588

Identification of novel therapeutic targets and design of effec-tive prevention strategies in cardiometabolic diseases require

the understanding of the molecular pathways linking metabolic perturbations to inflammatory responses, especially in the vascu-lar wall. One potential molecular link integrating metabolic and inflammatory signals is fatty acid–binding protein 4 (FABP4; adipocyte-FABP or aP2 in mice), a lipid chaperone highly expressed both in adipocytes and in macrophages.1–9 In mice,

FABP4 deficiency or pharmacological inhibition of FABP4 protein with an orally active small-molecule protected against atherosclerosis5,6,10,11 and reduced obesity-related metabolic dis-orders and type 2 diabetes mellitus.1,6,12,13 The proatherogenic effects of FABP4 in atherosclerosis have been attributed almost solely to its actions in macrophages.5,10,11 Specifically, FABP4

Background—Fatty acid–binding protein 4 (FABP4 or aP2 in mice) has been identified as a key regulator of core aspects of cardiometabolic disorders, including lipotoxic endoplasmic reticulum stress in macrophages. A functional promoter polymorphism (rs77878271) of human FABP4 gene has been described resulting in reduced FABP4 transcription.

Methods and Results—We investigated the effects of this low-expression variant of FABP4 on cardiovascular morbidity and carotid atherosclerosis on a population level (n=7491) and in patient cohorts representing endarterectomized patients with advanced carotid atherosclerosis (n=92) and myocardial infarction (n=3432). We found that the low-expression variant was associated with decreased total cholesterol levels (P=0.006) with the largest reduction in variant allele homozygotes. Obese variant allele carriers also showed reduced carotid intima-media thickness (P=0.010) and lower prevalence of carotid plaques (P=0.060). Consistently, the variant allele homozygotes showed 8-fold lower odds for myocardial infarction (P=0.019; odds ratio, 0.12; 95% confidence interval, 0.003–0.801). Within the carotid plaques, the variant allele was associated with a 3.8-fold reduction in FABP4 transcription (P=0.049) and 2.7-fold reduction in apoptosis (activated caspase 3; P=0.043). Furthermore, the variant allele was enriched to patients with asymptomatic carotid stenosis (P=0.038). High FABP4 expression in the carotid plaques was associated with lipid accumulation, intraplaque hemorrhages, plaque ulcerations, and phosphoactivated endoplasmic reticulum stress markers.

Conclusions—Our results reveal FABP4 rs77878271 as a novel variant affecting serum total cholesterol levels and cardiovascular risk. A therapeutic regimen reducing FABP4 expression within the atherosclerotic plaque may promote lesion stability through modulation of endoplasmic reticulum stress signaling, and attenuation of apoptosis, lipid burden, and inflammation. (Circ Cardiovasc Genet. 2014;7:588-598.)

Key Words: apoptosis ◼ atherosclerosis ◼ carotid stenosis ◼ cholesterol ◼ coronary artery disease ◼ ER stress ◼ FABP4 protein, human ◼ genetics ◼ stroke

© 2014 American Heart Association, Inc.

Circ Cardiovasc Genet is available at http://circgenetics.ahajournals.org DOI: 10.1161/CIRCGENETICS.113.000499

Received December 5, 2012; accepted June 25, 2014.From the Research Programs Unit, Molecular Neurology, University of Helsinki, Helsinki, Finland (J.S., P.I., K.N., P.M.I., P.J.L.); HUSLAB, Division of

Pathology (M.I.M.), Division of Cardiology, Department of Medicine (J.S., M.S.N.), Department of Neurology (P.I., K.N., M.K., P.J.L.), Helsinki University Central Hospital, Helsinki, Finland; Department of Pathology (M.I.M.), Transplantation Laboratory (M.-L.L.), Haartman Institute, Helsinki University, Helsinki, Finland; Institute for Molecular Medicine (FIMM), University of Helsinki, Helsinki, Finland (M.P., J.K.); Department of Clinical Neurosciences, University of Helsinki, Helsinki, Finland (P.J.L.); Finnish Red Cross Blood Service, Helsinki, Finland (J.T.); Wihuri Research Institute, Helsinki, Finland (E.L.-S., P.T.K.); Department of Chronic Disease Prevention, National Institute for Health and Welfare, Turku, Finland (A.J.); Department of Chronic Disease Prevention, National Institute for Health and Welfare, Helsinki, Finland (A.J., M.P., A.S.H., V.S., J.K., M.J.); and The Estonian Genome Center, University of Tartu, Tartu, Estonia (M.P.).

The Data Supplement is available at http://circgenetics.ahajournals.org/lookup/suppl/doi:10.1161/CIRCGENETICS.113.000499/-/DC1.Correspondence to Jani Saksi, MSc, Research Programs Unit, Molecular Neurology, University of Helsinki, Biomedicum Helsinki, PO Box 700, FI-

00290 Helsinki, Finland. E-mail jani.saksi@helsinki.fi

Low-Expression Variant of Fatty Acid–Binding Protein 4 Favors Reduced Manifestations of Atherosclerotic

Disease and Increased Plaque StabilityJani Saksi, MSc; Petra Ijäs, MD, PhD; Mikko I. Mäyränpää, MD, PhD;

Krista Nuotio, MD, PhD; Pia M. Isoviita, MD; Jarno Tuimala, PhD; Erno Lehtonen-Smeds, MD; Markku Kaste, MD, PhD; Antti Jula, MD, PhD;

Juha Sinisalo, MD, PhD; Markku S. Nieminen, MD, PhD; Marja-Liisa Lokki, MD, PhD; Markus Perola, MD, PhD; Aki S. Havulinna, DSc (tech.); Veikko Salomaa, MD, PhD;

Johannes Kettunen, PhD; Matti Jauhiainen, PhD; Petri T. Kovanen, MD, PhD; Perttu J. Lindsberg, MD, PhD

Clinical Perspective on p 598

Original Article

at Helsinki University on January 6, 2015http://circgenetics.ahajournals.org/Downloaded from at Helsinki University on January 6, 2015http://circgenetics.ahajournals.org/Downloaded from at Helsinki University on January 6, 2015http://circgenetics.ahajournals.org/Downloaded from at Helsinki University on January 6, 2015http://circgenetics.ahajournals.org/Downloaded from at Helsinki University on January 6, 2015http://circgenetics.ahajournals.org/Downloaded from at Helsinki University on January 6, 2015http://circgenetics.ahajournals.org/Downloaded from at Helsinki University on January 6, 2015http://circgenetics.ahajournals.org/Downloaded from at Helsinki University on January 6, 2015http://circgenetics.ahajournals.org/Downloaded from at Helsinki University on January 6, 2015http://circgenetics.ahajournals.org/Downloaded from at Helsinki University on January 6, 2015http://circgenetics.ahajournals.org/Downloaded from at Helsinki University on January 6, 2015http://circgenetics.ahajournals.org/Downloaded from at Helsinki University on January 6, 2015http://circgenetics.ahajournals.org/Downloaded from at Helsinki University on January 6, 2015http://circgenetics.ahajournals.org/Downloaded from at Helsinki University on January 6, 2015http://circgenetics.ahajournals.org/Downloaded from at Helsinki University on January 6, 2015http://circgenetics.ahajournals.org/Downloaded from

Saksi et al FABP4 and Atherosclerotic Plaque Vulnerability 589

regulates the unfolded protein response (UPR),5 an adaptive pro-gram of the cell employed to protect against increased endoplas-mic reticulum (ER) stress.14 In the macrophages of the vascular wall ER stress may arise following the accumulation of free cholesterol into the ER, which distorts its function and activates the UPR pathways to recover ER functionality and to divert unsalvageable cells to apoptosis.15,16 In human atherosclerosis, increased ER stress has been reported in vulnerable coronary plaques in connection with acute coronary syndrome.17

High expression of FABP4 in atherosclerotic carotid plaques (CPs) has been linked with stroke-prone carotid atherosclero-sis.18–21 Interestingly, a functional polymorphism rs77878271 (T-87C) was recently described in the promoter of human FABP4 gene.22 This polymorphism disrupts a CAAT box/enhancer-binding protein α binding sequence and results in reduced transcription of FABP4 gene both in vitro and in adi-pose tissue of carriers in vivo.22 In that study, this low-expres-sion variant of FABP4 was associated with reduced risk of coronary heart disease.22 The same study also demonstrated that the low-expression variant reduced the risk for hypertriglyceri-demia and obesity-related type II diabetes mellitus in man.

We previously discovered FABP4 as one of the most upregu-lated genes in unstable, stroke-associated CPs when compared with clinically silent asymptomatic CPs.18 Although a proposed antiatherogenic mechanism of FABP4 deficiency has been dissected in mice and in in vitro studies, the evidence linking FABP4 to human atherosclerosis is largely missing. Because the promoter variant rs77878271 has been shown to reduce FABP4 expression, we considered this low-expression variant of FABP4 as a naturally occurring genetic model of reduced FABP4 expression in man. We could establish that the variant also decreased FABP4 expression in human atherosclerotic lesions. This enabled us to investigate the pathophysiological mecha-nisms of FABP4 expression in atheroma from the development of early stage atherosclerosis to its effects on lipid accumula-tion, ER stress, and apoptosis in advanced vulnerable lesions and associated atherothrombotic diseases on a population level.

MethodsThe Health 2000 SurveyThe Health 2000 was a large epidemiological health survey performed in Finland from fall 2000 to spring 200123 and included 8028 participants representing the Finnish population aged ≥30 years. In-depth cardiovas-cular examinations were performed in the cardiovascular disease and dia-betes subcohort (SVT-D, sample size 1867 and participation rate 82%), the participants of which were aged 45 to 74 years and living within the catchment areas of 5 Finnish University Hospitals. Detailed informa-tion on clinical examination and disease definitions are provided in the Data Supplement. Carotid ultrasound examinations were performed by high-resolution B-mode equipment as described.24 The reported carotid intima-media thickness (IMT) represents the mean of the common ca-rotid artery IMT and internal carotid IMT. Altogether 1448 genotyped in-dividuals with available carotid ultrasound data participated in the study.

The Helsinki Carotid Endarterectomy StudyPatients with advanced carotid stenosis belong to The Helsinki Carotid Endarterectomy study (HeCES).18,25–27 Briefly, the study included 92 consecutive patients with a high-grade carotid steno-sis (>70%)28 who underwent carotid surgery in Helsinki University Central Hospital during 1997 to 2000. All patients underwent neuro-logical examination, carotid ultrasound and preoperative digital sub-traction angiography, brain imaging, and blood sampling for DNA

and serum. Seventy-four patients had a history of transient isch-emic attack or stroke from carotid stenosis (symptomatic patients). Eighteen patients had never experienced cerebrovascular symptoms or had a history of stroke or transient ischemic attack of other cause (asymptomatic patients). In addition, quantitative real-time reverse transcriptase-polymerase chain reaction (qPCR) and some immuno-histochemical analyses were performed in a subgroup of patients with a clinically defined ischemic stroke or retinal infarct in the ipsilateral carotid territory (n=25) and asymptomatic patients (n=18).

