11β-Hydroxysteroid Dehydrogenase Type 1 Gene Knockout Attenuates Atherosclerosis and In Vivo Foam Cell Formation in Hyperlipidemic apoE−/− Mice

Background Chronic glucocorticoid excess has been linked to increased atherosclerosis and general cardiovascular risk in humans. The enzyme 11β-hydroxysteroid dehydrogenase type 1 (11βHSD1) increases active glucocorticoid levels within tissues by catalyzing the conversion of cortisone to cortisol. Pharmacological inhibition of 11βHSD1 has been shown to reduce atherosclerosis in murine models. However, the cellular and molecular details for this effect have not been elucidated. Methodology/Principal Findings To examine the role of 11βHSD1 in atherogenesis, 11βHSD1 knockout mice were created on the pro-atherogenic apoE−/− background. Following 14 weeks of Western diet, aortic cholesterol levels were reduced 50% in 11βHSD1−/−/apoE−/− mice vs. 11βHSD1+/+/apoE−/− mice without changes in plasma cholesterol. Aortic 7-ketocholesterol content was reduced 40% in 11βHSD1−/−/apoE−/− mice vs. control. In the aortic root, plaque size, necrotic core area and macrophage content were reduced ∼30% in 11βHSD1−/−/apoE−/− mice. Bone marrow transplantation from 11βHSD1−/−/apoE−/− mice into apoE−/− recipients reduced plaque area 39–46% in the thoracic aorta. In vivo foam cell formation was evaluated in thioglycollate-elicited peritoneal macrophages from 11βHSD1+/+/apoE−/− and 11βHSD1−/−/apoE−/− mice fed a Western diet for ∼5 weeks. Foam cell cholesterol levels were reduced 48% in 11βHSD1−/−/apoE−/− mice vs. control. Microarray profiling of peritoneal macrophages revealed differential expression of genes involved in inflammation, stress response and energy metabolism. Several toll-like receptors (TLRs) were downregulated in 11βHSD1−/−/apoE−/− mice including TLR 1, 3 and 4. Cytokine release from 11βHSD1−/−/apoE−/−-derived peritoneal foam cells was attenuated following challenge with oxidized LDL. Conclusions These findings suggest that 11βHSD1 inhibition may have the potential to limit plaque development at the vessel wall and regulate foam cell formation independent of changes in plasma lipids. The diminished cytokine response to oxidized LDL stimulation is consistent with the reduction in TLR expression and suggests involvement of 11βHSD1 in modulating binding of pro-atherogenic TLR ligands.


control.In th
aortic root, plaque size, necrotic core area and macrophage content were reduced ,30% in 11bHSD1 2/2 /apoE 2/2 mice.Bone marrow transplantation from 11bHSD1 2/2 /apoE 2/2 mice into apoE 2/2 recipients reduced plaque area 39-46% in the thoracic aorta.In vivo foam cell formation was evaluated in thioglycollate-elicited peritoneal macrophages from 11bHSD1 +/+ /apoE 2/2 and 11bHSD1 2/2 / apoE 2/2 mice fed a Western diet for ,5 weeks.Foam cell cholesterol levels were reduced 48% in 11bHSD1 2/2 /apoE 2/2 mice vs. control.Microarray profiling of peritoneal macrophages revealed differential expression of genes involved in inflammation, stress response and energy metabolism.Several toll-like receptors (TLRs) were downregulated in 11bHSD1 2/2 /apoE 2/2 mice including TLR 1, 3 and 4. Cytokine release from 11bHSD1 2/2 /apoE 2/2 -derived peritoneal foam cells was attenuated following challenge with oxidized LDL.Conclusions: These findings suggest that 11bHSD1 inhibition may have the potential to limit plaque development at th vessel wall and regulate foam cell formation independent of changes in plasma lipids.The diminished cytokine response to oxidized LDL stimulation is consistent with the reduction in TLR expression and suggests involvement of 11bHSD1 in modulating binding of pro-atherogenic TLR ligands.

Introduction

Glucocorticoids are ubiquitous mammalian hormones involved in the regulation of several fundamental biological processes including energy metabolism, inflammation, arousal, cognition and the response to physiological stress.In humans, the primary active glucocorticoid hormone, cortisol, binds to intracellular glucocorticoid and mineralocorticoid receptors found in target tissues.Dysfunctional regulation of glucocorticoid metabolism resulting in excess cortisol in tissues such as adipose, liver and the vasculature has been implicated as a key mediator in the pathogenesis of obesity, type 2 diabetes and cardiovascular disease [1][2][3].Growing epidemiological evidence suggests that glucocorticoid excess may also contribute to the progression of atherosclerosis [3].Although not completely characterized, the evidence suggests that direct interactions of glucocorticoids with cells in the vasculature may potentiate plaque development [4] [3] [5] [6] independent of changes in plasma r sk factors such as cholesterol.

Endogenous cortisol levels (corticosterone in mice) are regulated by two distinct pathways.The hypothalamic-pituitary-adrenal axis is the pathway classically associated with regulation of plasma cortisol.In addition, cortisol tone is also regulated intracellularly by the enzymatic activities of two isoforms of 11b-hydroxysteroid dehydrogenase, type 1 (''11bHSD1'') and type 2 (''11bHSD2'').The latter, 11bHSD2, is mainly expressed in aldosterone-target tissues such as kidney, colon and salivary glands [7].11bHSD2 converts active cortisol/corticosterone to the inactive form cortisone/11-dehydrocorticosterone, thereby limiting ligand availability for mineralocorticoid receptor binding in these tissues.By contrast, 11bHSD1 is expressed in tissues with high sensitivity to glucocorticoids such as liver, adipose, brain and lung [7].11bHSD1 converts inactive substrate to the active hormone thereby stimulating glucocorticoid and mineralocorticoid receptor activation.In addition, 11bHSD1 has also been shown to possess oxysterol metabolizing properties via its keto-reductase activity [8][9].Although the pro-atherogenic properties of oxysterols such as 7-ketocholesterol hav been described, [10][11] the overall impact of this enzymatic activity on disease development is unclear.

