Journal of the American College of Cardiology Vol. 62, No. 15, 2013� 2013 by the American College of Cardiology Foundation ISSN 0735-1097/$36.00Published by Elsevier Inc. http://dx.doi.org/10.1016/j.jacc.2013.07.031

Genetics/Genomics

The 9p21 Locus Does Not Affect Riskof Coronary Artery Disease ThroughInduction of Type 1 Interferons

Clett Erridge, PHD,*y Jay Gracey, BA, RN,*y Peter S. Braund, MSC,*y Nilesh J. Samani, MD*yLeicester, United Kingdom

From the *D

Heart Foun

United Kin

Cardiovascu

Kingdom. T

National In

Disease. Jay

a chair supp

that they ha

Manuscri

accepted Jul

Objectives T

epartment of Cardiovas

dation Cardiovascular R

gdom; and the yNatio

lar Biomedical Research

his study is part of the

stitute for Health Researc

Gracey is funded by the

orted by the British Hea

ve no relationships relevan

pt received April 10, 20

y 9, 2013.

he study objective was to determine whether the coronary artery disease (CAD)-associated genotype atchromosome 9p21 modulates basal or induced expression of type I interferons (IFN-I).

Background T

he mechanism responsible for the association between common variants in chromosome 9p21.3 and CADremains unclear. It has been reported that the CAD risk locus is rich in enhancer-like elements and thatchromosome looping can lead to its physical proximity with the IFN-I gene cluster, raising the possibility that thelocus influences CAD risk by modulating expression of IFN-Is.

Methods W

e examined whether genotype at the lead CAD-associated single nucleotide polymorphism (rs1333049) in 9p21was associated with: 1) basal levels of IFN-I in plasma from 148 healthy male subjects; 2) induction of IFN-I by Toll-like receptor stimulants in peripheral blood mononuclear cells of 60 healthy volunteers assessed by enzyme-linkedimmunosorbent assay, quantitative polymerase chain reaction, Western blot, and IFN-I bioassay; and 3) enhanceractivity of predicted IFN regulatory factor 3/7 binding sites within the 9p21 CAD risk region in reporter assays.

Results N

o significant effects of 9p21 genotype were observed for plasma levels of IFN-a, IFN-a21, or CXCL10, or leukocyteinduction of IFN-a, IFN-a21, IFN-b, CXCL10, or total IFN-I measured at the mRNA, protein, and biological activitylevels. There was also no enhancement of reporter activity by predicted IFN regulatory factor 3/7 binding sites in theCAD risk locus of either genotype.

Conclusions T

he mechanism underlying the association between common 9p21 variants and CAD does not involve differentialregulation of IFN-I responses. (J Am Coll Cardiol 2013;62:1376–81) ª 2013 by the American College ofCardiology Foundation

Common variants in the 9p21.3 gene desert are stronglyassociated with the risk of coronary artery disease (CAD) (1),although the mechanisms responsible for this associationremain poorly understood. Harismendy et al. (2) recently re-ported that chromosome looping permits physical proximityof the CAD risk locus to a region downstream of the IFNA21gene, raising the possibility that enhancers in this locus mayregulate expression of the type I interferon (IFN-I) gene cluster.

The IFN-Is, which comprise 13 interferon (IFN)-a iso-types, IFN-b, IFN-k, IFN-ε, and IFN-u, are expressed

cular Sciences, University of Leicester, British

esearch Centre, Glenfield Hospital, Leicester,

nal Institute for Health Research Leicester

Unit, Glenfield Hospital, Leicester, United

research portfolio supported by the Leicester

h Biomedical Research Unit in Cardiovascular

British Heart Foundation. Dr. Samani holds

rt Foundation. All other authors have reported

t to the contents of this paper to disclose.

13; revised manuscript received July 6, 2013,

from a single gene cluster on chromosome 9 located ap-proximately 800 kb from the CAD risk interval. IFN-I isco-induced in response to bacterial or viral nucleic acidmotifs via pattern recognition receptors, such as endosomalToll-like receptors (TLRs), and signal through the sharedIFN-I receptor to promote expression of genes involved inantiviral defense.

