This journal is c The Royal Society of Chemistry 2011 Mol. BioSyst., 2011, 7, 3271–3279 3271

Cite this: Mol. BioSyst., 2011, 7, 3271–3279

Application of plasma lipidomics in studying the response of patients with

essential hypertension to antihypertensive drug therapyw

Chunxiu Hu,aHongwei Kong,

aFengxue Qu,

bYong Li,

aZhenqiu Yu,

bPeng Gao,

a

Shuangqing Peng*cand Guowang Xu*

a

Received 23rd August 2011, Accepted 20th September 2011

DOI: 10.1039/c1mb05342f

Hypertension is a key risk factor in the progression of cardiovascular disease (CVD).

Dyslipidemia, a strong predictor of CVD, frequently coexists with hypertension. Therefore,

the control of hypertension and dyslipidemia may help reduce CVD morbidity and mortality.

In the present study, the therapeutic effects of antihypertensive agents on blood pressure control

and plasma lipid metabolism were evaluated. The plasma lipid profiles of patients with treated

(n = 25) or untreated (n = 30) essential hypertension as well as of subjects with normotension

(n = 28) were analyzed using liquid chromatography mass spectrometry. Principal component

analysis of the lipidomics data revealed distinct clusters among studied subjects across three

human populations. Phosphatidylcholines and triacylglycerols (TG) dominated the pattern of

hypertension-influenced plasma lipid metabolism. Discriminatory lipid metabolites were analyzed

using one-way analysis of variance followed by a post hoc multiple comparison correction. TG

lipid class was significantly increased by 49.0% (p o 0.001) in hypertensive vs. normotensive

groups while tended to decrease (�21.2%, p = 0.054) in hypertensive patients after treatment.

Total cholesteryl esters were significantly decreased by �16.9% (p o 0.001) in hypertensive

patients after treatment. In particular, a large number of individual neutral lipid species were

significantly elevated in hypertensive subjects but significantly decreased after treatment with

antihypertensive agents. The present study applied, for the first time, a systems biology based

lipidomics approach to investigate differentiation among plasma lipid metabolism of patients with

treated/untreated essential hypertension and subjects with normotension. Our results demonstrate

that antihypertensive medications to lower blood pressure of hypertensive patients to target levels

produced moderate plasma lipid metabolism improvement of patients with hypertension.

Introduction

Hypertension is a leading cause of cardiovascular morbidity

and mortality worldwide.1,2 Individuals with hypertension often

exhibit abdominal obesity, decreased high density lipoprotein-

cholesterol (HDL-C), hyperglycemia and hyperlipidemia,3–6 all

of which may contribute to the onset of cardiovascular disease.

Because it exerts multiple harmful effects on the human body

without showing any symptoms, hypertension is known as a

‘‘silent killer’’ and presents a public health concern in modern

society.7 Recognition of the accumulation status of epicardial

and visceral fat deposits is considered crucial in diagnosing and

preventing hypertension at an early stage.8

Hypertension is generally recognised to be related to life-

style factors such as unhealthy dietary habits (e.g. excessive

intake of calories, alcohol and salt) and physical inactivity.9,10

Therefore, lifestyle modifications to correct these contributing

factors are frequently used as the initial treatment for subjects

with pre-hypertension. However, drug therapy is highly recom-

mended if the blood pressure cannot be adequately lowered

by these lifestyle modifications or the individual is in a more

advanced stage of hypertension. There is considerable evidence

that hypertensive patients must take two or more drugs to

achieve their target blood pressure.11 In clinical trials, three or

more therapeutic agents, including diuretics, beta-blockers

(BBs), angiotensin-converting enzyme inhibitors (ACEIs),

angiotensin receptor blockers (ARBs) and calcium channel

blockers (CCBs), are commonly used in combination for the

treatment of hypertension.12,13

a CAS Key Laboratory of Separation Science for AnalyticalChemistry, Dalian Institute of Chemical Physics, Chinese Academyof Sciences, Dalian 116023, PR China. E-mail: xugw@dicp.ac.cn;Fax: +86 411 84379559; Tel: +86 411 84379530

b Beijing Anzhen hospital, Capital University of Medical Sciences,Bejing 100029, PR China

c Evaluation and Research Center for Toxicology, Institute of DiseaseControl and Prevention, Academy of Military Medical Sciences,Beijing 100071, PR China. E-mail: pengsq@hotmail.com;Fax: +86 10 66948462; Tel: +86 10 66948462w Electronic supplementary information (ESI) available. See DOI:10.1039/c1mb05342f

MolecularBioSystems

Dynamic Article Links

www.rsc.org/molecularbiosystems PAPER

Publ

ishe

d on

18

Oct

ober

201

1. D

ownl

oade

d by

Uni

vers

ity o

f M

ichi

gan

Lib

rary

on

30/1

0/20

14 1

1:56

:53.

View Article Online / Journal Homepage / Table of Contents for this issue

3272 Mol. BioSyst., 2011, 7, 3271–3279 This journal is c The Royal Society of Chemistry 2011

Large population-based cohort studies have shown that

dyslipidemia, which causes endothelial dysfunction,14 plays a

key role in the development of hypertension.15–19 Evidence

from epidemiology and clinical trials has demonstrated that

dyslipidemia is frequently coexistent with hypertension.20–22

However, most of the medical treatments for hypertension usually

focus on achieving optimal blood pressure. Sufficient attention has

not been paid to whether these therapeutic approaches produce

any benefits for other hypertension-associated risk factors

(e.g. lipid metabolic abnormalities). Occasionally, single targeted

lipid-related biomarkers, such as total plasma triacylglycerols

(TG) and/or cholesterol (Cho), have been determined clinically

to evaluate individuals’ conditions or the effects of antihyper-

tensive agents on plasma lipids in hypertensive patients. Despite

its expedience, such an approach fails to provide in-depth

insights into basic metabolism, the set of important chemical

reactions that are very close to the phenotype of living systems.

