Quantitative trait locus mapping in mice identifies phospholipase Pla2g12a as novel atherosclerosis modifier

Accepted Manuscript Quantitative trait locus mapping in mice identifies phospholipase Pla2g12a as novel atherosclerosis modifier Alexandros Nicolaou, Bernd H. Northoff, Kristina Sass, Jana Ernst, Alexander Kohlmaier, Knut Krohn, Christian Wolfrum, Daniel Teupser, Lesca M. Holdt PII:

S0021-9150(17)31252-2

DOI:

10.1016/j.atherosclerosis.2017.08.030

Reference:

ATH 15185

To appear in:

Atherosclerosis

Received Date: 28 February 2017 Revised Date:

23 August 2017

Accepted Date: 24 August 2017

Please cite this article as: Nicolaou A, Northoff BH, Sass K, Ernst J, Kohlmaier A, Krohn K, Wolfrum C, Teupser D, Holdt LM, Quantitative trait locus mapping in mice identifies phospholipase Pla2g12a as novel atherosclerosis modifier, Atherosclerosis (2017), doi: 10.1016/j.atherosclerosis.2017.08.030. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT Quantitative trait locus mapping in mice identifies phospholipase Pla2g12a as novel atherosclerosis modifier

RI PT

Alexandros Nicolaou a, Bernd H. Northoff a, Kristina Sass a, Jana Ernst b, Alexander

SC

Kohlmaier a, Knut Krohn c, Christian Wolfrum d, Daniel Teupser a, Lesca M. Holdt a

a

Institute of Laboratory Medicine, Ludwig Maximilians University Munich, Munich, Germany,

b

M AN U

Department of Anatomy and Cell Biology, Faculty of Medicine, Martin Luther University

Halle-Wittenberg, Halle (Saale), Germany, cInterdisciplinary Center for Clinical Research Leipzig (IZKF), Core-Unit DNA Technologies, University of Leipzig, Leipzig, Germany, d

TE D

Institute of Food, Nutrition and Health, ETH Zurich, Schwerzenbach, Switzerland

Correspondence: Institute of Laboratory Medicine, Ludwig-Maximilians-University Munich (LMU),

Marchioninistr.15,

81377

Munich,

Germany.

E-mail:

[email protected]

AC C

EP

muenchen.de (L. M. Holdt)

1

ACCEPTED MANUSCRIPT

Abstract

Background and aims: In a previous work, a female-specific atherosclerosis risk locus on

atherosclerosis-susceptible

C57BL/6

(B6)

mice

on

RI PT

chromosome (Chr) 3 was identified in an intercross of atherosclerosis-resistant FVB and the

LDL-receptor

deficient

(Ldlr-/-) background. It was the aim of the current study to identify causative genes at this

SC

locus.

Methods: We established a congenic mouse model, where FVB.Chr3B6/B6 mice carried an

M AN U

80 Mb interval of distal Chr3 on an otherwise FVB.Ldlr-/- background, to validate the Chr3 locus. Candidate genes were identified using genome-wide expression analyses. Differentially expressed genes were validated using quantitative PCRs in F0 and F2 mice and their functions were investigated in pathophysiologically relevant cells.

for

female

TE D

Results: Fine-mapping of the Chr3 locus revealed two overlapping, yet independent subloci atherosclerosis

susceptibility:

when

transmitted

by

grandfathers

to

granddaughters, the B6 risk allele increased atherosclerosis and downregulated the

EP

expression of the secreted phospholipase Pla2g12a (2.6 and 2.2 fold, respectively); when inherited by grandmothers, the B6 risk allele induced vascular cell adhesion molecule 1

AC C

(Vcam1). Down-regulation of Pla2g12a and up-regulation of Vcam1 were validated in female FVB.Chr3B6/B6 congenic mice, which developed 2.5 greater atherosclerotic lesions compared to littermate controls (p=0.039). Pla2g12a was highly expressed in aortic endothelial cells in vivo, and knocking-down Pla2g12a expression by RNAi in cultured vascular endothelial cells or macrophages increased their adhesion to ECs in vitro. Conclusions: Our data establish Pla2g12a as an atheroprotective candidate gene in mice, where high expression levels in ECs and macrophages may limit the recruitment and accumulation of these cells in nascent atherosclerotic lesions. 2

ACCEPTED MANUSCRIPT Keywords: Quantitative trait locus (QTL) mapping; inbred mouse strains; secreted phospholipases;

AC C

EP

TE D

M AN U

SC

RI PT

vascular endothelial cells; cell adhesion

3

ACCEPTED MANUSCRIPT 1. Introduction

In the last decades, the identification of genomic loci that govern atherosclerosis has been promoted by the development of quantitative trait locus (QTL) mapping approaches in inbred mouse strains on the atherosclerosis-sensitizing LDL-receptor (Ldlr-/-)1 or apolipoprotein E

RI PT

(ApoE-/-) deficient backgrounds2, 3. So far, 46 loci have been identified in different mouse intercrosses4, 5. In parallel approaches, genome-wide association (GWAS) studies in human

SC

cohorts have revealed 57 loci that are associated with coronary artery disease in humans6, 7.

In a previous work, we have identified 3 atherosclerosis risk loci located on mouse

M AN U

chromosomes (Chr) 3, 10, and 12 based on QTL mapping, in an intercross of atherosclerosis-resistant FVB and atherosclerosis-susceptible C57BL/6 (B6) mice on the LDL-receptor deficient (Ldlr-/-) background8. For the initial QTL mapping, F1 animals from reciprocal crosses of male and female B6 and FVB mice were intercrossed and atherosclerosis lesion size quantified in the F2 generation8. In subsequent studies, we have

TE D

identified the metalloprotease Adam17 as candidate atherosclerosis modifier on Chr12 and have recently validated its anti-atherogenic role using a hypomorphic mouse model9. In parallel, increased expression of the major histocompatibility complex-like molecule Raet1e

EP

has been suggested to be causal for the Chr10 risk QTL10.

AC C

In the current study, we report on fine-mapping and functional dissection of the Chr3 atherosclerosis susceptibility locus. This locus exhibited gender-dependent effects on atherosclerosis onset, such that it became apparent only in F2 female but not in F2 male mice8. Furthermore, two linkage peaks were observed in the female F2 population: the first at 78.5 cM on Chr3, and becoming apparent only in a cross where the atherosclerosissusceptible B6 locus was introduced via grandfathers (referred in the following to as “cross A”); the second peak at 55 cM on Chr3 and only becoming apparent when the B6 risk locus was introduced via grandmothers (referred in the following to as “cross B”; see Methods section for description of genotypes)8. The linkage peaks in crosses A and B displayed LOD 4

ACCEPTED MANUSCRIPT score maxima at 3.3 and 4.1, respectively. Together, this indicated that potentially independent candidate genes on the Chr3 risk QTL predisposed for atherosclerosis susceptibility in females.

RI PT

Using systematic mRNA expression profiling in congenic mice, expression QTL (eQTL) mapping in male and female F2 mice, and tissue and cell profiling in parental mice, we identified discrete candidate genes for subcrosses A and B, respectively, and tested their functions in atherosclerosis-relevant cellular processes using RNAi approaches. Our data

SC

define two atherosclerosis susceptibility genes on Chr3 and, extending earlier mapping studies, functionally explore the molecular roles of the encoded factors in physiologically

AC C

EP

TE D

M AN U

relevant cellular settings.

5

ACCEPTED MANUSCRIPT 2. Material and methods

2.1.