The Corogene StudyThe Corogene study included 5295 consecutive Finnish patients assigned to coronary angiogram in 4 hospitals servicing 1.5 million people in the Hospital District of Helsinki and Uusimaa. Of the Corogene study, 2500 patients with acute coronary syndrome (International Classification of Diseases-Tenth Revision: I20–I25) were included in a genome-wide as-sociation study with a case–control setting.29 Here, we used available genome-wide association data to test the association of rs77878271 with myocardial infarction (MI, International Classification of Diseases-Tenth Revision, I21) using 1565 MI cases from the Corogene cohort and 1867 controls selected from the FINRISK surveys.30

GenotypingThe FABP4 rs77878271 polymorphism was genotyped either by TaqMan SNP Genotyping Assay (Applied Biosystems), by sequenc-ing, or by Illumina 660K BeadChip array and imputational approach-es (in the Data Supplement).

Tissue SamplingThe CPs were removed en bloc in endarterectomy, rinsed with saline, and graded macroscopically. All plaques represented complicated American Heart Association-class VI lesions.31 The plaques were cut into 5 longi-tudinal segments, each containing a portion of the tightest stenosis and destined for a specific purpose: histological and immunohistochemical examinations or RNA and protein extraction as previously described.18,25–27

qPCRqPCR was performed using Assays-on-Demand Gene Expression Products and ABI PRISM 7000 Sequence Detection System (Applied Biosystems) as described.18

Protein Isolation and FABP4 ELISATotal cellular proteins were isolated from the carotid specimens us-ing the Trizol-reagent. FABP4 protein was quantified using a hu-man AFABP ELISA kit (BioVendor) and expressed as picograms of FABP4 per microgram of total protein.

Histology and ImmunohistochemistryPlaque neutral lipids were stained using Oil-red-O (ORO) and ferric iron using Perls’ Prussian blue stainings. Adjacent sections were used in the stainings against CD36 and ATP-binding cassette transporter A1 (ABCA1),27 as well as for colocalization of tissue iron, FABP4, and phosphoactivated ER stress markers. Details of immunostainings are given in the Data Supplement. Terminal deoxynucleotidyl trans-ferase mediated dUTP nick end labeling (TUNEL) analysis was per-formed using In situ Cell Death Detection Kit (Roche). Nonspecific IgGs or PBS served as negative controls.

Light microscopy (Axioplan 2, Carl Zeiss) was performed by in-vestigators blinded to the clinical data. Histological and immunohisto-chemical stainings were graded by eye, apart from the stainings against activated caspase 3 (aCASP3) where the amount of immunoreactive cells was counted by eye and divided by the total area of the specimen giving a relative density for aCASP3 immunoreactive cells.26

StatisticsDifferences in continuous variables were tested by Mann–Whitney U, Kruskal–Wallis, or independent samples t test when indicated and be-tween nominal variables by χ2, Jonckheere–Terpstra, or Fisher's exact test. Bivariate correlations were analyzed by Spearman correlation

at Helsinki University on January 6, 2015http://circgenetics.ahajournals.org/Downloaded from

590 Circ Cardiovasc Genet October 2014

(rs). P values ≤0.05 were considered significant. Data are expressed as

mean±SE unless otherwise indicated. All statistical analyses were per-formed using IBM SPSS Statistics version 19 unless otherwise indicated.

The association of FABP4 rs77878271 with serum total and low-density lipoprotein-cholesterol (total-C and LDL-C) levels and apolipoprotein B (apoB) concentration was examined using linear regression adjusting for age, sex, body mass index (BMI), the use of lipid-lowering medication, and FABP4 rs77878271 genotypes. Carotid IMT was modeled using linear regression adjusting the effect of rs77878271 genotype for age, sex, high-density lipoprotein-cholester-ol, apoB, triglycerides, fasting blood glucose, systolic blood pressure, and smoking. In addition, an interaction between rs77878271 genotype and obesity was included in the model. Variables entered in the models were checked for multicollinearity, and all tolerance values were >0.48 indicating no serious collinearity between explanatory variables.

The genetic association of rs77878271 with MI was tested in Health 2000 (cases, n=532; noncases, n=6959) and Corogene (cases, n=1565; controls, n=1867) data sets separately and by meta-analysis of these 2 cohorts. In these analysis, the matched case–control setting was not used. The association was tested with logistic regression for additive model and the exact Mantel–Haenszel test for the recessive model (as the num-ber of homozygous MI cases was small, n=1) using R (version 2.15.0). Logistic regression model was adjusted for age, sex, and BMI. The as-sociation of FABP4 rs77878271 with incident ischemic cardiovascular events was analyzed in the Health 2000 cohort using Cox proportional hazards regression with age as the time scale. End point data were gath-ered using hospital discharge and causes of death registers from 1971 to 2011. The model was adjusted for sex and rs77878271 genotypes.

Ethical IssuesStudy protocols of the Health 2000, HeCES, and Corogene studies were approved by appropriate Ethics Committees of the Helsinki and Uusimaa Hospital region, and all participants gave signed informed consent.

ResultsPrevalence of the Low-Expression Variant in the Finnish PopulationGenotyping of the FABP4 rs77878271 promoter polymor-phism was successful in 7491 individuals (99% of available DNA samples) of the Health 2000 Survey. The genotype frequencies were 88.7% (TT), 10.9% (TC), and 0.4% (CC; Table 1). The minor allele frequency was 5.8%. The values followed Hardy–Weinberg equilibrium (P=0.58).

Low-Expression Variant of FABP4 Associates With Reduced Serum Total Cholesterol LevelsThere were no significant differences in the baseline char-acteristics and cardiovascular risk factors between FABP4 rs77878271 genotypes (Table I in the Data Supplement). After adjusting for age, sex, BMI, and the use of lipid-lowering medi-cation the FABP4 rs77878271 variant allele demonstrated a sig-nificant dose-dependent association with lower serum total-C levels (P=0.006; n=6868), with consistent trends toward lower LDL-C (P=0.066; n=6840) and apoB (P=0.070; n=6868) con-centration (Table II in the Data Supplement). When compared with major allele homozygotes (TT), TC heterozygotes showed significantly reduced serum total-C (P=0.020; Figure 1A). The effect in TC heterozygotes and in the variant allele homozy-gotes (CC homozygotes) on serum total-C were ≈−2% and −6%, respectively. The effect was largest in CC homozygotes, with the foremost reduction in apoB levels (P=0.068; Fig-ure 1A). No significant associations with FABP4 rs77878271 and apolipoprotein A-I concentration or high-density lipopro-tein-cholesterol levels were detected (data not shown).

Because aP2 deficiency (aP2−/−) in mice is associated with reduced circulating cholesterol levels selectively in obesity,13 we examined whether body weight would have similar modulatory effect on this association in man. There was a clear enhance-ment in the reduction of total and LDL-C levels and apoB con-centration in CC homozygotes in the highest BMI quartile (Q4; BMI>29.5; Figure 1B–1D). In the regression analysis of BMI Q4 alone, CC homozygotes showed substantial reduction in serum total-C (≈−16%; P=0.034) and apoB (≈−20%; P=0.018) levels when compared with TT homozygotes (Figure 1E).

Low-Expression Variant of FABP4 Associates With Lower Carotid Artery IMT and Lower Prevalence of CPs in Obese CarriersNext we studied, whether the low-expression variant was associ-ated with carotid IMT or the frequency of CPs, early surrogate markers of cardiovascular risk. The mean IMT in the SVT-D sub-cohort was 0.94±0.23 mm (Table III in the Data Supplement).

Table 1. FABP4 rs77878271 Genotype Frequencies and Cardiovascular End Points in The Health 2000 Survey at Baseline and During 10-Year Follow-Up

Groups All TT TC CC P geno* P Call† P Tall‡

FABP4 rs77878271 genotype, n (%) 7491 6647 (88.7) 816 (10.9) 28 (0.4) ... ... ...

Baseline health examination, n (%)

Ischemic stroke 255 (3.4) 229 (3.4) 26 (3.2) 0 (0.0) 0.848 1.000 0.687

Ischemic stroke or TIA 414 (5.5) 376 (5.7) 38 (4.7) 0 (0.0) 0.272 0.402 0.175

Myocardial infarction 377 (5.0) 336 (5.1) 41 (5.0) 0 (0.0) 0.706 0.400 0.867

Coronary heart disease 933 (12.5) 829 (12.5) 102 (12.5) 2 (7.1) 0.816 0.569 0.956

Cumulative end points in 10-year follow-up, n (%)

Ischemic stroke 403 (5.4) 360 (5.4) 42 (5.1) 1 (3.6) 0.934 1.000 0.746

Ischemic stroke or TIA 655 (8.7) 590 (8.9) 64 (7.8) 1 (3.6) 0.491 0.509 0.272

Myocardial infarction 532 (8.1) 474 (8.2) 58 (7.9) 0 (0.0) 0.453 0.260 0.831

Major coronary event 771 (11.7) 688 (11.8) 83 (11.3) 0 (0.0) 0.202 0.109 0.674

Ischemic cardiovascular event, n (%) 1230 (16.4) 1102 (16.6) 127 (15.6) 1 (3.6) 0.131 0.073 0.324

Major coronary events include myocardial infarctions, unstable angina pectoris, history of angioplasty, or bypass surgery; ischemic cardiovascular events include major coronary events, ischemic stroke, or TIA. TIA indicates transient ischemic attack.

Fisher's exact test or χ2 test for the difference between genotypes *(TT vs TC vs CC) and alleles †(TT/TC vs CC), ‡(TT vs. TC/CC).

at Helsinki University on January 6, 2015http://circgenetics.ahajournals.org/Downloaded from

Saksi et al FABP4 and Atherosclerotic Plaque Vulnerability 591

There were no significant differences in the mean IMT values between genotypes (0.94±0.23 mm in TT, 0.93±0.22 mm in TC, and 0.97±0.32 mm in CC) or between T and C allele carri-ers. Similarly, no differences were seen in the frequency of CPs (15.9% in TT, 13.0% in TC, and 14.3% in CC).

Studies in aP2−/− mice suggest that the effects of aP2 defi-ciency in vivo manifest under cumulative metabolic strain (ie, in obesity).1,13 Therefore, we stratified the SVT-D sub-cohort according to obesity (BMI≥30; Table IV in the Data Supplement). In obese individuals, the low-expression vari-ant was associated with a significantly lower carotid IMT (0.98±0.22 mm in TT versus 0.90±0.16 mm in TC/CC; P=0.010), as well as a 3.4-fold reduction in the prevalence of CPs (17.1% versus 5.1%; P=0.060; Table 2). These findings were consistent in both sexes although the reduction in CPs was most prominent in obese women (CP prevalence’s were 16.6% in TT versus 0.0% TC/CC; P=0.029; Table 2).

In multivariate analysis of the whole SVT-D subcohort adjusting for known risk factors for atherosclerosis (age, sex,

high-density lipoprotein-cholesterol, apoB, triglycerides, fast-ing blood glucose, systolic blood pressure, and smoking), the low-expression variant or obesity alone had no significant effect on carotid IMT. However, if we allow for the interaction between the low-expression variant and obesity, it is signifi-cant (P=0.021; Table V in the Data Supplement), suggesting that the obese carriers of the low-expression variant have sig-nificantly lower carotid IMT when compared with obese TT homozygotes.