It is necessary to monitor alterations of individuals’ lipid meta-

bolites and their response to biological stimuli or genetic

manipulation to make improvements; for example, hyper-

tensive patients vs. normotensive subjects or hypertensive patients

with vs. without drug therapies should be compared to gain a

better understanding of interactions among the intricate lipid

networks at the systemic level indicative of disease or the

response to drugs. Advances in mass spectrometric analysis

have facilitated large-scale studies of lipids and their inter-

actions (i.e. lipidomics) at the molecular level and have aided

in the characterisation of biomarkers of health/disease and

drug/nutritional effect.23–27 Recently, several hypertension-

related studies have been reported in the literature.15,28–31

However, these studies focused only on investigating the differ-

entiation of target compound/metabolites between hypertensive

and normotensive subjects; none of the studies reported the

effects of antihypertensive agents by comparing the lipid meta-

bolism in hypertensive subjects before and after treatment.

In the present study, previously validated, state-of-the-art

liquid chromatography–mass spectrometry (LC–MS) was utilised

based on lipidomics technology32 to investigate differentiation

among the plasma lipid profiles of patients with treated/

untreated essential hypertension and subjects with normo-

tension. This profiling system enabled us (1) to determine

global changes in lipid metabolite levels in the plasma of

studied subjects; (2) to identify potential biomarkers that

may reveal interaction among networks of lipids during the

development of hypertension; and (3) to provide insights into

functionally relevant mechanisms of specific lipids contributing

to hypertension progression.

Materials and methods

Subjects and drug administration

Male subjects aged 35 to 55 were randomly selected from Beijing

Anzhen hospital (Beijing, China). The study population consis-

ted of 30 male patients with treated/untreated uncomplicated

primary (essential) hypertension and 28 men with normotension

used as the healthy controls. The subjects were instructed to

adhere to their normal diet during the study. The hypertensive

patients and the healthy controls underwent a standardised

clinical examination including body mass index (BMI), fasting

blood sugar, TG, Cho, high density lipoprotein cholesterol

level (HDL-C), low density lipoprotein cholesterol level (LDL-C),

systolic blood pressure (SBP) and diastolic blood pressure

(DBP) before treatment. Notably, only patients who did not

have secondary hypertension or pseudo-hypertension or who

did not manifest acute inflammatory processes, electrolyte dis-

turbances (blood potassium o 3.5 mol L�1 or >5.5 mol L�1),

diabetes mellitus, cardiovascular disease, chronic respiratory

disease, gastroenteric disease, or severe renal or hepatic diseases

participated in the study. The 30 patients with essential hyper-

tension followed a 90-day, orally administered antihypertensive

therapy with BBs or a combination of 2 to 4 agents including

diuretics, BBs, ACEIs, ARBs and CCBs, respectively. Twenty-

five patients completed the drug study. Detailed information

regarding the antihypertensive agents used is summarised in

Table S1 in the ESI.wThe experimental protocol was approved by Beijing Anzhen

hospital (Beijing, China), and all participants provided written

informed consents.

Blood pressure was measured in accordance with the World

Health Organization (WHO) guidelines. Diagnosis of hypertension

was based on SBP Z 140 mmHg and/or DBP Z 90 mmHg.

Sample collection and storage

The blood was sampled in Beijing Anzhen hospital (Beijing,

China) by an experienced nurse. Venous blood was drawn

from the antecubital vein in the sitting position after overnight

fasting and drawn into tubes containing heparin. Plasma for

lipid profiling was separated after being stored at room tempera-

ture for 2 h and then centrifuged at 3000g for 10 min at 4 1C. The

samples were stored at �80 1C until analysis.

Biochemical parameters

Plasma total Cho, TG, HDL-C, LDL-C, and glucose levels

were determined using a Hitachi 7600-020 automatic analyzer

(Hitachi, Japan). All measurements were performed on the

plasma samples taken after one night of fasting.

Chemicals and lipid standards

Synthetic lipid standards including 1-heptadecanoyl-2-hydroxy-

sn-glycero-3-phosphocholine (LPC (17 : 0)), 1,2-diheptadecanoyl-

sn-glycero-3-phosphoethanolamine (PE (34 : 0)) and 1,2-dihepta-

decanoyl-sn-glycero-3-phosphocholine (PC (34 : 0)) were obtained

from Avanti Polar Lipids, Inc. (Alabaster, Alabama, USA) and

1,2,3-triheptadecanoateglycerol (TG (51 : 0)) was purchased from

Sigma-Aldrich Shanghai Trading Co. Ltd (Shanghai, China).

Distilled water was purified using a Milli-Q system (Millipore,

Bedford, MA, USA). Dichloromethane (CH2Cl2), acetonitrile

(ACN), methanol (MeOH) and isopropanol (IPA) were of high

performance liquid chromatography (HPLC) grade purchased

from Tedia (Fairfield, OH, USA). Analytical-grade ammonium

formate (AmFm) was purchased from Sigma-Aldrich (St. Louis,

MO, USA).

Sample preparation

Plasma samples were thawed at room temperature and extracted

with 2 : 1 CH2Cl2/MeOH as described previously.32 Briefly, 30 mL

Publ

ishe

d on

18

Oct

ober

201

1. D

ownl

oade

d by

Uni

vers

ity o

f M

ichi

gan

Lib

rary

on

30/1

0/20

14 1

1:56

:53.

