ORIGINAL ARTICLE

Correction of endothelial dysfunction after selectivehomocysteine lowering gene therapy reduces arterialthrombogenicity but has no effect on atherogenesis

Frank Jacobs & Eline Van Craeyveld & Ilayaraja Muthuramu & Stephanie C. Gordts &

Jan Emmerechts & Marc Hoylaerts & Paul Herijgers & Bart De Geest

Received: 20 April 2011 /Revised: 26 May 2011 /Accepted: 30 May 2011 /Published online: 18 June 2011# Springer-Verlag 2011

Abstract Hyperhomocysteinemia is an independent riskfactor for ischemic cardiovascular diseases, but its causalrole in atherothrombosis remains controversial. Proathero-genic and/or prothrombotic effects may underlie thepotential causal relation between hyperhomocysteinemiaand cardiovascular events. Here, the effects of selectivelowering of plasma homocysteine, plasma cholesterol, orboth on endothelial function and on atherogenesis in malehyperlipidemic and hyperhomocysteinemic C57BL/6 low-density lipoprotein receptor (LDLr)−/−/cystathionine-β-syn-thase (CBS)+/−-deficient mice were investigated. Second,we evaluated whether selective homocysteine lowering hasanti-thrombotic effects in a model of arterial thrombosis. Ahyperhomocysteinemic and atherogenic diet was started atthe age of 12 weeks. Three weeks later, gene transfer wasperformed with E1E3E4-deleted adenoviral vectors forhepatocyte-restricted overexpression of CBS (AdCBS) orof the LDLr (AdLDLr), or with the control vector Adnull.In a fourth group, AdCBS and AdLDLr were co-administered. Selective homocysteine lowering but notselective cholesterol lowering restored endothelial function

at 6 weeks after gene transfer. Intimal area in the aortic rootand in the brachiocephalic artery at 13 weeks was morethan 100-fold (p<0.001) smaller in AdLDLr and AdCBS/AdLDLr mice than in control mice and AdCBS mice. Nodifferences in intimal area were observed between controlmice and AdCBS mice. In a model of carotid arterythrombosis, the average time to first occlusion and tostable occlusion were 1.9-fold (p<0.01) and 2.1-foldlonger (p<0.01), respectively, in AdCBS-treated mice thanin control mice. Taken together, these data show thatcorrection of endothelial dysfunction following selectivehomocysteine lowering has anti-thrombotic but no anti-atherogenic effects.

Keywords Homocysteine . Gene transfer. Endothelialdysfunction . Atherosclerosis . Lipoproteins . Thrombosis

Introduction

Homocysteine is a thiol-containing amino acid interme-diate that is generated during the conversion of methio-nine to cysteine. In humans, normal plasmahomocysteine concentrations range between 5 and12 μM. Hyperhomocysteinemia induces endothelial dys-function [1], which is related to increased oxidant stress[2]. Endothelial dysfunction is reflected by reducedbioavai labi l i ty of ni t r ic oxide and at tenuatedendothelium-dependent vasodilatation, increased tissuefactor expression [3], decreased thrombomodulin activity[4], and reduced fibrinolysis as a result of posttranslationmodification of annexin A2 [5, 6].

In line with these molecular effects, hyperhomocystei-nemia is considered to be a predisposing factor for ischemicstroke, peripheral arterial disease, coronary heart disease,

Electronic supplementary material The online version of this article(doi:10.1007/s00109-011-0778-7) contains supplementary material,which is available to authorized users.

F. Jacobs : E. Van Craeyveld : I. Muthuramu : S. C. Gordts :J. Emmerechts :M. Hoylaerts :B. De Geest (*)Center for Molecular and Vascular Biology, K.U. Leuven,Campus Gasthuisberg,Herestraat 49,3000 Leuven, Belgiume-mail: bart.degeest@med.kuleuven.be

P. HerijgersDivision of Experimental Cardiac Surgery, K.U. Leuven,Leuven, Belgium

J Mol Med (2011) 89:1051–1058DOI 10.1007/s00109-011-0778-7

and venous thromboembolism [7]. However, whether therelation between hyperhomocysteinemia and vascular dis-ease in the clinical setting reflects cause or confounding iscurrently unresolved. Multiple prospective epidemiologicalstudies have demonstrated an independent associationbetween hyperhomocysteinemia and cardiovascular events[8]. On the basis of these observational studies, severaldouble-blind randomized controlled trials have been per-formed to test whether supplementation with appropriatedoses of folic acid, vitamin B6, and vitamin B12 reducesthe incident risk of cardiovascular diseases [9]. Clinicaltrials and most recent meta-analyses showed that homo-cysteine lowering supplementation therapy yielded consis-tent negative results with regard to prevention of vascularoutcomes. The apparent discrepancy between observationaland intervention studies may be due to residual confound-ing factors such as renal function [10]. On the other hand, itmay also reflect that dietary supplementation therapyproduces adverse vascular effects unrelated to homocys-teine metabolism, which obscure the beneficial effects oflowered homocysteine [11].