Homozygote Carriers of the Low-Expression Variant Show Low Rate of Cardiovascular EventsIn the Health 2000 Survey, there were 28 individuals homo-zygous for the low-expression variant of FABP4 (CC). At the baseline health examination, they had no history of ischemic strokes (0% versus 3.4% in TT and 3.2% in TC genotype) or MIs (0% versus 5.1% in TT and 5.0% in TC genotype; Table 1). Similarly, during a 10-year follow-up, there were no new cases of MIs or major coronary disease events among the

-0.2

0

0.2

0.4

-0.4

-0.6

-0.8

-1.0

and

95%

CI

0.6

Total cholesterol(mmol/L)n=6868

LDL cholesterol(mmol/L)n=6840

ApoB(g/L)

n=6868

ref. ref. ref.

-0.357

-0.099-0.197

TC CCTTRs77878271 genotype

0.6

1.0 B

0

4.0

6.0

Tota

l cho

lest

erol

(mm

ol/L

)

Rs77878271TTTCCC

(n(n)(n

8.0

BMI quartiles

Q1 Q2 Q3 Q4

1679(193)

9

1598(200)

8

1626(221)

5

1692(197)

6

C

0

2.0

4.0

LDL

chol

este

rol (

mm

ol/L

)

Rs77878271TTTCCC

(n(n)(n

5.0

BMI quartiles

Q1 Q2 Q3 Q4

1670(193)

9

1592(200)

8

1620(219)

5

1686(197)

6

6.0

D

0

1.0

1.5

Apo

B (g

/L)

Rs77878271TTTCCC

(n(n)(n

2.0

BMI quartiles

Q1 Q2 Q3 Q4

1679(193)

9

1598(200)

8

1626(221)

5

1692(197)

6

A

E

ref. ref. ref.

-0.5

0

0.5

-1.0

-1.5

and

95%

CI

1.0

Total cholesterol(mmol/L)n=1790

LDL cholesterol(mmol/L)n=1784

ApoB(g/L)

n=1790

-0.963

-0.658

TC CCTTRs77878271 genotype

1.5

-0.273

-2.0

β

β

Figure 1. The low-expression variant of fatty acid–binding protein 4 (FABP4) associates with reduced serum total cholesterol levels. A forest plot of generalized linear regression analyses of total and low-density lipoprotein (LDL) cholesterol levels, and apolipoprotein B (apoB) concentration in the Health 2000 cohort adjusted for age, sex, body mass index (BMI), the use of lipid-lowering medication, and FABP4 rs77878271 genotypes (A). In the model, men were coded as 1 and women as 2; the use of lipid-lowering medication as 1 or 0 (no lipid-lowering medication); FABP4 rs77878271 genotypes as TT (major allele homozygotes), TC, and CC. Major allele homozygotes serve as the reference (ref.). β values represent the difference that one category change has in each independent variable. Data represent β±95% confidence interval (CI), the exact point estimates are shown for variant allele homozygotes (CC). Total and LDL cholesterol levels, and apoB concentration according to BMI quartiles (Q, the number of individuals in each genotype is given above the quartiles, B–D). BMI in Q4>29.5. Data represent mean±SE. Generalized linear regression (as in panel A) in BMI Q4 (E).

at Helsinki University on January 6, 2015http://circgenetics.ahajournals.org/Downloaded from

592 Circ Cardiovasc Genet October 2014

CC homozygotes (0% versus 2.5% and 5.2% in TT and 2.8% and 5.0% in TC genotype, respectively). One ischemic stroke occurred among the CC homozygotes. The cumulative preva-lences were 0% in CC homozygotes versus 11.8% in TT and 11.3% in TC genotype for major coronary disease events (TT/TC versus CC; P=0.109) and 3.6% in CC homozygotes when compared with 16.6% in TT and 15.6% in TC genotype for all major ischemic cardiovascular events (TT/TC versus CC; P=0.073; Table 1). Similarly, in Cox regression analysis, the hazard ratio for incident ischemic cardiovascular events for CC homozygotes was 0.33 (TT versus CC; 95% confidence interval [CI], 0.05 to 2.35; P=0.268) and for TC heterozygotes 0.90 (TT versus TC; 95% CI, 0.73 to 1.10; P=0.264; Figure I in the Data Supplement).

We further explored the association of rs77878271 with MI in a large case–control data set, derived from the Corogene study,29 with available genome-wide genotyping data using imputational approaches. In this data set, using a recessive model, the rs77878271 CC homozygotes showed 6-fold lower odds for MI (P=0.08; odds ratio, 0.17; 95% CI, 0.004–1.33; MI, n=1565; controls, n=1867; imputation info 0.97; minor allele frequency 5.9%; Hardy–Weinberg equilibrium from best guess genotypes P=0.90). In a pooled analysis of Health 2000 and Corogene data sets, using a recessive model, the rs77878271 CC homozygotes showed 8-fold lower odds for MI (P=0.019; odds ratio, 0.12; 95% CI, 0.003–0.801, MI, n=2097; controls, n=8826). Rs77878271 was not associated with MI using an additive model and logistic regression (P=0.41).

Low-Expression Variant of FABP4 Associates With Stable Carotid StenosisWe next analyzed the low-expression variant of FABP4 in the HeCES cohort consisting of patients with advanced (>70%)28 carotid atherosclerosis25 (n=92; Table VI in the Data Supple-ment). Among patients with advanced carotid atherosclerosis, 12.1% were heterozygote carriers of the low-expression vari-ant with a minor allele frequency of 6.0%, which did not differ from population controls (Table VII in the Data Supplement). Genotype frequencies followed Hardy–Weinberg equilibrium (P=0.54). However, patients with stable, clinically asymptom-atic carotid atherosclerosis demonstrated higher than expected heterozygosity frequency of the low-expression variant when compared with symptomatic patients (27.8% versus 8.2%; P=0.038; Table VII in the Data Supplement). In addition, patients with asymptomatic carotid disease showed a differ-ence in allele distribution when compared with the Health 2000 survey population controls (P=0.045).

FABP4 Expression Is Elevated in Unstable CPs and Reduced in the Carriers of the Low-Expression VariantPreviously, we identified significant overexpression of FABP4 mRNA in stroke-associated when compared with asymptom-atic CPs in a genome-wide analysis.18 A 2.2-fold overexpression was confirmed using qPCR in stroke-associated when compared with asymptomatic CPs (1.95±0.37 versus 0.90±0.32; P=0.015; Figure 2A). In line with the qPCR experiment, FABP4 protein quantification by ELISA from the same CPs revealed a marked 3.9-fold higher level of FABP4 protein in the stroke-associated when compared with asymptomatic CPs (30.2±8.0 versus 7.7±1.5 pg/μg of total protein; P=0.003; Figure 2B). FABP4 pro-tein levels and mRNA expression in the CPs were highly corre-lated (r

S=0.695; P<0.001 not shown). Because the low-expression

variant carriers display reduced FABP4 transcription in adipose tissue,22 we examined whether a similar effect could be observed in endarterectomized CPs. Patients carrying the low-expression variant of FABP4 showed a 3.8-fold reduction in FABP4 mRNA expression (0.44±0.20 versus 1.68±0.30; P=0.049) and a trend toward lower FABP4 protein levels (6.2±1.6 versus 22.5±5.5 pg/μg of total protein; P=0.107) in their CPs when compared with TT homozygotes (Figure 2C and 2D).

FABP4 Expression Associates With Lipid and Macrophage Accumulation Into Carotid AtheromaAnimal studies have suggested that FABP4 deficiency promotes cholesterol efflux and resistance toward cholesterol accumulation in macrophages.2,5 However, the relationship of FABP4 expres-sion with lipid accumulation and inflammation in carotid ather-oma has not been studied. We analyzed the relationship between intra- and extracellular neutral lipid accumulation (ORO) and FABP4 expression in the CPs. We found that FABP4 expres-sion was correlated with the degree of carotid stenosis (mRNA, r

S=0.334; P=0.037 and protein, r

S=0.521; P=0.001, not shown).

In the CPs, increasing FABP4 immunoreactivity was highly cor-related with increasing HAM56 immunoreactive area detect-ing macrophages (P=0.008; Figure II in the Data Supplement). Increasing FABP4 immunoreactivity correlated with increas-ing intra- and extracellular ORO staining grades (P=0.031 and P=0.002, respectively; Figure 3A–3C). In part, cellular lipid accumulation in atheroma is regulated by CD36-mediated LDL uptake in proportion to balancing efflux through ABCA1 and ATP-binding casette transporter G1 (ABCG1). We found that FABP4 mRNA expression and protein levels were correlated with CD36 mRNA levels (r

S=0.863; P<0.001 for mRNA expres-

sion and rS=0.786; P<0.001 for protein levels, respectively, not

shown) and with immunoreactive area of CD36 protein (P=0.110

Table 2. Association Between FABP4 rs77878271 Genotype, Carotid IMT, and Carotid Plaques in the Obese (body mass index, ≥30)

FABP4 Genotypes

Women Men All

TT TC/CC P Value TT TC/CC P Value TT TC/CC P Value

n 168 27 ... 136 15 ... 304 42 ...

Carotid IMT, mm* 0.96±0.22 0.88±0.15 NS 1.00±0.23 0.94±0.18 NS 0.98±0.22 0.90±0.16 0.010

Carotid plaques, % 16.6 0.0 0.029 17.8 15.4 NS 17.1 5.1 0.060

IMT indicates intima-media thicknesses; and NS, nonsignificant.*Mean of the IMTs of common and internal carotid arteries.

at Helsinki University on January 6, 2015http://circgenetics.ahajournals.org/Downloaded from

Saksi et al FABP4 and Atherosclerotic Plaque Vulnerability 593

and P=0.034, respectively; Figure 3D and 3E). This associa-tion could not be demonstrated between FABP4 expression and ABCA1 immunoreactivity. However, FABP4 protein levels were correlated with the ratio of CD36/ABCA1 immunoreactivity27 (r

S=0.595; P=0.002, not shown). FABP4 protein levels were 4.5-

fold higher in those CPs where the immunoreactivity for CD36 exceeded that for ABCA1 (55.8±18.3 versus 12.3±3.66 pg/μg of total protein; P=0.009; Figure 3F).