F2 intercross and fine-mapping of Chr3 locus

In a previous work, male Ldlr

-/-

mice on the C57BL/6 background (B6.129S7-Ldlr

tm1Her

/J

stock 002207 from Jackson Laboratory; B6) were crossed with female Ldlr-/- mice on the FVB

RI PT

background, and the resulting F1 animals were intercrossed to generate 107 female and 112 male F2 animals (referred to as “cross A” throughout the paper). In an independent cross, male Ldlr-/- mice on the FVB background were crossed with female Ldlr-/- mice on the B6

SC

background and the resulting F1 animals were intercrossed to generate 120 female and 120 male F2 animals8 (referred to as “cross B” throughout). F2 mice were weaned at 4 weeks of

M AN U

age and fed a semisynthetic modified AIN76 diet with 0.02% cholesterol. Atherosclerosis was quantified in the aortic roots of 20 weeks old animals8. To fine-map the Chr3 atherosclerosis risk locus, 7 additional single nucleotide polymorphisms (SNPs) were genotyped using melting curve analysis11 (Supplementary Table 1 for list of SNPs).

TE D

2.2. Generation of Chr3 congenic mice

Chr3 congenic mice were generated by crossing B6 and FVB mice and backcrossing the resulting heterozygous Chr3FVB/B6 F1 mice onto the FVB.Ldlr-/- background for 10 generations.

EP

Heterozygous FVB.Chr3FVB/B6 (carrying an interval from megabases (Mb) 80 to 160 from Chr3 of B6 mice in an otherwise FVB background) were intercrossed to generate

AC C

FVB.Chr3B6/B6, FVB.Chr3FVB/B6 and wild-type FVB.Chr3FVB/FVB control mice. Animal care and experimental procedures were approved by the responsible authorities of the Free State of Saxony (Reg. No. N6/11).

2.3.

Laboratory analyses

Clinical chemistry parameters, alanine transaminase (ALT), aspartate transaminase (AST), cholinesterase (ChE), urea, cholesterol and triglycerides were measured on automated analyzers (Modular PPE, Roche). Lipoproteins were isolated by sequential ultracentrifugation

6

ACCEPTED MANUSCRIPT (Beckman Coulter, Optima LE-80K Ultracentrifuge, Rotor Ti42.2) and cholesterol was determined as described previously12. Cholesterol distribution among lipoproteins was determined by fast protein liquid chromatography (FPLC). For this, plasma samples of 4 mice of each strain, C57BL/6.Ldlr-/-, chromosome 3 congenics (FVB.Chr3B6/B6.Ldlr-/-) and FVB.Ldlr/-

RI PT

, were pooled and FPLC elution profiles were measured. For determination of blood cell

parameters (white blood cells), an automated hematology analyzer (Sysmex) was used.

2.4.

Cell culture and adhesion assays

SC

Bone marrow cells of congenic FVB.Chr3B6/B6, C57BL/6 and FVB F0 mice, mouse cardiac endothelial cells (MCEC, C57/BL6 genotype, Cedarlane - Cat.# CLU510), primary vascular

M AN U

aortic smooth muscle cells (MOVAS, C57/BL6 genotype, ATCC CRL-2797) and RAW264.7 macrophages (BALB/c genotype) were cultivated as described9, 11. For quantifying adhesion of bone marrow-derived BMDMs to endothelial cells, two experimental setups were investigated: first, siRNA-treated ECs were preplated on uncoated 96 well plates at 28,000

TE D

cells per well, 24 h before the adhesion experiments. BMDMs were seeded on top of the MCECs, incubated for 45 min, washed twice with cell culture medium and BMDMs’ adhesion was measured with an IncuCyte Live Cell Analysis System (Essen Bioscience).

EP

Quadruplicate measurements per condition were performed. Second, BMDM were treated with siRNA for 24 h and subsequently allowed to adhere to non-treated ECs. Adherent cells

AC C

were quantified as described above. Cell adhesion assays of RAW264.7 macrophages were performed in 96-well plates coated with Matrigel (BD Biosciences) or on uncoated dishes. Cells

were

allowed

to

adhere

for

40

min

in

both

conditions,

quadruplicate

measurements/condition were performed. Numbers of adherent cells were determined in relation to standard curves using CellTiter-Glo (Promega). Caspase 3/7 activity was measured during apoptosis assays13.

7

ACCEPTED MANUSCRIPT 2.5.

RNAi

siRNA knockdown of mouse Pla2g12a (Cat. # s83067, Ambion) and scrambled (SCR, Cat. # 4390843, Ambion) control was performed in BMDM of congenic FVB.Chr3B6/B6, C57BL/6 and FVB mice, in mouse cardiac endothelial cells (MCEC) and RAW264.7 macrophages using

RI PT

the Amaxa Nucleofector Technology (4D-Nucleofector System, Lonza) by transfecting 0.5-1 Mio cells. For the BMDMs we used the P2 Primary Cell 4D-Nucleofector X Kit L (Catalog # V4XP-2024), for the MCECs the P5 Primary Cell 4D-Nucleofector X Kit L (Catalog # V4XP5024) and for the RAW264.7 macrophages the SF Cell Line 4D-Nucleofector X Kit L

Phenotyping and RNA expression analysis

M AN U

2.6.

SC

(Catalog # V4XC-2024). Knock-down efficiency was determined by qPCRs.

Atherosclerosis analysis and RNA expression analyses were conducted as described8, 14. In brief, atherosclerosis quantification of congenic FVB.Chr3B6/B6 and control mice was performed on 10 µm aortic root sections of hearts that were cut parallel to the valvular plane.

TE D

Five cuts were selected with a distance of 40 µm and quantification was performed using Oil Red O (Sigma-Aldrich) for staining neutral triglycerides and lipids, light green SF yellowish (Merck) for collagen and haematoxylin (Sigma-Aldrich) for cell nuclei. Before staining,

EP

cryosections were fixed in formalin solution overnight at 4°C. The staining was performed with an automatic slide stainer (Bavimed). Total RNA from livers and cells was isolated using

AC C

TRIzol reagent (Life Technologies), RNA from mouse aortas was isolated using the RNAqueous-Micro Kit (Thermo Fisher Scientific) according to the manufacturer’s instructions. cDNA was prepared using SuperScript II Reverse Transcriptase and random hexamer priming. qPCRs (see Supplementary Table 2 for primers and probes) were performed on a ViiA 7 Real-Time PCR-System (Life Technologies)15. For microarrays, total liver RNA from female congenic FVB.Chr3B6/B6 (n=4) and FVB (n=4) mice was labeled and hybridized to Illumina MouseWG-6 v2.0 Expression BeadChips and imaged using the iScan System. Bead level data were preprocessed in the Illumina Genome Studio. The Supplementary file (Array FVB.Chr3B6/B6 vs

FVB.Chr3FVB/FVB.txt)

contains

the 8

ACCEPTED MANUSCRIPT corresponding raw gene expression data after quantile normalization and background subtraction.

2.7.

Immunohistochemical staining

RI PT

Immunohistochemical staining of frozen and methanol-acetone (1:1)-fixed serial sections (10 µm) of the aortic root was performed after quenching peroxidases with 3% H2O2 for 3 min at room temperature and blocking samples with normal horse or goat serum: primary antibodies PLA2G12A (16009-1-AP, Proteintech), CD68 (MCA1957, Serotec), α-SMA

SC

(ab5694; Abcam) and von Willebrand Factor (vWF) (ab11713, Abcam) were incubated

M AN U

overnight (4˚C), horseradish peroxidase-conjugated secondary antibodies were incubated for 30 minutes and visualized with Nova Red (Vector Laboratories). Samples were counterstained with hematoxylin and embedded with VectaMount mounting medium.

2.8.

Western blotting

TE D

Immunoblots of PLA2G12A and GAPDH were performed in BMDMs pooled from 4 C57BL/6 mice, MCEC and MOVAS cells9 by loading 20 µg per sample. Protein bands were visualized by chemiluminescence (AceGlow-Solution, VWR). Densitometric analysis of western blots

Statistical analysis

AC C

2.9.

