Carcinoembryonic antigen-related cell adhesion molecule 1 (CEACAM1) promotes hepatic insulin clearance and endothelial survival. However, its role in the morphology of macrovessels remains unknown. Mice lacking Ceacam1 (Cc1−/−) exhibit hyperinsulinemia, which causes insulin resistance and fatty liver. With increasing evidence of an association among hyperinsulinemia, fatty liver disease, and atherosclerosis, we investigated whether Cc1−/− exhibited vascular lesions in atherogenic-prone aortae. Histological analysis revealed impaired endothelial integrity with restricted fat deposition and aortic plaque-like lesions in Cc1−/− aortae, likely owing to their limited lipidemia. Immunohistochemical analysis indicated macrophage deposition, and in vitro studies showed increased leukocyte adhesion to aortic wall, mediated in part by elevation in vascular cell adhesion molecule 1 levels. Basal aortic eNOS protein and NO content were reduced, in parallel with reduced Akt/eNOS and Akt/Foxo1 phosphorylation. Ligand-induced vasorelaxation was compromised in aortic rings. Increased NADPH oxidase activity and plasma 8-isoprostane levels revealed oxidative stress and lipid peroxidation in Cc1−/− aortae. siRNA-mediated CEACAM1 knockdown in bovine aortic endothelial cells adversely affected insulin's stimulation of IRS-1/PI 3-kinase/Akt/eNOS activation by increasing IRS-1 binding to SHP2 phosphatase. This demonstrates that CEACAM1 regulates both endothelial cell autonomous and nonautonomous mechanisms involved in vascular morphology and NO production in aortae. Systemic factors such as hyperinsulinemia could contribute to the pathogenesis of these vascular abnormalities. Cc1−/− mice provide a first in vivo demonstration of distinct CEACAM1-dependent hepatic insulin clearance linking hepatic to macrovascular abnormalities.
- metabolic syndrome
- fatty liver disease
the carcinoembryonic antigen-related cell adhesion molecule 1 (CEACAM1) is a membrane glycoprotein expressed in many cell types, including hepatocytes and angiogenically activated endothelial cells, but not skeletal myocytes or adipocytes. CEACAM1 promotes vascular morphogenesis (10, 17). Consistently, mice with endothelial-specific CEACAM1 overexpression exhibit improved ensheathment of the endothelial barrier by pericytes as well as by a well-structured vascular basement membrane (11). On the other hand, mice with global deletion of Ceacam1 (Cc1−/−) develop several endothelial and vascular abnormalities, including reduced endothelial cell response to vascular endothelial growth factor (VEGF) and increased basal endothelial permeability (24).
CEACAM1 also promotes hepatic insulin clearance, as demonstrated by the development of hyperinsulinemia resulting from impairment of insulin extraction in Cc1−/− mice (8, 39). Hyperinsulinemia causes systemic insulin resistance (8, 28, 39) and increased hepatic triglyceride synthesis, leading to hepatic steatosis and substrate redistribution to white adipose tissue to cause visceral obesity (23), followed by lipolysis and elevated plasma nonesterified fatty acid (NEFA) levels (8) in addition to a rise in adipose tissue-associated macrophage recruitment and adipokine release (12). Despite increased liver triglyceride and cholesterol synthesis, plasma triglyceride (8) and total cholesterol (39) levels are unchanged. Moreover, Cc1−/− mice develop features of nonalcoholic fatty liver disease (NAFLD), with mild basal fibrosis and inflammatory infiltration in liver that could in part arise from increased fat deposition (3). When fed a high-fat diet deriving 45% of calories from fat, a more progressive phenotype replicating all features of human nonalcoholic steatohepatitis (NASH) develops. This includes a more robust fibrosis and inflammation (12). Given that chronic hyperinsulinemia causes an increase in ectopic fat accumulation, including in large vessels, and that CEACAM1 is markedly reduced in the liver of obese subjects with NAFLD (19), a population at high risk for atherosclerosis (2), we investigated whether Cc1−/− mice also develop lipid accumulation and vascular abnormalities in atherosclerosis-prone large vessels.