Glucocorticoid amplification by increased 11bHSD1 activity in highly metabolic tissues such as adipose and liver is proposed to potentiate a phenotype resembling the metabolic syndrome [2] [7].This concept has been rigorously tested in preclinical models of disease.In the Zucker fatty rat, increased 11bHSD1 activity was shown to positively correlate with plasma corticosterone levels and omental fat mass [12].In the mouse, selective overexpression of 11bHSD1 in adipose tissue yields a distinctive phenotype of visceral obesity with many features f the metabolic syndrome [13].Conversely, 11bHSD1 knockout mice do not develop characteristics of the metabolic syndrome when placed on a high-fat diet.Fat-fed 11bHSD1 knockout mice have reduced visceral fat accumulation, increased insulin sensitization, enhanced glucose tolerance and improvements in lipid and lipoprotein profiles [14][15][16].11bHSD1 inhibitor studies in the mouse have shown that selective pharmacological inhibition can positively impact metabolic endpoints in the setting of diet-induced obesity [17] [7].These observations have stimulated significant efforts towards development of

to treat type 2
iabetes mellitus.

The beneficial effects of 11bHSD1 inhibition on atherosclerosis have been demonstrated in the mouse with small-molecule 11bHSD1 inhibitors [17] [18].In these studies, plaque reductions were observed in conjunction with improvements in metabolic endpoints and overall lowering of pro-atherogenic lipids.Thus, it remains unclear whether atheroprotection via 11bHSD1 inhibi-tion is attributed to improvements in glucose/insulin handling, lipid lowering, a direct effect of 11bHSD1 inhibition at the vessel wall or some combination of the above.In addition, the molecular pathways that mediate the atheroprotective effect of 11bHSD1 have not been elucidated.

To clarify these issues, 11bHSD1-deficient mice were created on the apoE knockout background to investigate the propensity of atherosclerosis development.The impact of 11bHSD1 deficiency on plasma lipid profiles, vessel wall atherosclerosis and macrophage function was evaluated in Western diet-fed mice.Intrinsic differences in gene expression profiles and biological pathways affected by 11bHSD1 gene deficiency were evaluated by microarray-based gene profiling of thioglycollate-elicited macrophages from hyperlipidemic mice.The results of these analyses and their implications are described herein.


Methods


Animals and Diet

Animal studies were performed according to guidelines es

blished by the American Asso
iation for Accreditation of Laboratory Animal Care and protocols were approved by the Bristol-Myers Squibb-Hopewell Animal Care and Use Committee.11bHSD1 knockout mice were created at Lexicon Pharmaceuticals (The Woodlands, TX) as described in Figure S1.11bHSD1 knockout mice were backcrossed onto a C57BL/6 background for more than 10 generations and mated with apoE knockout mice on the C57BL/6 background obtained from The Jackson Laboratory (Bar Harbor, ME) to create litters on an apoE-deficient background.Animal breeding was carried out at Bristol-Myers Squibb under pathogen-free barrier conditions.Genotypes of litters were characterized by isolated genomic DNA analysis and grouped according to sex and genotype.All mice were fed standard rodent chow during the acclimation phase.At ,8 weeks of age, mice were fed Western Diet D12079B (Research Diets, New Brunswick, NJ) containing 0.2% cholesterol and 20% milk fat for 5 weeks (peritonitis studies) or 14 weeks (diet-induced atherosclerosis studies).Mice were euthanized under isoflurane anesthesia followed by inhalation of CO 2 for collection of blood, hearts, aortas and archival tissues.All in vivo studies were repeated at least in duplicate to assess reproducibility and to increase statistical power.


Bo e Marrow Transplantation

Male apoE 2/2 mice (8-10 weeks of age; B6.129P2-Apoe tm1Unc /J) purchased from The Jackson Laboratory were subjected to whole body irradiation to induce bone marrow aplasia using a Mark I Model 30 irradia

r (JL Shepherd & Associates,
San Francisco, CA).Irradiation was carried out with a single exposure of 1000 rads using a cesium radiation source.Donor bone marrow cells were harvested from age-matched male 11b 2/2 /apoE 2/2 and 11b +/+ /apoE 2/2 littermate control mice by flushing isolated femurs and tibias with RPMI supplemented with 20 mM HEPES, 10 U/ml heparin and 100 U/ml penicillin/streptomycin. Approximately four hours after irradiation, mice were anesthetized with isoflurane (2-4%) and immediately reconstituted with 5-7 million bone marrow cells given intravenously (,200 ml) via the retro-orbital sinus.Following bone marrow transplantation, mice were housed in autoclaved cages supplied with HEP

filtered air and fed re
ular chow for 4 weeks to enable bone marrow reconstitution.A small group of irradiated control mice that were injected with cell-free vehicle did not survive beyond two weeks post irradiation.Bone marrow transplant-recipient mice were switched to Western diet (Research Diets, D12079B) for 12 weeks to stimulate atherogenesis.


Confirmation of bon

marrow reconstitution was
arreid out by

RT-PCR analysis of whole blood leukocytes, as described (vide infra).Atherosclerosis development was analyzed by en face evaluation of excised aortas using oil red O.


Blood Pressure Measurements

Blood pressures (systolic, diastolic and mean arteri

laque composition, histolo
ical analysis of lesions in the aortic root was performed via the Paigen method [19].Serial sections were stained with trichrome for lesion area measurements.Plaque necrosis area was analyzed from trichromestained aortic root images based on acellularity within the intima and/or presence of crystalline clefts [20].Macrophage content was evaluated by CD68-positive staining via immunohistochemistry.A rat anti-mouse CD68 antibody (Cat.#: MCA1957; AbD Serotec, Raleigh, NC) was applied to paraffin-embedded and sectioned samples.The antigen:antibody binding complex was detected using an anti-rat polymer system directly conjugated to horse radish peroxidase (BioCare Medical, Concord, CA).Sections were counterstained with Hematoxylin, Gills Formula (Vector Labs, Burlingame, CA), dehydrated and cleared with xylene.The antibody required heat induced epitope retrieval for one minute using a citrate-based retrieval solution at pH 6.0 (Biocar

Medical, Concord, CA) prio
to the start of the immunolabeling procedure.Isotyped-matched controls were simultaneously run for each sample.


