Role for hepatic CEACAM1 in regulating fatty acid metabolism along the adipocyte-hepatocyte axis[S]

Carcinoembryonic antigen-related cell adhesion molecule 1 (CEACAM1) regulates insulin sensitivity by promoting hepatic insulin clearance and mediating suppression of fatty acid synthase activity. Feeding C57BL/6J male mice with a high-fat (HF) diet for 3–4 weeks triggered a >60% decrease in hepatic CEACAM1 levels to subsequently impair insulin clearance and cause systemic insulin resistance and hepatic steatosis. This study aimed at investigating whether lipolysis drives reduction in hepatic CEACAM1 and whether this constitutes a key mechanism leading to diet-induced metabolic abnormalities. Blocking lipolysis with a daily intraperitoneal injection of nicotinic acid in the last two days of a 30-day HF feeding regimen demonstrated that white adipose tissue (WAT)-derived fatty acids repressed hepatic CEACAM1-dependent regulation of insulin and lipid metabolism in 3-month-old male C57BL/6J mice. Adenoviral-mediated CEACAM1 redelivery countered the adverse metabolic effect of the HF diet on insulin resistance, hepatic steatosis, visceral obesity, and energy expenditure. It also reversed the effect of HF diet on inflammation and fibrosis in WAT and liver. This assigns a causative role for lipolysis-driven decrease in hepatic CEACAM1 level and its regulation of insulin and lipid metabolism in sustaining systemic insulin resistance, hepatic steatosis, and other abnormalities associated with excessive energy supply.


Insulin and glucose tolerance tests
Mice were fasted for 6 h before being injected intraperitoneally with Human Regular Insulin (0.75 U/kg BW, Novo Nordisk) (for insulin tolerance) or glucose (1.5 g/kg BW of 50% dextrose solution, Dextrose Injection, USP). Glucose was measured in tail blood at 0-180 min postinjection.

Indirect calorimetry analysis
Mice were individually caged for 5 days (CLAMS system, Columbus Instrument), and their spontaneous physical activity was determined on the X-(locomotor), Y-(ambulatory), and Z-(standing) axes (17). Oxygen consumption (VO 2 ), CO 2 production (VCO 2 ), and heat production were sampled every 20 min and normalized to lean mass. The respiratory exchange rate (RER) was calculated as the VCO 2 /VO 2 ratio. Heat production was calculated as Cv × VO 2 normalized to lean body mass; with Cv (calorific value) being: 3.815 + 1.232 × RER. Data were represented as mean ± SEM of light (700 h-1900 h) and dark (1900 h-700 h) cycles.

Fatty acid synthase activity
As previously described (11), livers were homogenized in buffer containing 20 mM Tris (pH 7.5), 1.0 mM EDTA, 1.0 mM DTT, and phosphatase and protease inhibitors (11). Following centrifugation at 12,500 g for 30 min, 10 l of the supernatant was added to 65 l of the reaction mix. This contains: 166.6 M acetyl-CoA, 100 mM potassium phosphate (pH 6.6), 0.1 Ci [ 14 C] malonyl-CoA (Perkin Elmer, Waltham, MA), and 25 nmol malonyl-CoA, in the absence (negative controls) or presence of 500 M NADPH (Sigma Aldrich). A 1:1 chloroform-methanol solution was used to stop the reaction. Following centrifugation, the supernatant was vacuum-dried, and the pellet resuspended in 200 l of water-saturated butanol. To reextract the upper layer, 200 l of ddH 2 O was added followed by vortexing and spinning for 1 min. The butanol layer was extracted, dried, and counted for the incorporation of radiolabeled malonyl-CoA into palmitate. Fasn activity was calculated as cpm of [ 14 C] incorporated per microgram of cell lysates, and the protein concentration was determined by Bio-Rad protein assay.
(Fasn) activity in response to acute rise in insulin in the first hour of refeeding following an overnight fast (11). This positions CEACAM1 to contribute to the regulation of fatty acid oxidation until glycogen repletion is complete (12). Mice with null deletion of Ceacam1 (Cc1 / ) or with liver-specific inactivation of CEACAM1 develop hyperinsulinemia, caused by impaired insulin clearance, followed by insulin resistance, hepatic steatosis, and visceral obesity (13)(14)(15)(16). Normal acute-phase insulin release to glucose and intact -cell function support the observation that chronic hyperinsulinemia is mainly driven by impaired insulin clearance in Cc1 / mutants (15).
Studies in mice (17), dogs (18), and humans (19) demonstrated that defective hepatic insulin clearance is involved in diet-induced insulin resistance. Providing a Western-style diet caused rapid hepatic insulin resistance in healthy young South Asian men but not Caucasians, in association with altered insulin clearance in the Asian group (19). We have shown that high fat (HF) reduced hepatic CEACAM1 level by >60% and impaired insulin clearance in WT mice within 3 weeks to introduce metabolic abnormalities that were prevented by forced transgenic fat-inducible CEACAM1 overexpression in liver (17). The clinical implication of our studies is underscored by the observed low hepatic CEACAM1 level in obese, insulinresistant subjects with fatty liver disease (20).
Dietary fatty acids can reach the liver via chylomicrons in addition to lipolysis in WAT. In uncomplicated human obesity with low-grade insulin resistance, lipolysis-derived fatty acids are mainly removed by oxidation in the liver (21). Lipolysis occurs within few days of the initiation of HF intake, owing to dysregulated hypothalamic control, even in the absence of insulin resistance in WAT (22). This could cause rapid hepatic insulin resistance (portal hypothesis) via several mechanisms (23), including protein kinase C-mediated pathways (24). With persistent nutritional burden, hepatic lipotoxicity and systemic insulin resistance develop in parallel to the progression of a proinflammatory state in WAT (25).
Release of fatty acids during an acute intralipid infusion reduced hepatic CEACAM1 levels and impaired insulin clearance (24). Thus, we herein investigated the role of lipolysis-derived fatty acids in the suppression of CEACAM1 by HF diet and assessed the significance of this mechanism in diet-induced metabolic abnormalities.