Proatherogenic and/or prothrombotic effects may under-lie the potential causal relation between elevated homo-cysteine levels and cardiovascular events. Nygard et al. [12]did not find a relation between homocysteine levels and theextent of coronary atherosclerosis in patients with angio-graphically confirmed coronary artery disease. Moreover,vascular complications in patients with homocystinuria arerelated to thrombosis rather than to atherosclerosis [13, 14].Therefore, major aims of the current study were tocritically reevaluate the homocysteine hypothesis ofatherosclerosis [15] and also the proposed pivotal role ofendothelial dysfunction in atherosclerosis initiation andprogression [16].

Till now, no experimental animal study has beenperformed that investigates the effect of a selectivehomocysteine lowering therapy on atherogenesis and onthe susceptibility to arterial thrombosis. In the current study,a selective homocysteine lowering gene transfer strategywithout any dietary changes during the intervention phasewas evaluated in C57BL/6 low-density lipoprotein receptor(LDLr)−/−/cystathionine-β-synthase (Cbs)+/−-deficient micefed an atherogenic and hyperhomocysteinemic diet. Sincethe same diet is maintained during the intervention phase,dietary effects unrelated to homocysteine lowering areexcluded by the study design. Our hypothesis was thatselective homocysteine lowering would correct endothelialdysfunction and attenuate the susceptibility to arterialthrombosis. In order to disentangle the relation betweenhomocysteine, endothelial function, and atherogenesis, amore complex study design was used that evaluated theeffects of selective lowering of plasma cholesterol, plasmahomocysteine, or both.

Materials and methods

For detailed methodology, please see the SupplementaryMaterials and Methods in the Electronic SupplementaryMaterial.

In vivo experiments All animal procedures were approvedby the ethical committee for animal experimentation of theKatholieke Universiteit Leuven, Leuven, Belgium. MaleC57BL/6 Ldlr−/− Cbs+/− mice were fed a folate-depleted,methionine-enriched diet (TD00205; 0.2 mg/kg folic acid,4.1 g/kg L-methionine; Harlan Teklad, Horst, the Nether-lands) supplemented with 0.2% (w/w) cholesterol and 10%(v/w) coconut oil ad libitum, starting from the age of12 weeks. Three weeks after initiation of the diet, micewere injected with 5×1010 adenoviral particles of AdCBS.Control mice received an equivalent dose of Adnull [17]or saline buffer. To evaluate effects of selective homo-cysteine lowering on endothelial function and atherogen-esis, additional groups were injected with 5×1010

adenoviral particles of AdLDLr [18] or a with combina-tion of both AdCBS and AdLDLr. All experimental dietswere maintained throughout the entire duration of theexperiments.

End-point analyses Total plasma homocysteine was mea-sured with an automated Abbott IMX immunoanalyzer(Abbott Diagnostics, Abbott Park, IL, USA) [19]. Mea-surement of Thiobarbituric Acid Reactive Substances(TBARS) was performed according to the instructions ofthe manufacturer (Cayman Chemical, Ann Arbor, MI,USA). Mouse lipoproteins were separated by densitygradient ultracentrifugation in a swing-out rotor. Coagula-tion assays were performed on the BCS-XP coagulationanalyzer (Siemens, Hamburg, Germany) using modifiedsettings for murine samples [20]. Endothelial function wasdetermined 6 weeks after intervention. Thirteen weeks aftergene transfer, mice were euthanized for histological andimmunohistochemical characterization of aortic lesions[21]. Photochemically induced carotid artery thrombosiswas performed essentially as described [22].

Statistical analysis All data are expressed as means±standard error of the means (SEM). Areas under the curvewere calculated using Prism4 (GraphPad Software, SanDiego, CA, USA). Lipid, homocysteine, and histologicalparameters were compared by ANOVA followed by Tukeymultiple-comparison post-test. Responses to vasodilatorswere analyzed by two-way repeated-measures ANOVAwithBonferroni multiple-comparison analysis at specific con-centrations of vasodilators using SPSS 16.0 (SPSS Inc.,Chicago, IL, USA). When indicated, a logarithmic trans-formation or non-parametric test was performed. A two-

1052 J Mol Med (2011) 89:1051–1058

sided p value of less than 0.05 was considered statisticallysignificant.