MATERIALS AND METHODS
C57BL/6 Cc1−/− and their Cc1+/+ littermates (39) were kept on a 12:12-h dark-light cycle and fed standard chow ad libitum. Male mice at 6 mo of age (unless otherwise mentioned) were examined. All procedures were approved by the Animal Care and Utilization Committee in each participating institution. Plasma levels of insulin, triglycerides, and NEFA were determined as described (8), of thiobarbituric acid reactive substance (TBARS) biochemically (25), and of reduced glutathione (GSH) by the Bioxytech GSH-400 kit (OxisResearch).
Plasma was purified in 4xV cold ethanol and centrifuged at 1500 g for 10 min, and the supernatant was collected and ethanol vacuum-evaporated before acidification to pH 4.0 with ∼50 μl of 30% acetic acid followed by purification on preactivated SPE cartridges (C-18) (Item no. 400020, Cayman Chemical, Ann Arbor, MI). 8-Isoprostane was eluted at 4°C with 5 ml of ethyl acetate containing 1% methanol, vacuum-dried, reconstituted in 200 μl of EIA buffer, and assayed (50 μl) in triplicates using the 8-Isoprostane EIA Kit (Item no. 516351, Cayman Chemical). At the end of the incubation period with 8-isoprostane tracer and 8-isoprostane EIA antiserum at 4°C for 18 h, samples were rinsed five times with buffer, and Ellman's Reagent was added in the dark at room temperature for 120 min. Absorbance was read at 420 nm and data wereplotted as %B/B0 vs. log concentration using a four-parameter logistic fit.
Lipoproteins (VLDL, intermediate-density lipoprotein plus LDL, and HDL) were separated by sequential density ultracentrifugation of plasma in a TLA100 rotor, and their cholesterol content was determined by colorimetric assays and measurement on the SpectraMax 250 system (13).
Plasma fatty acid analysis.
Fatty acid distribution in whole plasma was assayed as described (31). Briefly, each sample was subjected to direct transesterification and injected into a gas chromatograph by using a (90 m × 0.32 mm) WCOT-fused silica capillary column VF-23ms coated with 0.25 mm film thickness (Varian, Canada).
Transfection of endothelial cells.
Bovine aortic endothelial (BAE) cells were maintained in MCDB-131 medium (Vec Technologies, Rensselaer, NY). Cells were transfected with 100 pmol of scrambled or CEACAM1-specific siRNA, using Lipofectamine 2000 reagent (Invitrogen, Carlsbad, CA) as previously described (24).
Nitric oxide release analysis in cell media.
Nitric oxide (NO) level was assessed in 20 μl of medium using a Nitrate/Nitrite Fluorometric Assay Kit (catalog no. 780051, Cayman Chemical), per the manufacturer's instructions. Fluorescence was read using the Synergy H1 Hybrid Multi-Mode Microplate Reader (BioTek Instruments, Winooski, VT) at 360 nm excitation and 430 nm emission wavelengths.
Analysis of NO production in isolated aortic rings.
Thoracic aorta segments were removed and cut into four rings (2.5 mm each) before concentration-response studies of vasorelaxation stimulated by acetylcholine and sodium nitroprusside were performed (29).
NADPH oxidase activity.
Aortic tissue was homogenized in lysis buffer (20 mmol/l KH2PO4, 1 mmol/l EGTA, and protease inhibitors, pH 7.4) and subjected to a lucigenin-derived luminescence assay in the presence of NADPH (0.1 mM) (33). Luminescence was measured every 1.8 s for 5 min in a luminometer (Veritas Microplate Luminometer; Turner Biosystems, Sunnyvale, CA).
Toluidine blue staining and histological examination by light microscopy.
Aortic arch (3 mice/group) was serially sectioned (4–5 μm thick), and every 10th section was H&E stained. To identify plaque area, the internal elastic membrane of the aortic wall marking the border between the tunica intima (endothelial layer) and the tunica media (smooth muscle layer) was used as a reference point. Additionally, the morphology of cells under the endothelial layer in relation to the following smooth muscle cells and to the internal elastic membrane was considered to determine the plaque border within the aortic wall. Measurements were done under (Keyence, BZ 9000) light microscope using the BZ-II image analysis software (Keyence, Neu-Isenburg, Germany). The automatically calculated plaque area was recalculated based on the final magnification at ×200. Measurements were performed on 15 H&E-stained sections (5 per mouse).