EP

was performed using ImageJ Software (NIH).

All data are given as mean ±SEM unless otherwise indicated. Normality of distribution was assessed using Kolmogorov-Smirnov testing (PRISM6 statistical software; GraphPad). Multiple groups were compared by ANOVA using Tukey tests as post-tests. Comparison of two groups of normally distributed samples was done using the t test. Linkage analysis for single QTLs was performed using the QTL mapping software, R/qtl, implemented as an addon package for the statistical software R16 17.

9

ACCEPTED MANUSCRIPT 3. Results

3. Fine-mapping of Chr3 subloci in female F2 mice and validation of the Chr3 atherosclerosis locus in congenic mice

RI PT

To fine map the two previously identified female-specific atherosclerosis risk subloci within the Chr3 QTL8, we genotyped 6 additional SNPs in 107 female mice of cross A and in 120 female mice of cross B (Fig. 1). QTL analysis revealed LOD score maxima for

SC

atherosclerosis at the initially identified markers d3mit45 (LOD 3.3, cross A) and d3mit57 (LOD 4.1, cross B), where homozygosity for the B6 alleles correlated with 2.3-fold and 2.2-

M AN U

fold more atherosclerosis8. Applying a conservative cutoff of 1.5 LOD when defining the confidence intervals18, we narrowed down the genetic regions potentially containing candidate genes of atherosclerosis risk from 44 to 37 Mb in cross A and 55 to 44 Mb in cross B, respectively (Supplementary Table 3). The two loci now spanned from SNP rs36523879 to rs31598641 (Fig. 1A) and from SNP rs4224041 to d3mit254 (Fig. 1B), containing 221 genes

TE D

(cross A) and 593 genes (cross B), respectively. To validate that the B6 allele on Chr3 was causal for promoting atherosclerosis, we generated Chr3 congenic mice by introducing an 80 Mb long genomic stretch of Chr3 from

EP

atherosclerosis-susceptible B6 mice into the FVB.Ldlr-/- background (Fig. 1C). By this, we produced FVB mice homozygous for the B6-derived Chr3 locus (FVB.Chr3B6/B6) or

AC C

heterozygous for the locus (FVB.Chr3B6/FVB). Since these mice had been derived by crossing heterozygotes, FVB.Chr3FVB/FVB littermates were used as controls. Clinical chemistry parameters and plasma lipid levels showed no differences between FVB.Chr3B6/B6, FVB.Chr3B6/FVB and FVB.Chr3FVB/FVB mice (Supplementary Table 4, Supplementary Fig. 1). Yet, we found that female homozygous FVB.Chr3B6/B6 mice showed robustly enhanced 2.5fold larger atherosclerotic lesions in the aortic root compared to heterozygous FVB.Chr3B6/FVB (p=0.02) or control FVB.Chr3FVB/FVB mice (p=0.039) (Fig. 1D and E). These data confirmed that atherosclerosis-susceptibility was contained within the 80 Mb of genomic DNA from Chr3

10

ACCEPTED MANUSCRIPT of B6 mice, and that atherogenesis was likely triggered independent of cholesterol and triglyceride metabolism.

3.1.

Identification of candidate genes by transcriptome-wide expression profiling in

RI PT

Chr3 congenic and F0 mice We next aimed to prioritize candidate genes underlying the QTL on Chr3 of B6 mice for subsequent functional characterization using a multistep process, which is detailed in Supplementary Fig. 2. We first performed transcriptome-wide expression analyses in RNA

SC

from livers of female congenic FVB.Chr3B6/B6 mice and in FVB control mice (n=4/4). We found that 19 genes were differentially expressed with a fold change (FC) ratio >1.4 and <0.7,

M AN U

which we chose as a biologically relevant arbitrary cutoff (Fig. 2A and Table 1). 9 of these genes were located in the confidence interval of Cross A, 13 genes in the interval of Cross B, and 3 of these genes were in the overlap of the two intervals (Fig. 2A and Table 1). Using qPCRs we validated the differential expression of these genes by comparing their mRNA

TE D

abundance firstly in preparations of livers from congenic FVB.Chr3B6/B6 and FVB (n=8/9) mice, and secondly by analyzing the parental F0 B6 and FVB female mice (n=5/5) (Table 1). To examine whether the identified candidates were also differentially expressed in vascular

EP

tissue or cells known to contribute to atherosclerotic plaque formation, we next quantified candidate gene expression in parental F0 aortas (n=6/6) and, secondly, in bone marrow-

AC C

derived macrophages (BMDM, n=6/6) (Table 1). In total, 7 genes passed the first two validation stages of expression analysis in the liver, out of which 5 were also expressed in aortic tissue and/or BMDMs: tropomodulin-4 (Tmod4), vascular cell adhesion molecule 1 (Vcam1), fatty acid elongase 6 (Elovl6), the group XIIA secretory phospholipase A2 (Pla2g12a) and cystathionine gamma lyase (Cth) (Table 1). Of these, Vcam1, Elovl6 and Cth were induced, and Tmod4 and Pla2g12a were repressed in mice carrying the B6 allele in the Chr3 interval (Table 1). To remain conservative in our gene prioritization strategy we took all 7 candidates to further analyses stages.

11

ACCEPTED MANUSCRIPT 3.2.

Validation of Pla2g12a and Vcam1 candidate genes through expression QTL mapping in female F2 mice

To next dissect, if indeed the B6-derived genetic variants that promoted atherosclerosis also regulated the expression of our candidate genes in cis, we obtained mRNA preparations

RI PT

from livers of female F2 mice that were derived from cross A or from cross B, and performed expression quantitative trait locus (eQTL) mapping for Pklr (pyruvate kinase liver and RBC), Vcam1, Elovl6, Agxt2l1 (alanine glyoxylate aminotransferase 2-like1), Cth, Tmod4 and Pla2g12a (Fig. 2B, C). Unfortunately, no other tissues were available from the F2 mice.

SC

Since eQTLs can principally arise due to genetic variants contained either gene-proximally (within the transcribed RNA or the promoter/promoter-proximal regions, changing the

M AN U

transcription, stability or degradation kinetics of the mRNA by various mechanisms)19 or in more distant regulatory regions20, we performed an identity by descent analysis (IBD) of chromosome 3 in C57BL/6 and FVB mice to explore if any genomic regions in the candidate gene interval could be excluded due to sharing of haplotypes (Fig. 2A, Supplementary Table

TE D

5). The IBD analysis revealed that all 7 candidate genes indeed resided in regions that were inherited independently in both strains, and thus none of these candidates could be excluded. Also, none of the 7 candidate genes contained non-synonymous protein-

EP

sequence-changing nsSNPs directly within their coding regions (Supplementary Table 6). For Elovl6, Agxt2l1 and Tmod4 no eQTLs were detected in any subcross

21-23

and

AC C

these genes were thus not further considered as candidate genes (Fig. 2B, C). On the contrary, we observed a LOD score maximum of 3.6 at SNP rs36523879 for Pla2g12a expression in F2 females of cross A, which is 7.4 Mb 5’ to the Pla2g12a gene and the top atherosclerosis marker d3mit45 in this subcross (Fig. 2B). In the same subcross, we observed a significant LOD score for Cth expression at marker d3mit19, which is located 9.8 Mb distal to marker d3mit45 and only 88.7 kb 3’ of the Cth gene. Applying the same conservative cutoff of 1.5 LOD when defining the confidence interval for the Cth cis eQTL18, Cth’s spans from rs30102565 to rs31598641 (150.234.187-159.008.817 Mb). Thus, there is

12

ACCEPTED MANUSCRIPT no overlap between the Cth cis eQTL and the maximum LOD score for atherosclerosis in this subcross making Cth an unlikely, yet not impossible, candidate. In subcross B, a LOD score maximum of 4.4 for Vcam1 was detected and overlapped most closely the location of atherosclerosis-susceptibility (Fig. 2C). We also observed a

RI PT

significant eQTL for Pklr at marker d3mit63, however, as with Cth in subcross A, no overlap of the confidence interval of the Pklr cis eQTL (d3mit203 to rs3683384; 26.8–110 Mb) with the maximum atherosclerosis LOD score was detected. Females, carrying two B6-alleles at d3mit57 had 1.5-fold higher Vcam1 expression levels (p= 0.001; Supplementary Fig. 3) and

SC

developed 1.9-fold more atherosclerosis (p= 0.004; Supplementary Fig. 3A, B, C). Since Vcam1 is a well-known regulator of atherogenesis24 and high Vcam1 levels have been shown 26

, our data are consistent with the published literature and

M AN U

to promote atherosclerosis25,

make Vcam1 a likely effector gene in cross B. Thus, from the analysis of cross A, Pla2g12a emerged as novel candidate atherosclerosis modifier.