View Article Online

This journal is c The Royal Society of Chemistry 2011 Mol. BioSyst., 2011, 7, 3271–3279 3273

of plasma was placed in a new 1.5 mL Eppendorf vial (Eppendorf,

Hamburg, Germany) and 30 mL of a lipid internal standard (I.S.)

mixture consisting of 15 mg mL�1 of LPC (17 : 0), 20 mg mL�1 of

PC (34 : 0), 35 mg mL�1 of PE (34 : 0) and 40 mg mL�1 of TG

(51 : 0) dissolved in 2 : 1 CH2Cl2/MeOH was added, followed by

190 mL of MeOH. The solution was thoroughly mixed by vortex-

ing for 30 s. Afterwards, 380 mL of CH2Cl2 was added and the

solution was mixed for another 30 s by vortexing. Subsequently,

120 mL of water was added to the solution and thoroughly mixed,

resulting in a two-phase system. After centrifuging (5415 R,

Eppendorf, Hamburg, Germany) for 10 min at 6000g, 200 mLof lipid extract from the lower organic phase was transferred to a

new 1.5 mL brown glass autosampler vial and stored at �20 1C

prior to analysis. For LC–MS analysis, 80 mL of the lipid extract

was diluted with 320 mL of IPA/ACN/water at a ratio of 30/65/5

(v/v/v).

LC–MS analysis

A LC–MS lipidomics analysis was performed using an ultra-

fast liquid chromatography system (Shimadzu, Kyoto, Japan)

coupled with an ion-trap time-of-flight mass spectrometer

(IT-TOF MS) equipped with an electrospray ion source

(Shimadzu, Kyoto, Japan). An Ascentiss Express C8 column

(2.7 mm particle size, 90 A, 2.1 � 150 mm) (Sigma-Aldrich,

Munich, Germany) was used for the LC chromatographic

separation. The separation conditions were based on a pre-

viously published method.32 A binary solvent consisted of

water/ACN [(2 : 3 (v/v), 10 mM AmFm) and IPA/ACN

(9 : 1 (v/v), 10 mM AmFm) was used for LC–MS separation.

The chromatographic auto-injector plate containing diluted

lipid extracts was maintained at 12 1C. MS survey scans were

acquired in the positive ion mode. The voltages of the interface

and the detector of the TOF analyzer were set to 4.5 kV and

1.6 kV, respectively. The temperatures of the curved

desorption line and heat block were both set at 200 1C. The

flow rate of the nebulising gas was 1.5 L min�1, and the dry gas

pressure was 0.2 MPa. The flight tube temperature was stable

at 40 1C, and the ion trap pressure was maintained at 1.6 �10�2 Pa. Ultra-high purity argon was used for collision and

ion cooling. The data were collected at a mass range of m/z

400–1500 with an ion scan duration of 20 ms using LC–MS

solution software (Shimadzu, Kyoto, Japan). Each sample

was prepared in duplicate, and each extraction was injected

once.

The identification of lipids was achieved based on MS/MS

fragments, accurate masses, and comparing relevant informa-

tion of observed peaks with that in our in-house lipid library

which includes MS/MS fragmentation data, retention time,

theoretical masses and observed masses established on the

basis of our previous study.32 The current lipidomic analysis

covered 81 lipid species distributed among 7 major lipid classes

including lyso-phosphocholines (LPC), phosphatidylcholines

(PC), sphingomyelins (SPM), phosphatidylethanolamines

(PE), cholesteryl esters (ChoE), diacylglycerols (DG) and TG.

Data pre-processing

LC–MS data acquired from all analyzed samples were first

converted into the NetCDF format and then processed using

XCMS,33 in which the filtering of the raw data, the retention

time correction, noise correction and peak alignment were per-

formed automatically. The lipid species were quantified based

on peak area ratios to an appropriate I.S. based on a previously

described strategy.32 The results from the duplicate analyses for

each sample were averaged before further statistical data

analysis.

Statistical analysis

Principal Component Analysis (PCA) was performed using

Matlab (version 7.7.0.471, the Mathworks) with the PLS

toolbox (version 5.0.3, Eigenvector Research, Inc.) to visualise

clusters of the study samples among the three groups under

current LC–MS conditions. Analysis of variance (ANOVA)

for repeated measures was used followed by a Bonferroni

post hoc performance test for multiple comparisons to further

examine any statistically significant difference in the individual

lipids and lipid classes among the three (i.e. normotensive,

hypertensive and treated-hypertensive) groups. Specifically, a

paired t-test was performed for evaluation of the statistical

significance of two substrate lipid ratios such as PC (38 : 4)/

LPC (18 : 0) and PC (36 : 4)/LPC (16 : 0) in 25 hypertensive

patients before and after drug treatment.

Data were expressed as the mean � SD. A value of po 0.05

was considered statistically significant. All statistical analyses

were performed using the SPSS statistical package (V.17.0 for

Windows; SPSS, Chicago, IL, USA).

Results

Basal clinical characteristics of the studied population before

treatment

Basal clinical data showed that patients with hypertension had

significantly higher BMI, total plasma TG and fasting plasma

glucose levels and significantly lower HDL-C levels than did

the healthy cohort (Table 1). The average ages of the hyper-

tensive and normotensive groups were comparable. As expected,

the older hypertensive cohort (47.2 � 3.9 years old) had more

plasma metabolic risk factors in terms of fasting blood glucose

(5.38 � 1.04 vs. 4.92 � 0.61 mM), plasma TG (2.54 � 2.02 vs.

1.69 � 0.91 mM), Cho (5.05 � 1.37 vs. 4.36 � 1.60 mM) and

LDL-C (3.52 � 1.84 vs. 3.11 � 0.86 mM) levels compared with

those of the younger patient cohort (37.4 � 1.5 years old).