Aortic arch was sectioned, fixed in phosphate-buffered glutaraldehyde (5.5%) for 18 h immediately after removal, embedded in Epon 812, cut in semithin sections (0.5 μm thick), and stained with Toluidine blue before analysis with a Leica microscope equipped with a digital camera (Leica, Bennsheim, Germany) and the software Leica Application Suite v. 2.7.
Goldner trichrome staining.
Paraffin aortic arch sections of 5 μm thick were rehydrated in ethanol and treated with iron hematoxylin stain for 2 min, washed in water for 10 min, and exposed to Mason-Goldner (MG) mixture for 7 min and sequentially to a short treatment with 1% acetic acid, Phosphormolybden-Orange-G solution for 10 min, 1% acetic acid, 0.1% light green for 8–10 min, triple treatment with 100% ethanol (1 min each time) and a double treatment with Xylol (5 min each time).
Oil red O staining.
Aortic roots (n > 5 mice/group) were fixed overnight in 4% paraformaldehyde and frozen in OCT embedding medium prior to staining with Oil red O. To quantify the positive lipid-stained areas, the NIS-Elements Imaging Software 3.0 system was used.
Frozen sections of aortic roots were immunostained with MOMA-2 antibody (1:500; Abcam, Cambridge, UK) overnight at 4°C prior to incubation with secondary antibody for 1 h according to Dako IHC LSAB kit instructions. To quantify positive-stained areas, the NIS-Elements Imaging Software 3.0 system was used.
Tissue pieces of aortic arch were fixed in Bouin's solution and embedded in paraffin before immunostaining with polyclonal antibodies against endothelial NO synthase (eNOS, 1:100; R&D Systems; Minneapolis, MN). The specific staining was developed by glucose-peroxidase technique (17). Aortic fragments were cut, fixed immediately in 4% paraformaldehyde for 2 h, and whole-mounted on glass slides. Following an overnight incubation with goat polyclonal vascular endothelial (VE)-cadherin antibody (1:100; Santa Cruz Biotechnology, Santa Cruz, CA) at 4°C, aortae were washed twice in PBS for 10 min, treated with FITC-conjugated anti-goat IgG for 60 min, washed three times, and analyzed using a Leica microscope.
Adhesion of isolated mouse leukocytes to mouse aorta in vitro.
As previously described (26), the thoracic aorta and blood-resident leukocytes were isolated from anesthetized donor littermate Cc1+/+ and Cc1−/− mice (8–9 mo of age). Following a midline thoracotomy, the aorta was quickly removed and placed in cold, oxygenated phosphate-buffered saline (PBS). After careful removal of the adventitia, aortas were cut into 4-mm segments under a dissecting microscope (PZMIV; WPI, Sarasota, FL). Aortic segments were then carefully opened longitudinally and placed in culture dishes containing 1 ml of Krebs-Henseleit solution with their luminal surface facing up.
Whole blood was obtained from anesthetized mice through a cannula inserted in the carotid artery. Leukocytes were isolated from whole blood by gradient centrifugation as previously described (29). Isolated leukocytes were then fluorescently labeled using a PKH26GL staining kit (Sigma-Aldrich, St. Louis, MO). Briefly, a suspension of leukocytes was incubated with 4 μM PKH26GL for 5 min prior to adding PBS containing 10% fetal calf serum. Leukocytes were then incubated with the aortic segments at a concentration of 105 cells/aortic segment in incubation wells for 60 min at 37°C in an orbital shaker platform. The aortic segments were then removed, gently washed in fresh Krebs-Henseleit buffer, and placed lumen side up on microscope slides and treated with a drop of immersion oil followed by a glass coverslip. The number of leukocyte adhering to the endothelial surface was counted in 20 separate microscopic fields under epifluorescent microscopy at a magnification of ×200. Results are expressed as total number of cells per microscopic field.
Western blot analysis.