Results

AdCBS gene transfer normalizes homocysteine levelsin C57BL/6 LDLr−/− Cbs+/− mice whereas AdLDLr genetransfer drastically lowers non-HDL cholesterol levels

A folate-depleted, methionine-enriched diet supplementedwith cholesterol and coconut oil was initiated in C57BL/6Ldlr−/− Cbs+/− mice at the age of 12 weeks to inducehyperhomocysteinemia and hypercholesterolemia. Threeweeks later, gene transfer was performed with AdCBS tolower homocysteine, or with AdLDLr to decrease plasma

cholesterol levels, or with a combination of both vectors.Compared to the homocysteine plasma level at the day ofgene transfer (88±5.0 μM), AdCBS gene transfer resulted ina 5.6-fold (p<0.0001) decrease of plasma homocysteine atday 14 (16±1.1 μM) after gene transfer (Fig. 1a). This effectwas stable for the entire duration of the experiment and wassimilar in the combined gene transfer group (Fig. 1a).AdLDLr gene transfer did not affect homocysteine levels(Fig. 1a). Plasma cholesterol levels were similar before (680±39 mg/dl; Fig. 1a) and after AdCBS gene transfer (710±28 mg/dl) whereas AdLDLr transfer and the combination ofboth vectors induced 6.2-fold (p<0.0001) lower cholesterollevels in the period of 13 weeks after gene transfer comparedto control mice (Fig. 1b; Table 1). This major reduction ofcholesterol occurred predominantly in very low-densitylipoprotein (VLDL), intermediate-density lipoprotein (IDL),and LDL (Table 1). To investigate the effect of selective

Fig. 1 Time course of plasma homocysteine levels (a), plasmacholesterol levels (b), and body weights (d) in male C57BL/6 Ldlr−/−

Cbs+/− control mice (black circle; n=14), or in male C57BL/6 Ldlr−/−

Cbs+/− mice injected with 5×1010 particles of AdCBS (blue circle; n=35), AdLDLr (green circle; n=8), or the combination of AdCBS andAdLDLr (pink circle; n=9). A hyperhomocysteinemic and atherogenicdiet (0.2 mg/kg folic acid, 4.1 g/kg L-methionine, 1.25% cholesterol (w/

w), and 10% coconut oil (v/w)) was initiated 3 weeks before adenoviralgene transfer and maintained throughout the experiment. The 0-weektime point corresponds to the time point of gene transfer in theintervention groups. c TBARS expressed as plasma malondialdehydeequivalents in mice from control mice at 3 weeks after the start of thediet and in intervention groups at 6 weeks after gene transfer. All datarepresent means±SEM

J Mol Med (2011) 89:1051–1058 1053

cholesterol lowering and selective homocysteine lowering onoxidative stress, TBARS were quantified in plasma (Fig. 1c).Malondialdehyde (MDA) equivalents in C57BL/6 Ldlr−/−

Cbs+/− mice after 3 weeks on diet (Fig. 1c) were 2.1-fold(p<0.001) and 2.6-fold (p<0.01) higher than in C57BL/6Ldlr−/− Cbs+/− mice at baseline (3.0±0.17 μM; n=11) andC57BL/6 mice (2.4±0.019; n=4), respectively. Gene transferwith AdCBS and AdCBS/AdLDLr resulted in a 1.7-fold (p<0.01) and a 2.8-fold (p<0.001) decrease of MDA equiv-alents, respectively, at 6 weeks after gene transfer comparedto MDA equivalents before gene transfer (Fig. 1c). MDAequivalents were not significantly decreased after AdLDLrgene transfer. No significant alteration of TBARS wasobserved 6 weeks after saline injection or Adnull transferin control mice. Since severe hyperhomocysteinemia hasbeen shown to result in a reduced body weight, longitudinalfollow-up of body weight was performed. Figure 1dillustrates that body weights were similar in different groupsthroughout the entire duration of the experiment.

Selective homocysteine lowering but not selectivecholesterol lowering corrects endothelial dysfunction

To perform ex vivo vasomotor experiments, mice wereeuthanized 9 weeks after initiation of the hyperhomocys-teinemic and atherogenic diet. Vasorelaxation of thoracicaorta segments was measured by performing dose–responseexperiments using the endothelium-dependent vasodilatoracetylcholine and the endothelium-independent vasodilatorsodium nitroprusside. Relaxation was expressed as thepercentage of the maximal contracted tension obtained byaddition of 10−6 M phenylephrine. Figure 2a shows theendothelium-dependent relaxation in C57BL/6 Ldlr−/−

Cbs+/− control mice and in the three intervention groups.Endothelium-dependent relaxation was severely impaired inthe case of severe hyperhomocysteinemia compared toconditions of normal homocysteine levels (Fig. 2a). Maximal

relaxation to the highest dose of acetylcholine was signifi-cantly (p<0.001) decreased in control (61±6.5%) andAdLDLr (65±5.8%) injected mice compared to AdCBS(97±1.8%)- and AdCBS/AdLDLr (97±1.9%)-treated mice.No differences were observed in vasodilator responses tosodium nitroprusside between any conditions (Fig. 2b),suggesting that the sensitivity of soluble guanylyl cyclaseto nitric oxide in smooth muscle cells remained normalunder conditions of severe hyperhomocysteinemia. Of note,no differences in the relaxation of aortic segments toacetylcholine or sodium nitroprusside were observed betweenC57BL/6 mice and AdCBS- or AdCBS/AdLDLr-treatedC57BL/6 Ldlr−/− Cbs+/− mice (data not shown).