Plaque Ulcerations and Intraplaque Hemorrhages Associate With FABP4 ExpressionMacroscopic ulcerations are strongly associated with symp-tomatic carotid disease because they indicate a recent throm-boembolic event.32 We discovered a 2.8-fold upregulation of FABP4 mRNA (2.2±0.44 versus 0.78±0.24; P=0.005) and a 4.4-fold elevation of FABP4 protein (32.5±9.4 versus 7.4±1.1 pg/μg of total protein; P=0.001) in CPs with macroscopic ulcer-ations when compared with nonulcerated CPs (Figure 4A and 4B). Moreover, CPs that were both ulcerated and ipsilateral to a carotid territory cerebral infarction had 4-fold higher FABP4 protein levels when compared with ulcerated plaques in asymp-tomatic patients (41.0±12.2 versus 10.2±4.4 pg/μg of total protein, respectively; P=0.046; Figure 4C). FABP4 immunore-activity was also increased in CPs with intraplaque hemorrhages (IPHs) when compared with CPs without IPHs (P<0.001; Fig-ure 4D). Frequent IPHs promote in situ accumulation of remnant tissue iron that may further provoke oxidative stress, another recognized atheromatous UPR activator.33 We found a striking colocalization between FABP4 immunoreactivity and tissue iron (Prussian blue staining), the surrogate marker of previous IPHs (Figure 5F). Consistently, there was a strong association between Prussian blue staining and FABP4 immunoreactivity (P<0.001; Figure 4E). Increasing FABP4 immunoreactivity was also highly correlated with platelet (CD42b; P=0.001) and with fibrin immunoreactivity (P<0.001), both distinctive for more recent IPHs (Figure 4F and 4G). Because the atherogenic effects of FABP4 in the atheroma are almost completely mediated via macrophage functions,5,10,11 we investigated the connection between FABP4 expression and CD163-mediated macrophage hemoglobin scavenging in the CPs and found that the expression

of FABP4 and CD163 mRNA (rS=0.828; P<0.001, not shown)

and their immunoreactivities (P<0.001; Figure 4H) were corre-lated. The carriers of the low-expression variant also had 1.6-fold lower CD163 mRNA expression in their CPs when compared with major allele homozygotes (1.33±0.09 versus 0.81±0.19, respectively; P=0.022; Figure 4I), despite equal frequency of IPHs in both genotypes (45% versus 52%, respectively).

FABP4 Expression Associates With ER Stress in Carotid AtheromaIn mice, aP2−/− has been shown to reduce macrophage apoptosis in atherosclerotic lesions in vivo through alleviation of lipotoxic ER stress.5 The low-expression variant of FABP4 enables similar investigations on the potential relation between FABP4 expres-sion and chronic ER stress in advanced CPs. For this purpose, we quantified apoptosis markers aCASP3 and TUNEL reactivity in the CPs. We found that increasing FABP4 immunoreactivity was associated with higher relative density of aCASP3 immu-noreactive cells and TUNEL reactivity in the CPs (P=0.019 and P=0.008, respectively; Figure 5A and 5B). This association was also evident between FABP4 protein levels in the CPs and the relative density of aCASP3 immunoreactive cells and TUNEL reactivity (r

S=0.487; P=0.003 and P=0.023, respectively; Fig-

ure IIIA and IIIB in the Data Supplement). The relative density of aCASP3 immunoreactive cells and TUNEL reactivities also strongly associated with increasing intracellular ORO staining grades (P=0.003 and P=0.006, respectively; Figure IIIC and IIID in the Data Supplement). In line, FABP4 immunoreactiv-ity, intracellular ORO staining grades, and the relative density of aCASP3 immunoreactive cells were positively associated with each other (Figure 5C). In the low-expression variant car-riers, the relative density of aCASP3 immunoreactive cells was reduced by 2.7-fold (1.10±0.37 versus 3.00±0.60; P=0.043), and a trend toward less TUNEL reactivity was observed when compared with TT homozygotes (P=0.095; Figure 5D and 5E). Immunohistochemical stainings of FABP4 together with suc-cessively phosphoactivated ER stress markers, phosphorylated pancreatic ER kinase and phosphorylated eukaryotic translation initiation factor 2 alpha, and apoptosis marker, aCASP3 from consecutive CP sections, revealed the presence of these UPR

0

0.5

1.0

1.5

2.0

TT TC

Rel

ativ

e FA

BP

4/A

CTB

mR

NA

C P=0.049

TT TC0

10

20

30

FAB

P4

prot

ein

pg/µ

g of

tota

l pro

tein

D P=0.107

0

10

20

30

40

FAB

P4

prot

ein

pg/µ

g of

tota

l pro

tein

B P=0.003

0

0.5

1.0

1.5

2.0

2.5

Rel

ativ

e FA

BP

4/A

CTB

mR

NA

A P=0.015

Figure 2. Fatty acid–binding protein 4 (FABP4) expression is induced in unstable carotid plaques and reduced in the low-expression variant carriers (TC). Relative FABP4 mRNA expression (A; n=39) and protein levels (B; n=39) in stroke-associated when compared with asymptomatic carotid plaques. FABP4 mRNA expression (C) and protein levels in the carotid plaques of the low-expression variant carri-ers (n=5; D) when compared with major allele homozygotes (TT, n=34). Data represent mean±SE.

at Helsinki University on January 6, 2015http://circgenetics.ahajournals.org/Downloaded from

594 Circ Cardiovasc Genet October 2014

markers with apoptosis in macrophage-rich areas of the CPs (Figure 5F).

DiscussionGenetic models and in vitro studies have outlined a crucial role for aP2, the mouse orthologue of human FABP4, in the cross-roads of metabolic and cardiovascular diseases.1,5,6,10–13 However, data on FABP4 in human atherosclerosis are scarce. The present study shows that FABP4 expression is high in unstable CPs and associates with the cardinal features of plaque vulnerability—lipid and macrophage accumulation, ulcerations, IPHs—and with ER stress and increased apoptosis. Our results suggest that a naturally occurring genetic low-expression variant of FABP4 associates not only with lower circulating serum total-C levels and reduced FABP4 expression and apoptosis within the carotid atheroma but also with lower risk for developing large artery

atherosclerosis in metabolically strained obese individuals on a population level. In line, a greater proportion of patients with clinically asymptomatic advanced carotid disease were carriers of the low-expression variant when compared with patients with symptomatic disease. Furthermore, CC homozygotes, with the largest reduction in serum total-C levels, also showed an 8-fold reduced odds for MI and a trend toward lower frequency of all ischemic cardiovascular events. Therefore, FABP4 rs77878271 may represent a novel variant affecting both serum total-C levels and cardiovascular risk and suggests that genetically reduced FABP4 expression within the atherosclerotic plaque may modulate ER stress signaling and attenuate apoptosis, lipid burden, and inflammation with a combined effect of promoting atherosclerotic plaque stability.

Circulating total and LDL-C levels are heritable risk fac-tors for cardiovascular diseases, and as such subject for active

0

20

40

60

80

100

0-2525-5050-7575-100

FABP4 immunoreactive area %

% B

Intracellular ORO staining area (%)0 0-33 33-66 66-100

P=0.031

0

20

40

60

80

100

0-2525-5050-7575-100

FABP4 immunoreactive area %

% C

Extracellular ORO staining area (%)0 0-33 33-66 66-100

P=0.002

A Intracellular ORO staining pattern Extracellular ORO staining pattern

Rel

ativ

e FA

BP

4/A

CTB

mR

NA

D

0

1.0

4.0

0 1 2 3CD36 immunoreactive area (%)

P=0.110

2.0

3.0

FAB

P4

prot

ein

pg/µ

g of

tota

l pro

tein

E

0

10

40

50

0 1 2 3CD36 immunoreactive area (%)

P=0.034

20

30

60

0

20

40

60

FAB

P4

prot

ein

pg/µ

g of

tota

l pro

tein

CD36/ABCA1immunoreactivity

FP=0.009

80

>1 <1

Figure 3. Fatty acid–binding protein 4 (FABP4) expression associates with lipid accumulation into carotid atheroma. FABP4 immunore-activity and intra- and extracellular neutral lipid accumulation (A) defined by the association or absence of nuclei with Oil-red-O (ORO) positive lipid droplets, scale bar, 20 μm (B and C; n=72). FABP4 mRNA expression and protein amount and CD36 mRNA expression and immunoreactivity (D, n=39; E, n=28) in carotid plaques. FABP4 protein levels in carotid plaques according to the ratio of CD36/ATP-bind-ing cassette transporter A1 (ABCA1) immunoreactivity (ratio >1; F, n=24). ACTB indicates beta-actin; and Data, mean±SE.

at Helsinki University on January 6, 2015http://circgenetics.ahajournals.org/Downloaded from

Saksi et al FABP4 and Atherosclerotic Plaque Vulnerability 595

therapeutic intervention.34 We report here a novel association between the low-expression variant of FABP4 and reduced serum total-C levels, with the most prominent effect in CC homozygotes (Figure 1A). This is in line with a previous result suggesting a trend toward lower apoB levels in the low-expression variant carriers.22 The reduction in serum total-C and apoB levels was most prominent, ≈−16% and -20%, respec-tively, in the highest BMI quartile among the CC homozygotes (Figure 1E). These results suggest that the reduction in serum total-C might, in part, be mediated via other apoB-containing particles than LDL. In aP2−/− mice, on either normal chow or on high fat diet, circulating cholesterol levels remain unaffected.1 However, when crossed to a leptin-deficient background (ob/

ob-aP2−/−), a genetic model of obesity, the mice show a sig-nificant reduction in circulating cholesterol and triglyceride lev-els.13 The enhanced reduction in serum total-C and apoB levels in the highest BMI quartile of CC homozygotes suggests that body weight and the cholesterol-lowering effect might be mech-anistically linked also in man. One such mechanistic option is that the adipose tissue–specific reduction of FABP4 expression seen in the low-expression variant carriers22 mediates analogous enhancement of adipocyte de novo lipogenesis as reported in the lipid chaperone-deficient mice with reduced cholesterol levels and highly beneficial metabolic phenotype.4,35 At pres-ent, no evidence exists linking reduced FABP4 expression in white adipose tissue to enhancement of de novo lipogenesis and

0

20

40

60

80

100

0-2525-5050-7575-100

Prussian blue staining area %

%

EP<0.001

FABP4 immunoreactive area % 0-25 25-50 50-75 75-100

0

1.0

2.0

3.0

Rel

ativ

e FA

BP

4/A

CTB

mR

NA

AP=0.005

0

20

40

60

80

100

00-5050-100

CD163 staining area %

% H

P<0.001

FABP4 immunoreactive area %0-25 25-50 50-75 75-100

0

20

40

60

80

100

00-11-10>10

CD42b staining area %

%

FABP4 immunoreactive area %0-25 25-50 50-75 75-100

P=0.001

F

0

10

20

30

40

50

FABP

4 pr

otei

n pg

/µg

of to

tal p

rote

in

BP=0.001

0

10

20

30

40

50

60

Nonulcerated Ulcerated

FABP

4 pr

otei

n pg

/µg

of to

tal p

rote

in

Asymptomatic

Stroke-associated

CP=0.046

0

20

40

60

80

100

<1010-50>50

Fibrin staining area %

%

GP<0.001

FABP4 immunoreactive area %0-25 25-50 50-75 75-100

D

0

20

40

60

80

100%

0-2525-5050-7575-100

FABP4 immunoreactive area %

P<0.001

0

0.5

1.0

1.5

TT TC

Rel

ativ

e C

D16

3/A

CTB

mR

NA

I P=0.022

Figure 4. Plaque ulcerations and intraplaque hemorrhages associate with increased fatty acid–binding protein 4 (FABP4) expression. FABP4 mRNA expression and protein levels in carotid plaques with macroscopic ulcerations (A, n=19 vs 19; B, n=20 vs 18) and in ulcer-ated stroke-associated when compared with ulcerated asymptomatic carotid plaques (C, n=13 vs n=5). FABP4 immunoreactivity and intraplaque hemorrhages (IPHs; D, n=76), remnant tissue iron (Prussian blue staining; E, n=75 and Figure 5F), platelets (CD42b immuno-reactivity; F, n=68) and fibrin immunoreactivity (G, n=65) in the carotid plaques. FABP4 immunoreactivity and CD163 immunoreactivity (H, n=34) in the carotid plaques. CD163 mRNA expression in the carotid plaques of the low-expression variant carriers (TC; I, n=11) when compared with major allele homozygotes (TT; n=79). Data represent mean±SE.