Table 1 Clinical parameters in normotensive and hypertensive men

Normotensive(n = 28)

Hypertensive(n = 30)

Significancep

Age/years 44.86 � 5.32 44.27 � 5.64 0.684BMI/kg m�2 23.86 � 2.77 25.72 � 2.07 0.002Triacylglycerols/mM 1.48 � 0.60 2.29 � 1.79 0.026Total cholesterol/mM 5.11 � 1.02 4.84 � 1.45 0.415HDL-cholesterol 1.24 � 0.23 1.09 � 0.19 0.010LDL-cholesterol 3.36 � 0.98 3.40 � 1.60 0.916Fasting blood-glucose/mM

4.74 � 0.78 5.24 � 0.95 0.020

Note: all values are expressed as mean � SD; p o 0.05 was considered

to be statistically significant.

Publ

ishe

d on

18

Oct

ober

201

1. D

ownl

oade

d by

Uni

vers

ity o

f M

ichi

gan

Lib

rary

on

30/1

0/20

14 1

1:56

:53.

View Article Online

3274 Mol. BioSyst., 2011, 7, 3271–3279 This journal is c The Royal Society of Chemistry 2011

Drug therapies significantly improve blood pressure in

hypertensive patients

It can be seen from Table S1 in the ESIw that patients took 1 to 4

types of antihypertensive drugs in different doses for 90 days. To

simplify the study, we did not distinguish the individual effects of

the various antihypertensive medications in the trial group.

Before treatment, SBP and DBP in hypertensive patients

were significantly higher than those in the normotensive sub-

jects (Table 2). After a 90-day study with oral medication,

SBP in hypertensive patients was significantly decreased from

150.2 � 14.3 to 125.0 � 8.9 mmHg (p o 0.001), DBP from

101.8 � 9.3 to 78.6 � 6.0 mmHg (p o 0.001), dynamic SBP

from 134.9 � 12.5 to 118.7 � 7.8 mmHg (p o 0.001), and

dynamic DBP from 87.5 � 7.8 to 74.6 � 3.2 mmHg

(po 0.001) (Table 2), as compared to those in the normotensive

control group. These data indicate that, after a 90-day treat-

ment with antihypertensive medication, the blood pressure of

the hypertensive patients returned to normal.

Plasma lipidomics reveals the details of the systemic changes in

lipid homeostatic response to antihypertensive drugs

PCA was performed to study the cluster data from the subjects

(i.e. the scores plot) with normotension, hypertension under-

going treatment, and hypertension without treatment and to

identify which lipid species contributed most to the clusters

(i.e. the loading plot). The first two principal components of

the established PCA model described 61.58% of the total

variance of the plasma lipidomics data set (Fig. 1a). The scores

plot showed very clear differentiation between the normo-

tensive and hypertensive groups, indicating striking changes

of plasma lipid metabolites between these two groups

(Fig. 1a). Another finding from the scores plot was that an

obvious trend of separation was observed between the hyper-

tensive and treated-hypertensive groups (although slight over-

lapping was seen), indicating an influence of antihypertensive

agents on the plasma lipid metabolism of patients. In addition,

the loading plot clearly revealed that the two most abundant

lipid classes in the hypertensive patients, PC and TG, domi-

nated in the differentiation between the normotensive and

hypertensive groups and between the hypertensive and treated-

hypertensive groups (Fig. 1b).

To investigate quantitative changes of the lipid profiles in the

observed clustering patterns (i.e. hypertensive vs. normotensive

and treated-hypertensive vs. hypertensive groups), statistically

significant differences were assessed in terms of individual lipid

molecular species and different lipid classes across groups using

one-way ANOVA with a post hoc multiple comparison correc-

tion. The lipid molecular species of LPC (22 : 6), PC (40 : 6),

SPM (16 : 1), ChoE (20 : 4), ChoE (22 : 6), TG (48 : 0 - : 3),

TG (50 : 2), TG (50 : 4), and TG (56 : 5 - : 8) were signifi-

cantly increased in hypertensive vs. normotensive groups but

significantly decreased after medication administration; SPM

(24 : 2), TG (50 : 0, 50 : 1, 50 : 3, 50 : 5), TG (52 : 1 - : 6),

TG (54 : 2 - : 6), and TG (56 : 5 - : 9) lipids were signifi-

cantly increased in the hypertensive vs. normotensive groups,

and most of these began to decrease after drug administration

(Fig. 2a–c). In addition, ChoE (18 : 1) and ChoE (20 : 5) lipids

were significantly decreased in hypertensive subjects after treat-

ment (Fig. 2b). With regard to the quantitative changes of

different lipid classes (i.e. the summation of the individually

measured lipids into different lipid classes) under the current

analytical conditions, total phospholipids (PLs) and total

TGs were significantly increased by 8.3% (p = 0.019) and by

49.0% (p o 0.001), respectively, in hypertensive vs. normo-

tensive groups, total TGs tended to decrease (�21.2%, p=0.054)

in hypertensive patients after treatment. In addition, total

ChoEs were significantly decreased by �16.9% (p o 0.001)

Table 2 Systolic and diastolic blood pressures of the investigatedpopulation

Normotensive(n = 28)

Hypertensive(n = 30)

Drug treatedhypertensive(n = 25)

SBP/mmHg 116.6 � 10.7a 150.2 � 14.3 125 � 8.9b

DBP/mmHg 78.6 � 7.5a 101.8 � 9.3 78.6 � 6.0b

Dynamic SBP/mmHg 134.9 � 12.5 118.7 � 7.8b

Dynamic DBP/mmHg 87.5 � 7.8 74.6 � 3.2b

Note: values are expressed as mean � SD. a Significant difference

between normotensives and hypertensives (p o 0.001). b Significant

difference between hypertensives and drug treated hypertensives

(p o 0.001).