Aortic Oxysterol Analysis

Samples were reconstituted in 150 mL of methanol and briefly vortexed.A volume of 130 mL was transferred to 96 well plates for LCMS analysis.A Waters Acquity UPLC System (Milford, MA) in line with a Thermo Fisher LTQ Orbitrap mass spectrometer (Waltham, MA) was used to separate and detect components with mass accuracy ,5 ppm.The mass spectrometer was operated in full scan, positive electrospray (+ESI) mode.Reverse-phase gradient LC conditions were employed using a Water Acquity BEH C18 2.16100 mm 1.7 mm column and acetonitrile-water mobile phases.Component peak integration was performed using Thermo-Fisher Xcalibur Quan Browser software.An accurate mass window of 10 ppm was used to process the following components: cholesterol = 369.3504(ms source water loss), 7bhydroxycholesterol = 385.3494(ms source water loss) 7-ketocholesterol = 401.3405;D7-7b-hydroxycholesterol (internal standard) = 392.3898(ms source water loss).


In Vivo Foam Cell Studies

In vivo foam cell studies were performed with mice starting at 8 weeks of age and fed a Western diet for 4 weeks (Western diet, 12079B, 20% fat 0.2% cholesterol, Research Diets).Following the dieta

lead-in phase, m
ce were injected with 1.5 ml of a 4% thioglycollate solution to induce peritonitis.Four days postinjection, mice were euthanized and the abdominal cavity was lavaged with 10 ml of ice-cold phosphate-buffered saline to collect peritoneal macrophages.Approximately 100 ml of each cell suspension was analyzed for total cell count via automated cell counting and cell type using modified Wright-Giemsa staining and light microscope evaluation.For analysis of lipid droplet accumulation, cells were stained using oil red O/hematoxylin [21].Remaining cell suspensions were transferred to pre-weighed glass vials and centrifuged at 1200 rpm for 4 minutes.Supernatant devoid of cells was removed and each cell pellet was mixed with ethyl acetate:acetone (2:1 vol/vol) supplemented with 0.01% 2,6,di-tert-butyl-4-methylphenol overnight at 37u.Samples were blown dry under nitrogen and resuspended in 100 ml of 10% Tr

on X-100 with agitation a
37uC for 90 minutes.Samples were analyzed for total cholesterol content by enzymatic analysis using a Roche Cobas Mira Chemistry Analyzer.


Cytokine Studies

Thioglycollate-elicited peritoneal macrophage/foam cells were prepared as described above.Macrophages were plated in DMEM (1 million cells/ml) supplemented with 20% serum and 200 nM 11-dehydrocorticosterone at 37uC for 5 hours.Adherent cells

re washed twice
ith phosphate-buffered saline and incubated with serum-free DMEM containing 200 nM 11-dehydrocorticosterone.Macrophages were incubated overnight in the presence or absence of ,30 mg/ml copper-oxidized LDL (#BT-910, Biomedical Technologies, Stoughton, MA).Cell media samples were analyzed for cytokine levels via multiplexing immunoassay (#MPXMCYTO-70 K, Millipore, Bedford, MA) using a Luminex 200 System (Austin, TX).Cells were lysed with CelLytic M reagent (#C2978, Sigma, Creamridge, NJ) and total protein content per well was determined by Bradford assay.Cytokine concentrations were normalized to the cellular protein content per well.


Macrophage RNA Isolation

Peritoneal macrophages were collected by lavage as described in the preceding section.Total RNA was isolated from the resulting cell pellet using 1 ml of Tri-Reagent (#AM9738, Ambion, Inc., Austin, TX) according to the manufacturer's instructions (Manual 9738M RevC).The integrity of RNA samples was assessed under non-denaturing conditions on 2% E-gels (#G5018-02, Invitrogen Inc., Carlsbad, CA).


RT-PCR Analysis

For whole blood expression analysis, cDNA was synthesized using qScript cDNA SuperMix (#95048, Quanta Biosciences Inc., Gaither

urg, MD).PCR reactions were performed in an ABI7900HT PCR system us
ng Applied Biosystems Taqman Gene Expression Assay Mm00476182_m1 (for 11bHSD1) and Mm01611464_g1 (for ribosomal protein L30, the normalization control) in Taqman Universal PCR Mix (#4304337, Applied Biosystems Inc.).For tissue and peritoneal macrophage expression analyses, cDNA was synthesized with High Capacity cDNA Reverse Transcription Kit (#4368813, Applied Biosystems Inc.) PCR reactions were performed via TaqMan assay using ABI7900HT PCR system.PPIA was used as a control for normalization.For the SYBRH Green assay, total RNA was reverse transcribed according to the manufacturer's instructions (#95048, Quanta Biosciences Inc.).PCR reactions were carried out using Power SYBRH Green PCR Master Mix according to the manufacturer's instructions (#4367659, Applied Biosystems Inc.).Ribosomal protein L30 was used as a control.For both protocols, relative mRNA expression was calculated by fold change using the comparative Ct method (2-DDCt) [22].Sequences of primer sets used are listed (Table S1).


Affy

trix Target Label
ng, Chip Hybridization and Data Processing

All target labeling reagents and GeneChipH HT One-Cycle Target Labeling kits were purchased from Affymetrix (Santa Clara, CA).Mouse Genome HT_MG-430 arrays were purchased from Affymetrix.Double-stranded complementary DNAs (cDNAs) were synthesized from 1.2 mg total RNA from each tissue sample through reverse transcription with an ol

o-dT primer containing the T7
NA polymerase promoter using the cDNA Synthesis System from Affymetrix.Biotin-labeled cRNAs were generated from the cDNAs and were processed on a Caliper GeneChip Array Station from Affymetrix.Labeled cRNAs were hybridized on Affymetrix Mouse Genome HT_MG-430 arrays.Array hybridization, washing

were performed according to Af
ymetrix protocol recommendations.Scanned images were subjected to visual inspection and chip quality reports were generated by Expression console (Affymetrix).The image data was processed using the Robust Multichip Average (RMA) method to determine the specific hybridizing signal for each probe set.


Pathway Analysis

Ingenuity Systems Pathway Analysis (IPA) software version 9.0 (Ingenuity Systems Incorporated; Redwood City, CA; www.ingenuity.com)was used to annotate the Affymetrix probesets for the genes they represent and the functional pathways to which they belong.The canonical pathways and the pathway genes listed in the data analyses were obtained from the Ingenuity IPA systems.