Pla2g12a downregulation correlates with increased atherosclerosis

TE D

3.3.

To improve eQTL mapping accuracy in cross A, we increased the marker coverage at the chromosomal location of Pla2g12a by genotyping SNP rs13464244, located only 0.24 Mb 5’

EP

of the gene. This led to a co-localization of the LOD score maximum of atherosclerosis (3.7) with an increased LOD score maximum for Pla2g12a expression of 5.01 (Fig. 3A).

AC C

Homozygosity for the B6 allele at rs13464244 in F2 females from cross A was associated with decreased mRNA abundance of Pla2g12a compared to the FVB genotype at this position (p=9.6 x 10-07) and with enlarged atherosclerotic lesions (p=0.0009) (Fig. 3B and C). Furthermore, we observed an inverse correlation of high Pla2g12a mRNA expression with smaller atherosclerotic lesion sizes (r2=0.18, p=0.002) indicating that Pla2g12a protects from atherosclerosis (Fig. 3D). Consistently, we observed significantly lower expression levels in aortic tissue of B6 compared to FVB F0 mice (Fig. 3E). On the contrary, Cth in cross A, and Pklr in cross B, revealed no correlation with atherosclerosis (Supplementary Fig. 4). Together, eQTL mapping in female F2 progeny from cross A identified Pla2g12a as an 13

ACCEPTED MANUSCRIPT independent candidate gene for atherosclerosis susceptibility in this locus and indicated that low Pla2g12a expression in the arterial wall might promote atherosclerosis.

3.4.

Gender-dependence of eQTLs in F2 mice

RI PT

Corroborating the known gender-dependence of this Chr3 QTL (see Fig. 1), eQTL fine mapping with SNP rs13464244 in F2 males of cross A revealed LOD scores for Pla2g12a and atherosclerosis below the significance limit of 3 (Supplementary Fig. 5A) and consistently, Pla2g12a mRNA levels were not correlated with atherosclerosis lesions size

SC

(r2=0.00006, p=0.87) (Supplementary Fig. 5B-D). Similarly, Vcam1 was not induced and did not correlate with atherosclerosis levels in male cross B animals (Supplementary Fig. 6A-D).

M AN U

Together, eQTL mapping in F2 mice indicated that downregulation of Pla2g12a and upregulation of Vcam1 might be causal for promoting atherosclerosis in female but not male F2 mice carrying the B6 alleles at the two atherosclerosis subloci of the Chr3 risk QTL.

Pla2g12a-deficiency macrophages

promotes

adhesion

between

endothelial

cells

and

TE D

3.5.

Whereas Vcam1 is an established atherosclerosis modifier gene27, the potential role of

EP

Pla2g12a, which belongs to the secreted phospholipase A2 enzymes (sPLA2), has not been investigated in atherogenesis, yet. We thus focussed on PLA2G12A and performed

AC C

immunohistochemical stainings, qPCR analyses and western blots to prioritize which vascular cell type to study in subsequent functional analyses. Immunohistochemical stainings in consecutive sections in the aortic root of a C57BL/6 mouse of PLA2G12A protein ,CD68positive macrophages, von Willebrand-factor-positive (vWF) endothelial cells and α-SMApositive smooth muscle cells showed that PLA2G12A was highly expressed in vascular endothelial cells (ECs) and to a lesser extent in macrophages (Fig. 4A). Also in cultured cell lines, Pla2g12a was expressed most strongly in endothelial cells when compared to cultured macrophages or vascular smooth muscle cells (Fig. 4B). The expression results were

14

ACCEPTED MANUSCRIPT validated on protein level, where we observed highest protein levels of PLA2G12A in MCECs (Fig. 4C). EC activation is one of the first cellular changes in the vasculature en route to atherosclerosis28,

29

and during atherosclerosis development ECs control adhesion of

RI PT

circulating leukocytes to nascent atherosclerotic lesions30. We thus investigated, if Pla2g12a was important for the adhesive capacity of ECs (Fig. 4D-F). To this end, we established a cell culture model, where bone marrow-derived macrophages (BMDMs) were seeded on top of pre-plated ECs (Fig. 4D). Here, we found that BMDMs from B6 mice, as well as BMDMs

SC

derived from Chr3 congenic mice (both with lower Pla2g12a expression) were more adhesive than FVB BMDMs (relative higher Pla2g12a expression; Fig. 4E). Importantly, we found that

M AN U

depleting Pla2g12a from ECs in this adhesion assay further increased adhesion of (untreated) BMDMs to ECs, with a comparable relative increase in adhesion of both B6derived and FVB-derived BMDMs (Fig. 4E, F). These results indicated that PLA2G12A was exhibiting parallel atheroprotective roles in macrophages and in ECs. In this adhesion assay

TE D

upon loss of Pla2g12a, mRNA levels of the integrin ligand Vcam1 did not increase (Fig. 4F) suggesting that the increase in adhesion was due to Pla2g12a-dependent processes and independent of indirect transcriptional regulation of Vcam1. To ask if Pla2g12a levels were

EP

determining the adhesive capacity also in bone marrow-derived macrophages, we depleted Pla2g12a in BMDMs (Fig. 4G-I) that we had derived from FVB mice, FVB.Chr3B6/B6

AC C

congenics and C57BL/6 mice. We observed an increased adhesion after Pla2g12a knockdown in cells from all genetic backgrounds compared to cells treated with scrambled RNAi oligonucleotides (Fig. 4G). Again, Vcam1 expression was not induced during the adhesion experiment (Fig. 4I). To test whether Pla2g12a was, thus, a more general determinant of adhesion, we depleted Pla2g12a in a cultured macrophage cell line of independent genetic origin (RAW264.7 cells from BALB/C mice). Also in this cell line we observed increased adhesion to both untreated culture plates as well as to matrigel-coated surfaces after Pla2g12a knockdown compared to cells treated with scrambled RNAi oligonucleotides (Supplementary Fig. 7A-C). These data indicate that PLA2G12A is normally 15

ACCEPTED MANUSCRIPT functioning as anti-adhesive factor irrespective of the genetic background, which is consistent with the increased sensitivity to atherosclerosis when PLA2G12A levels are low (Fig. 3). We think that this role in curbing adhesion was a specific regulatory process, because depletion of Pla2g12a did, for example, not alter other cellular regulatory modes in

RI PT

ECs such as cell survival properties (Supplementary Fig. 8). Summarizing, in vivo as well as in vitro data indicate a role for Pla2g12a in protecting from

atherosclerosis-relevant

cellular

functions

highlight

Pla2g12a

as

novel

AC C

EP

TE D

M AN U

SC

atherosclerosis modifier in mice.

and

16

ACCEPTED MANUSCRIPT 4. Discussion

Our study is the first to dissect two overlapping, yet independent female-specific QTL loci for atherosclerosis susceptibility by investigating inheritance patterns from the grandparental F0