Correction of endothelial dysfunction after selectivehomocysteine lowering has no effect on atherogenesis

To evaluate whether endothelial function predicts differ-ences in atherosclerosis progression, intimal area wasquantified in the aortic root and the brachiocephalic arteryat 16 weeks after the start of diet, i.e., 13 weeks afterintervention. Intimal area in AdLDLr and AdCBS/AdLDLrmice was more than 100-fold (p<0.001) smaller than incontrol mice and AdCBS mice (Fig. 3a). Neither differ-ences in intimal area nor a trend of difference was observedbetween control mice and AdCBS mice (Fig. 3a), indicatingthat correction of endothelial function after homocysteinelowering has no effect on atherogenesis. Since profoundinhibition of atherogenesis was observed in the AdLDLrgroup in the presence of persistent endothelial dysfunction,the overall results show that endothelial function is not anadequate predictor of atherosclerosis progression. Analysisof the intimal area in the brachiocephalic artery showedentirely similar results (Fig. 3b). Further morphometric andhistological data are provided in Supplementary Table 1.Importantly, no significant differences of macrophage andsmooth muscle cell content of plaques were observed between

Table 1 Average total, non-HDL, VLDL, IDL, LDL, and HDL cholesterol levels (milligrams per deciliter) in plasma of male C57BL/6 Ldlr−/−

Cbs+/− mice for the 13-week post-injection period in the control group and in AdCBS, AdLDLr, and combined gene transfer groups

Control AdCBS AdLDLr Combined

Total cholesterol 710±32 710±28 110±2.3***,### 110±11***,###

Non-HDL cholesterol 620±29 620±26 54±2.3***,### 53±4.7***,###

VLDL cholesterol 99±9.2 120±14 7.7±2.3***,### 8.9±2.0***,###

IDL cholesterol 270±11 260±12 16±2.2***,### 16±1.4***,###

LDL cholesterol 250±14 250±10 31±4.9***,### 28±5.4***,###

HDL cholesterol 87±3.5 84±3.0 59±2.3***,### 61±6.4**,##

Data are expressed as means±SEM (n=10 for control and AdCBS; n=8 for AdLDLr; n=9 for combined). Lipoproteins were isolated byultracentrifugation at the time points shown in Fig. 1a and average values for the time period between 1 and 13 weeks after gene transfer wereobtained by dividing the area under the curve by the time of follow-up

**: p<0.01; ***: p<0.001 versus Control. ##: p<0.01 ; ###: p<0.001 versus AdCBS

1054 J Mol Med (2011) 89:1051–1058

control mice and AdCBS-treated mice (SupplementaryTable 1). Representative histological stainings are illustratedin Supplementary Fig. 1.

Correction of endothelial dysfunction after homocysteinelowering reduces arterial thrombogenicity

Compared to C57BL/6 reference mice, a procoagulant statewas observed in C57BL/6 Ldlr−/− Cbs+/− control mice fedthe atherogenic and hyperhomocysteinemic diet for 3 (data

not shown) and 9 weeks (Supplementary Table 2).However, coagulation parameters were not significantlyaltered after AdCBS gene transfer (Supplementary Table 2).To evaluate whether selective homocysteine loweringaffects arterial thrombosis susceptibility, carotid arterythrombosis was induced by photochemical injury. Theaverage time to first intermittent occlusion and to stableocclusion were 1.9-fold (p<0.01) and 2.1-fold longer (p<0.01), respectively, in AdCBS-treated mice than in controlmice (Fig. 4a). Comparison of stable occlusion events in

Fig. 2 Relaxation of descending thoracic aorta segments in responseto the endothelium-dependent vasodilator acetylcholine (a) and theendothelium-independent vasodilator sodium nitroprusside (b). Vaso-motor experiments were performed 6 weeks after the time point ofgene transfer in male C57BL/6 Ldlr−/− Cbs+/− control mice (blackcircle), or in male C57BL/6 LDLr−/− Cbs+/− mice injected with 5×1010

particles of AdCBS (blue circle), AdLDLr (green circle), or thecombination of AdCBS and AdLDLr (pink circle). Data are shown asmeans±SEM (n=8 to 9 for each condition) and are expressed aspercentages of the maximal contracted tension obtained by addition of10−6 M phenylephrine. *p<0.05; **p<0.01; ***p<0.001 versuscontrol