Data and Statistical Analysis

Data are reported as mean 6 standard error of the mean.Statistical analyses were performed using a Student's unpaired Table 1.Differentially expressed genes related to toll-like receptor signaling in peritoneal macrophages from Western-diet fed 11b +/+ /apoE 2/2 and 11b 2/2 /apoE 2/2 mice.


Gene Symbol

Gene N

e 11b


Results


11bHSD1 mR
A Tissue Expression

Mice deficient in 11bHSD1 were created by targeted disruption of the 11bHSD1 gene locus (Fig. S1).The resulting knockout animals were generated on the apoE 2/2 background to create the 11b 2/2 /apoE 2/2 strain.These mice were compared to littermate  11bHSD1 wild-type controls (11b +/+ /apoE 2/2 ).Expression of 11bHSD1 mRNA in selected tissues was determined by quantitative PCR analysis and data are shown in Figure 1A-C.11bHSD1 gene expression was undetectable in tissues taken from 11b 2/2 /apoE 2/2 mice vs. littermate wild-type controls.Lack of 11bHSD1 gene expression was also confirmed in thioglycollateelicited peritoneal macrophages taken from 11b 2/2 /apoE 2/2 mice vs. control (Fig. 1D).Ablation of the 11bHSD1 gene diminished 11bHSD1 cortisone reductase activity in vivo via examination of conversion of an exogenous oral dose of cortisone to cortisol in intact mice (Fig. S2A).In addition, inhibition of reductase activity was also observed in vitro using liver microsomes taken from 11bHSD1 deficient mice (Fig. S2B).


Atherosclerosis Development

To study the effects of 11bHSD1 ge e deficiency on atherosclerosis development, mice were fed a Western-type diet starting at eight weeks of age and continued for an additional 14 weeks.All mice gained weight normally (Fig. S3), blood pressures were within a normal range (Fig. S4) and no overt derangements in their basic physiology or behavior were detected.Plasma lipids were measured at the end of the dietary feeding phase and results are shown in Figure 2. All mice displayed elevated plasma total cholesterol and triglycerides as expected for the hyperlipidemic apoE knockout model [23].Total cholesterol and triglyceride levels were similar between male (Fig. 2A-B) and female (Fig. 2C-D) groups of mice, and between genotypes, indicating no statistically significant effect of 11bHSD1 deficiency on circulating lipid levels.In female mice, however, a non-statistically significant trend towards decreased triglycerides was noted in the 11b 2/2 / apoE 2/2 group.No significant changes were detected in plas a HDL-cholesterol, non-HDL cholesterol or fasting glucose in any of the groups of mice (Fig. S5).

Cholesterol accumulation in the thoracic aortae of mice was evaluated as a measure of atherosclerosis development.In males, aortic total cholesterol levels were reduced by 50% (p,0.01) in 11b 2/2 /apoE 2/2 mice vs. control (Fig. 3A).Similar effects were observed in female mice (Fig. 3B): 55% (p,0.01)reduction in 11b 2/2 /apoE 2/2 vs. control, respectively.In an analysis of mice of both genders, aortic 7-ketocholesterol levels were measured and revealed a decrease of 40% in 11b 2/2 /apoE 2/2 mice vs. control (Fig. 3C).A similar reduction was observed in aortic 7bhydroxycholesterol (Fig. 3D).Atherosclerosis was also evaluated in the aortic sinus of male mice using the method described by Paigen [19].Cross-sectional lesion areas were examined following Masson's trichrome staining of serial sections.Representative images and quantitative data (Fig. 4A) show a reduction in aortic root lesion area of 31% in 11b 2/2 /apoE 2/2 mice vs. control (11b 2/2 /apoE 2/2 :

,454,000 mm 2 ; 11b +/+ /apoE 2/2 : ,661,000 mm 2 ; p,0.05).Figures 4A and 4B show necrotic core sizes within the intima of aortic root lesions (denoted by asterisks).As shown in Figure 4B, necrotic core area revealed a 30% decrease in 11b 2/2 /apoE 2/2 mice vs. control (11b 2/2 / apoE 2/2 : ,106,000 mm 2 ; 11b +/+ /apoE 2/2 : 151,000 mm 2 ; p,0.05).Lesional macrophage content measured in the aortic root via CD68 immunohistochemistry is shown in Figure 4C.Analysis of CD68 positive stain area revealed a 32% reduction in 11b 2/2 /apoE 2/2 mice vs. control (Fig. 4C; p,0.05).

The role of leukocyte-derived 11bHSD1 on plaque development was evaluated using the bone marrow transplantation model.Bone marrow-derived stem cells from 11b 2/2 /apoE 2/2 and 11b +/+ /apoE 2/2 littermate control mice were harvested and injected into irradiated apoE-deficient recipient male mice.Following a 4 week engraftment phase, mice were placed onto Western diet for 12 weeks.Conf

mation of efficient transplantation
as carried out by quantitative PCR analysis of 11bHSD1 mRNA in whole blood from recipient mice.11bHSD1 mRNA was essentially undetectable in recipients transplanted with 11b 2/2 /apoE 2/2 -derived stem cells vs. control (Fig. 5A).Similar to findings in the Western diet-fed atherosclerosis model (Fig. 2), plasma total cholesterol was similar between both groups of mice (Fig. 5B).En face analysis of the whole thoracic aorta revealed that mice transplanted with 11b 2/2 /apoE 2/2 -derived stem cells developed 39% less atherosclerosis than control mice (Fig. 5C).Plaque area in the descending aorta was reduced by 46% in 11b 2/2 /apoE 2/2 -transplanted mice vs. control (Fig. 5D).Although not statistically significant, a strong trend towards reduced plaque area in the aortic arch of 11b 2/2 /apoE 2/2 -transplanted mice was noted (35% reduction vs. control, p = 0.06; Fig. 5E).In addition, histological analysis of atherosclerosis in the aortic root revealed a reduction in plaque area of 18% vs. control (Fig. 5F).