RI PT

generation together with systematic expression QTL analyses in male and female F2 mice. Thereby, we identify Pla2g12a as novel atherosclerosis modifier gene in a female subcross (cross A), where the atherosclerosis-susceptible locus was introduced via B6.Ldlr-/grandfathers. PLA2G12A prevented leukocyte adhesion to endothelial cells in FVB mice,

SC

which is one of the earliest events in atherogenesis24, 25, 31, 32, and establishes Pla2g12a as

M AN U

atheroprotective factor in this mouse intercross. In parallel, eQTL analysis identified Vcam1 as independent candidate effector of the locus in subcross B, where high Vcam1 expression and atherosclerosis were associated with the B6 risk allele when introduced via grandmothers. The Chr3 locus was correlated with atherosclerosis risk only in F2 female progeny of either cross, but not in male F2 mice and, consistently, we observed no eQTLs in

TE D

male F2 mice for Pla2g12a or Vcam1. Given that the DNA sequence was identical in mice of the grandparents’ generation, an epigenetic regulation and sex-specific regulation of its transmission through the F1 and F2 generations is likely. It might, thus, become an important

EP

factor in the experimental design for gene discovery in inbred mouse models to consider male or female germline transmission in QTL studies. Furthermore, future profiling of

AC C

epigenetic imprints in such models may explain at least part of the missing causality of complex traits in existing studies33, 34 35, 36. Although it goes beyond the scope of this paper to identify the exact DNA position or

epigenetic mark that causes Pla2g12a and Vcam1 misregulation through the female F2 risk alleles, our study establishes the possibility that expression changes in VCAM1 and PLA2G12A may modulate atherosclerosis susceptibility by independently controlling a common effector pathway: induction of the immunoglobulin-superfamily adhesion molecule Vcam1 on endothelial cells25 and binding of VCAM1 to integrin α4β137 on homing blood mononuclear leukocytes38 recruits monocytes to growing lesions

29

, which is a key step in 17

ACCEPTED MANUSCRIPT initiating and promoting atherosclerosis32. Consistently, blocking of VCAM1 by an antibody39 or in a hypomorphic mouse mutant in vivo24, reduced monocyte adhesion on the early atherosclerotic endothelium and reduced lesion size. On the contrary, Pla2g12a has not been decisively functionally explored before40 and no knockout data exist. It belongs to the

RI PT

group of secreted phospholipase A2 enzymes41. Likely due to the secretory nature of this family, existing mouse models of other sPLA2s have revealed diverse, tissue-specific and sometimes contradicting cellular functions42-46. Our data now indicate that PLA2G12A may dampen the onset of atherosclerosis via reducing cell-cell adhesive properties in the vascular

SC

endothelium. How adhesion changes upon alteration of PLA2G12A levels remains unknown, but we can at least exclude that Vcam1 expression itself is regulated by Pla2g12a.

M AN U

Interestingly, previous reports have shown that some soluble PLA2s (group IIA) can bind to selective integrin subtypes (αvβ3, α4β1), where they compete with the integrin ligand VCAM1 for binding47,

48

. Given that adhesion of macrophages to endothelial cells was

increased when Pla2g12a was depleted in our own experiments (Fig. 4), it is conceptually

TE D

possible, that PLA2G12A impairs adhesion via impairing integrin signaling, by virtue of blocking the access of the integrin ligand VCAM1 to integrins. Supporting this integrinbinding hypothesis for altering adhesive properties, phospholipase catalytic activity is not

EP

needed for this type of interaction of sPLA2’s with integrins47. This is relevant, because PLA2G12A’s phospholipase activity is known to be very weak compared to other members of

AC C

this family of secreted phospholipases49, 50. Together, genetically determined parallel Vcam1 upregulation and Pla2g12a down-regulation may hypothetically converge on activation of integrin-dependent adhesion. Epistasis analyses in the F2 generation of our intercrosses revealed no significant interaction between SNPs in the candidate interval on Chr38, thus rather speaking against a coordinate regulatory function of Pla2g12a and Vcam1 in atherogenesis. Furthermore, despite the cell-proximal role of PLA2G12A in adhesion control (shown by the RNAi experiments) and the absence of relevant lipid changes, we can at this point not exclude that still unknown factors might change atherosclerosis onset by additional routes. Moreover, we have focused mainly on factors that changed their gene expression 18

ACCEPTED MANUSCRIPT dependent on atherosclerosis risk genotype, but we want to note that a number of nonsynonymous SNPs exist in the coding regions of the interval in genes that are not identical by descent. Bioinformatic predictions indicate that some may be detrimental to protein function, and Nes, Gon4l, Tchhl1, Hsd3b2, Pdlim5, Erich3 and Tnni3k are potential

RI PT

candidates at this level of analysis (Supplementary Table 6). Nevertheless this does not necessarily affect our finding that the change in Pla2g12a mRNA abundance is an independent important factor, as demonstrated by functional RNAi experiments in different genetic backgrounds.

SC

Supporting a broader relevance of our findings, increased secretory phospholipase A2 as well as lipoprotein-associated phospholipase A2 have already been identified as biomarkers of

M AN U

cardiovascular risk in population studies with patients with established coronary artery disease (CHD)51. This has to be qualified, however, as no SNPs within PLA2G12A itself have been associated with atherosclerosis in humans. Instead, other human SNPs contained in the mouse Chr3 syntenic regions are indeed associated with different endpoints of the

TE D

disease and some of these associate with loci for atherosclerosis risk factors, such as blood pressure and relevant lipids (Supplementary Table 7). Since the orthologues of the relevant human genes were not differentially expressed in our mouse model (Supplementary Table

EP

7), and since we found no co-localization of the Chr3 QTL, at least not with lipid phenotypes in mouse, it is thus, however, rather likely that the human genetic variants control susceptibility

through

other

molecular

mechanisms.

AC C

atherosclerosis

To date, pharmacological inhibition of conventional sPLA2s (not including PLA2G12A) have been clinically tested and failed in clinical phase III trials in cardiovascular patients and even increased myocardial infarction risk52. Therefore, to conclude, atheroprotective roles of some secreted phospholipases like Pla2g12a will have to be considered, and stimulation of Pla2g12a, not inhibition, might be medically relevant, but whether Pla2g12a or also other candidates from the mouse Chr3 interval have to be coordinately induced and repressed remains to be tested.

19

ACCEPTED MANUSCRIPT Conflict of interest The authors declared they do not have anything to disclose regarding conflict of interest with respect to this manuscript.

RI PT

Financial support This work was funded by the University of Leipzig, the University Hospital Munich, by the German Research Foundation (DFG) as part of the Collaborative Research Center CRC1123 “Atherosclerosis - Mechanisms and Networks of Novel Therapeutic Targets”

SC

(project B1- L.M.H., D.T) and by a grant from the Foundation Leducq (CADgenomics

M AN U

network) (L.M.H. and D.T.).

Author contributions

A.N., K.S., D.T. and L.M.H. conceived and designed the experiments. A.N., B.H.N., K.S., J.E. and K.K. performed the experiments.

TE D

A.N., B.H.N., A.K. and K.K. analyzed the data. A.N., A.K., D.T. and L.M.H. wrote the paper.

EP

Acknowledgments

AC C

We thank Wolfgang Wilfert and Anika Stahringer for their technical assistance.