Fig. 3 Individual value bar graph of the intimal area in the ascendingaorta (a) and the brachiocephalic artery (b) of male C57BL/6 LDLr−/−

Cbs+/− control mice (black circle; n=14), or in male C57BL/6 LDLr−/−

Cbs+/− mice injected with 5×1010 particles of AdCBS (blue circle; n=35), AdLDLr (green circle; n=8), or the combination of AdCBS andAdLDLr (pink circle; n=9). Analysis was performed 13 weeks after

gene transfer with continuation of the hyperhomocysteinemic andatherogenic diet. Quantifications were measured on hematoxylin andeosin stained sections at 70-μm-spaced intervals along the entire aorticroot or brachiocephalic artery. Data points show the individual values.Means are shown by the horizontal lines

J Mol Med (2011) 89:1051–1058 1055

control mice and AdCBS mice by log-rank test yielded ahighly significant difference (p<0.0001; Fig. 4b).

Discussion

The main findings of the present study are that (1) selectivehomocysteine lowering but not selective cholesterol lower-ing corrects endothelial dysfunction in a model of diet-induced hyperhomocysteinemia and hypercholesterolemia,(2) correction of endothelial dysfunction by selectivehomocysteine lowering has no effect on atherogenesiswhereas profound inhibition of atherogenesis is observedafter selective cholesterol lowering in the presence ofpersistent endothelial dysfunction, and (3) selective homo-cysteine lowering reduces arterial thrombogenicity in amodel of photochemically induced carotid artery thrombo-sis. Taken together, the current study does not support thehomocysteine hypothesis of atherosclerosis [15] and chal-lenges the proposed pivotal role of endothelial dysfunctionin atherogenesis [16].

The current study represents the first experimentalstudy in which the effects of selective homocysteinelowering intervention were investigated. The design ofthe current study does not involve a difference in dietbefore and after the intervention and between the controland intervention groups. Therefore, dietary effects unre-lated to homocysteine lowering cannot have an impact onend-points in our study. Furthermore, weights and weightgain were similar in different groups, indicating thatthere were no developmental abnormalities and no toxiceffects associated with hyperhomocysteinemia in contrastto prior studies [23].

Selective homocysteine lowering induced by AdCBSgene transfer in C57BL/6 Ldlr−/− Cbs+/− had no effect onatherogenesis. In apparent contrast, two studies comparingdifferent transgenic lines in the C57BL/6 apo E−/−

background suggested that hyperhomocysteinemia acceler-

ates atherosclerosis progression. In the first study, apo E−/−

Cbs−/− mice and apo E−/− Cbs+/+ mice were compared [23].However, the apo E−/− Cbs−/− model is associated withsignificant morbidity and a mortality of 95% at 15 weeks[23]. In addition, plasma cholesterol levels in apo E−/−

Cbs−/− mice are higher compared to apo E−/− Cbs+/+ mice,indicating that the comparison of these two transgenic linesis not suited to study the effect of homocysteine onatherogenesis. In the second study, Tg-hCbs apo E−/−

Cbs−/− mice, which contain an inducible human Cbstransgene to prevent neonatal lethality, were compared withTg-hCbs apo E−/− Cbs+/+ and Tg-hCbs apo E−/− Cbs+/−

sibling controls [24]. Increased atherosclerosis in Tg-hCbsapo E−/− Cbs−/− mice should be interpreted in light of verysevere degree of hyperhomocysteinemia (levels of 180 μM)and the use of an aggressive hyperlipidemic diet (contain-ing 21.2% fat). More significantly, reduced body weightgain and a tendency for reduced spleen weight wasobserved in Tg-hCbs apo E−/− Cbs−/− mice compared tosibling controls. Reduced body weights and lower spleenweights may be indicative of global alterations in thedevelopmental program in these mice that may affectatherosclerosis progression.

Endothelial dysfunction is considered to be pivotal in theinitiation and progression of atherosclerosis [16]. Endothe-lial function not only involves continuous regulation ofvascular tone by the endothelium [25] but also affectsleukocyte adhesion, hemostasis (exerting control on plate-lets, on the coagulation system, and on the fibrinolyticsystem), endothelial permeability, and medial smoothmuscle cell growth. In the current study, correction ofendothelial dysfunction by selective homocysteine loweringhas no effect on atherogenesis whereas profound inhibitionof atherogenesis was observed after selective cholesterollowering in the presence of persistent endothelial dysfunc-tion. Therefore, endothelial dysfunction appears neither tobe necessary nor sufficient for progression of atherosclero-sis in the current model. Although we cannot entirely