Fig. 1 PCA scores (a) and loading (b) plots of mean centered plasma

lipidomics data from three study groups. N: normotensive subjects;

H: hypertensive patients; T: treated hypertensive patients.

Publ

ishe

d on

18

Oct

ober

201

1. D

ownl

oade

d by

Uni

vers

ity o

f M

ichi

gan

Lib

rary

on

30/1

0/20

14 1

1:56

:53.

View Article Online

This journal is c The Royal Society of Chemistry 2011 Mol. BioSyst., 2011, 7, 3271–3279 3275

in hypertensive patients after treatment (Fig. 3). Furthermore,

two substrate ratios such as PC (38 : 4)/LPC (18 : 0) and PC

(36 : 4)/LPC (16 : 0) regulated by enzyme phospholipase A

(PLA) were investigated due to their pathophysiological signi-

ficance in inflammatory status including dyslipidemia, obesity,

arthrosclerosis and hypertension.34 A paired t-test was per-

formed to determine how the anti-hypertensive treatment

upon 25 hypertensive patients influenced the values of PC

(38 : 4)/LPC (18 : 0) and PC (36 : 4)/LPC (16 : 0). The means

of both of the ratio differences between before and after drug

treatment were 0.220 and 0.274, two-paired p values were

o0.001 and =0.001, and 95% confidence intervals about

mean ratio differences are (0.107, 0.333) and (0.090, 0.312),

respectively.

Correlation networks of specific lipid metabolites during the

progression of hypertension

To examine the functionally relevant mechanisms of specific

lipids contributing to the progression of hypertension, we studied

the correlation networks of lipids that showed significant changes

between the studied subgroups. Specific lipid metabolites were

linked according to their Pearson correlation coefficient (Cij),

Fig. 2 Comparison of the content of specific lipid molecular species from PLs (a), ChoE (b) and TG (c) in plasma samples of all study

populations. N: normotensive subjects; H: hypertensive patients; T: treated hypertensive patients. Values are means � SD; *p o 0.05, **po 0.01,

***po 0.001: significant differences either between groups of normotensive and hypertensive subjects or between hypertensive patients before and

after drug treatment groups.

Fig. 3 Comparison of the content of total PLs (closed diamonds),

TGs (closed triangles) and ChoEs (closed squares) of plasma samples

from three study populations. Values are means � SD. *p o 0.05

in hypertensive vs. normotensive subjects; (a) ***p o 0.001 in hyper-

tensive vs. normotensive subjects; (b) ***p o 0.001 in hypertensive

patients before vs. after treatment. N: normotensive subjects; H: hyper-

tensive patients; T: treated hypertensive patients.

Publ

ishe

d on

18

Oct

ober

201

1. D

ownl

oade

d by

Uni

vers

ity o

f M

ichi

gan

Lib

rary

on

30/1

0/20

14 1

1:56

:53.

View Article Online

3276 Mol. BioSyst., 2011, 7, 3271–3279 This journal is c The Royal Society of Chemistry 2011

i.e., when the absolute value of Cij was >0.8. The statistical

significance of the connection was set at p o 0.05. Three

correlation networks were constructed based on the data from

the hypertensive vs. the normotensive subjects (Fig. 4a), the

hypertensive patients with vs. without treatment (Fig. 4b), and

the hypertensive patients with treatment vs. the normotensive

subjects (Fig. 4c), respectively. The correlation networks

revealed that lipids, especially the neutral lipids, TG species,

were correlated more tightly in hypertensive patients because

they had a higher number of connections (Fig. 4a). Comparing

Fig. 4a with Fig. 4b it is observed that all the specific TG, LPC

and PC lipid molecular species that significantly accumulated

in the hypertensive patients were either regulated to target

levels or showed a tendency towards down-regulation after the

medication had been administered. Comparing Fig. 4a with

Fig. 4c, we found that the antihypertensive drugs were able

to decrease most of the specific lipids that were significantly

accumulated in the hypertensive patients to normal levels. In

the meantime, many tight correlations disappeared after medi-

cation had been administered. The comparison among Fig. 4a, b

and c clearly reveals that the antihypertensive administration

had minor effects on the regulation of lipid species such as PC

(38 : 6), TG (52 : 3, 52 : 4, 52 : 5) and TG (50 : 3). Collec-

tively, our results indicate that the antihypertensive prescrip-

tion used had positive effects on modifying lipid metabolism,

especially TG species in the hypertensive patients with under-

lying dyslipidemia, suggesting that TG molecules may play

important roles in the progression of hypertension.

Discussion

Routine clinical parameters showed that antihypertensive

administration led to a significant increase in SBP, DBP, BMI,

plasma TG and fasting blood glucose levels and a significant

decrease in the levels of HDL-C in patients with hypertension as

compared to those in normotensive subjects; these data indicate a

combination of hypertension and dyslipidemia in the hyper-

tensive patients. In addition, the older hypertensive patients

(41–55 years old) exhibited a higher risk for plasma metabolic

factors, manifested as higher levels of fasting blood glucose,

plasma TG, Cho and LDL-C than did younger patients (35–39

years old). After a 90-day antihypertensive administration, levels

of SBP and DBP in the hypertensive subjects returned to normal

and the dynamic SBP and DBP were significantly improved as

compared to those before treatment (Table 2), indicating that

these antihypertensive therapies are effective in the regulation of

blood pressure.