In Vivo Inflammatory Cell Migration

Given the reduction in

crophage content in aortic r
ot lesions, the results of the bone marrow transplantation studies and the central role of the macrophage in the pathophysiology of atherosclerosis, two important aspects of macrophage function that are known to contribute to plaque formation were further evaluated in vivo: inflammatory cell migration and cholesterol loading capacity.Inflammatory cell migration was evaluated in vivo using Western diet-fed 11b +/+ /apoE 2/2 and 11b 2/2 / apoE 2/2 mice subjected to intraperitoneal thioglycollate challenge.The ,5 week protocol shown in Figure 6A was followed.Total peritoneal cell counts were measured and no statistically significant differences were detected between groups (Fig. 6B).Phenotyping of the peritoneal lavage showed that greater than 95% of the cells were macrophages (Fig. S6).No significant differences in cell counts for either macrophages or neutrophils were detected between groups.These data s

gest that under these experimental
onditions, 11bHSD1 does not modulate inflammatory cell migration.


In Vivo Foam Cell Formation

Thioglycollate-elicited peritoneal macrophages were also evaluated for the extent of cholesterol loading following thioglycollate challenge.Prior studies using pro-atherogenic hyperlipidemic mice have shown that thioglycollate-induced peritoneal macrophages can accumulate lipid sterols in vivo and possess a phenotype suggestive of foam cells [24][25].Following the protocol shown in Figure 6A, peritoneal macrophages harvested from 11b +/+ / apoE 2/2 and 11b 2/2 /apoE 2/2 mice were subjected to organic extraction of cellular lipids followed by quantification of total cholesterol.As shown in Figure 7A, macrophage cholesterol content was reduced by 48% in 11b 2/2 /apoE 2/2 mice vs. control.Consistent with these findings, microscopic evaluation of peritoneal macrophages stained with oil red O revealed higher levels of lipid droplets in macrophages from 11b +/+ /apoE 2/2 than 11b 2/2 /apoE 2/2 (Fig. 7B).


Gene Expression Profiling Analysis

Affymetrix gene expression profiling was carried out with thioglycollate-elicited peritoneal macrophages from 11b +/+ / apoE 2/2 and 11b 2/2 /apoE 2/2 mice.Gene expression profiles were compared by Two-Way ANOVA using PartekH Discovery Suite software.Only the Affymetrix probe sets that displayed a maximum anti-log RMA signal intensity of greater than 16 were considered for post hoc analyses.Differential changes in gene expression levels ranged from ,+2.3 fold for up-regulated genes (428 genes) and ,23.8 fold for down-regulated genes (434 genes).Interestingly, given the observation of decreased in viv cholesterol loading with 11bHSD1 deficiency, there were no significant changes in transcript levels of genes associated with oxidized LDL scavenging/uptake in macrophages, for example, scavenger receptor A1(SR-A1), A2(SR-A2), B1(SR-B1) and thrombospondin receptor CD36.Similarly, there were no transcript level changes in genes classically associated with macrophage cholesterol homeostasis: acetyl-CoA acetyltransferase 1 (ACAT1), cholesteryl ester hydrolase (CEH), nuclear hormone receptors LXRa and LXRb and ABC transporter proteins ABCA1 and ABCG1.However, a significant 1.8 fold decrease (p = 0.001) in toll-like receptor 4 (TLR 4) gene expression was detected in 11b 2/2 / apoE 2/2 mice vs. control.Further inspection of the TLR gene family revealed signific nt decreases in TLR 1, 3, 4, 8 and 13 (Table 1) in 11b 2/2 /apoE 2/2 mice.These data indicate that 11bHSD1 gene knockout leads to significant down regulation of selected TLRs in peritoneal macrophages.

Further analysis of peritoneal macrophage microarray data revealed down regulation of genes associated with TLR signal transduction.As shown in Table 1, TNF receptor-associated factor 5 (TRAF5) and TRAF3 interacting protein 3 (TRAF3IP3) were decreased in 11b 2/2 /apoE 2/2 mice vs. control.Other downstream TLR signal transduction pathway genes including Map3k2 and JAK2 were also down regulated relative to control (Table 1).In addition, endoplasmic reticulum degradation enhancer, mannosidase alpha-like 3 (Edem3) mRNA levels were also downregulated 1.6 fold in 11b 2/2 /apoE 2/2 vs. control.Real-time quantitative PCR of several key genes from the microarray studies confirmed the magnitude and direction of changes observed by microarray analysis (Fig. 8).

An evalu

ion of significantly affected biological pathways with 11bHSD
gene deficiency was carried by Ingenuity Pathway Analysis (IPA).IPA revealed reductions in genes associated with inflammatory/immune function, suggesting that the overall inflammatory state of 11bHSD1-deficient macrophages was attenuated (Table 2).Inflammatory pathways that were attenuated included; chemokine signaling, JAK/Stat signaling, pattern recognition receptor signaling (Table S2A-C), NF-kB signaling (Table S2D) and glucocorticoid receptor signaling (Table S2E).In addition, IPA revealed significant differential expression of genes associated with mitochondrial function/energy generation: oxidative phosphorylation was the most impacted pathway (Table 2 and Table S3).Mitochondrial genes involved in the electron transport chain (i.e., ATP production), NADH dehydrogenase, ubiquinolcytochrome C reductase and ATP synthase were significantly upregulated in 11b 2/2 /apoE 2/2 mice.


Cytokine Secretion with Pro-Atherogenic TLR Ligand Challenge

Cytokine release from cultured thioglycollate-elicited peritoneal macrophages from Western diet-fed 11b +/+ /apoE 2/2 and 11b 2/2 /apoE 2/2 mice was examined by antibody array analysis following stimulation with oxidized LDL.Incubation conditions and challenge with oxidized LDL occurred in the presence of the 11bHSD1 substrate 11-dehydrocortic

terone.As s
own in Figure 9, treatment with oxidized LDL stimulated the release of several pro-inflammatory cytokines from cultured cells of both strains of mice vs. untreated control, including granulocyte colony stimulating factor (G-CSF), keratinocyte-derived chemokine (KC; murine homologue of IL-8), monocyte chemotactic protein-1 (MCP-1) and tumor necrosis factor-alpha (TNF-a).Release of these cytokines into cell culture media was reduced with 11bHSD1 deficiency (Fig. 9).Significant reductions in GCSF, KC and MCP-1 ranged from 41-46% vs. control.Release of TNF-a by 11bHSD1-deficient peritoneal macrophages was reduced by 30% vs. control.Seven additional cytokines that were measured yielded variable responses in replicate studies: IL-1b, IL-6, IP-10, MIP-1A, MIP-1B, MIP-2, RANTES (data not shown).In addition, cytokine levels for peritoneal macrophages not challenged with oxidized LDL were at or below the detection limit of the assay.