Fig. 4 Carotid artery thrombo-sis after photochemical injury inmale C57BL/6 LDLr−/− Cbs+/−

mice. a Time to first occlusion(open bars) or stable occlusion(closed bars) in control (n=8)and AdCBS-treated (n=9) maleC57BL/6 LDLr−/− Cbs+/− mice.Values are presented as means±SEM. b Percentage of mice witha patent carotid artery as afunction of time after adminis-tration of Rose Bengal in control(black circle; n=8) and AdCBS-treated (blue circle; n=9) maleC57BL/6 LDLr−/− Cbs+/− mice

1056 J Mol Med (2011) 89:1051–1058

exclude that acetylcholine-induced vasodilatation is notrepresentative of the global endothelial function, the resultsof the current study challenge the dogma that endothelialdysfunction obligatory precedes atherosclerosis develop-ment. The absence of an effect of hypercholesterolemia onacetylcholine-induced vascular relaxation is in agreementwith a prior study that showed that endothelial functionbefore plaque development is not impaired in hypercholes-terolemic C57BL/6 apo E−/− mice compared to C57BL/6mice [26]. In humans, flow-mediated dilatation, but not riskfactors or Framingham Risk Score, was associated withaverage annual progression of carotid intima–media thick-ness in late middle-aged individuals at low to intermediatecardiovascular risk, suggesting a potential role of endothe-lial dysfunction in progression of atherosclerosis in laterstages.

Hyperhomocysteinemia but not hypercholesterolemiainduced increased TBARS in plasma in the currentmouse model. Therefore, this parameter of oxidativestress was not an adequate predictor of the initiation andprogression of atherosclerosis. It should be noted thatlevels of TBARS in plasma are an indirect marker ofoxidative stress. Although we did not determine directproduction of reactive oxygen species in this study,previous reports in hyperhomocysteinemic mouse modelshave shown that increased oxidative stress is predomi-nantly mediated by superoxide production [27, 28].According to the response to retention hypothesis, sub-endothelial retention of atherogenic lipoproteins is thecentral pathogenic process in atherogenesis [29, 30]. Ourresults are in agreement with the view that other processesthan retention, such as endothelial dysfunction or lipopro-tein oxidation, have a contributory role or are secondaryphenomena but are neither sufficient nor required. Fur-thermore, the plasma TBARS data in AdCBS mice are notin contradiction with lipoprotein oxidation occurring in thearterial wall after retention of apo B containing particleswithin this sequestered microenvironment and do notexclude a role of minimally oxidized LDL [31]. TheTBARS data are concordant with data on endothelialfunction. Redox effects of hyperhomocysteinemia areparticularly important in mediating the adverse impact ofhyperhomocysteinemia on the endothelium, inducing lossof endothelium-derived nitric oxide and endothelial dys-function. Superoxide produced by NADPH oxidase,uncoupled nitric oxide synthase, and as a result ofdepletion of intracellular glutathione can react with nitricoxide, leading to reduced nitric oxide bioavailability.

Our study shows that selective homocysteine lower-ing decreases the susceptibility to arterial thrombosis.Prior experimental observations have demonstrated thathyperhomocysteinemia increases the susceptibility toarterial thrombosis [28]. We did not find any significant

difference in coagulation parameters between hyperho-mocysteinemic mice and AdCBS-treated mice. Further-more, Wilson et al. [22] previously reported no effect ofhomocysteine levels in apo E−/− mice on ex vivo plateletactivation responses. Therefore, correction of endothelialdysfunction after homocysteine lowering may underliethe reduced arterial thrombogenicity after AdCBS genetransfer. Oxidative inactivation of thrombomodulin [32]resulting in decreased thrombomodulin activity [4] andincreased tissue factor expression [33] may promotethrombosis under conditions of hyperhomocysteinemia.In addition, hyperhomocysteinemia may impair fibrino-lysis. Reduced fibrinolysis may occur as a result ofposttranslation modification of annexin A2 [5, 6],secondary to posttranslational modification of fibrinogen[34], and increased activity of thrombin-activatablefibrinolysis inhibitor [35].

The current study may shed a new light on theapparent discrepancy between observational [8] andhomocysteine lowering clinical intervention studies [9].Our results suggest that the arterial thrombogenicityassociated with moderate or severe hyperhomocysteinemiamay be reduced as a result of a selective homocysteinelowering intervention. In contrast, current clinical inter-ventions based on dietary supplementation therapy aregenerally performed in patients with mild hyperhomocys-teinemia and result in relatively small and non-selectivereductions of homocysteine levels. In addition, whereasthe results of the current study are in line with thehypothesis that selective homocysteine lowering mightreduce clinical events, an effect on the extent of athero-sclerosis is not anticipated.

In conclusion, the current study comparing selectivecholesterol lowering and homocysteine lowering genetransfer strategies strongly suggests that a major role ofhyperhomocysteinemia in atherogenesis is unlikely. Incontrast, selective homocysteine lowering reduces arterialthrombogenicity. One of the strengths of the current study isthat the diet was exactly the same in different groups.Therefore, our study corroborates the view that the lack ofsuccess of homocysteine supplementation therapies in aclinical setting may be due to effects unrelated tohomocysteine lowering.