Cells were serum starved overnight prior to insulin treatment (100 nM) for 5–30 min and protein analysis by 4–12% SDS-PAGE (Invitrogen, Carlsbad, CA), followed by Western blot with antibodies against phospho-Ser1177 eNOS, eNOS, phospho-Ser473 Akt, Akt A (Cell Signaling Technologies, Beverly, MA), and CEACAM1 (Ab 3759, a custom-made rabbit polyclonal). Proteins were detected using LiCOR secondary antibodies per manufacturer's instructions (LiCOR Biosciences, Lincoln, NE). Aortae were perfused with 10 ml of PBS after opening of the left cardiac chamber. Fragments of the aortic arch were homogenized using RIPA buffer on ice for 15 min. Proteins were analyzed by 8% reducing SDS-PAGE and Western blot using 1:100 vascular endothelial growth factor receptor (VEGFR)-2 (Santa Cruz Biotechnology) and eNOS antibodies followed by β-actin monoclonal antibody (1:50, Sigma-Aldrich) for protein loading. Proteins were detected by enhanced chemiluminescence (ECL; Amersham Biosciences, Sunnyvale, CA) and imaged using Image Reader LAS-3000 (Fujifilm), and the intensity was determined by comparable densitometric analyses via MultiGauge software (Fujifilm).
Semiquantitative real-time PCR analysis.
Total RNA was prepared using a PerfectPure RNA Tissue kit (5Prime, Gaithersburg, MD) following the manufacturer's instructions. cDNA was synthesized with oligo(dT) primers and Improm II Reverse Transcriptase (Promega), using 1 μg of total RNA and primers (Table 1). cDNA was evaluated with semiquantitative real-time PCR (qRT-PCR) using the StepOne Plus Real Time PCR system (Applied Biosystems, Foster City, CA). The relative amount of mRNA was normalized relative to glyceraldehyde-3-phosphate dehydrogenase (GAPDH). Results are expressed as means ± SE relative to controls.
Statistical analysis was determined by unpaired two-tailed Student's t-test with GraphPad Prism software. Statistical significance was set at 0.01–5%.
Development of morphological vascular abnormalities in Cc1−/− aortae.
Histological analysis on H&E-stained sections from 6-mo-old mice revealed small plaque-like lesions in the aortae of Cc1−/− mice (aortic arch, ascending, thoracic, and abdominal parts; Fig. 1B). High-power light microscopic (LM) analysis on Toluidine blue (TB)-stained semithin sections demonstrated formation of lesions in the aortic intima with thickening and marked cell accumulation in the opposite aortic adventitia of Cc1−/− mice (Fig. 1A.ii). MG trichrome staining of aortic arch areas with lesions revealed subendothelial accumulation of fibrotic materials (Fig. 1A.iii, *green stains). Consistently, aortic mRNA levels of fibrosis markers, transforming growth factor-β (Tgfβ) connective tissue growth factor (Ctgf), and fibronectin were also elevated (Table 2). Oil red O (ORO) staining of cross-sections of aortic roots showed evidence of lipid deposition in the intima of Cc1−/− mice (Fig. 1A.i and accompanying graph).
Absence of proatherogenic hypercholesterolemia in Cc1−/− mice.
The area of aortic lesions ranged from 33.7 to 218.3 μm2, with a mean value approximately sixfold smaller than the mean plaque area in age-matched apolipoprotein E-deficient mice (ApoE−/−), a genetic model of atherosclerosis (Fig. 1B and accompanying graph). Unlike ApoE−/− mice, Cc1−/− mice exhibited normal total and LDL-cholesterol levels in addition to human-like HDL-cholesterol compared with wild-type littermates (Table 3). As previously observed (8, 39), these mice also exhibited normal plasma total and VLDL-triglyceride levels despite increased hepatic lipid production and mildly elevated plasma apoB-48/100 (8), consistent with increased lipid redistribution from liver to white adipose tissue (WAT) and elevated NEFA levels (Table 3) (8). However, this did not translate into changes in the relative composition of fatty acids that are commonly associated with macrovascular abnormalities (21, 37), such as saturated and transsaturated fat, and in the ratio of omega 6 to omega 3 polyunsaturated fatty acids (Table 3). Because wild-type mice resist atherosclerosis due to their low circulating apoB-containing lipoproteins, it is likely that limited circulating cholesterol and triglyceride levels, together with unfavorable changes in fatty acid composition, restricted vascular lesions in Cc1−/− mice.