at Helsinki University on January 6, 2015http://circgenetics.ahajournals.org/Downloaded from

596 Circ Cardiovasc Genet October 2014

improved insulin sensitivity in man. However, recent evidence indicates that activated de novo lipogenesis as an independent mechanism predicts metabolic health in humans.36

In the Health 2000 population cohort, we discovered that the low-expression variant of FABP4 was associated with reduced carotid IMT among obese (BMI≥30) carriers (Table 2). Higher

FABP4 P-PERK P-eIF2α

PRUSSIAN BLUE CD68aCASP3

F

Rel

ativ

e de

nsity

of a

CA

SP

3 im

mun

orea

ctiv

e ce

lls

A

0

20

40

60

80

100

0%0-33%33-66%66-100%

TUNEL staining area %

% B

FABP4 immunoreactive area %0-25 25-50 50-75 75-100

P=0.008

0

2.0

4.0

6.0

8.0

10.0

0-25 25-50 50-75 75-100FABP4 immunoreactive area %

P=0.019

Rel

ativ

e de

nsity

of a

CA

SP

3 im

mun

orea

ctiv

e ce

lls

TT TC0

1.0

2.0

3.0

4.0 P=0.043

D E

0

20

40

60

80

100

TT TC

%

0%0-33%33-66%66-100%

TUNEL staining area %

P=0.095

Rel

ativ

e de

nsity

of a

CA

SP

3

imm

unor

eact

ive

cells

C

0

2

4

6

8

0-25 25-50 50-75 75-100FABP4 immunoreactive area %

10

1200-33%33-66%66-100%

Intracellular ORO staining %

Figure 5. Fatty acid–binding protein 4 (FABP4) expression associates with endoplasmic reticulum (ER) stress markers in carotid atheroma. FABP4 immunoreactivity and the relative density of activated caspase 3 (aCASP3) immunoreactive cells (A, n=72) and TUNEL reactivity (B, n=75) in the carotid intima. Positive association between FABP4 immunoreactivity, intracellular Oil-red-O staining grades, and relative density of aCASP3 immunoreactive cells (C). Relative density of aCASP3 immunoreactive cells (D) and TUNEL reactivity (E) in the carotid plaques of the low-expression variant carriers (TC; n=11) when compared with major allele homozygotes (TT; n=76 and n=73, respectively). Immunohistochemi-cal stainings of FABP4, successively phosphorylated pancreatic ER kinase (P-PERK), and eukaryotic translation initiation factor 2α (P-eIF2α), tis-sue iron (Prussian blue), aCASP3, and CD68 (macrophages) in consecutive carotid sections (scale bar, 50 μm; F). Data represent mean±SE.

at Helsinki University on January 6, 2015http://circgenetics.ahajournals.org/Downloaded from

Saksi et al FABP4 and Atherosclerotic Plaque Vulnerability 597

carotid IMT is a marker of subclinical atherosclerosis that directly associates with increased risk for MI and stroke.37,38 In line with reduced IMT, the obese low-expression variant car-riers also demonstrate 3× less atherosclerotic CPs (Table 2). Importantly, these atheroprotective effects of the low-expres-sion variant manifest during obesity, a metabolically stressed state. Consistently in aP2-deficient mice, obesity has been shown to trigger the emergence of a favorable metabolic pheno-type characterized with improved lipid metabolism, peripheral insulin action, and preserved pancreatic β-cell function result-ing in protection against obesity-induced insulin resistance and type 2 diabetes mellitus.1,13 Our results indicate a previously unrecognized role for the low-expression variant of FABP4 in reducing pathological arterial remodeling and retarding ather-oma progression in metabolically challenged high-risk individ-uals. However, additional studies in larger cohorts are needed to confirm the association between reduced carotid IMT and the low-expression variant of FABP4 in obese individuals.

The CC homozygotes showed a low rate of cardiovascular events at baseline and there was only 1 new case of stroke and no new cases of MI or major coronary events during 10-year follow-up (Table 1). However, the small homozygosity fre-quency (0.4%) limited statistical power in the end point analyses of the population cohort, and the cardioprotective effect of this genotype was borderline significant (P=0.073; Table 1). When pooling all prevalent cases in the Health 2000 and the Corogene data sets, CC homozygotes showed an 8-fold reduced odds for MI (P=0.019; odds ratio, 0.12; 95% CI, 0.003–0.801). This find-ing is substantiated by our novel observation of reduced serum total-C levels, one of the most important risk factors of MI, in the low-expression variant carriers with the most prominent cho-lesterol-lowering effect coinciding with the 8-fold lower odds for MI. This is also in line with the reported atheroprotective effect of the low-expression variant on coronary heart disease22 and the atheroprotective role of FABP4 deficiency in mice.5,10,11 The rs77878271 variant has not been included in the recent large meta-analyses of genome-wide association studies analyz-ing genetic variants associated with MI, stroke, or plasma lipid traits, and hence explains why it has remained undetected.34,39,40

FABP4 expression in the CPs was detected in the same his-tological areas together with successively phosphoactivated ER stress markers and positively associated with apoptosis mark-ers, aCASP3 and TUNEL reactivity (Figure 5). Consistently, the carriers of the low-expression variant of FABP4 had a 2.7-fold reduction in the relative density of aCASP3 immunoreac-tive cells and showed a trend toward lower TUNEL reactivity in their CPs (Figure 5D and 5E). The genetic deficiency or chemi-cal inhibition of FABP4 has been shown to reduce lipotoxic ER stress and macrophage apoptosis in atherosclerotic lesions of mice.5 In humans, UPR-mediated macrophage apoptosis in advanced atherosclerotic lesions has been postulated to require several simultaneous noxious ER stress signals to trigger cell death.16,33 Perhaps, by reducing FABP4 expression and con-sequently FABP4-mediated lipotoxic ER stress, the atheroma cells in the low-expression variant carriers are able to evade apoptosis in situations that otherwise would lead to cell death. It is intriguing to speculate that similar enhancement of de novo lipogenesis with generation of protective bioactive lipids, such as C16:1n7-palmitoleate, that is afforded by FABP4 deficiency

in mice could also promote macrophage survival in athero-sclerotic lesions of the low-expression variant carriers.5 These results suggest that the low-expression variant is not only asso-ciated with reduced FABP4 transcription but also could confer an effect on cell viability in the atherosclerotic lesion.

In conclusion, we report here novel data on multiple levels of evidence suggesting an important role for FABP4 in pathophysi-ological cascades of human atherosclerotic disease. This evi-dence suggests that a naturally occurring genetic low-expression variant is atheroprotective and associates with reduced serum total-C levels and FABP4 expression in atherosclerotic lesions thus promoting plaque stability and reducing the risk of cardio-vascular events. Our data on human atherosclerotic plaques sup-port the view that genetically reduced FABP4 expression leads to an alleviation of ER stress as proposed by previous studies in deficient animal models. In mice, an orally active small-mole-cule inhibitor of aP2 has been demonstrated not only to prevent atherosclerosis development but also to retard already formed vascular lesions.6 Thus, pharmacological inhibition of FABP4 may be an effective strategy to manage atherosclerosis progres-sion and to prevent its serious thromboembolic complications, stroke, and premature cardiovascular death.

AcknowledgmentsNancy Lim and Taru Puhakka are thanked for skilful technical assistance.

Sources of FundingThis study was funded by Helsinki University Central Hospital re-search grants, the Academy of Finland and the Sigrid Jusélius, Lundbeck, Aarne and Aili Turunen, Finnish Angiology, The Ida Montin, The Finnish Medical, Finnish Brain, Maire Taponen and Paavo Nurmi Foundations. Wihuri Research Institute is maintained by the Jenny and Antti Wihuri Foundation.

DisclosuresNone.

References 1. Hotamisligil GS, Johnson RS, Distel RJ, Ellis R, Papaioannou VE, Spiegelman

BM. Uncoupling of obesity from insulin resistance through a targeted mutation in aP2, the adipocyte fatty acid binding protein. Science. 1996;274:1377–1379.

2. Makowski L, Brittingham KC, Reynolds JM, Suttles J, Hotamisligil GS. The fatty acid-binding protein, aP2, coordinates macrophage cholesterol trafficking and inflammatory activity. Macrophage expression of aP2 im-pacts peroxisome proliferator-activated receptor gamma and IkappaB ki-nase activities. J Biol Chem. 2005;280:12888–12895.

3. Hotamisligil GS. Inflammation and metabolic disorders. Nature. 2006;444:860–867.

4. Cao H, Gerhold K, Mayers JR, Wiest MM, Watkins SM, Hotamisligil GS. Identification of a lipokine, a lipid hormone linking adipose tissue to sys-temic metabolism. Cell. 2008;134:933–944.

5. Erbay E, Babaev VR, Mayers JR, Makowski L, Charles KN, Snitow ME, et al. Reducing endoplasmic reticulum stress through a macrophage lipid chaperone alleviates atherosclerosis. Nat Med. 2009;15:1383–1391.

6. Furuhashi M, Tuncman G, Görgün CZ, Makowski L, Atsumi G, Vaillan-court E, et al. Treatment of diabetes and atherosclerosis by inhibiting fatty-acid-binding protein aP2. Nature. 2007;447:959–965.

7. Cabré A, Babio N, Lázaro I, Bulló M, Garcia-Arellano A, Masana L, et al. FABP4 predicts atherogenic dyslipidemia development. The PREDIMED study. Atherosclerosis. 2012;222:229–234.

8. Cabré A, Lázaro I, Girona J, Manzanares JM, Marimón F, Plana N, et al. Fatty acid binding protein 4 is increased in metabolic syndrome and with thiazolidin-edione treatment in diabetic patients. Atherosclerosis. 2007;195:e150–e158.

9. Aragonès G, Ferré R, Lázaro I, Cabré A, Plana N, Merino J, et al. Fatty acid-binding protein 4 is associated with endothelial dysfunction in pa-tients with type 2 diabetes. Atherosclerosis. 2010;213:329–331.

at Helsinki University on January 6, 2015http://circgenetics.ahajournals.org/Downloaded from

598 Circ Cardiovasc Genet October 2014

10. Makowski L, Boord JB, Maeda K, Babaev VR, Uysal KT, Morgan MA, et al. Lack of macrophage fatty-acid-binding protein aP2 protects mice defi-cient in apolipoprotein E against atherosclerosis. Nat Med. 2001;7:699–705.

11. Layne MD, Patel A, Chen YH, Rebel VI, Carvajal IM, Pellacani A, et al. Role of macrophage-expressed adipocyte fatty acid binding protein in the development of accelerated atherosclerosis in hypercholesterolemic mice. FASEB J. 2001;15:2733–2735.

12. Scheja L, Makowski L, Uysal KT, Wiesbrock SM, Shimshek DR, Meyers DS, et al. Altered insulin secretion associated with reduced lipolytic ef-ficiency in aP2-/- mice. Diabetes. 1999;48:1987–1994.