Acknowledgments This work was supported by the Fonds voorWetenschappelijk Onderzoek-Vlaanderen (grant G.0533.08). The Centerfor Molecular and Vascular Biology is supported by the Excellentiefinan-ciering KU Leuven (EF/05/013). Frank Jacobs is a postdoctoral fellow ofthe Fonds voor Wetenschappelijk Onderzoek-Vlaanderen. Eline VanCraeyveld is a Research Assistant of the Fonds voor WetenschappelijkOnderzoek-Vlaanderen. Stephanie C. Gordts is a Research Assistant of theAgentschap voor Innovatie door Wetenschap en Technologie.

Disclosures None. There is no conflict of interest.

J Mol Med (2011) 89:1051–1058 1057

References

1. Eberhardt RT, Forgione MA, Cap A, Leopold JA, Rudd MA,Trolliet M, Heydrick S, Stark R, Klings ES, Moldovan NI et al(2000) Endothelial dysfunction in a murine model of mildhyperhomocyst(e)inemia. J Clin Invest 106:483–491

2. Dayal S, Lentz SR (2008) Murine models of hyperhomocysteine-mia and their vascular phenotypes. Arterioscler Thromb Vasc Biol28:1596–1605

3. Undas A, Brozek J, Szczeklik A (2005) Homocysteine andthrombosis: from basic science to clinical evidence. ThrombHaemost 94:907–915

4. Dayal S, Bottiglieri T, Arning E, Maeda N, Malinow MR,Sigmund CD, Heistad DD, Faraci FM, Lentz SR (2001)E n d o t h e l i a l d y s f u n c t i o n a n d e l e v a t i o n o f S -adenosylhomocysteine in cystathionine beta-synthase-deficientmice. Circ Res 88:1203–1209

5. Hajjar KA (1993) Homocysteine-induced modulation of tissueplasminogen activator binding to its endothelial cell membranereceptor. J Clin Invest 91:2873–2879

6. Jacovina AT, Deora AB, Ling Q, Broekman MJ, Almeida D,Greenberg CB, Marcus AJ, Smith JD, Hajjar KA (2009)Homocysteine inhibits neoangiogenesis in mice through blockadeof annexin A2-dependent fibrinolysis. J Clin Invest 119:3384–3394

7. Homocysteine Studies Collaboration (2002) Homocysteine andrisk of ischemic heart disease and stroke: a meta-analysis. Jama288:2015–2022

8. Wald DS, Law M, Morris JK (2002) Homocysteine andcardiovascular disease: evidence on causality from a meta-analysis. BMJ (Clinical research ed) 325:1202

9. Marti-Carvajal AJ, Sola I, Lathyris D, Salanti G (2009)Homocysteine lowering interventions for preventing cardiovascu-lar events. Cochrane Database Syst Rev: CD006612

10. Potter K, Hankey GJ, Green DJ, Eikelboom JW, Arnolda LF(2008) Homocysteine or renal impairment: which is the realcardiovascular risk factor? Arterioscler Thromb Vasc Biol28:1158–1164

11. Bonaa KH, Njolstad I, Ueland PM, Schirmer H, Tverdal A,Steigen T, Wang H, Nordrehaug JE, Arnesen E, Rasmussen K(2006) Homocysteine lowering and cardiovascular events afteracute myocardial infarction. N Engl J Med 354:1578–1588

12. Nygard O, Nordrehaug JE, Refsum H, Ueland PM, Farstad M,Vollset SE (1997) Plasma homocysteine levels and mortality inpatients with coronary artery disease. N Engl J Med 337:230–236

13. Mudd SH, Skovby F, Levy HL, Pettigrew KD, Wilcken B, PyeritzRE, Andria G, Boers GH, Bromberg IL, Cerone R (1985) Thenatural history of homocystinuria due to cystathionine beta-synthase deficiency. Am J Hum Genet 37:1–31

14. Schimke RN, McKusick VA, Huang T, Pollack AD (1965)Homocystinuria. Studies of 20 families with 38 affected members.Jama 193:711–719

15. McCully KS (1969) Vascular pathology of homocysteinemia:implications for the pathogenesis of arteriosclerosis. Am J Pathol56:111–128

16. Ross R (1999) Atherosclerosis—an inflammatory disease. N EnglJ Med 340:115–126

17. Van Linthout S, Lusky M, Collen D, De Geest B (2002) Persistenthepatic expression of human apo A-I after transfer with a helper-virus independent adenoviral vector. Gene Ther 9:1520–1528