Elevated proinflammatory state in Cc1−/− mice.
Consistent with increased inflammation in visceral obesity, mRNA levels of WAT-associated F4/80 and the proinflammatory interleukin-6 (IL-6) and TNFα cytokines, in addition to monocyte chemotactic protein-1 (MCP-1) and interferon (IFN)γ chemokines were elevated (∼2-fold or more) in Cc1−/− mice (Table 4). Because CEACAM1 is not expressed in adipocytes, the rise in WAT-associated inflammation likely results from a hyperinsulinemia-driven increase in lipid production in liver and redistribution to WAT for storage (8).
Immunostaining of cross-sections of aortic roots with monocyte and macrophage antibody 2 (MOMA-2) showed inflammatory infiltration in the aortic wall of Cc1−/− mice at 6–12 mo of age (Fig. 2A and graph). qRT-PCR analysis of Cc1−/− aortae revealed increased mRNA levels of F4/80 (a macrophage marker), T cell markers (CD3+, CD4+, and CD8+T), and Toll-like receptor 2 and 4 (TLR-2 and -4) with their target gene, the TNFα inflammatory cytokine (Table 4).
Elevated Iκbα phosphorylation (Fig. 2B) indicates activation of the NF-κB inflammatory pathway (20), which induces the transcription of TNFα and the vascular cell adhesion molecule-1 (VCAM-1), as shown by increased mRNA (Table 4) and protein levels (Fig. 2B) in aortae derived from 6-mo-old Cc1−/− mice. Consistent with VCAM-1 playing an important role in attracting leukocytes to endothelium (6, 7), leukocyte adhesion to aortic wall was elevated in Cc1−/− mice (Fig. 2C).
Impaired endothelial membrane integrity in Cc1−/− aortae.
We then examined the level of VE-cadherin, a major transmembrane adhesion molecule of the endothelial adherent junctions that preserves endothelial integrity and regulates leukocyte adhesion (1). En face staining revealed elongated endothelial cells that were oriented longitudinally to the aortic axis in Cc1+/+ mice, as opposed to Cc1−/− aortae, in which the majority of endothelial cells were mostly round and with a cell border exhibiting more meander (Fig. 2D, arrows). Cc1−/− aortae also exhibited increased number and size of areas with separation of the VE-cadherin barrier and altered junction between neighboring endothelial cells (Fig. 2D, *). Because CEACAM1 regulates the activity of β-catenin (16), which in turn regulates VE-cadherin (18), it is possible that the decrease in β-catenin mRNA levels (Table 5) contributes to the aberrant intercellular junctions seen in Cc1−/− aortae.
In support of a role for CEACAM1 in regulating the expression of several factors relevant in endothelial barrier and vascular integrity in cultured endothelial cells (17), Cc1−/− aortae showed a more than threefold decrease in the mRNA (Table 5) of VEGF-A, its receptors VEGFR-1 and -2, and angiopoietins (Ang-1 and -2). Western blot analysis demonstrated an ∼40% reduction of aortic VEGFR-2 (Fig. 2E). In contrast, mRNA levels of VEGF-C and VEGF-D, which are mainly involved in the formation of lymphatic vessels, were not altered (Table 5).
Reduced endothelial relaxation in Cc1−/− aortae.
Because loss of NO initiates the development of proatherogenic risk factors, including increased vascular permeability, VCAM-1 levels and chronic inflammation of the arterial wall (15), we next examined NO levels in Cc1−/− aortae. Basal endothelial NO content was lower (by ∼60%) in the aortae of 6-mo-old Cc1−/− mice (Fig. 3A). This could in part result from a reduced eNOS level, as shown by Western (Fig. 3B) and immunohistochemical (Fig. 3C) analyses, and eNOS activation (phosphorylation) by Akt-dependent pathways (Fig. 3D). Reduction of eNOS levels in turn could result, in part, from decreased transcription by activated (dephosphorylated) Foxo1 (Fig. 3D) (27, 34) in these insulin-resistant mice (8).
