure S2), and these bacteria are known to be essential
for healthy status. For example, reduced levels of Faeca-
libacterium and Roseburia in the intestines are as-
sociated with Crohn’s disease and ulcerative colitis [30,
31]. Both bacteria are crucial for butyric acid production
[30, 32]. Moreover, Bifidobacterium is an important pro-
biotic necessary to intestinal microbial homeostasis, gut
barrier, and lipopolysaccharide (LPS) reduction [33].
The divergence of GM composition in each sample
was assessed to explore the correlation of microbial
abundance with body mass index (BMI), age, and gender
(Additional file 5: Figure S3). Although the gender ratio
is discrepant among groups (Additional file 1: Table S1),
we found no remarkable regularity of bacterial abun-
dance based on BMI, age or gender.
To further validate the bacterial alterations in HTN,
an independent metagenomic analysis was performed
using the sequencing data generated from a previous
study of type 2 diabetes [2]. From a total of 174 non-
diabetic controls in the study, normotensive controls
with SBP ≤125 mmH g or DBP ≤80 mmHg were en-
rolled, and HTN were elected with the inclusion criteria
of SBP ≥150 mmHg or DBP ≥100 mmHg. The FBG
levels between normotensive controls and HTN were
similar. Finally, six subjects (HTNs, n = 3; normotensive
controls, n = 3) were included in our analysis (Additional
file 2: Table S6). A s expected, the microbial diversity was
decreased in HTN (Additional file 6: Figure S4a), and
there were at least 20 genera showing consistent trends
with our findings, includin g decreased Butyrivibrio
,
Clost
ridium, Faecalibacterium, Enterococcus, Roseburia,
Blautia, Oscillbacter, and ele vated Klebsiella, Prevotella,
and Desulfovibrio (Additional file 6: Figure S4b,
Additional file 2: Table S7).
Collectively, these results supported our hypothesis that
bacteria associated with healthy status were reduced in pa-
tients with HTN. This phenomenon together with the
overgrowth of bacteria such as Prevotella and Klebsiella
may play important role in the pathology of HTN.
Co-abundance groups enriched in pHTN and HTN
Firstly, for each gene, an OR score was calculated ac-
cording to the abundance of each gene. Then, for the
comparative analysis between control and HTN samples,
the HTN-associated genes were classified as HTN-
enriched (OR >2) or HTN-depleted (OR <0.5) as previ-
ously described [34]. When calculating HTN-associated
ORs, samples of pHTN wer e excluded, and samples
Li et al. Microbiome (2017) 5:14 Page 4 of 19
labeled as HTN were excluded as well when calculating
pHTN-associated ORs. A total of 1,120,526 genes signifi-
cantly different in relative abundance across groups were
identified (Additional file 7: Table S8). Secondly, we
clustered genes by a rather high threshold (Spearman’s
correlation coefficient ≥0.7) according to previous
methods [4, 35]. Spearman’s correlation coefficient was
analyzed by R. The cluster groups with at least 50 genes
were defined as co-abundance groups (CAGs) [4], and
used for further analysis [35]. One thousand ninety-nine
distinct CAGs were obtained (Additional file 2: Table S9–
S11 and Additional file 8: Figure S5a). Seven hundred
fourteen CAGs were assigned to known bacterial genera
based on the tracer genes, with at least 80% of the genes
mapped to the reference genome at an identity higher
than 85% (Additional file 8: Figure S5b).
CAGs were further clustered by Spearman’scorrelation
based on the abundance. Compared with the controls,
there were 316 CAGs and 372 CAGs enriched in pHTN
and HTN, respectively (Additional file 2: Table S12). In
the control group, Firmicutes and Roseburia were more
abundant (Fig. 3a, b). Most CAGs enriched in pre-
hypertensive samples were originated from Enterobacter,a
disease-causing bacteria linked to obesity. Klebsiella, caus-
ally implicated in various infections, was also overrepre-
sented in pre-hypertensive and hypertensive patients [27].
Additionally, most recent studies revealed that Fusobac-
terium was enriched in the fecal samples of patients with
liver cirrhosis, colorectal carcinoma, or ulcerative colitis
[4, 36, 37]. We also detected several clusters of CAGs
assigned to Fusobacterium enriched in pHTN and HTN
groups. About 200 CAGs were different in pHTN and
HTN. Most of them in pHTN were from Enterobacter
and Klebsiella,whilePrevotella and Fusobacterium were
more abundant in HTN.
To further examine the relationship between clinical
indices and CAGs of GM, physiological parameters of
SBP, DBP, BMI, FBG, total cholesterol (TC), triglyceride
Fig. 2 Genera strikingly different across groups. a Relative abundance of the top 44 most different genera across groups at the criteria of P value
<0.1 by Wilcoxon rank sum test. C, control; P, pHTN; H, HTN. The abundance profiles are transformed into Z scores by subtracting the average
abundance and dividing the standard deviation of all samples. Z score is negative (shown in blue) when the row abundance is lower than the
mean. Genera at P value <0.01 are marked with dark green star, P value <0.05 with light green star, and P value ≥0.05 with gray circle. b The box
plot shows the relative abundance of four genera enriched in pHTN and HTN patients, and 11 genera abundant in control. Genera are colored
according to the phylum. Boxes represent the inter quartile ranges, lines inside the boxes denote medians, and circles are outliers
Li et al. Microbiome (2017) 5:14 Page 5 of 19