Recent epidemiological and clinical data have demonstrated

that hypertension is frequently accompanied by obesity, insulin

resistance, increased inflammatory mediators, hyperglycemia,

and atherogenic dyslipidemia.1,16,35–37 All of these factors may

contribute to the highly increased risk of cardiovascular disease

and type 2 diabetes associated with hypertension.38,39 There-

fore, the therapeutic effect of medications in the management of

hypertension should be investigated in terms of blood pressure

and other risk factors. Because dyslipidemia plays an important

role in the pathogenesis of hypertension and frequently coexists

with hypertension,12,17,40 monitoring global changes of lipids

in individuals and their response to biological stimuli, genetic

manipulation or drug therapy may reveal intricate interactions

among lipid networks at a level indicative of disease or response

to drugs.

Using advanced LC–MS-based plasma lipidomics technology,

a more in-depth analysis was conducted to study lipids on a

metabolic basis, rather than as single target lipid biomarkers

(e.g. total plasma TG and Cho levels are traditionally used in

clinics). Such a lipidomics approach has great advantage over

the traditional method (i.e.measuring total lipid related items)

in the aspect of exploring the details of lipids and unveiling the

mechanisms of lipid function. It enables characterization of

dynamic changes in individual lipid metabolites and their

interactions in a systems-integrated context. Using this holistic

approach, we are able to build a comprehensive picture of

lipid metabolic interconnections, discover new molecular

species and determine how lipids function the way they do.

Our lipidomics study demonstrated that there was a redistri-

bution of different lipid classes among the hypertensive cohort

with/without treatment and the normotensive subjects. To our

knowledge, this is the first study of lipidomics that has been

applied to hypertension in human subjects after medication,

and a specific association between hypertension, lipid meta-

bolism and anti-hypertensive drugs has been determined. The

outcome from the multivariate analysis of plasma lipidomics

based on LC–IT-TOF MS revealed that (1) lipid metabolism

in the hypertensive patients is clearly different from that in the

normotensive subjects, and PC and TG lipids are highly

abundant in the plasma of hypertensive patients; and (2) lipid

metabolism is remarkably changed in the hypertensive patients

after administration of antihypertensive medication. The

univariate analysis of plasma lipidomics data elucidated the

lipid metabolism and redistribution of different lipid classes

introduced by hypertension or medications more than the use

of a single-target lipid biomarker did; the data highlighted

changes in the lipid profile at the molecular level in the hyper-

tensive patients vs. the normotensive subjects and the response

to antihypertensive drugs. Our plasma lipidomics data on a

univariate basis revealed that hypertension coexisted with dys-

lipidemia, manifested as a significant increase in the levels of a

large number of TG and PC lipid species (see Table S2, ESIw)and the summation of individually measured lipids into neutral

TG lipids and polar PLs. The 90-day antihypertensive treat-

ment induced a significant reduction in some moieties of neutral

lipid species, such as TG and ChoE lipids, and a tendency to

decrease the summation of all individually measured TG lipids

in the hypertensive patients. It can be observed that TGs

containing three or two saturated fatty acyl chains (i.e. TG

48 : 0, 48 : 1, 50 : 0, 50 : 1, 52 : 1) were significantly accumu-

lated in hypertensive patients (vs. the normotensive subjects),

indicating the possible lipotoxic effects.41 After anti-hypertensive

administration, TGs (48 : 0, 48 : 1, 50 : 0, 50 : 1, 52 : 1) were

either significantly reduced or showed a tendency to decrease,

showing the detoxification by desaturation, that is, the accumula-

tion of TGs containing three or two saturated fatty acyl chains.42

Collectively, our lipidomics result suggests that the antihyper-

tensive prescription used in the present study had moderate effects

in modifying lipid levels in patients with hypertension.

It is well recognized that inflammation is a typical feature of

hypertension. Arachidonic acid has been identified to be a potent

Publ

ishe

d on

18

Oct

ober

201

1. D

ownl

oade

d by

Uni

vers

ity o

f M

ichi

gan

Lib

rary

on

30/1

0/20

14 1

1:56

:53.

View Article Online

This journal is c The Royal Society of Chemistry 2011 Mol. BioSyst., 2011, 7, 3271–3279 3277

inflammatory mediator and closely involved in the inflammatory

response and lipid signalling. In detail, various enzyme PLA

types such as PLA1 and PLA2 are implicated in the pathology of

the inflammatory case,34 in which they catalyze the hydrolysis

of the sn-1 (PLA1) and sn-2 (PLA2) position of specific PC

molecular species (i.e. PC 16 : 0/20 : 4 or PC 20 : 4/16 : 0 and

PC 18 : 0/20 : 4 or PC 20 : 4/18 : 0, respectively) to release

arachidonic acid and LPCs. Therefore, the activity of PLA can

be estimated by calculating the substrate ratios of PC (38 : 4)/

LPC (18 : 0) and PC (36 : 4)/LPC (16 : 0). It was found that the

means of the ratio differences of PC (38 : 4)/LPC (18 : 0) (mean=

0.220, SD = 0.274, n = 25) and PC (36 : 4)/LPC (16 : 0)

Fig. 4 Correlation networks of specific lipid species that may play roles as potential biomarkers in the progression of hypertension. Lipid

metabolites were associated based on their Pearson correlation coefficient (Cij), i.e., when the absolute value of Cij was more than 0.8. Statistical

significance was set at p o 0.05. (a) A network constructed using the data from the hypertensive and normotensive subjects; (b) a network

constructed using the data from the hypertensive patients with or without treatment; (c) a network constructed using the data from the treated

hypertensive subjects and normotensive subjects. Red nodes represent significantly increased lipids as compared to those in the normotensive

subjects in (a and c); light red nodes represent an increased tendency without a significant change compared with the normotensive subjects in

(a and c); green nodes represent significantly decreased lipids as compared to the normotensive subjects in (a and c) and in the treated hypertensive

subjects in (b); white nodes represent a decreased tendency without a significant change compared with the treated hypertensive subjects in (b) and

the normotensive subjects in (c).