Functional studies in aortic rings revealed an approximately twofold increase in the EC50 of endothelial NO-dependent relaxation in response to acetylcholine (0.170 ± 0.023 vs. 0.085 ± 0.003 μM in Cc1+/+, P < 0.05) but not in the EC50 of endothelial-independent relaxation in response to nitroprusside (P > 0.05) in Cc1−/− mice (Fig. 3E). This demonstrates a moderate decrease in basal and ligand-stimulated relaxation in Cc1−/− aortae.
Elevated lipid peroxidation and oxidative stress in Cc1−/− mice.
In addition to lower eNOS activation, increased catalytic NO consumption by a high oxidative environment could reduce NO levels, and elevation of NADPH oxidase activity contributes to increased macrophage adherence to the artery wall and lipid peroxidation. Thus, we examined whether Cc1−/− aortae develop oxidative stress. NADPH oxidase activity and mRNA of gp91-phox (Fig. 4, A and B) and Nox4 (Table 2) were elevated in aortic lysates derived from 6-mo-old Cc1−/− mice. Together with elevated concentrations of plasma TBARS and 8-isoprostane (Fig. 4, C and D), this suggests that Cc1−/− mice developed oxidative stress and lipid peroxidation.
Moreover, reduction in Niemann-Pick type C1 protein (NPC-1) represses GSH, an important mitochondrial defence system against the cytotoxic effect of TNFα. qRT-PCR analysis revealed a significant (∼2-fold) reduction in NPC-1 mRNA levels in Cc1−/− aortae (Fig. 4E) in association with a marked (∼6-fold) reduction in GSH (Fig. 4F). The decrease in GSH could have caused a robust response to the TNFα cytotoxic effect.
Potential underlying endothelial cell autonomous mechanisms.
Next, siRNA-reduction of CEACAM1 in bovine aortic endothelial cells (BAEC) was employed to determine the endothelial cell autonomous mechanisms underlying changes in VCAM-1 and NO levels in Cc1−/− aortae. Decreasing CEACAM1 protein levels by >90% (Fig. 5A) induced the transcriptional activity of NF-κB, as inferred from elevated Iκb-α phosphorylation (Fig. 5B), to elevate VCAM-1 protein levels about twofold.
Because acetylcholine-mediated relaxation requires eNOS phosphorylation, which is regulated, among other things, by insulin (41), we next investigated whether Ceacam1 deletion in endothelial cells adversely affects insulin's regulation of eNOS phosphorylation and NO production. As previously reported (24), Western blot analysis revealed that siRNA knockdown of CEACAM1 (sibCC1) in BAEC did not significantly affect basal phosphorylation (activation) of the Akt/eNOS pathway and, consequently, basal NO level relative to control scrambled RNA cells (Scr) (Fig. 5C). However, it impeded the ability of insulin to activate the Akt/eNOS pathway and stimulate NO production (Fig. 5C). Coimmunoprecipitation experiments revealed binding of CEACAM1 to the Src homology tyrosine phosphatase-2 (SHP2) in Scr cells, as previously shown (14), but not in sibCC1 cells (Fig. 5D). This caused a decrease in SHP2 binding to insulin receptor substrate-1 (IRS-1) in response to insulin and, consequently, an increase in insulin-stimulated tyrosine phosphorylation of IRS-1 in Scr but not sibCC1 cells, in which insulin induced an approximately twofold higher IRS-1 binding to SHP2 than in Scr cells (Fig. 5D). As summarized in Fig. 5E, CEACAM1 binding to SHP2 caused its sequestration, reciprocally limiting IRS-1 binding to SHP2 and its dephosphorylation. This led to a higher activation of IRS-1 and the downstream PI3K/Akt/eNOS pathway in response to insulin. Thus, CEACAM1 appears to play an important role in insulin's regulation of NO production via the Akt/eNOS pathway in aortic endothelial cells.