Discussion

Treatment of hyperlipidemic mice with pharmacological inhibitors of 11bHSD1 has been shown to reduce plaque in the presence of reduced plasma lipids.For example, a study in the Western diet-fed apoE knockout mouse showed that eight weeks of treatment with a selective 11bHSD1 inhibitor decreased aortic cholesterol burden by 84% vs. vehicle control [17].Plaque lowering was observed along with reductions in circulating total cholesterol (,30%), triglycerides (,60%) and glucose (44%).Similarly, a study performed in a variant o the LDLr knockout mouse, that exhibits characteristics of the metabolic syndrome, showed that four weeks of treatment with carbenoxolone, a pan-11bHSD inhibitor, decreased aortic root plaque size by ,30% vs. vehicle control [18].In addition, carbenoxolone treatment significantly improved plasma lipid profiles, enhanced VLDL clearance rates and lowered fasting plasma insulin levels.Full interpretation of the latter study, however, is limited by the nonselectivity of carbenoxolone, which inhibits both 11bHSD1 and 11bHSD2 with similar affinity.While these reports were important in first demonstrating the intriguing potential of 11bHSD1 inhibitors to ameliorate atherosclerosis, the results could not distinguish whether the reduction in plaque burden was due to a direct effect on the vessel wall, an indirect effect via improvement in the lipid profile and/or glucose metabolism, changes in blood pressure, or a combination of the above.

In order to elucidate the physiological and molecular mechanisms underlying the anti-atherosclerotic effect of 11bHSD1 inhibition, the studies described herein were performed.The results indicate hat inactivation of 11bHSD1 can positively impact aortic plaque by decreasing i) aortic cholesterol and 7oxysterol accumulation, ii) aortic root plaque area iii) necrotic core intimal area and iv) CD68-positive macrophage inflammation.These data support a rapidly evolving concept that inhibition of 11bHSD1 can ameliorate atherosclerotic disease via direct effects at the arterial wall, without impacting plasma lipids or other metabolic parameters [3].Furthermore, bone marrow transplantation experiments revealed that the absence of leukocyte-specific 11bHSD1 reduces plaque burden, suggesting direct involvement of monocyte/macrophage 11bHSD1 in atherosclerosis development.Upon closer inspection of macrophage phenotype, the observed downregulation of several TLRs, including TLR4 and the attenuated cytokine response to oxidized LDL challenge suggests a key role of 11bHSD1 in modulating TLR interactions with pro-atherogenic ligands.

Analyses of cholesterol levels in the thoracic aortae of mice revealed significant decreases in cholesterol (Fig. 3A-B) and 7oxysterols (Fig. 3C-D) with 11bHSD1 deficiency.Oxysterols such as 7-ketocholesterol and its metabolite 7b-hydroxycholesterol are among the most abundant oxysterols in human foam cell macrophages and advanced human lesions [10][11].Aortic oxysterols of mice of both genotypes revealed detectable amounts of 7-ketocholesterol and 7b-hydroxycholesterol.11b 2/2 /apoE 2/2 mice had significantly lower absolute levels of 7-ketocholesterol and 7b-hydroxycholesterol than 11b +/+ /apoE 2/2 control mice (240% and 245%, respectively, Fig. 3C-D).Studies in culture have shown that mouse peritoneal macrophages loaded with 7-ketocholesterol-enriched LDL particles are impaired in their ability to efflux cholesterol to apoA-I acceptor particles [26].Thus, the reduction in aortic 7ketocholesterol levels observed with decreasing 11bHSD1 gene levels may promote an improvement in the ability of macrophages to unload cholesterol and thereby attenuate foam cell formation in the vessel wall.

Histological analyses of structure and composition of the atherosclerotic plaques present in the aortic root revealed a 30% reduction in necrotic core area with 11bHSD1 gene deficiency (Fig. 4B).We performed CD68 immunohistochemistry to determine whether changes in macrophage activation state and/or macrophage numbers could account for the reductions in necrotic core area.CD68 is a useful marker of both macrophage number and also the state of macrophage activation in the context of the atherosclerotic plaque.CD68 is highly upregulated as macrophages take up oxidized lipids in the process of becoming foam cells [27].Aortic root samples from homozygous 11bHSD1 knockout mice had 32% less staining for CD68 than samples taken from control mice.We asked whether this reduction was due to i) differences in circulating monocyte levels between groups, ii) impairment in monocyte infiltration or iii) a potential change in cell surface CD68 expression due to a reduction in foam cell formation.

The first possibility was addressed by analyzing blood samples from 11b +/+ /apoE 2/2 and 11b 2/2 /apoE 2/2 mice (naı ¨ve and thioglycollate injected): no differences in monocyte or total leukocyte counts were detected between groups (Fig. S7).To address the second possibility, macrophage infiltration capacity was tested via the ability of macrophages to migrate into the peritoneum in response to thioglycollate challenge.The data presented in Figure 6B indicate that 11bHSD1 gene deficiency does not modulate inflammatory cell migration into the peritoneum within the context of the experimental design.Cell counts are similar regardless of the genotype.Similar results have been reported in the literature for independently derived 11bHSD1 knockout mice [28].While migration into the peritoneum is not precisely the same as migration into the vessel wall, it does indicate that there is no major change in macrophage migration capacity in response to an inflammatory stimulus.It therefore suggests that 11bHSD1-mediated regulation of inflammatory cell migration/ infiltration is not likely to have a substantial effect on CD68 staining and atherosclerosis in this model.