18. Van Craeyveld E, Gordts SC, Nefyodova E, Jacobs F, De Geest B(2011) Regression and stabilization of advanced murine athero-sclerotic lesions: a comparison of LDL lowering and HDL raisinggene transfer strategies. J Mol Med 89:555–567

19. Shipchandler MT, Moore EG (1995) Rapid, fully automatedmeasurement of plasma homocyst(e)ine with the Abbott IMxanalyzer. Clin Chem 41:991–994

20. Peters LL, Cheever EM, Ellis HR, Magnani PA, Svenson KL, VonSmith R, Bogue MA (2002) Large-scale, high-throughput screen-ing for coagulation and hematologic phenotypes in mice. PhysiolGenomics 11:185–193

21. Feng Y, Lievens J, Jacobs F, Hoekstra M, Van Craeyveld E,Gordts SC, Snoeys J, De Geest B (2010) Hepatocyte-specificABCA1 transfer increases HDL cholesterol but impairs HDLfunction and accelerates atherosclerosis. Cardiovasc Res88:376–385

22. Wilson KM, McCaw RB, Leo L, Arning E, Lhotak S, BottiglieriT, Austin RC, Lentz SR (2007) Prothrombotic effects of hyper-homocysteinemia and hypercholesterolemia in ApoE-deficientmice. Arterioscler Thromb Vasc Biol 27:233–240

23. Wang H, Jiang X, Yang F, Gaubatz JW, Ma L, Magera MJ, Yang X,Berger PB, Durante W, Pownall HJ et al (2003) Hyperhomocys-teinemia accelerates atherosclerosis in cystathionine beta-synthaseand apolipoprotein E double knock-out mice with and withoutdietary perturbation. Blood 101:3901–3907

24. Zhang D, Jiang X, Fang P, Yan Y, Song J, Gupta S, Schafer AI,Durante W, Kruger WD, Yang X et al (2009) Hyperhomocystei-nemia promotes inflammatory monocyte generation and acceler-ates atherosclerosis in transgenic cystathionine beta-synthase-deficient mice. Circulation 120:1893–1902

25. Furchgott RF, Zawadzki JV (1980) The obligatory role ofendothelial cells in the relaxation of arterial smooth muscle byacetylcholine. Nature 288:373–376

26. Fransen P, Van Assche T, Guns PJ, Van Hove CE, De KeulenaerGW, Herman AG, Bult H (2008) Endothelial function in aortasegments of apolipoprotein E-deficient mice before developmentof atherosclerotic lesions. Pflugers Arch 455:811–818

27. Eberhardt RT, Forgione MA, Cap A, Leopold JA, Rudd MA,Trolliet M, Heydrick S, Stark R, Klings ES, Moldovan NI et al(2000) Endothelial dysfunction in a murine model of mildhyperhomocyst(e)inemia. J Clin Investig 106:483–491

28. Dayal S, Wilson KM, Leo L, Arning E, Bottiglieri T, Lentz SR(2006) Enhanced susceptibility to arterial thrombosis in a murinemodel of hyperhomocysteinemia. Blood 108:2237–2243

29. Williams KJ, Tabas I (1995) The response-to-retention hypothesisof early atherogenesis. Arterioscler Thromb Vasc Biol 15:551–561

30. Tabas I, Williams KJ, Boren J (2007) Subendothelial lipoproteinretention as the initiating process in atherosclerosis: update andtherapeutic implications. Circulation 116:1832–1844

31. Steinbrecher UP, Lougheed M (1992) Scavenger receptor-independent stimulation of cholesterol esterification in macro-phages by low density lipoprotein extracted from human aorticintima. Arterioscler Thromb 12:608–625

32. Glaser CB, Morser J, Clarke JH, Blasko E, McLean K, Kuhn I,Chang RJ, Lin JH, Vilander L, Andrews WH (1992) Oxidationof a specific methionine in thrombomodulin by activatedneutrophil products blocks cofactor activity. A potential rapidmechanism for modulation of coagulation. J Clin Investig90:2565–2573

33. Loscalzo J (2009) Homocysteine-mediated thrombosis and angio-stasis in vascular pathobiology. J Clin Invest 119:3203–3205

34. Sauls DL, Lockhart E, Warren ME, Lenkowski A, Wilhelm SE,Hoffman M (2006) Modification of fibrinogen by homocysteinethiolactone increases resistance to fibrinolysis: a potential mech-anism of the thrombotic tendency in hyperhomocysteinemia.Biochemistry 45:2480–2487

35. Colucci M, Cattaneo M, Martinelli I, Semeraro F, Binetti BM,Semeraro N (2008) Mild hyperhomocysteinemia is associatedwith increased TAFI levels and reduced plasma fibrinolyticpotential. J Thromb Haemost 6:1571–1577

1058 J Mol Med (2011) 89:1051–1058