See page 1382

The potential role played by IFN-I in atheroscleroticdisease is currently debated, because both pro- and anti-atherogenic properties of IFN-I have been reported (3–5).Moreover, the autoimmune disease systemic lupus eryth-ematosus (SLE), which is thought to be driven largely byoverproduction of IFN-a, is associated with an elevated riskof CAD (6). We therefore tested the hypothesis that 9p21genotype regulates plasma levels of IFN-I or the capacity ofleukocytes to produce IFN-I in response to TLR stimulants.

Table1

Clin

ical

Cha

racteristics

ofSub

jects

Coh

ort

rs1333049

N(M

ale:Fe

male)

Age(yrs)

BMI(kg/m

2)

LDL(m

g/dl)

HDL(m

g/dl)

TAG(m

g/dl)

SBP(m

mHg)

DBP(m

mHg)

Current

Smok

erCRP(m

g/dl)

1CC

49(49:0)

21.4

�2.4

24.8

�4.5

91.4

�25.0

49.7

�10.7

143�

71

126�

9.6

72�

7.1

24%

0.19�

0.36

1CG

51(51:0)

21.4

�2.4

23.0

�3.2

86.4

�21.6

50.3

�9.6

108�

54

125�

8.3

72�

6.3

37%

0.11�

0.16

1GG

48(48:0)

21.5

�2.4

23.3

�2.5

84.9

�23.3

52.7

�10.7

120�

53

124�

7.4

71�

7.3

17%

0.11�

0.18

pvalue

dN/A

0.851

0.053

0.194

0.173

0.072

0.372

0.411

0.341

0.164

2CC

32(9:23)

37.7

�12.2

27.2

�5.3

105.4

�24

57.8

�12.4

141�

63

122�

10.2

76�

7.2

19%

0.29�

0.41

2GG

28(7:21)

40.3

�13.0

25.2

�3.7

115.8

�31

58.5

�13.7

119�

41

120�

10.5

75�

8.4

4%

0.14�

0.21

pvalue

d0.785

0.425

0.101

0.168

0.831

0.143

0.436

0.856

0.068

0.097

3CC

11(7:4)

37.1

�11.9

ND

ND

ND

ND

ND

ND

18%

ND

3CG

12(7:5)

27.6

�9.6

ND

ND

ND

ND

ND

ND

8%

ND

3GG

9(6:3)

32.8

�11.9

ND

ND

ND

ND

ND

ND

11%

ND

pvalue

d0.888

0.432

dd

dd

dd

0.660

d

Coh

orts

referto:1

)stored

plasmasamples

from

theGen

eticReg

ulationof

ArterialB

lood

Pressurein

Hum

ansin

theCom

mun

ity(GRAPHIC)stud

y(results

show

nin

Fig.

1);2)GRAPHIC

participan

tsrecalledto

providefreshbloo

dsamples

forleuk

ocytestim

ulation(Fig.2

);an

d3)he

althyvoluntee

rsrecruitedby

localad

vertisem

entfortran

scrip

tiona

lstud

ies(Fig.3),as

describ

edin

the“M

etho

ds”section.

Mea

ns�

SDareshow

n.BMI¼

body

massinde

x;CRP¼

C-re

activeprotein;

DBP¼

diastolic

bloo

dpressure;H

DL¼

high

-den

sity

lipop

rotein;LD

L¼

low-den

sity

lipop

rotein;N/A

¼no

tavailable;

ND¼

notde

term

ined

;SB

P¼

systolic

bloo

dpressure;TA

G¼

triglycerid

es.