20

ACCEPTED MANUSCRIPT Figure legends

Fig. 1. Fine-mapping of the Chr3 atherosclerosis susceptibility locus and validation in

RI PT

congenic mouse model. (A) LOD score plot of female F2 progeny from cross A (originating from C57BL/6 (B6) grandfathers and FVB grandmothers; see Materials and methods section for crossing

SC

scheme). The confidence interval (CI, green area) spans from SNP rs36523879 to SNP rs31598641 containing 37 Mb. cM, centimorgans. Black lines: microsatellite markers 8; blue

M AN U

lines: SNPs used for fine-mapping. (B) LOD score plot of female F2 progeny from cross B (originating from FVB grandfathers and B6 grandmothers). The CI of cross B (blue) spans from SNP rs4224041 to microsatellite marker d3mit254 containing 44Mb. (C) Schematic of Chr3 congenic mice carrying an 80 Mb interval of Chr3 from B6 mice (black chromosomal segment) on the FVB.Ldlr-/- background (red segments). CIs of cross A and cross B are

TE D

shown below in green and blue, respectively. (D) Representative oil red O stainings of atherosclerotic plaques in the aortic root of a homozygous Chr3 congenic (FVB.Chr3B6/B6) and an FVB.Chr3FVB/FVB mouse. (E) Atherosclerosis in homozygous Chr3 congenics

EP

(FVB.Chr3B6/B6; B6/B6), heterozygous Chr3 congenics (FVB.Chr3B6/FVB; B6/FVB) and FVB.Ldlr-/- (FVB.Chr3FVB/FVB; FVB/FVB) mice. The number of mice used for the analysis is

AC C

indicated in each column. *, p<0.05

Fig. 2. Differentially expressed candidate genes in Chr3 congenic mice and expression QTLs analyses in female F2 mice. (A) List of genes, which are differentially expressed in homozygous Chr3 congenics (FVB.Chr3B6/B6) compared to FVB.Chr3FVB/FVB controls (see also Table 1). Top panel: the genomic location of confidence intervals from cross A and B is depicted as green and blue bar, respectively. Bottom panel: identity by descent (IBD) analysis in the interval of 80-160 21

ACCEPTED MANUSCRIPT Mb of chromosome 3 in C57BL/6 and FVB mice (see Supplementary Table 5 for genomic coordinates of IBD regions shared (black) between the two strains). Identification of cisregulated candidate genes by expression QTL (eQTL) mapping in cross A (C) and B (D).

RI PT

Black lines- microsatellite markers 8; blue lines- SNPs used for fine-mapping.

Fig. 3. Correlation of Pla2g12a expression with Chr3 genotype and atherosclerosis.

SC

(A) Chr3 LOD score plot for atherosclerosis (red) and Pla2g12a mRNA expression (blue) in female F2 progeny of cross A after fine-mapping with SNP rs13464244. (B) Genotypic

M AN U

effects at rs13464244 for atherosclerotic lesion size. (C) Genotypic effect at rs13464244 for Pla2g12a mRNA abundance. (D) Correlation of Pla2g12a mRNA abundance with atherosclerotic lesion size in females of cross A. (E) Pla2g12a mRNA expression in aortic tissue of B6 and FVB F0 mice. mRNA expression was normalized to 106 copies of β-actin

TE D

house-keeping gene. *, p<0.05; **, p<0.01; ***, p<0.001.

Fig. 4. Enhanced Pla2g12a expression in endothelial cells and increased macrophages

EP

adhesion after Pla2g12a siRNA mediated knock-down. (A) Representative immunohistochemical staining in the aortic root of a C57BL/6 mouse of

AC C

PLA2G12A protein, von Willebrand-factor (vWF)-positive vascular endothelial cells, α-SMA smooth muscle actin and CD68-positive macrophages. Negative control (-CTR). Staining was performed on 10-µm consecutive sections of the same mouse. Overview staining (left) and zoom-ins of the boxed area are shown. PLA2G12A, CD68, vWF and α-SMA are stained red (Vector® NovaRED™) and hematoxylin was used to counterstain cell nuclei. (B) Pla2g12a mRNA abundance in C57BL/6 BMDMs, mouse cardiac endothelial MCEC cells (EC) and vascular smooth muscle MOVAS cells (SMC). mRNA expression was normalized to 106 copies of β-actin house-keeping gene. (C) Western blot of PLA2G12A protein in C57BL/6 mouse cardiac endothelial cells MCEC (EC), BMDMs and vascular smooth muscle MOVAS 22

ACCEPTED MANUSCRIPT cells (SMC). GAPDH was used as control. (D) Adhesion of bone marrow-derived macrophages

(BMDM)

from

FVB

(n=4),

homozygous

chromosome

3

congenics

(FVB.Chr3B6/B6 (n=4)) or C57BL/6 (n=4) mice on mouse cardiac endothelial cells that were treated with siRNAs against Pla2g12a or with scrambled (SCR) siRNA control. Statistical

RI PT

differences are depicted relative to SCR control. (E) qPCR analysis of Pla2g12a and (F) Vcam1 mRNA expression in mouse cardiac endothelial cells (MCEC) after transfection with Pla2g12a siRNA. (G) Adhesion of bone marrow-derived macrophages (BMDM) from FVB (n=4), homozygous chromosome 3 congenics (FVB.Chr3B6/B6 (n=4)) or C57BL/6 (n=4) mice

SC

treated with siRNAs against Pla2g12a or with scrambled (SCR) siRNA controls. BMDMs were seeded on to of untreated mouse cardiac endothelial cells. (H) qPCR analysis of

M AN U

Pla2g12a and (I) Vcam1 mRNA expression in BMDMs of FVB (n=4), homozygous chromosome 3 congenics (FVB.Chr3B6/B6 (n=4)) or C57BL/6 (n=4) mice after transfection with

AC C

EP

TE D

Pla2g12a siRNA. *, p<0.05; **, p<0.01; ***, p<0.001.

23

ACCEPTED MANUSCRIPT

Table 1: Identification of candidate genes through transcriptome-wide expression arrays and qPCRs. F0 Aorta:

F0 BMDM:

qRT-PCR

qRT-PCR

qRT-PCR

FC B6/FVB

FC B6/FVB

FC B6/FVB

(n=5/5)

(n=6/6)

(n=6/6)

n.f.

n.f.

n.f.

n.f.

2.22**

not expr.

not expr.

4.4*

n.f.

n.f.

n.f.

n.f.

1.13

n.f.

n.f.

n.f.

n.f.

0.67

n.f.

n.f.

n.f.

n.f.

0.72 ↑↓

n.f.

n.f.

n.f.

n.f.

0.25***

0.23***

0.17***

0.32***

0.7

1.59

n.f.

n.f.

n.f.

n.f.

1.73*

2.14**

1.6**

not expr.

4.4*

1.14

n.f.

n.f.

n.f.

n.f.

2.53*

3.09*

2.2***

1.02

0.95 A; 2.4 B

0.77*

0.53 ***

0.46**

0.71*

3.6* A, 2.4 B

2.1*

3.31***

2.01*

not expr.

not expr.

0.13 A; 0.5 B

3.51***

0.62 ↑↓

n.f.

n.f.

n.f.

n.f.

0.69 **

0.83

n.f.

n.f.

n.f.

n.f.

1.45

1.01

n.f.

n.f.

n.f.

n.f.

Position Gene

Cross

Cross

Congenic: Illumina WG-6

Congenic: qRT-PCR

FC FVB.Chr3B6/B6/FVB.Chr3FVB/FVB

FC FVB.Chr3B6/B6/FVB.Chr3FVB/FVB

[Mb] A

B

(n=4/4)

(n=8/9)

88.8

X

1.86

1.14

Pklr

88.9

X

1.52

2.16**

Pmvk

89.2

X

1.59

1.24

Them4

94.1

X

1.47*

Tuft1

94.1

X

0.65**

Selenbp2

94.4

X

1.64

Tmod4

94.9

X

0.58***

Gm129

95.6

X

1.85*

Vcam1

115.8

X

1.46*

Abcd3

121.4

X

1.41*

Elovl6

129.2

X

X

1.73*

Pla2g12a

129.5

X

X

0.49***

Agxt2l1

130.3

X

X

Adh7

137.8

X

Adh4

138.1

X

Gbp3

142.2

X

AC C

EP

TE D

M AN U

Fdps

SC

CI

RI PT

F0 Liver: CI

LOD Score F2 Liver (n=107/120)

24

ACCEPTED MANUSCRIPT

142.3

X

4.85 ***

1.48 *

1.08

n.f.

n.f.

n.f.