Publ

ishe

d on

18

Oct

ober

201

1. D

ownl

oade

d by

Uni

vers

ity o

f M

ichi

gan

Lib

rary

on

30/1

0/20

14 1

1:56

:53.

View Article Online

3278 Mol. BioSyst., 2011, 7, 3271–3279 This journal is c The Royal Society of Chemistry 2011

(mean = 0.201, SD= 0.269, n= 25) between before and after

drug treatments were significantly greater than zero, and the

two-tailed p o 0.001 and p = 0.001, respectively, providing

evidence that the anti-hypertensive agents produce the

reduction of PLA activity of the hypertensive patients after

anti-hypertensive medication.

In addition, the correlation networks of plasma lipidomic

profiling of the hypertensive patients (vs. the normotensive

subjects), of the treated hypertensive patients (vs. the untreated

hypertensive patients) and of the normotensive subjects (vs.

the treated hypertensive patients) contributed to a systematic

identification and understanding of ‘‘specific’’ lipid metabolites

that may play key roles (and can therefore act as potential

biomarkers) in the development of hypertension with under-

lying dyslipidemia or response to antihypertensive drugs. A

systems biology-based metabolomics approach is quite promis-

ing in fields such as early diagnosis, disease prevention or drug

therapy. The correlation networks constructed on the basis of

the alterations of all specific lipid metabolites in the present

study clearly revealed that the lipid regulation effect induced by

the prescribed antihypertensive agents was moderate under the

experimental conditions. It substantiates the important roles of

specific lipid molecules in the progression of hypertension and

provides insights into the functionally relevant lipid metabolic

pathways/networks associated with hypertensive risk factors.

Collectively, our results demonstrate that the antihypertensive

medications used in our study not only lower blood pressure

of hypertensive patients to target levels but also led to a mild

modification of the metabolism of neutral plasma lipids, such as

ChoE and TG, in the patients. Our study provides experimental

evidence that, despite the moderate benefits of using individual

antihypertensive drugs for plasma lipid metabolism, a combi-

nation of multiple agents aimed at the simultaneous management

of both hypertension and dyslipidemia is highly recommended

for treatment.

In conclusion, this study provides a systematic and accurate

characterisation of lipid metabolism in relation to hypertension

and drug therapy and reveals that the LC–MS lipidomics

approach is promising for discovery of potential lipid bio-

markers in relation to health/disease or drug/nutritional effects.

This study may provide the basis for metabolomics-based

approaches in the treatment of epidemics to help understand

the intricate interactions, pathways/networks that are influenced

by different lifestyles, environments and genes.

Acknowledgements

The authors gratefully acknowledge support from the China

International Science and Technology Cooperation program

(2009DFA41250) and National Key Project on Drug Develop-

ment (2009ZX09501-034) funded by the Ministry of Science

and Technology of China, the key foundation (No. 20835006)

and the creative research group project (No. 21021004) from

National Natural Science Foundation of China and the

Netherlands Genomics Initiative. We acknowledge Dr Mei

Wang (SU BioMedicine and TNO, the Netherlands) for useful

discussions on drug therapy and biological interpretation, and

Heng Wei (TNO, The Netherlands) for help with multiple

comparison performances.

References

1 S. Bo, R. Gambino, L. Gentile, G. Pagano, R. Rosato, G. M. Saracco,M. Cassader, M. Durazzo and P. Cavallo-Perin, J. Hypertens., 2009,27, 102–108.

2 R. Padwal, S. E. Straus and F. A. McAlister, Br. Med. J., 2001,322, 977–980.

3 K. Aizawa, J. K. Shoemaker, T. J. Overend and R. J. Petrella,Diabetes Res. Clin. Pract., 2008, 81, 72–78.

4 S. Harding, M. J. Maynard, K. Cruickshank and A. Teyhan, Int. J.Obes., 2008, 32, 82–90.

5 F. Natale, M. A. Tedesco, R. Mocerino, V. de Simone, G. M. DiMarco, L. Aronne, M. Credendino, C. Siniscalchi, P. Calabro,M. Cotrufo and R. Calabro, Eur. J. Echocardiogr., 2009, 10, 549–555.

6 C. Serban, T. Nicola, R.Mateescu, L. Noveanu, L. Susan, A. Pacurari,A. Caraba, I. Romosan and A. Cristescu, Rev. Med.-Chir. Soc. Med.Nat. Iasi, 2010, 114, 798–802.

7 J. Moore, Nurse Pract., 2005, 30, 16–35.8 A. M. Sironi, A. Pingitore, S. Ghione, D. De Marchi, B. Scattini,V. Positano, E. Muscelli, D. Ciociaro, M. Lombardi, E. Ferranniniand A. Gastaldelli, Hypertension (Dallas), 2008, 51, 282–288.

9 C. M. Reid and A. G. Thrift, Clin. Exp. Pharmacol. Physiol., 2005,32, 374–376.

10 T. Tsuchihashi, Nippon Rinsho, 2008, 66, 1537–1541.11 K. A. Jamerson, G. L. Bakris, C. C. Wun, B. Dahlof, M. Lefkowitz,

S. Manfreda, B. Pitt, E. J. Velazquez and M. A. Weber, Am. J.Hypertens., 2004, 17, 793–801.

12 K. C. Ferdinand and M. A. Kleinpeter, Curr. Hypertens. Rep.,2006, 8, 489–496.

13 J. J. Torre, Z. T. Bloomgarden, R. A. Dickey, M. J. Hogan, J. J.Janick, S. G. Jyothinagaram and H. M. Siragy, Endocr. Pract., 2006,12, 193–222.