Abbreviationsand Acronyms

CAD = coronary artery

disease

ELISA = enzyme-linked

immunosorbent assay

IFN = interferon

IFN-I = type I interferon

IRF = interferon regulatory

factor

LD = linkage disequilibrium

ODN = oligodeoxynucleotide

PBMC = peripheral blood

mononuclear cell

SLE = systemic lupus

erythematosus

SNP = single nucleotide

polymorphism

TLR = Toll-like receptor

JACC Vol. 62, No. 15, 2013 Erridge et al.October 8, 2013:1376–81 9p21 Variants Do Not Regulate Type I Interferon

1377

Methods

Subjects and samples. To as-sess the effect of 9p21 genotypeon plasma levels of IFN-I, weanalyzed stored samples from 148healthy male subjects previouslyrecruited into the GRAPHIC(Genetic Regulation of ArterialBlood Pressure in Humans inthe Community) study (cohort 1)(7). Sixty GRAPHIC partici-pants homozygous for rs1333049variants were invited to donate afresh blood sample for leukocyteIFN-I production studies (cohort2), and 32 healthy volunteerswere recruited by local advertise-ment with approval from theUniversity of Leicester College ofMedicine Research Ethics Com-

mittee to study IFN-I transcriptional responses (cohort 3).All subjects provided written, informed consent, and proce-dures were carried out in accordance with institutionalguidelines and the Declaration of Helsinki. 9p21 CAD-riskgenotype was determined in all subjects by genotypingthe lead CAD-associated single nucleotide polymorphism(SNP) rs1333049.

Enzyme-linked immunosorbent assays, quantitativepolymerase chain reactions, and bioassays for IFN-I.Cytokine and C-reactive protein levels were measured byenzyme-linked immunosorbent assay (ELISA) (R&D,Minneapolis, Minnesota; Pierce, Waltham, Massachusetts;and Caltag-Medsystems, Buckingham, United Kingdom).For stimulation experiments, peripheral blood mononuclearcells (PBMCs) (106 cells/ml) isolated from fresh venousblood were cultured with medium alone, 50 mg/ml PolyI:C,1 mg/ml R848, or 1.5 mmol/l CpG oligodeoxynucleotide(ODN)-2216 (InvivoGen, San Diego, California), with orwithout 2-hour pre-treatment with 100 ng/ml IFN-g(PeproTech, Rocky Hill, New Jersey). IFN-I bioactivitywas measured at 24 h using human embryonic kidney 293cells stably transfected with STAT2, interferon regulatoryfactor (IRF)-9, and an IFN stimulated response element-9driven reporter that is sensitive to IFN-I signaling (Inviv-oGen Inc., San Diego, California) and compared witha recombinant IFN-a standard curve. Immunoblots wereprobed with anti–IFN-a21 (Sigma, St. Louis, Missouri)or antiglyceraldehyde-3-phosphate dehydrogenase (SantaCruz Biotechnology, Inc., Santa Cruz, California). PBMCcDNA was quantified using a Rotor-Gene Q real-time-polymerase chain reaction cycler (Qiagen, Venlo, theNetherlands).Reporter assays. Four regions of the 9p21 CAD risk locuswere cloned from carriers homozygous for the risk or non-risk rs1333049 variants into pGL3-promoter (Promega,

Figure 1 Effect of 9p21 Genotype on Plasma IFN-a, IFN-a21, CXCL10, and C-Reactive Protein

Total IFN-a (A), IFN-a21 (B), CXCL10 (C), and C-reactive protein (D) were measured in plasma of 148 healthy male volunteers (cohort 1) with CC (homozygous risk), CG, and GG

rs1333049 genotypes by enzyme-linked immunosorbent assay (ELISA); N ¼ 32, 33, and 23, respectively, for interferon (IFN)-a21.

Erridge et al. JACC Vol. 62, No. 15, 20139p21 Variants Do Not Regulate Type I Interferon October 8, 2013:1376–81

1378

Fitchburg, Wisconsin) and sequenced for verification ofvariants in high linkage disequilibrium (LD) (R2 > 0.8)with CAD-associated SNPs (2). Human embryonickidney 293 cells were transfected with TLR4/MD2,CD14, and each reporter, and challenged with mediumalone (control), 10 mg/ml PolyI:C, or 10 ng/ml lipopoly-saccharide for 24 h.Statistics. The effect of genotype on plasma cytokine levelswas assessed by linear regression of log-transformed re-sponses. Stepwise regression was used to assess the effect ofage, low-density lipoprotein cholesterol, high-density lipo-protein cholesterol, triglycerides, systolic and diastolic bloodpressures, body mass index, and smoking status on cytokinelevels, and those found to be significant (p < 0.05) wereadjusted for when examining the effect of genotype. PBMCexpression of cytokine mRNA and protein was analyzed by2-way analysis of variance with Tukey’s post hoc test.Significance was assumed at p < 0.05.