13. Uysal KT, Scheja L, Wiesbrock SM, Bonner-Weir S, Hotamisligil GS. Improved glucose and lipid metabolism in genetically obese mice lacking aP2. Endocrinology. 2000;141:3388–3396.

14. Hotamisligil GS, Erbay E. Nutrient sensing and inflammation in metabolic diseases. Nat Rev Immunol. 2008;8:923–934.

15. Feng B, Yao PM, Li Y, Devlin CM, Zhang D, Harding HP, et al. The endo-plasmic reticulum is the site of cholesterol-induced cytotoxicity in macro-phages. Nat Cell Biol. 2003;5:781–792.

16. Tabas I. Macrophage apoptosis in atherosclerosis: consequences on plaque progression and the role of endoplasmic reticulum stress. Antioxid Redox Signal. 2009;11:2333–2339.

17. Myoishi M, Hao H, Minamino T, Watanabe K, Nishihira K, Hatakeyama K, et al. Increased endoplasmic reticulum stress in atherosclerotic plaques as-sociated with acute coronary syndrome. Circulation. 2007;116:1226–1233.

18. Saksi J, Ijäs P, Nuotio K, Sonninen R, Soinne L, Salonen O, et al. Gene ex-pression differences between stroke-associated and asymptomatic carotid plaques. J Mol Med (Berl). 2011;89:1015–1026.

19. Agardh HE, Folkersen L, Ekstrand J, Marcus D, Swedenborg J, Hedin U, et al. Expression of fatty acid-binding protein 4/aP2 is correlated with plaque instability in carotid atherosclerosis. J Intern Med. 2011;269:200–210.

20. Peeters W, de Kleijn DP, Vink A, van de Weg S, Schoneveld AH, Sze SK, et al. Adipocyte fatty acid binding protein in atherosclerotic plaques is associated with local vulnerability and is predictive for the occurrence of adverse cardiovascular events. Eur Heart J. 2011;32:1758–1768.

21. Holm S, Ueland T, Dahl TB, Michelsen AE, Skjelland M, Russell D, et al. Fatty Acid binding protein 4 is associated with carotid atherosclerosis and outcome in patients with acute ischemic stroke. PLoS One. 2011;6:e28785.

22. Tuncman G, Erbay E, Hom X, De Vivo I, Campos H, Rimm EB, et al. A genetic variant at the fatty acid-binding protein aP2 locus reduces the risk for hypertriglyceridemia, type 2 diabetes, and cardiovascular disease. Proc Natl Acad Sci U S A. 2006;103:6970–6975.

23. Aromaa A, Koskinen S, eds. Health and Functional capacity in Finland. Baseline results of the Health 2000 Health Examination Survey. Helsinki: Publication of the National Public Health Institute B12/2004; 2004. http://www.terveys2000.fi/julkaisut/baseline.pdf. Accessed June 18, 2013.

24. Niiranen T, Jula A, Kantola I, Moilanen L, Kähönen M, Kesäniemi YA, et al. Home-measured blood pressure is more strongly associated with ath-erosclerosis than clinic blood pressure: the Finn-HOME Study. J Hyper-tens. 2007;25:1225–1231.

25. Nuotio K, Lindsberg PJ, Carpén O, Soinne L, Lehtonen-Smeds EM, Saimanen E, et al. Adhesion molecule expression in symptomatic and as-ymptomatic carotid stenosis. Neurology. 2003;60:1890–1899.

26. Nuotio K, Mäyränpää MI, Saksi J, Ijas P, Sairanen T, Carpén O, et al. Endothelial apoptosis does not determine symptom status in carotid artery disease. Cerebrovasc Dis. 2007;23:27–34.

27. Isoviita PM, Nuotio K, Saksi J, Turunen R, Ijäs P, Pitkäniemi J, et al. An im-balance between CD36 and ABCA1 protein expression favors lipid accumu-lation in stroke-prone ulcerated carotid plaques. Stroke. 2010;41:389–393.

28. NASCET Collaborators. North American Symptomatic Carotid Endarte-roctomy Trial; Beneficial effect of carotid endarterectomy in symptomatic patients with high-grade carotid stenosis. NEJM. 1991;325:445–453.

29. Vaara S, Nieminen MS, Lokki ML, Perola M, Pussinen PJ, Allonen J, et al. Cohort Profile: the Corogene study. Int J Epidemiol. 2012;41:1265–1271.

30. Vartiainen E, Jousilahti P, Alfthan G, Sundvall J, Pietinen P, Puska P. Car-diovascular risk factor changes in Finland, 1972-1997. Int J Epidemiol. 2000;29:49–56.

31. Stary HC, Chandler AB, Dinsmore RE, Fuster V, Glagov S, Insull W Jr, et al. A definition of advanced types of atherosclerotic lesions and a his-tological classification of atherosclerosis. A report from the Committee on Vascular Lesions of the Council on Arteriosclerosis, American Heart Association. Circulation. 1995;92:1355–1374.

32. Sitzer M, Müller W, Siebler M, Hort W, Kniemeyer HW, Jäncke L, et al. Plaque ulceration and lumen thrombus are the main sources of cerebral microemboli in high-grade internal carotid artery stenosis. Stroke. 1995;26:1231–1233.

33. Tabas I. The role of endoplasmic reticulum stress in the progression of atherosclerosis. Circ Res. 2010;107:839–850.

34. Teslovich TM, Musunuru K, Smith AV, Edmondson AC, Stylianou IM, Koseki M, et al. Biological, clinical and population relevance of 95 loci for blood lipids. Nature. 2010;466:707–713.

35. Cao H, Maeda K, Gorgun CZ, Kim HJ, Park SY, Shulman GI, et al. Regu-lation of metabolic responses by adipocyte/macrophage fatty acid-binding proteins in leptin-deficient mice. Diabetes. 2006;55:1915–1922.

36. Eissing L, Scherer T, Tödter K, Knippschild U, Greve JW, Buurman WA, et al. De novo lipogenesis in human fat and liver is linked to ChREBP-β and metabolic health. Nat Commun. 2013;4:1528.

37. Bots ML, Hoes AW, Koudstaal PJ, Hofman AG, Diederick E. Common carotid intima-media thickness and risk of stroke and myocardial infarc-tion the rotterdam study. Circulation. 1997;96:1432–1437.

38. O’Leary DH, Polak JF, Kronmal RA, Manolio TA, Burke GL, Wolfson SK Jr. Carotid-artery intima and media thickness as a risk factor for myo-cardial infarction and stroke in older adults. Cardiovascular Health Study Collaborative Research Group. N Engl J Med. 1999;340:14–22.

39. CARDIoGRAMplusC4D Consortium. Large-scale association analysis iden-tifies new risk loci for coronary artery disease. Nat Genet. 2013;45:25–33.

40. International Stroke Genetics Consortium (ISGC) and Wellcome Trust Case Control Consortium 2 (WTCCC2). Genome-wide association study identifies a variant in HDAC9 associated with large vessel ischemic stroke. Nat Genet. 2012;44:328–333.

CLINICAL PERSPECTIVEFatty acid–binding protein 4 (FABP4 or aP2 in mice) has been identified as a key regulator of core aspects of cardiometabolic disorders, including lipotoxicity-mediated endoplasmic reticulum stress in macrophages. However, the significance of FABP4 in human atherosclerosis remains unclear. Here, we demonstrate that FABP4 is overexpressed in atherosclerotic carotid plaques causing thromboembolic strokes. We show that a naturally occurring low-expression variant of FABP4 (rs77878271) is associ-ated with reduced FABP4 transcription within the atherosclerotic plaque and that it restricts the development of large artery atherosclerosis during obesity and reduces lesion vulnerability and associated atherotrombotic complications on a population level. We also found that the low-expression variant was associated with reduced total cholesterol levels with the largest reduc-tion in variant allele homozygotes. Consistently, the variant allele homozygotes showed 8-fold lower odds for MI. The choles-terol-lowering effect in the variant allele homozygotes was enhanced by almost 3-fold in the highest body mass index quartile. These data reveal FABP4 rs77878271 as a novel variant affecting serum total cholesterol levels and cardiovascular risk. This is the first evidence connecting genetically reduced expression of the main adipocyte lipid chaperone, FABP4, to cholesterol metabolism in man. This raises the notion that downregulation of FABP4 could mediate analogous enhancement of adipocyte de novo lipogenesis in man as reported in murine combined lipid chaperone deficiency, with the largest effect in obese individu-als. A therapeutic regimen reducing FABP4 expression within the atherosclerotic plaque may promote lesion stability through modulation of endoplasmic reticulum stress signaling, attenuation of apoptosis, lipid burden, and inflammation.

at Helsinki University on January 6, 2015http://circgenetics.ahajournals.org/Downloaded from

SUPPLEMENTAL MATERIAL

Supplemental Methods

Disease definition in The Health 2000 Survey

Diagnoses at baseline were based on data gathered by a health interview, clinical examination,

available documents on medical history and data on hospital discharge registers, or patients' rights

for drug reimbursements. At 10 years follow-up, the diagnoses were gathered from the registers on

hospital discharge and causes of death.

Hypertension was defined as systolic blood pressure ≥ 140 mm Hg and/or diastolic blood

pressure ≥ 90 mm Hg and/or use of blood pressure lowering drugs. Diabetes mellitus was defined

as fasting serum glucose level ≥ 7.0 mmol/L, the 2 hour oral glucose tolerance test glucose level ≥

11.1 mmol/l or a history of use of oral hypoglycemic agents or insulin injections. Smoking was

defined as the daily use of tobacco products.

Classification of myocardial infarction required either a clinical diagnosis of old myocardial

infarction by the examining physician, large Q-waves in resting ECG, a previous discharge

diagnosis of myocardial infarction (ICD-8/9 code 410 or ICD-10 codes I21–I22) or typical self-

reported history of myocardial infarction treated in hospital.

Classification as CHD required at least one of the following: diagnosis of myocardial infarction

and/or angina pectoris during the field health examination by a physician, large Q-waves in the

1

resting ECG, hospitalization for CHD (ICD-8/9 codes 410–414 or ICD-10 codes I20–I25), a history

of coronary revascularization, the right to drug reimbursement for CHD or the use of nitroglycerine

combined with an anticoagulant, acetyl salicylic acid or beta-blocker.

Classification of ischemic stroke required hospitalization for a plausible period with a specific

discharge diagnosis (ICD-8/9 codes 433-4, ICD-10 code I63). Classification of TIA required

hospitalization with a TIA diagnosis (ICD-9 435, ICD-10 G45) or with a diagnosis of ischemic

stroke but only for 24 hours. Also a well-documented history of ischemic stroke or TIA by a

physician at the health examination was included.

Genotyping

The FABP4 rs77878271 polymorphism was genotyped either by TaqMan SNP Genotyping Assay

(C__25762406_10, Applied Biosystems) or by sequencing. In sequencing a 318 bp fragment

containing the polymorphism was amplified from genomic DNA (F:5’-

TCCCCTCTCTACACTGGGATA-3’, R:5’-GCTGTGACCCTCTTGAGTCC-3’) and sequenced

using the forward primer. The Corogene cohort was genotyped with Illumina 660K BeadChip array

at the Sanger Intitute (Hixton, Cambridge, UK) and data for rs77878271 was imputed using

IMPUTE software and 1000 genomes imputation April 2012 reference panel.