Mice with global Ceacam1 deletion (Cc1−/−) display impaired endothelial barrier resulting from increased basal endothelial permeability (24). They also exhibit hyperinsulinemia at 2 mo of age resulting from impaired insulin clearance (8, 28, 39) in addition to liver steatosis and fibrosis with predisposition to diet-induced nonalcoholic steatohepatitis (12). Because hyperinsulinemia drives global ectopic fat accumulation (4), we investigated whether Ceacam1 deletion also causes fat deposition in large vessels together with an associated rise in inflammatory infiltration (3). We herein present a first in vivo demonstration linking global Ceacam1 deletion to the formation of small intimal plaque-like lesions exhibiting subendothelial deposition of lipids and fibrotic materials with increased inflammatory infiltration and leukocyte adhesion, and elevated oxidative stress in large vessels in the absence of proatherogenic hypercholesterolemia. Eventually, these vascular derangements are accompanied by a reduced endothelial NO-dependent relaxation in aortic rings.
The vascular abnormalities in Cc1−/− aortae are associated with reduced basal NO level, which could be caused by multiple mechanisms, including lower eNOS levels (likely resulting from Foxo1-dependent downregulation of eNOS transcription) (27, 34) and eNOS inactivation via an Akt-dependent pathway. Both events could, in turn, be caused, at least partly, by the systemic insulin resistance state resulting from impaired hepatic insulin clearance in these mice (8). Hyperinsulinemia and elevated plasma NEFA could contribute to systemic regulation of eNOS level/activation, although the relative composition of NEFA in these mice is not commonly associated with marked vascular anomalies (21, 37). For instance, elevated plasma NEFA could reduce eNOS mRNA stability by increasing TLR-2/4-mediated TNFα transcription (30, 40), which is elevated in Cc1−/− aortae. On the other hand, reduction in basal Akt activation in parallel to that of eNOS precludes a significant role for a NEFA-induced ROS-mediated (9), Akt-independent (32) pathway regulating NO production in Cc1−/− aortae.
Normal basal eNOS activation and NO level in BAEC with siRNA-mediated reduction of CEACAM1 supports a role for a negative systemic effect on endothelial NO production. However, impediment of insulin-induced Akt/eNOS-mediated NO production in these cells demonstrated an additional cell autonomous negative effect of CEACAM1 deletion on insulin signaling. This appears to be mediated by reduction in CEACAM1 sequestration of SHP2 phosphatase and a reciprocal increase in its binding to IRS-1, causing IRS-1 dephosphorylation and deactivation of downstream Akt signaling pathways. Because CEACAM1 deletion caused a similar negative effect on Akt/eNOS-mediated NO production in response to VEGF-A in BAEC (24), CEACAM1, a common substrate of insulin receptor and VEGFR-2 (22, 24), could link insulin to VEGF actions with regard to Akt2/eNOS or Akt1/eNOS activation, respectively (32), in an endothelial cell autonomous manner.
Global Ceacam1 deletion induced leukocyte adhesion to the aortic wall, which could be partly mediated by the elevated level of the VCAM-1 leukocyte adhesion molecule. Several endothelial cell autonomous and nonautonomous (systemic) factors could contribute to this molecular event. For instance, siRNA reduction of CEACAM1 induced VCAM-1 expression in BAEC by activating NF-κB-mediated transcriptional activity. Systemic factors such as hyperinsulinemia could also be involved in VCAM-1 regulation. For instance, hyperinsulinemia induces fat production, which alters the inflammatory milieu (3), causing macrophage recruitment and a rise in TNFα and other proinflammatory cytokines. TNFα induces expression of VCAM-1 directly (36, 42) and indirectly, by reducing the repressive effect of NO (7). In the presence of low GSH, the response to TNFα activation of IκKβ/NF-κB-dependent oxidative stress and inflammatory pathways is more robust (5). Given that oxidative stress and inflammation were also detected in the liver of Cc1−/− mice (12), we postulate that hyperinsulinemia, caused by impaired hepatic insulin clearance, links lipid accumulation, inflammation, and oxidative stress to global Ceacam1 deletion.