Results

Subject characteristics. The clinical characteristics of the148 healthy male volunteers (cohort 1) whose plasma sampleswere analyzed for basal levels of cytokines, 60 subjectsrecruited for IFN-I secretion studies (cohort 2), and32 healthy volunteers recruited for IFN-I transcriptionstudies (cohort 3) are summarized in Table 1. There wereno significant differences by genotype of the lead CAD-associated 9p21 SNP rs1333049 in any of the demographic,clinical, or biochemical characteristics of these subjects.

Effect of genotype on plasma levels of IFN-Is. Because ofthe extensive (>80%) similarity among the 13 IFN-asubtypes, total plasma IFN-a was quantified using pan-subtype reactive ELISAs. These assays revealed no effectof 9p21 genotype on plasma total IFN-a in healthy men(Fig. 1A). There was also no effect of genotype on plasmaIFN-a21 or the chemokine CXCL10, which is a sensitive

Figure 2 Effect of 9p21 Genotype on Induction of IFN-a

Peripheral blood mononuclear cells (PBMCs) from 60 healthy volunteers homozy-

gous for the risk (CC) or nonrisk (GG) rs1333049 variants (cohort 2) were chal-

lenged with medium alone (control), PolyI:C, R848, CpG oligodeoxynucleotide

(ODN), or CpG ODN with IFN-g priming (I-CpG). Total IFN-a was measured by ELISA

(A), and IFN-I bioactivity was measured by bioassay (B) at 24 h. IFN-a21 was

quantified in PBMC lysates by densitometry of immunoblots relative to glyceral-

dehyde-3-phosphate dehydrogenase (mean � SD) (C). GAPDH ¼ glyceraldehyde-

3-phosphate dehydrogenase; IFN ¼ interferon.

JACC Vol. 62, No. 15, 2013 Erridge et al.October 8, 2013:1376–81 9p21 Variants Do Not Regulate Type I Interferon

1379

and stable marker of IFN-I induction, or C-reactive protein(Figs. 1B to 1D). Because ELISAs for IFN-a are known tooverestimate the fraction of IFN-I that remains biologicallyactive in plasma (8), we also measured IFN-I biologicalactivity in plasma samples by bioassay. In confirmation ofearlier studies, biologically active IFN-I was below the limitof detection (<5 pg/ml) in plasma samples from healthysubjects in any genotype group (6).Effect of genotype on induction of IFN-a and IFN-b.We next examined the effect of 9p21 genotype on inductionof IFN-a by PBMCs of 60 healthy volunteers in responseto stimulants ofTLR3 (PolyI:C), TLR7/8 (R848), andTLR9(CpG ODN). Induction of total IFN-a or CXCL10 asmeasured by ELISA or total IFN-I measured by bioassay wasnot modified by genotype (Figs. 2A and 2B). Because 2 SNPs