Given this conclusion, the third possibility was tested.Mice of both genotypes were placed on a Western diet for 4 weeks.At the end of this period, they were subjected to a peritoneal thioglycollate challenge to elicit recruitment of macrophages into the eritoneum and stimulate foam cell formation with hyperlipidemia.Staining and microscopic evaluation of macrophages showed a distinct foamy appearance and the presence of intracellular lipid droplets due to the accumulation of cholesteryl ester (Fig. 7B).Quantification of cellular lipid content showed that the macrophages isolated from 11b 2/2 /apoE 2/2 mice had 48% less cholesterol than macrophages isolated from control mice (Fig. 7A).It therefore is likely that reduced CD68 levels in aortic root plaque of 11bHSD1 knockout mice is at least partially related to the reduced foam cell-like character of macrophages isolated from these animals leading to either downregulation of CD68 expression or a decrease in total foam cell area.Regardless, the reduction in aortic CD68 staining is indicative of a decrease in the foam cell state generated by macrophages in the plaque.

Mouse bone marrow transplantation studies have been used to study the role of leukocyte-associated genes on the pathophysiology of atherosclerosis.The results of the analysis described herein clearly indicate the involvement of leukocyte associated-11bHSD1 in atherosclerosis development.In these studies, plaque burden in the thoracic aorta was reduced 39% with 11bHSD1 deficiency vs. control (Fig. 5C).These changes are consistent with findings obtained in the diet-induced atherosclerosis experiments described in this report (50% reduction in aortic cholesterol content of male 11b 2/2 /apoE 2/2 mice; Fig. 3A).In both cases, total circulating cholesterol levels were similar between groups despite differences in 11bHSD1 expression levels (compare Fig. 2A and Fig. 5B).Thus, both atherosclerosis studies provide direct evidence that 11bHSD1 deficiency can attenuate vessel wall atherosclerosis independent of changes in plasma cholesterol.The observation that monocyte/macrophage-specific 11 HSD1 deficiency is associated with decreased plaque burden may relate, in part, to macrophage-associated 11bHSD1 function.Foam cells are largely derived from circulating monocytes that differentiate into proinflammatory macrophages upon infiltration into the vascular subendothelium.Studies in culture with human monocytes have shown that monocyte to macrophage differentiation is a strong stimulus for 11bHSD1 expression [29].Moreover, 11bHSD1 expression is completely absent in pre-differentiated human monocytes, whereas macrophages differentiated with pro-inflammatory agents (i.e., LPS, cytokines) express high levels of 11bHSD1 [29].Although we cannot rule out the impact of other immune cells lacking 11bHSD1 in the bone marrow transplantation studies (i.e., lymphoid and other myeloid types), the association between 11bHSD1 and macrophage function is likely.

The critical observations that leukocyte-specific 11bHSD1 deficiency reduces plaque burden and in vivo cholesterol accumulation in peritoneal macrophages is reduced in 11b 2/2 / apoE 2/2 mice suggest that inhibition of 11bHSD1 can modulate an important function of the macrophage central to atherosclerosis pathophysiology.Prior studies by other investigators have shown that hyperlipidemic peritoneal macrophages possess phenotypic and functional attributes of lesional macrophages and appear to be relevan surrogates of macrophages present in atheroma [25] [24].Based on those previous findings, microarray studies of in vivo peritoneal foam cells were performed to elucidate the potential pathways that drive reduced foam cell formation and atherosclerosis in 11bHSD1 deficient mice.While many differences in the gene expression profile were noted, of particular interest to this discussion is the observation that ablation of the 11bHSD1 gene resulted in downregulation of several toll-like receptors (TLRs) including TLR 1, 3, 4, 8 and 13.In particular, the mRNA for TLR4 was downregulated approximately two fold as confirmed by quantitative PCR.Moreover, the impact of decreased TLR transcript levels translated to a diminished functional cytokine response following challenge with oxidized LDL, a known proinflammatory and pro-atherogenic ligand that binds TLR4.

TLR4 is expressed by foam cells, is present in human coronary artery plaque, is up regulated by oxidized LDL and, most relevant to the results presented herein, is involved in macrophagemediated lipid accumulation [30][31][32][33].Indeed, Higashimori et al. [34] showed that TLR4/apoE double knockout mice accumulated ,75% less cholesterol in the aortic arch vs. control without changes in circulating lipids.Down regulation of the TLR family of genes, including TLR4, in 11bHSD1-deficient macrophages therefore represents one likely mechanistic explanation for the reduced cholesterol accumulation observed in our in vivo foam cell experiments and the reduction in atherosclerosis wi hout a concomitant change in the plasma lipid profiles.The attenuated cytokine response from 11bHSD1-deficient macrophages following TLR4 agonist stimulation with oxidized LDL suggests that 11bHSD1 is important for potentiating TLR4-dependent activation and signaling in the setting of atherosclerosis.This observation is without precedent and suggests a significant interaction between foam-cell associated TLRs and 11bHSD1 function.Interestingly, studies in culture using rat perit neal macrophages have shown that exposure to corticosterone can modulate TLR2 mRNA levels without directly impacting TLR4 expression [35].These results suggest a more complex interaction with TLR4 beyond regulation of cortisol/corticosterone levels by 11bHSD1.Additional studies are needed to fully understand this relationship.

To discern additional mechanistic hypotheses for the antiatherosclerotic effect of the 11bHSD1 gene knockout, the microarray data were further analyzed via an Ingenuity Pathway Analysis (IPA).The results of this analysis (Table S2) revealed significant modulation of genes related to inflammatory function (chemokine signaling, JAK/STAT signaling, pattern recognition, NF-kB signaling, and glucocorticoid receptor signaling) in 11b 2/2 /apoE 2/2 peritoneal macrophages vs. control.