Immunohistochemistry

The primary antibodies used were: FABP4 (Sigma, product number: HPA002188, working

concentration/dilution: 0.3 µg/ml), HAM56 (GenTex, GTX72010, 0.7 µg/ml), CD36 (Chemicon,

CBL168, 2.0 µg/ml), ABCA1(Abcam, ab18180, 20 µg/ml), CD42b (Novocastra, NCL-CD42b,

1:400), CD163 (R&D Systems, AF1607, 10 µg/ml), activated caspase 3 (aCASP3, Cell Signalling,

2

9664, 1:150), phosphorylated pancreatic ER kinase (PERK[pY981], Santa Cruz, sc-32577, 1.0

µg/ml), phosphorylated eukaryotic translation initiation factor 2 alpha (eIF2a[pS52], Invitrogen,

44728G, 6.0 µg/ml) and CD68 (Dako, M0876, 20µg/ml). Appropriate unspecific IgGs in equivalent

concentration or PBS were used as negative controls.

3

Table I. Demographics of The Health 2000 Survey participants by gender and FABP4 rs77878271 genotype.

Variable All Females Males TT TC CC P*

Number 7491 4153 3338 6647 816 28Gender, % of males 44.6 44.8 41.9 53.6 0.172Age, years 55.0±15.3 56.3±16.0 53.4±14.2 55.0±15.2 54.9±15.7 52.0±12.2 -

Age, range 30-98Cardiovascular risk factors Smoking, % 20.5 16.0 26.2 20.6 19.9 14.3 - Obesity, % 22.5 24.1 20.6 22.6 22.1 17.9 - BMI, kg/m2 27.0±4.7 26.9±5.1 27.1±4.1 27.0±4.6 26.9±4.8 27.4±5.4 - Hypertension, % 50.3 48.0 53.1 50.4 49.7 50.0 - Diabetes type I and II, % 8.3 7.5 9.2 8.4 7.4 14.3 -Medication, % Antihypertensive medication 24.5 24.7 24.3 24.3 26.6 16.0 - Diabetes medication 4.6 4.0 5.3 4.6 4.0 12.0 0.134 Lipid lowering medication 7.6 6.7 8.7 7.8 6.2 8.0 - Antithrombotic medication 18.4 17.6 19.5 18.6 17.4 4.0 0.069Laboratory measures Total cholesterol, mmol/l 5.94±1.11 5.96±1.12 5.93±1.12 5.95±1.12 5.88±1.10 5.68±1.18 0.099 HDL, mmol/l 1.34±0.39 1.44±0.39 1.21±0.34 1.34±0.39 1.33±0.38 1.29±0.41 - LDL, mmol/l 3.75±1.07 3.68±1.07 3.84±1.06 3.76±1.07 3.71±1.07 3.63±0.94 - Triglycerides, mmol/l 1.63±1.04 1.47±0.80 1.82±1.26 1.63±1.05 1.60±1.00 1.54±0.78 - ApoA-I, g/l 1.58±0.29 1.65±0.30 1.49±0.26 1.58±0.29 1.58±0.29 1.51±0.28 - ApoB, g/l 1.23±0.29 1.19±0.29 1.28±0.29 1.23±0.29 1.22±0.29 1.17±0.26 - Fasting glucose, mmol/l 5.59±1.41 5.47±1.28 5.74±1.54 5.60±1.44 5.50±1.09 5.90±2.27 0.091 Fasting insulin, mU/l 9.94±47.81 8.59±7.98 11.58±70.71 10.05±50.67 9.06±7.05 9.64±7.47 - CRP, mg/l 2.27±6.31 2.24±5.57 2.30±7.10 2.34±6.63 1.80±2.90 1.02±0.97 0.069

*Difference between genotypes by one-way ANOVA (P<0.2 are shown). The c2 or Fisher’s exact test was used for categorical variables.vContinuous variables are presented as mean ± SD.

4

Table II. Linear regression analysis of total and LDL cholesterol levels and apolipoptotein B concentrations.

*Total cholesterol *LDL cholesterol *Apolipoprotein BR2 = 0.048 R2 = 0.058 R2 = 0.138

N=6868 N=6840 N=6868

Coefficients b (95% CI) P b (95% CI) P b (95% CI) P

(Intercept) 4.80 (4.60 to 5.00) 2.83 (2.63 to 3.02) 0.750 (0.699 to 0.801)Age 0.011 (0.009 to 0.013) <0.001 0.010 (0.009 to 0.012) <0.001 0.003 (0.003 to 0.003) <0.001Gender 0.010 (-0.042 to 0.063) 0.698 -0.168 (-0.218 to -0.119) <0.001 -0.092 (-0.105 to -0.079) <0.001BMI 0.025 (0.020 to 0.031) <0.001 0.027 (0.022 to 0.032) <0.001 0.018 (0.016 to 0.019) <0.001Use of lipid lowering medication -0.579 (-0.678 to -0.480) <0.001 -0.623 (-0.716 to -0.529) <0.001 -0.073 (-0.097 to -0.048) <0.001FABP4 Rs77878271 genotypes -0.108 (-0.186 to -0.031) 0.006 -0.069 (-0.143 to 0.005) 0.066 -0.018 (-0.037 to 0.001) 0.070

*Dependent variable. In the models males, were coded as 1 and females as 2; the use of lipid lowering medication as 1 or 0 (no lipid lowering medication);

FABP4 rs77878271 genotypes as 1 (TT, major allele homozygotes), 2 (TC) and 3 (CC).

5

Table III. Demographics of The Health 2000 Survey SVT-D cohort participants by gender and FABP4 rs77878271 genotype.

Variable All Females Males TT TC CC P*

Number 1448 786 662 1276 165 7Gender,% of males 45.7 46.0 43.6 42.9 -Age, years 57.0±8.1 57.3±8.2 56.7±7.8 57.1±8.1 56.3±7.8 56.9±10.0 - Age, range 45-74Cardiovascular risk factors Smoking, % 18.0 14.8 21.8 17.6 21.8 0.0 - Obesity, % 24.0 24.9 23.0 23.9 24.5 28.6 - BMI, kg/m2 27.3±4.5 27.1±4.9 27.5±4.1 27.2±4.5 27.3±4.6 28.5±6.5 - Hypertension, % 54.4 51.9 57.4 54.7 52.1 57.1 - Diabetes, type I and II, % 11.0 9.0 13.4 10.9 12.7 0.0 -Medication, % Antihypertensive medication 31.2 30.9 31.6 31.5 29.1 28.6 - Diabetes medication 4.4 2.7 6.3 4.7 1.8 0.0 - Lipid lowering medication 12.6 10.8 14.7 12.6 12.1 14.3 - Antithrombotic medication 15.5 12.7 18.9 16.0 12.7 0.0 -Laboratory measures Total cholesterol, mmol/l 5.57±0.96 5.65±0.93 5.48±0.98 5.58±0.97 5.52±0.90 5.47±1.06 - HDL, mmol/l 1.57±0.43 1.70±0.43 1.42±0.38 1.57±0.43 1.59±0.42 1.57±0.35 - LDL, mmol/l 3.38±1.03 3.77±1.01 3.85±1.05 3.82±1.04 3.73±0.97 3.61±0.79 - Triglycerides, mmol/l 1.40±0.93 1.29±0.67 1.53±1.15 1.41±0.95 1.37±0.76 1.10±0.28 - ApoA-I, g/l 1.74±0.31 1.84±0.30 1.63±0.28 1.74±0.31 1.76±0.32 1.67±0.30 - ApoB, g/l 1.17±0.25 1.15±0.25 1.19±0.25 1.17±0.25 1.15±0.23 1.15±0.19 - Fasting glucose, mmol/l 5.86±1.28 5.62±0.99 6.16±1.49 5.88±1.31 5.76±0.92 5.59±0.44 - Fasting insulin, mU/l 9.70±7.01 8.83±5.86 10.81±8.09 9.77±7.11 9.53±6.50 7.9±5.24 - CRP, mg/l 2.90±4.27 2.99±4.52 2.79±3.97 2.94±4.36 2.67±3.54 1.53±0.82 -Carotid IMT and plaques Carotid IMT, mm 0.94±0.23 0.91±0.22 0.97±0.24 0.94±0.23 0.93±0.22 0.97±0.32 - Carotid plaques, % 15.5 13.6 17.8 15.9 13.0 14.3 -

*Difference between genotypes by one-way ANOVA (P<0.2 are shown). The c2 or Fisher’s exact test was used for categorical variables.xContinuous variables are presented as mean ± SD.

6

Table IV. Demographics of the obese (BMI≥30) Health 2000 Survey SVT-D cohort participants by gender and FABP4 rs77878271 genotype.

Variable All Females Males TT TC CC P*

Number 346 195 151 304 40 2Gender, % of males 43.6 44.7 35.0 50.0 -Age, years 57.8±7.8 58.7±8.2 56.7±7.1 58.0±7.9 57.2±7.2 45.5±0.7 0.071 Age, range 45-74Cardiovascular risk factors Smoking, % 16.5 14.9 18.7 16.8 15.0 0.0 - BMI, kg/m2 33.4±3.5 33.7±3.8 33.1±3.2 33.4±3.6 33.7±3.3 36.0±6.4 - Hypertension, % 72.5 72.3 72.8 72.7 70.0 100.0 - Diabetes type I and II, % 22.3 18.5 27.2 22.0 25.0 0.0 -Medication, % Antihypertensive medication 46.0 45.6 46.4 46.4 42.5 50.0 - Diabetes medication 6.9 5.1 9.3 7.2 5.0 0.0 - Lipid lowering medication 15.0 12.3 18.5 16.1 7.5 0.0 - Antithrombotic medication 19.9 16.9 23.8 20.4 17.5 0.0 -Laboratory measures Total cholesterol, mmol/l 5.60±0.99 5.74±0.99 5.43±0.97 5.62±1.01 5.53±0.78 4.7±1.27 - HDL, mmol/l 1.40±0.37 1.50±0.36 1.27±0.33 1.40±0.37 1.41±0.37 1.25±0.44 - LDL, mmol/l 3.83±1.07 3.89±0.99 3.77±1.16 3.87±1.09 3.58±0.84 3.23±0.73 0.187 Triglycerides, mmol/l 1.77±1.31 1.63±0.83 1.95±1.74 1.77±1.36 1.75±0.96 1.40±0.14 - ApoA-I, g/l 1.69±0.30 1.78±0.29 1.57±0.26 1.68±0.29 1.71±0.33 1.44±0.40 - ApoB, g/l 1.23±0.25 1.24±0.25 1.23±0.25 1.24±0.25 1.21±0.24 1.04±0.07 - Fasting glucose, mmol/l 6.36±1.64 5.99±1.07 6.83±2.07 6.40±1.70 6.10±1.02 5.50±0.56 - Fasting insulin, mU/l 14.54±9.76 12.70±8.11 16.93±11.12 14.51±9.83 14.84±9.42 13.10±8.77 - CRP, mg/l 4.45±5.70 4.65±5.29 4.18±6.19 4.54±5.95 3.86±3.33 2.20±0.20 -Carotid IMT and plaques Carotid IMT, mm 0.97±0.22 0.95±0.21 1.00±0.23 0.98±0.22 0.91±0.18 0.78±0.21 0.087 Carotid plaques, % 15.7 14.2 17.6 17.1 5.4 0.0 0.150

*Difference between genotypes by one-way ANOVA (P<0.2 are shown). The c2 or Fisher’s exact test was used for categorical variables.xContinuous variables are presented as mean ± SD.