in high LD with rs1333049 disrupt an IFN-g–responsiveSTAT1 binding site (2), we also examined the effects of2-hour pre-incubation with IFN-g on production of IFN-a.These experiments revealed no capacity of IFN-g to prime forproduction of IFN-a or IFN-I biological activity (Fig. 2A).9p21 genotype also did not specifically affect induction ofIFN-a21 by CpG-ODN in cell lysates (Fig. 2C). There wasalso no effect of genotype on the production of IFN-b, thedownstream chemokine CXCL10, or the inflammatorycytokines interleukin-1b and interleukin-6 (Figs. 3A to 3D).Examination of the transcriptional response to TLR4- orTLR9-stimulation also revealed no effect of genotype onbasal levels of IFN-I mRNA or on induction of IFN-b, totalIFN-a, IFN-u, or the downstream markers of transientIFN-I production CXCL10 and ISG-54 (Figs. 3E to 3H)(and data not shown).Enhancer activity of IRF3/7-rich regions in 9p21. Se-quence analysis revealed 4 regions rich in consensus sites forthe principle IFN-I–inducing transcription factors IRF3 andIRF7 (GAAANNGAAA), which are close to or containSNPs in high LD (R2 > 0.8) with CAD-associated 9p21SNPs (Fig. 4A). However, the selected regions of eithergenotype did not demonstrate appreciable enhancer activityin unstimulated cells. Moreover, activation of IRF3 andIRF7 (by PolyI:C) or of IRF3, IRF5, and nuclear factorkappa B (by lipopolysaccharide) also failed to stimulatereporter activation (Fig. 4B).

Discussion

Harismendy et al. (2) recently reported that chromosomelooping induced by IFN-g permits close physical proximitybetween the 9p21 CAD risk interval and certain proximalgenes, such as CDKN2A and CDKN2B, and also thedistant IFN-I gene cluster, raising the possibility thatenhancers in the CAD risk locus may modify expression ofthese genes in a genotype-dependent manner. However, wefound no effect of genotype on plasma levels of IFN-I, thecapacity of PBMC to produce IFN-a, IFN-b, or theirdownstream markers via the 2 major pathways of theirinduction (i.e., TLR3/4/IRF3 or TLR9/IRF7) byELISA, bioassay, immunoblot, and quantitative polymerasechain reaction. As viral infections or medication could haveimpacted on our findings, we took care to only includesubjects that were free of any current medication, disease, orinfection. Moreover, although putative binding sites forIRF3 and IRF7 are present in the CAD risk region, these donot seem to possess enhancer activity basally or on TLRstimulation. It was also shown recently that IFN-g–depen-dent induction of CDKN2A and CDKN2B occurs inde-pendently of 9p21 genotype (9). Our findings are supportedby recent genetic evidence from studies of SLE, a disease thatis driven by genetic propensity to increased IFN-a produc-tion (6,10). SLE symptoms can be triggered in humans bythe administration of IFN-a, and approximately 60% ofSLE-associated SNPs are linked to IFN-I-related genes

Figure 3 Effect of 9p21 Genotype on Induction of IFN-I mRNA and Protein

PBMCs from 32 carriers of the CC, CG, and GG variants of rs1333049 (cohort 3) were challenged with medium alone (control), 100 ng/ml lipopolysaccharide, or 1.5 mmol/l CpG

ODN. Cytokines were measured by ELISA at 24 h (A to D). Expression of IFN-b, total IFN-a, CXCL10, and ISG54 mRNA was measured at 2 h and normalized to expression of

36B4 (E to H, log2y-axis). Ctrl ¼ control; IFN ¼ interferon; IL ¼ interleukin; LPS ¼ lipopolysaccharide; ns ¼ not significant.

Erridge et al. JACC Vol. 62, No. 15, 20139p21 Variants Do Not Regulate Type I Interferon October 8, 2013:1376–81

1380

(6,10). Nevertheless, despite numerous robust genome-wideassociation studies and meta-analyses for SLE in multiplepopulations, 9p21 has not emerged as a genetic risk factor forthis disease (10).