14 G. Nickenig, J. Hum. Hypertens., 2002, 16(Suppl 3), S26–S33.15 J. Graessler, D. Schwudke, P. E. H. Schwarz, R. Herzog,

A. Shevchenko and S. R. Bornstein, PLoS One, 2009, 4, e6261.16 R. O. Halperin, H. D. Sesso, J. Ma, J. E. Buring, M. J. Stampfer

and J. M. Gaziano, Hypertension (Dallas), 2006, 47, 45–50.17 M. L. Johnson, K. Pietz, D. S. Battleman and R. J. Beyth, Am. J.

Manag. Care, 2004, 10, 926–932.18 M. McDonald, R. P. Hertz, A. N. Unger and M. B. Lustik,

J. Gerontol., Ser. A, 2009, 64, 256–263.19 H. Yanai, Y. Tomono, K. Ito, N. Furutani, H. Yoshida and

N. Tada, Nutr. J., 2008, 7, 10.20 M. P. Dutro, T. D. Gerthoffer, E. D. Peterson, S. S. Tang and

G. A. Goldberg, J. Clin. Hypertens. (Greenwich), 2007, 9, 684–691.21 R. J. Petrella and E. Merikle, Clin. Ther., 2008, 30, 1145–1154.22 X. G. Yao, F. Frommlet, L. Zhou, F. Zu, H. M. Wang, Z. T. Yan,

W. L. Luo, J. Hong, X. L. Wang and N. F. Li, BMC Public Health,2010, 10, 91.

23 K. Duffin, M. Obukowicz, A. Raz and J. J. Shieh, Anal. Biochem.,2000, 279, 179–188.

24 X. Han, Curr. Opin. Mol. Ther., 2007, 9, 586–591.25 F. F. Hsu, A. Bohrer, M. Wohltmann, S. Ramanadham,

Z. M. Ma, K. Yarasheski and J. Turk, Lipids, 2000, 35, 839–854.26 J. Van der Greef, T. Hankemeier and R. N. McBurney, Pharmaco-

genomics, 2006, 7, 1087–1094.27 M. R. Wenk, Nat. Rev. Drug Discovery, 2005, 4, 594–610.28 K. Akira, S. Masu, M. Imachi, H. Mitome, M. Hashimoto and

T. Hashimoto, J. Pharm. Biomed. Anal., 2008, 46, 550–556.29 K. Akira, H. Mitome, M. Imachi, Y. Shida, H. Miyaoka and

T. Hashimoto, J. Pharm. Biomed. Anal., 2010, 51, 1091–1096.30 J. T. Brindle, J. K. Nicholson, P. M. Schofield, D. J. Grainger and

E. Holmes, Analyst, 2003, 128, 32–36.31 Y. Lu, J. A. G. Wang, H. Hao, Q. Huang, B. Yan, W. Zha, S. Gu,

H. Ren, Y. Zhang, X. Fan, M. Zhang and K. Hao, Rapid Commun.Mass Spectrom., 2008, 22, 2882–2888.

32 C. Hu, J. van Dommelen, R. van der Heijden, G. Spijksma,T. H. Reijmers, M. Wang, E. Slee, X. Lu, G. Xu, J. van der Greefand T. Hankemeier, J. Proteome Res., 2008, 7, 4982–4991.

33 C. A. Smith, E. J. Want, G. O’Maille, R. Abagyan and G. Siuzdak,Anal. Chem., 2006, 78, 779–787.

34 M. Murakami and I. Kudo, Phospholipase A(2), J. Biochem.,2002, 131, 285–292.

35 P. Mathieu, P. Poirier, P. Pibarot, I. Lemieux and J. P. Despres,Hypertension (Dallas), 2009, 53, 577–584.

Publ

ishe

d on

18

Oct

ober

201

1. D

ownl

oade

d by

Uni

vers

ity o

f M

ichi

gan

Lib

rary

on

30/1

0/20

14 1

1:56

:53.

View Article Online

This journal is c The Royal Society of Chemistry 2011 Mol. BioSyst., 2011, 7, 3271–3279 3279

36 N. T. Nguyen, C. P. Magno, K. T. Lane, M. W. Hinojosa andJ. S. Lane, J. Am. Coll. Surg., 2008, 207, 928–934.

37 M. F. Saad, M. Rewers, J. Selby, G. Howard, S. Jinagouda,S. Fahmi, D. Zaccaro, R. N. Bergman, P. J. Savage andS. M. Haffner, Hypertension (Dallas), 2004, 43, 1324–1331.

38 P. Deedwania, J. Clin. Hypertens. (Greenwich), 2011, 13, 52–59.39 J. Scholze, E. Alegria, C. Ferri, S. Langham, W. Stevens, D. Jeffries

and K. Uhl-Hochgraeber, BMC Public Health, 2010, 10, 529.

40 SHEP Cooperative Research Group, JAMA, J. Am. Med. Assoc.,1991, 265, 3255–3264.

41 C. Hu, M. Hoene, X. Zhao, H. U. Haring, E. Schleicher,R. Lehmann, X. Han, G. Xu and C. Weigert, PLoS One, 2010,5, e13318.

42 A. Peter, C. Weigert, H. Staiger, K. Rittig, A. Cegan, P. Lutz,F. Machicao, H. U. Haring and E. Schleicher, Am. J. Physiol.Endocrinol. Metab., 2008, 295, E339–E349.

Publ

ishe

d on

18

Oct

ober

201

1. D

ownl

oade

d by

Uni

vers

ity o

f M

ichi

gan

Lib

rary

on

30/1

0/20

14 1

1:56

:53.

View Article Online