7

Table V. Linear regression analysis of carotid IMT.

Model 1 Model 2 Model 3 Model 4R2 = 0.292 R2 = 0.292 R2 = 0.293 R2 = 0.296

Coefficients* b (95% CI) P b (95% CI) P b (95% CI) P b (95% CI) P

(Intercept) 0.095 (-0.079 to 0.268) 0.095 (-0.081 to 0.271) 0.083 (-0.097 to 0.263) 0.062 (-0.120 to 0.244)Age 0.454 (0.012 to 0.014) <0.001 0.454 (0.012 to 0.014) <0.001 0.454 (0.012 to 0.014) <0.001 0.452 (0.012 to 0.014) <0.001Gender -0.113 (-0.076 to -0.029) <0.001 -0.113 (0.076 to -0.029) <0.001 -0.113 (-0.076 to -0.029) <0.001 -0.111 (-0.075 to -0.028) <0.001HDL cholesterol -0.056 (-0.062 to 0.002) 0.068 -0.056 (-0.062 to 0.002) 0.070 -0.056 ( -0.062 to 0.002) 0.067 -0.056 (-0.062 to 0.002) 0.067ApoB 0.027 (-0.028 to 0.077) 0.362 0.027 (-0.028 to 0.077) 0.363 0.027 (-0.028 to 0.078) 0.354 0.027 (-0.029 to 0.077) 0.077Triglycerides† 0.060 (-0.003 to 0.062) 0.079 0.060 (-0.004 to 0.062) 0.081 0.060 (-0.004 to 0.062) 0.082 0.061 (-0.003 to 0.063) 0.075Fasting blood glucose† 0.013 (-0.054 to 0.090) 0.622 0.013 (-0.055 to 0.091) 0.628 0.013 (-0.053 to 0.090) 0.614 0.012 (-0.057 to 0.089) 0.665Systolic blood pressure 0.102 (0.001 to 0.002) <0.001 0.102 (0.001 to 0.002) <0.001 0.102 (0.001 to 0.002) <0.001 0.104 (0.001 to 0.002) <0.001Smoking 0.079 (0.019 to 0.076) 0.001 0.079 (0.019 to 0.076) 0.001 0.079 (0.019 to 0.075) 0.001 0.077 (0.018 to 0.074) 0.001Obesity (BMI>30) <0.001 (-0.027 to 0.027) 0.998 -0.150 (-0.154 to -0.007) 0.031rs77878271 alleles 0.013 (-0.024 to 0.042) 0.585 0.045 (-0.006 to 0.070) 0.102Product term‡ 0.162 (0.014 to 0.168) 0.021

*Dependent variable: mean of intima-media thicknesses (IMTs) of common and internal carotid arteries, in the model males, obesity, and rs77878271 T allele

were coded as 1; †log-transformed values were used in the model; ‡two-way interaction between obesity and rs77878271 alleles was implemented in the mod-

el as a product term denoted as 1 for obese (BMI≥30) T allele carriers and 0 for obese C allele and non-obese T and C allele carriers. Obesity, rs77878271 al-

leles, or the product term are added sequentially to the regression model as shown in models 1-4. Neither obesity nor rs77878271 T-allele alone but rather

these two factors in interaction (the product term) are associated with a higher IMT in obese T allele carriers.

8

Table VI. Patient characteristics and macroscopic plaque features by FABP4 rs77878271 genotypes in

the HeCES cohort.

FABP4 rs77878271All TT TC P*

N, (%) 91 80 (87.9) 11 (12.1)Gender, % of males 64.8 67.5 45.5 0.185Age, years† 64.6±0.85 64.7±0.93 63.5±2.27Degree of ICA stenoses†‡ 77.8±0.88 77.6±0.95 77.3±1.84Cerebrovascular symptoms, %§ 80.2 83.8 54.5 0.038Comorbidities, % Diabetes mellitus, type I or II 25.3 27.5 9.1 Dyslipidemia 62.2 64.6 45.5 Peripheral arterial disease 30.8 33.8 9.1 0.162 Coronary heart disease 37.4 35.0 54.5 Arterial hypertension 67.0 63.7 90.9 0.093Medications, % Antiaggregatory 28.1 30.4 10.0 ACE inhibitor 19.3 17.9 30.0 Statin 42.9 43.8 36.4Laboratory measures† Hematocrit 40.7±0.37 41.0±0.40 38.9±0.81 0.046 LDL 3.4±0.12 3.5±0.13 3.0±0.36 0.193 HDL 1.3±0.04 1.3±0.04 1.5±0.17 0.112 Triglycerides 1.9±0.12 1.9±0.14 1.3±0.10 0.094 hsCRP 7.9±1.30 8.0±1.43 5.4±1.40Macroscopic plaque features, %

Ulceration 42.5 46.1 18.2 0.108Intraplaque hemorrhage 52.2 53.2 45.5Intramural thrombus 15.6 15.2 18.2Loose atheroma 22.7 22.1 27.3Calcification 62.2 60.8 72.2

*Differences in interval variables were tested by Mann-Whitney U test and dichotomical variables by

Fisher’s exact test. P<0.2 are shown; †age, degree of internal carotid artery (ICA) stenosis, and labora-

tory measurements are given as mean and standard error; ‡degree of ICA stenosis according to the

NASCET criteria; §a subcohort of patients with ipsilateral stroke and asymptomatic patients (n=43) was

used in some analysis (see methods for details).

9

Table VII. FABP4 rs77878271 genotype frequencies in patients with advanced carotid disease.

Groups TT TC CC All P*

Carotid disease†, n (%) 80 (87.9) 11 (12.1) 0 (0) 91Symptomatic 67 (91.8) 6 (8.2) 0 (0) 73Asymptomatic 13 (72.2) 5 (27.8) 0 (0) 18

*Symptomatic compared to asymptomatic (Fisher’s exact test); †symptomatic patients have suffered a cerebrovascular event during their lifetime, while

asymptomatic patients have remained event free despite significant carotid stenosis.

0.038

10

Figure I

Figure I. Cox adjusted survival curves for ischemic cardiovascular events according to

FABP4 rs77878271 genotypes in the Health 2000 cohort. Analysis was performed with age as

the time scale adjusting for gender (males were coded as 1 and females as 2) and FABP4

rs77878271 genotypes (coded as 1: TT, 2: TC, and 3: CC). The hazard ratio (HR) and 95%

confidence intervals (CIs) are given in comparisons to major allele homozygotes (TT).

Ischemic cardiovascular events include myocardial infarctions, unstable angina pectoris,

history of angioplasty or bypass surgery, ischemic stroke or TIA.

Cum

ulat

ive

surv

ival

1.0

0.8

0.6

0.4

0.2

0.0

Age (years)

0 20 40 60 80 100

TTTCCC

TT vs. TCHR = 0.90 (0.73 to 1.10 95% CI)P=0.264

TT vs. CCHR = 0.33 (0.05 to 2.35 95% CI)P=0.268

FABP4 rs77878271

11

Figure IIH

AM

56im

mun

orea

ctiv

ear

ea(%

)

0

1.0

2.0

3.0

4.0

0-25 25-50 50-75 75-100FABP4 immunoreactive area (%)

P=0.008

Figure II. Correlation between FABP4 and HAM56 immunoreactivity (macrophages) in ca-

rotid plaques (n=66).

12

Figure III

0.0 2.0 4.0 6.0

FAB

P4pr

otei

npg

/µg

ofto

talp

rote

in

A

0

50

100

150

Relative density of aCASP3 immunoreactive cells

n=36, rs=0.487, P=0.003

8.0 10.0

0

20

40

60

80

100

0-2525-5050-7575-100

TUNELstaining (%)

%

D

Intracellular ORO staining area (%)0 33-66 33-66 66-100

P=0.006

0 0-33 33-66 66-100

FAB

P4pr

otei

npg

/µg

ofto

talp

rote

in

B

0

2.0

4.0

6.0

TUNEL staining area (%)

P=0.023

0 0-33 33-66 66-100

Rel

ativ

ede

nsity

ofaC

ASP3

imm

unor

eact

ive

cells

C

0

2.0

4.0

6.0

Intracellular ORO staining area (%)

P=0.003

Figure III. Correlations between FABP4 protein levels (ELISA) and (A, n=36) relative density of activated

caspase 3 (aCASP3) immunoreactive cells and (B, n=37) TUNEL reactivity in carotid plaques as well as be-

tween intracellular Oil-red-O (ORO) staining area and (C, n=85) relative density of aCASP3 immunoreactive

cells and (D, n=81) TUNEL reactivity in carotid plaques.

13

Jauhiainen, Petri T. Kovanen and Perttu J. LindsbergLokki, Markus Perola, Aki S. Havulinna, Veikko Salomaa, Johannes Kettunen, Matti

Lehtonen-Smeds, Markku Kaste, Antti Jula, Juha Sinisalo, Markku S. Nieminen, Marja-Liisa Jani Saksi, Petra Ijäs, Mikko I. Mäyränpää, Krista Nuotio, Pia M. Isoviita, Jarno Tuimala, Erno

of Atherosclerotic Disease and Increased Plaque StabilityBinding Protein 4 Favors Reduced Manifestations−Low-Expression Variant of Fatty Acid

Print ISSN: 1942-325X. Online ISSN: 1942-3268 Copyright © 2014 American Heart Association, Inc. All rights reserved.

Dallas, TX 75231is published by the American Heart Association, 7272 Greenville Avenue,Circulation: Cardiovascular Genetics

doi: 10.1161/CIRCGENETICS.113.0004992014;7:588-598; originally published online August 13, 2014;Circ Cardiovasc Genet. 

http://circgenetics.ahajournals.org/content/7/5/588World Wide Web at:

The online version of this article, along with updated information and services, is located on the

http://circgenetics.ahajournals.org/content/suppl/2014/08/13/CIRCGENETICS.113.000499.DC1.htmlData Supplement (unedited) at:

  http://circgenetics.ahajournals.org//subscriptions/

is online at: Circulation: Cardiovascular Genetics Information about subscribing to Subscriptions: 

http://www.lww.com/reprints Information about reprints can be found online at: Reprints:

  document. Permissions and Rights Question and Answer information about this process is available in the

requested is located, click Request Permissions in the middle column of the Web page under Services. FurtherCenter, not the Editorial Office. Once the online version of the published article for which permission is being

can be obtained via RightsLink, a service of the Copyright ClearanceCirculation: Cardiovascular Geneticsin Requests for permissions to reproduce figures, tables, or portions of articles originally publishedPermissions:

at Helsinki University on January 6, 2015http://circgenetics.ahajournals.org/Downloaded from