Figure 4 Enhancer Activity of Putative IRF3/7 Binding Site Clusters

Four regions (R1–R4) of the 9p21 coronary artery disease (CAD) risk locus were cloned fro

enhancer-dependent reporter construct (A). CAD-associated single nucleotide polymorphi

respectively. Human embryonic kidney 293 cells were transfected with each reporter and

enhancer. pEL is the positive control nuclear factor kappa B–sensitive reporter. Results n

0.01 versus medium alone. CAD ¼ coronary artery disease; Ctrl ¼ control; IRF ¼ interfer

Study limitations. It is possible that the observed lack ofeffect of genotype on IFN-I responses is due to limited studypower. On the basis of the mean and SD of log-transformedplasma IFN-a levels in the nonrisk group, our study had an

in 9p21

m subjects homozygous for the risk (CC) or nonrisk (GG) rs1333049 variants into an

sm (SNPs) and predicted IRF3/IRF7 binding sites are marked in red and green,

challenged as described in the Methods section. pGL3P is the control vector without

ormalized to co-transfected renilla reporter are presented as mean � SD (B). **p <

on regulatory factor; LPS ¼ lipopolysaccharide; NF-kB ¼ nuclear factor kappa B.

JACC Vol. 62, No. 15, 2013 Erridge et al.October 8, 2013:1376–81 9p21 Variants Do Not Regulate Type I Interferon

1381

80% power at an alpha of 5% to detect an 18.2% differencein log-plasma IFN-a in the risk genotype group (11.0% forplasma CXCL10) and 80% power to detect a difference of8.2% or 4.5% between log-IFN-a or log-IFN-I bioactivitylevels, respectively, in supernatants of CpG-ODN stimu-lated cultures of CC and GG genotypes. Because all thesubjects studied were of European Caucasian origin, we alsocannot exclude the possibility that 9p21 genotype modulatesIFN-I production in populations or ethnicities distinct fromthose examined here.

Conclusions

Our findings suggest that the mechanism by which 9p21genotype modulates cardiovascular risk does not involveregulation of IFN-I responses.

Reprint requests and correspondence: Dr. Clett Erridge,Department of Cardiovascular Sciences, University of Leicester,Glenfield Hospital, Leicester LE3 9QP, United Kingdom. E-mail:ce55@le.ac.uk.

REFERENCES

1. Schunkert H, König IR, Kathiresan S, et al. Large-scale associationanalysis identifies 13 new susceptibility loci for coronary artery disease.Nat Genet 2011;43:333–8.

2. Harismendy O, Notani D, Song X, et al. 9p21 DNA variants associ-ated with coronary artery disease impair interferon-g signallingresponse. Nature 2011;470:264–8.

3. Goossens P, Gijbels MJ, Zernecke A, et al. Myeloid type I interferonsignaling promotes atherosclerosis by stimulating macrophage recruit-ment to lesions. Cell Metab 2010;12:142–53.

4. Döring Y, Manthey HD, Drechsler M, et al. Auto-antigenic protein-DNA complexes stimulate plasmacytoid dendritic cells to promoteatherosclerosis. Circulation 2012;125:1673–83.

5. Zhang LN, Velichko S, Vincelette J, et al. Interferon-beta attenu-ates angiotensin II-accelerated atherosclerosis and vascular remod-eling in apolipoprotein E deficient mice. Atherosclerosis 2008;197:204–11.

6. Niewold TB, Hua J, Lehman TJ, et al. High serum IFN-alpha activityis a heritable risk factor for systemic lupus erythematosus. GenesImmun 2007;8:492–502.

7. Tobin MD, Raleigh SM, Newhouse S, et al. Association of WNK1gene polymorphisms and haplotypes with ambulatory blood pressure inthe general population. Circulation 2005;112:3423–9.

8. Jabs WJ, Hennig C, Zawatzky R, Kirchner H. Failure to detect anti-viral activity in serum and plasma of healthy individuals displaying highactivity in ELISA for IFN-alpha and IFN-beta. J Interferon CytokineRes 1999;19:463–9.

9. Almontashiri NA, Fan M, Cheng BL, et al. Interferon-g activatesexpression of p15 and p16 regardless of 9p21.3 coronary artery diseaserisk genotype. J Am Coll Cardiol 2013;61:143–7.

10. Bronson PG, Chaivorapol C, Ortmann W, et al. The genetics of type Iinterferon in systemic lupus erythematosus. Curr Opin Immunol 2012;24:530–7.

Key Words: coronary artery disease - 9p21 - Toll-like receptor - type Iinterferon.