Acadm

153.5

X

0.68 *

0.78

n.f.

n.f.

n.f.

n.f.

Cth

157.5

X

1.83 *

2.76***

2.6***

0.85

3.1***

5.6*

AC C

EP

TE D

M AN U

SC

RI PT

Ccbl2

25

ACCEPTED MANUSCRIPT Differentially expressed candidate genes identified by genome-wide expression arrays in homozygous chromosome 3 congenic FVB.Chr3B6/B6 and FVB.Chr3FVB/FVB mice in the confidence intervals of cross A and B, respectively. As a cut-off value, a fold change (FC) of congenic FVB.Chr3B6/B6 to FVB.Chr3FVB/FVB mice greater than 1.4 and less 0.7 was chosen.

RI PT

All fold changes of the validation stages are given. CI, confidence Interval; n.f., not followed; not expr., not expressed; BMDM , bone marrow-derived macrophages; A Cross A; B Cross B. Fdps, farnesyl pyrophosphate synthase; Pklr, pyruvate kinase, liver and RBC; Pmvk,

SC

phosphomevalonate kinase; Them4, thioesterase superfamily member 4; Tuft1, tuftelin 1; Selenbp2, selenium binding protein 2; Tmod4, tropomodulin-4; Gm129, hypothetical protein

M AN U

LOC229599; Vcam1, vascular cell adhesion molecule 1; Abcd3, ATP-binding cassette subfamily D member 3; Elovl6, fatty acid elongase 6; Pla2g12a, group XIIA secretory phospholipase A2; Agxt2l1, alanine-glyoxylate aminotransferase 2-like 1; Adh7, alcohol dehydrogenase class 4 mu/sigma chain; Adh4, alcohol dehydrogenase 4; Gbp3, guanylatebinding protein 4; Ccbl2, kynurenine-oxoglutarate transaminase 3; Acadm, medium-chain

TE D

specific acyl-CoA dehydrogenase; Cth, cystathionine gamma-lyase. * p<0.05; ** p<0.01; ***

AC C

EP

p<0.001; * LOD > 3.3 (corresponding to a genome-wide significance level of 0.05 53).

26

ACCEPTED MANUSCRIPT References [1]

Ishibashi, S, Brown, MS, Goldstein, JL, et al., Hypercholesterolemia in low density

lipoprotein receptor knockout mice and its reversal by adenovirus-mediated gene delivery, The Journal of clinical investigation, 1993;92:883-893. Zhang, SH, Reddick, RL, Piedrahita, JA, et al., Spontaneous hypercholesterolemia

RI PT

[2]

and arterial lesions in mice lacking apolipoprotein E, Science, 1992;258:468-471. [3]

Plump, AS, Smith, JD, Hayek, T, et al., Severe hypercholesterolemia and

SC

atherosclerosis in apolipoprotein E-deficient mice created by homologous recombination in ES cells, Cell, 1992;71:343-353.

Bennett, BJ, Davis, RC, Civelek, M, et al., Genetic Architecture of Atherosclerosis in

Mice:

A

Systems

Genetics

Analysis

2015;11:e1005711. [5]

M AN U

[4]

of

Common Inbred Strains,

PLoS Genet,

von Scheidt, M, Zhao, Y, Kurt, Z, et al., Applications and Limitations of Mouse Models

for Understanding Human Atherosclerosis, Cell Metab, 2016.

Nikpay, M, Goel, A, Won, HH, et al., A comprehensive 1,000 Genomes-based

TE D

[6]

genome-wide association meta-analysis of coronary artery disease, Nature genetics, 2015;47:1121-1130.

Consortium, CAD, Deloukas, P, Kanoni, S, et al., Large-scale association analysis

EP

[7]

identifies new risk loci for coronary artery disease, Nature genetics, 2013;45:25-33. Teupser, D, Tan, M, Persky, AD, et al., Atherosclerosis quantitative trait loci are sex-

AC C

[8]

and lineage-dependent in an intercross of C57BL/6 and FVB/N low-density lipoprotein receptor-/- mice, Proceedings of the National Academy of Sciences of the United States of America, 2006;103:123-128. [9]

Nicolaou, A, Zhao, Z, Northoff, BH, et al., Adam17 Deficiency Promotes

Atherosclerosis by Enhanced TNFR2 Signaling in Mice, Arteriosclerosis, thrombosis, and vascular biology, 2017;37:247-257.

27

ACCEPTED MANUSCRIPT [10]

Rodriguez, JM, Wolfrum, S, Robblee, M, et al., Altered expression of Raet1e, a major

histocompatibility complex class 1-like molecule, underlies the atherosclerosis modifier locus Ath11 10b, Circ Res, 2013;113:1054-1064. [11]

Holdt, LM, Thiery, J, Breslow, JL, et al., Increased ADAM17 mRNA expression and

RI PT

activity is associated with atherosclerosis resistance in LDL-receptor deficient mice, Arteriosclerosis, thrombosis, and vascular biology, 2008;28:1097-1103. [12]

Teupser, D, Pavlides, S, Tan, M, et al., Major reduction of atherosclerosis in

fractalkine (CX3CL1)-deficient mice is at the brachiocephalic artery, not the aortic root,

SC

Proceedings of the National Academy of Sciences of the United States of America, 2004;101:17795-17800.

Holdt, LM, Hoffmann, S, Sass, K, et al., Alu elements in ANRIL non-coding RNA at

M AN U

[13]

chromosome 9p21 modulate atherogenic cell functions through trans-regulation of gene networks, PLoS genetics, 2013;9:e1003588. [14]

Teupser, D, Persky, AD and Breslow, JL, Induction of atherosclerosis by low-fat,

TE D

semisynthetic diets in LDL receptor-deficient C57BL/6J and FVB/NJ mice: comparison of lesions of the aortic root, brachiocephalic artery, and whole aorta (en face measurement), Arteriosclerosis, thrombosis, and vascular biology, 2003;23:1907-1913. Holdt, LM, Beutner, F, Scholz, M, et al., ANRIL expression is associated with

EP

[15]

atherosclerosis risk at chromosome 9p21, Arteriosclerosis, thrombosis, and vascular biology,

[16]

AC C

2010;30:620-627.

Broman, KW, Wu, H, Sen, S, et al., R/qtl: QTL mapping in experimental crosses,

Bioinformatics, 2003;19:889-890. [17]

R Core Team, RFfSC, Vienna, Austria, R: A Language and Environment for Statistical

Computing, 2016. [18]

Dupuis, J and Siegmund, D, Statistical methods for mapping quantitative trait loci

from a dense set of markers, Genetics, 1999;151:373-386. [19]

Bregman, A, Avraham-Kelbert, M, Barkai, O, et al., Promoter elements regulate

cytoplasmic mRNA decay, Cell, 2011;147:1473-1483. 28

ACCEPTED MANUSCRIPT [20]

Price, AL, Helgason, A, Thorleifsson, G, et al., Single-tissue and cross-tissue

heritability of gene expression via identity-by-descent in related or unrelated individuals, PLoS Genet, 2011;7:e1001317. [21]

Mani, S, Li, H, Untereiner, A, et al., Decreased endogenous production of hydrogen

[22]

RI PT

sulfide accelerates atherosclerosis, Circulation, 2013;127:2523-2534. Cheung, SH, Kwok, WK, To, KF, et al., Anti-atherogenic effect of hydrogen sulfide by

over-expression of cystathionine gamma-lyase (CSE) gene, PloS one, 2014;9:e113038. [23]

Du, HP, Li, J, You, SJ, et al., DNA methylation in cystathionine-gamma-lyase (CSE)

SC

gene promoter induced by ox-LDL in macrophages and in apoE knockout mice, Biochemical and biophysical research communications, 2016;469:776-782.