Despite developing a constellation of proatherogenic factors, such as hyperinsulinemia, oxidative stress, reduced NO production, and endothelial NO-dependent relaxation, in addition to increased vascular permeability and chronic inflammation, Cc1−/− mice exhibited restricted plaque-like lesions. These morphological abnormalities were limited to the formation of small subintimal foci of lipid accumulation and macrophage infiltration, likely owing to the absence of proatherogenic cholesterolemia (38) as well as to Foxo1 activation in aortae, which has been associated with restricted atheroma (35).
In summary, the present studies provide the first in vivo evidence that Ceacam1 deletion causes the formation of small plaque-like intimal lesions accompanied by adventitial reactions in large vessels and that decreased insulin-stimulated NO production in aortic endothelial cells could contribute to altered ligand-induced endothelial cell-dependent relaxation along the large vessel wall. Given the early onset of morphological lesions in Cc1−/− aortae, further studies using tissue-targeted Ceacam1 deletion along the liver-endothelial cell axis are needed to understand whether these are caused by deletion of Ceacam1 in endothelial cells or are related to hyperinsulinemia caused by impaired insulin clearance in liver. Nonetheless, the similarity of this vascular phenotype to that in the liver (fatty liver, inflammation, oxidative stress, and fibrosis) and progression of the hepatic phenotype to advanced NASH in response to high fat intake (12) predict a progression of vascular abnormalities to more robust vascular lesions of earlier onset than their wild-type animals, if mice were fed an atherogenic diet. However, this would not promote a role for hyperinsulinemia distinct from hypercholesterolemia in the pathogenesis of early vascular abnormalities, which is the most remarkable attribute of the current findings. Thus, the Cc1−/− mouse provides a unique in vivo demonstration of distinct CEACAM1-dependent hepatic insulin clearance linking hepatic to early macrovascular abnormalities.
This work was supported by grants from the National Institutes of Health: R01 DK-054254, R01 DK-083850, and R01 HL-112248 (S. M. Najjar), P01 HL-36573 (S. V. Pierre and S. M. Najjar), R01 HL-111877 (G. Vazquez), R01 HL-45095 (I. J. Goldberg), and R01 DK-064344 (R. Scalia); American Heart Association-Great Rivers Affiliate (075100B, G. Heinrich); US Department of Agriculture (USDA 38903-19826, S. M. Najjar); Canadian Institutes of Health Research (CIHR MOP-86582, N. Beauchemin), and Deutsche Forschungsgemeinschaft (DFG ER 276 4-4 and DFG TI 690 2-1 ER 276 7-1) (S. Ergün).
No conflicts of interest, financial or otherwise, are declared by the author(s).
Author contributions: S.M.N., G.H., S.V.P., R.B., A.A.J., G.V., R.S., and S.E. conception and design of research; S.M.N., K.J.L., L.R., S.V.P., R.B., A.A.J., E.L., G.V., I.J.G., R.S., and S.E. analyzed data; S.M.N., K.J.L., S.L.A., G.V., R.S., and S.E. interpreted results of experiments; S.M.N., S.L.A., A.P., L.R., and S.E. prepared figures; S.M.N., K.J.L., S.L.A., and S.E. drafted manuscript; S.M.N., G.V., R.S., and S.E. edited and revised manuscript; S.M.N., K.J.L., S.L.A., A.P., L.R., M.K.K., S.R., H.T.M., C.K.R., S.G.L., G.H., S.V.P., R.B., V.K., A.A.J., E.L., G.V., I.J.G., N.B., R.S., and S.E. approved final version of manuscript; K.J.L., S.L.A., A.P., L.R., M.K.K., S.R., H.T.M., C.K.R., S.G.L., G.H., V.K., E.L., I.J.G., N.B., and R.S. performed experiments.
We thank M. Kopfman and J. Kalisz (Najjar laboratory) as well as D. Schünke (Institute of Anatomy, University Hospital Essen), and G. Landesberg (Temple University) for excellent technical assistance. We also thank A.-L. Nouvion (Beauchemin laboratory), K. Preston (Scalia laboratory), and H. Jastrow and B. B. Singer (University Hospital Essen) for their technical assistance and scientific discussions. We also thank Dr. Z. A. Shah (University of Toledo College of Pharmacy) for the use of the Synergy H1 Hybrid Multi-Mode Microplate Reader, and D. Accili (Columbia University) for helpful scientific discussions.