The pattern for chemokine, NF-kB and JAK/STAT signaling, combined with the downregulation of the pattern recognition receptors is consistent with an anti-inflammatory response to 11bHSD1 gene ablation and the relative anti-atherosclerotic phenotype observed in 11b 2/2 /apoE 2/2 mice.On the other hand, the broad downregulation in the glucocorticoid signaling pathway is not, on the surface, consistent with an anti-atherosclerotic effect given the known anti-inflammatory effects of pharmacological glucocorticoid therapy.One possibility to explain this discrepancy is that the decrease in signaling via the glucocorticoid receptor (GR) is balanced or dominated by a similar decrease in signaling via the aldosterone-mineralocorticoid receptor pathway.It is well known that aldosterone is a pro-inflammatory agent in macrophages, [36] an effect that can be blocked with mineralocorticoid receptor (MR) antagonists such as spironolactone or eplerenone.In the vasculature, aldosterone infusion has been shown to stimulate neo-intimal hyperplasia and vascular remodeling following endothelial injury [37].Also, it is well known that cortisol has a higher affinity for the MR than it does for the GR.Therefore, it is possible that reducing intracellular glucocorticoid tone via the 11bHSD1 knockout reduces pro-inflammatory signaling via the MR to a greater extent than it reduces the anti-inflammatory signaling through the GR.
upport for this concept comes from studies in 11bHSD2/apoE double knockout mice, where loss of inactivation of corticosterone to 11dehydrocorticosterone accelerates plaque development and intimal inflammation [38].In the same study, MR antagonism with eplerenone reduced plaque development despite an increase in tissue glucocorticoids.Importantly, treatment with eplerenone did not lower blood pressure, indicating that aldosterone function was not significantly impacted.Similar to these findings, our studies did not reveal an impact on blood pressure with 11bHSD1 deficiency (Fig. S4).Thus the anti-atherosclerotic activity was likely mediated by inhibition of corticosterone-MR interactions, further supporting the concept that reduction in glucocorticoid tone at the MR is anti-atherosclerotic.

The IPA analysis also revealed significant differential changes in expression of genes related to mitochondrial oxidative phosphorylation (Table 2 and Table S3).Most of these genes were up regulated.These findings are intriguing and suggest that ATP production is enhanced in thioglycollate-elicited peritoneal macrophages deficient in 11bHSD1.The increase in oxidative phosphorylation genes in 11bHSD1-deficient macrophages may relate to decreases in intracellular glucocorticoid tone that promote a less inflammatory macrophage phenotype, as suggested by our inflammation-related gene profiling data.Interestingly, published studies by Vats et al. showed that polarization of macrophages to the less inflammatory M2 type using IL4 stimulation induced robust increases in genes associated with mitochondrial energy metabolism, including those involved in uptake and oxidation of fatty acids [39].These changes in oxidative metabolism genes translated to increases in fatty acid uptake and b-oxidation activity by M2 macrophages [39].By contrast, in the same report, pro-inflamma ory M1 macrophages were shown to preferentially utilize glucose and suppress boxidation.The preference for glucose and glycolytic ATP and concomitant arrest of oxidative phosphorylation by pro-inflammatory M1 macrophages has also been demonstrated in murine J774 macrophages [40][41].Taken together, these studies indicate that oxidative phosphorylation is inherently linked to the less inflammatory phenotype associated with M2 macrophages.While there were no significant differences in other well-recognized M2 marker genes (e.g., arginase I, mannose receptor C-type I) relative to 11bHSD1 gene expression levels in our microarray studies (data not shown), the increase in xidative phosphorylation genes may regulate overall macrophage activities that favor attenuated plaque development.

Besides reducing the cholesterol content within the vessel wall, reduction in macrophage foam cell formation can also reduce the progression from simple foam-cell laden fatty streaks to lesions with necrotic cores.It is generally accepted that accumulation of cholesterol and oxidized lipids drives foam cell formation, the subsequent sequelae of pro-inflammatory changes and eventually, apoptotic and/or necrotic cell death [42][43][44].Applying these observations to the current data, evaluation of plaque phenotype in the current study revealed a significant reduction in average necrotic area n the aortic root of 11b 2/2 /apoE 2/2 mice (230% vs. control; Fig. 4A-B).

We therefore propose a model whereby the absence of 11bHSD1 attenuates macrophage foam cell formation via downregulation of the toll-like receptors.This in turn reduces uptake of cytotoxic levels of sterols and oxidized lipids, lessening the insult to the cell and diminishing the propensity of macrophages to proceed to necrosis.As fewer macrophages enter necrosis, there will be reduced substrate for necrotic core formation.Besides directly reducing necrotic core formation, reducing macrophage necrosis will also reduce the pro-inflammatory signal that stimulates further macrophage infiltration.This would result in prolongation of the rate at which the plaque matures to the point of being pathological.

In addition, given recent advances in the understanding of the natural history of macrophages in atherogenesis, [45] it is also possible that reducing macrophage foam cell formation may not just reduce necrosis, but also enhance clearance of necrotic cells (efferocytosis) as part of the process of resolution of inflammation [46].It has been reported that impaired clearance of dying cells in atherosclerotic plaque may contribute to necrotic core formation [47] [46].It is possible that the 11bHSD1 knockout, by reducing the extent of macrophage foam cell formation, allows macrophages to proceed down a pathway leading to enhanced efferocytosis.Our data indicate i) a switch to oxidative phosphorylation for increased ATP production and ii) overall attenuation of pro-inflammatory pathways by 11bHSD1-deficient macrophages, suggesting the develop ent of a macrophage phenotype that is more resolution-competent and less inflammatory [39].The resulting removal of necrotic material from the plaque would provide a protective function to guard the surrounding intimal environment from harmful exposure to pro-atherogenic reactive oxygen species, pro-inflammatory cytokines, pro-thrombotic proteins, extracellular matrix degrading enzymes and other factors that contribute to the vicious cycle of atherogenesis leading to plaque instability.The end result of this would be a slowing of atherogenesis cou led with enhanced resolution of inflammation, leading to a decrease in overall plaque area with the potential for overall improvement in the structure and function of the vessel wall.

The findings reported herein demonstrate that 11bHSD1 inhibition can attenuate development of atherosclerotic disease directly at the vessel wall without necessarily affecting plasma lipid profiles.Decreasing cholesterol loading in macrophages via downregulation of toll-like receptors offers one plausible mechanistic explanation for atheroprotection observed in this model.It is clear that additional investigation at the molecular level and via pharmacological intervention is required to confirm the hypotheses generated by these in vivo studies.

The utility of 11bHSD1 inhibition as a treatment for atherosclerosis is becoming increasingly evident.At present, however, efforts to develop 11bHSD1 inhibitors in the clinic have primarily focused on management of type 2 diabetes [48].Our results suggest that direct therapeutic targeting of the vessel wall with 11bHSD1 inhibitors may also offer benefit.This raises the intriguing possibility of a single therapeutic agent tha