Cybulsky, MI, Iiyama, K, Li, H, et al., A major role for VCAM-1, but not ICAM-1, in

M AN U

[24]

early atherosclerosis, J Clin Invest, 2001;107:1255-1262. [25]

Cybulsky, MI and Gimbrone, MA, Jr., Endothelial expression of a mononuclear

leukocyte adhesion molecule during atherogenesis, Science, 1991;251:788-791. Nakashima, Y, Raines, EW, Plump, AS, et al., Upregulation of VCAM-1 and ICAM-1

TE D

[26]

at atherosclerosis-prone sites on the endothelium in the ApoE-deficient mouse, Arteriosclerosis, thrombosis, and vascular biology, 1998;18:842-851. Galkina, E and Ley, K, Vascular adhesion molecules in atherosclerosis,

EP

[27]

Arteriosclerosis, thrombosis, and vascular biology, 2007;27:2292-2301. Gimbrone, MA, Jr. and Garcia-Cardena, G, Vascular endothelium, hemodynamics,

AC C

[28]

and the pathobiology of atherosclerosis, Cardiovasc Pathol, 2013;22:9-15. [29]

Mestas, J and Ley, K, Monocyte-endothelial cell interactions in the development of

atherosclerosis, Trends Cardiovasc Med, 2008;18:228-232. [30]

Tabas, I, Garcia-Cardena, G and Owens, GK, Recent insights into the cellular biology

of atherosclerosis, The Journal of cell biology, 2015;209:13-22. [31]

Conway, DE and Schwartz, MA, Flow-dependent cellular mechanotransduction in

atherosclerosis, Journal of cell science, 2013;126:5101-5109.

29

ACCEPTED MANUSCRIPT [32]

Hajra, L, Evans, AI, Chen, M, et al., The NF-kappa B signal transduction pathway in

aortic endothelial cells is primed for activation in regions predisposed to atherosclerotic lesion formation, Proceedings of the National Academy of Sciences of the United States of America, 2000;97:9052-9057. Rakyan, VK, Down, TA, Balding, DJ, et al., Epigenome-wide association studies for

common human diseases, Nat Rev Genet, 2011;12:529-541. [34]

Farh, KK, Marson, A, Zhu, J, et al., Genetic and epigenetic fine mapping of causal

autoimmune disease variants, Nature, 2015;518:337-343.

Ost, A, Lempradl, A, Casas, E, et al., Paternal diet defines offspring chromatin state

SC

[35]

RI PT

[33]

and intergenerational obesity, Cell, 2014;159:1352-1364.

Johannes, F, Porcher, E, Teixeira, FK, et al., Assessing the impact of

M AN U

[36]

transgenerational epigenetic variation on complex traits, PLoS genetics, 2009;5:e1000530. [37]

Elices, MJ, Osborn, L, Takada, Y, et al., VCAM-1 on activated endothelium interacts

with the leukocyte integrin VLA-4 at a site distinct from the VLA-4/fibronectin binding site,

[38]

TE D

Cell, 1990;60:577-584.

Luster, AD, Alon, R and von Andrian, UH, Immune cell migration in inflammation:

present and future therapeutic targets, Nat Immunol, 2005;6:1182-1190. Huo, Y, Hafezi-Moghadam, A and Ley, K, Role of vascular cell adhesion molecule-1

EP

[39]

and fibronectin connecting segment-1 in monocyte rolling and adhesion on early

[40]

AC C

atherosclerotic lesions, Circulation research, 2000;87:153-159. Murakami, M, Masuda, S, Shimbara, S, et al., Cellular arachidonate-releasing

function of novel classes of secretory phospholipase A2s (groups III and XII), J Biol Chem, 2003;278:10657-10667. [41]

Murakami, M, Sato, H, Miki, Y, et al., A new era of secreted phospholipase A(2), J

Lipid Res, 2015;56:1248-1261. [42]

Murase, R, Sato, H, Yamamoto, K, et al., Group X Secreted Phospholipase A2

Releases omega3 Polyunsaturated Fatty Acids, Suppresses Colitis, and Promotes Sperm Fertility, J Biol Chem, 2016;291:6895-6911. 30

ACCEPTED MANUSCRIPT [43]

Sato, H, Taketomi, Y, Ushida, A, et al., The adipocyte-inducible secreted

phospholipases PLA2G5 and PLA2G2E play distinct roles in obesity, Cell Metab, 2014;20:119-132. Taketomi, Y, Ueno, N, Kojima, T, et al., Mast cell maturation is driven via a group III

phospholipase

A2-prostaglandin

D2-DP1

receptor

paracrine

2013;14:554-563. [45]

axis,

Nat

Immunol,

RI PT

[44]

Yamamoto, K, Miki, Y, Sato, M, et al., The role of group IIF-secreted phospholipase

A2 in epidermal homeostasis and hyperplasia, J Exp Med, 2015;212:1901-1919.

Cormier, RT, Hong, KH, Halberg, RB, et al., Secretory phospholipase Pla2g2a

SC

[46]

confers resistance to intestinal tumorigenesis, Nat Genet, 1997;17:88-91.

Saegusa, J, Akakura, N, Wu, CY, et al., Pro-inflammatory secretory phospholipase A2

M AN U

[47]

type IIA binds to integrins alphavbeta3 and alpha4beta1 and induces proliferation of monocytic cells in an integrin-dependent manner, J Biol Chem, 2008;283:26107-26115. [48]

Fujita, M, Zhu, K, Fujita, CK, et al., Proinflammatory secreted phospholipase A2 type

TE D

IIA (sPLA-IIA) induces integrin activation through direct binding to a newly identified binding site (site 2) in integrins alphavbeta3, alpha4beta1, and alpha5beta1, J Biol Chem, 2015;290:259-271.

Gelb, MH, Valentin, E, Ghomashchi, F, et al., Cloning and recombinant expression of

EP

[49]

a structurally novel human secreted phospholipase A2, J Biol Chem, 2000;275:39823-39826. Singer, AG, Ghomashchi, F, Le Calvez, C, et al., Interfacial kinetic and binding

AC C

[50]

properties of the complete set of human and mouse groups I, II, V, X, and XII secreted phospholipases A2, J Biol Chem, 2002;277:48535-48549. [51]

Rosenson, RS and Hurt-Camejo, E, Phospholipase A2 enzymes and the risk of

atherosclerosis, European heart journal, 2012;33:2899-2909. [52]

Nicholls, SJ, Kastelein, JJ, Schwartz, GG, et al., Varespladib and cardiovascular

events in patients with an acute coronary syndrome: the VISTA-16 randomized clinical trial, JAMA, 2014;311:252-262.

31

ACCEPTED MANUSCRIPT [53]

Ott, J, Wang, J and Leal, SM, Genetic linkage analysis in the age of whole-genome

AC C

EP

TE D

M AN U

SC

RI PT

sequencing, Nat Rev Genet, 2015;16:275-284.

32

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

ACCEPTED MANUSCRIPT

QTL mapping revealed a female-specific atherosclerosis risk locus on mouse Chr3. Locus fine-mapping identified Vcam1 and Pla2g12a as parallel risk-determining factors. Pla2g12a is a novel atherosclerosis modifier. Transcriptional downregulation of the secreted phospholipase Pla2g12a confers risk. Pla2g12a is atheroprotective by limiting adhesion of macrophages to the endothelium.

RI PT

    

AC C

EP

TE D

M AN U

SC