Adenovirus preparation and cell transduction
The construction of the adenovirus vectors harboring WT and the phosphorylation-defective serine503-to-alanine (S503A) rat CEACAM1 mutant was described (15). Adenoviruses were produced in 293AD packaging cells (Agilent Technologies) following LipofectAMINE-mediated transfection (Invitrogen) and concentrated by ultracentrifugation in cesium chloride gradient (31). Adenoviral-mediated CEACAM1 overexpression was confirmed by transducing primary mouse hepatocytes prepared from C57BL/6J mice at 1 × 10 7 particles/well of a 6-well plate for 36 h. CEACAM1 level was determined by immunoblotting with anti-rat CEACAM1 antibody (24).

Tail-vein injection of recombinant adenovirus to redeliver CEACAM1 to the liver
Two-month-old C57BL/6J mice were HF fed for 20 days before being subjected to adenoviral infection by tail-vein injection of 1 × 10 10 particles/mouse and housed individually for 21 days. The control group was injected with adenovirus construct bearing green fluorescent protein control cDNA (Ad-GFP) at the same viral titer. Insulin and glucose tolerance were performed 13 and 16 days postinjection, as described below. Mice were subjected to indirect calorimetry before being euthanized at 21 days postinjection.

Real-time quantitative RT-PCR
Total RNA was isolated with PerfectPure RNA Tissue Kit (5 PRIME Inc.) and cDNA was synthesized by iScript cDNA Synthesis Kit (BIO-RAD), using 1 g of total RNA and oligodT primers (supplemental Table S1). cDNA was evaluated with quantitative RT-PCR (qRT-PCR; StepOne Plus, Applied Biosystems), and mRNA was normalized to 18S or Gapdh, as previously described (17).

Statistical analysis
Data were analyzed by one-way ANOVA with Bonferroni correction, using GraphPad Prism 6 software. P < 0.05 was statistically significant.

NA inhibits lipolysis and causes fat accumulation in WAT
To investigate the role of adipocyte-derived fatty acids in diet-induced metabolic abnormalities, mice were HF fed and treated with NA or saline (S) in the last 2 days of feeding. Relative to RD, HF induced BW, visceral and total fat mass, including brown adipose tissue (BAT) mass with a reciprocal decrease in lean mass ( Table 1, HF-S vs. RD-S).

Gomori-trichrome staining
Trichrome stain was performed on formalin-fixed adipose tissue using the Thermo Scientific Richard-Allan Scientific Chromaviewadvanced Testing (30). Adipose tissue was fixed in 10% formalin and replaced by 70% ethanol before undergoing blocking in paraffin. Sections were deparaffinized at 60°C and hydrated in deionized water. Slides were then stained with Bouin's Fluid at 56°C for 45 min, followed by rinsing in deionized water to remove the yellow color. Slides were placed in Working Weigert's Iron Hematoxylin at room temperature for 10 min followed by Trichrome Stain for 15 min, dehydrated sequentially in 1% acetic acid solution for 1 min, 95% ethanol for 30 s, and 100% ethanol for 1 min (twice). Sections were cleared in xylene 3 times for 1 min each and mounted.

Immunofluorescence
Whole WAT was formalin-fixed for 24 h, transferred to Dulbecco's PBS (Sigma-Aldrich) at 4°C, and permeabilized in 1% Tri-tonX-100/PBS for 15 min before staining macrophages with rat anti-mouse F4/80 (Invitrogen) and detecting with donkey anti-rat IgG-conjugated to Alexa-Fluor488. Tissues were incubated with primary antibodies overnight at 4°C then washed 3× with PBS-Tween20 before applying the secondary stain for 2 h at room temperature and washing 3× in PBS-Tween. Stained samples were then counterstained for 25 min at room temperature with 5 µM BODIPY-558/568 (Molecular Probes) and Hoechst stain/4',6-diamidino-2-phenylindole (DAPI) (Invitrogen) to visualize lipid and nuclei, respectively. Samples were placed on a coverslip and imaged using a Leica-TCS-SP5 laser-scanning microscope equipped with conventional solid-state and a Ti-sapphire tunable multiphoton laser (Coherent). Images were acquired in the 3D-XYZ plane in 4 µm steps with a 63× objective (NA-0.70) using the sequential scan mode to eliminate any spectral overlap in individual fluorophores. Specifically, Alexa-Fluor488 was excited at 488 nm with collection at 500-558 nm. The BODIPY-558/568 dye was excited at 561 nm and collected at 567-609 nm. Selected images are a 2D-representation of the 3D-laser scanning confocal microscopy-image stack as labeled.

Primary hepatocytes isolation and fatty acid treatment
Hepatocytes were isolated by perfusing liver (1 ml/min) with Collagenase-Type II (1 mg/ml) (Worthington) (13). Cells were dispensed in Williams-E complete media (Gibco) containing 10 mM lactate, 10 nM dexamethasone, 100 nM insulin (Sigma Aldrich), 10% FBS, and 1% penicillin-streptomycin (Gibco). Cells were plated onto 12-well plates at 2.5 × 10 5 /well and incubated at 37°C in Williams-E medium for 24 h before switching to phenol red-free Williams-E medium-supplemented with 10% dialyzed IL-6, and IFN- (supplemental Table S2). NA further induced the mRNA levels of these markers by ~2-to 3-fold. Immunofluorescence analysis of F4/80 (green) in WAT showed evidence of macrophage inflammation in HF-S and HF-NA compared with RD-S mice ( Fig. 1C-viii and ix vs. vii). In contrast to RD-fed mice ( Fig. 1C-x), HF induced macrophage staining around the adipocytes (gray), but to a higher extent in NA-treated versus saline-treated mice ( Fig. 1C-xii vs. xi), indicating increased macrophage inflammation by NA. Despite fewer macrophages; however, HF-S mice displayed higher plasma NEFA levels than HF-NA (Table 1).

NA reverses diet-induced insulin resistance
Consistent with our previous report (17), HF caused insulin resistance, as indicated by fed hyperglycemia (Table  1) and intolerance to exogenous insulin in HF-S versus RD-S [ Fig. 2A and accompanying graph of the area under the curve (AUC)]. Both glucose levels (Table 1)

NA restores insulin clearance in HF-fed mice
Western blot analysis of liver lysates revealed reduction of CEACAM1 (CC1) protein level by ~>60% in HF-S versus RD-S mice ( Fig. 2C-i). This was restored by NA treatment (Fig. 2C-i, HF-NA vs. RD-S). Subsequently, this normalized insulin clearance, assessed by the higher steady-state C-peptide/insulin molar ratio (Table 1; HF-NA vs. HF-S) and Consistently, HF increased fasting plasma NEFA (Table 1) and the mRNA level of hormone sensitive lipase (Hsl) in WAT (12.7 ± 0.56 in HF-S vs. 6.77 ± 1.07 in RD-S mice; P < 0.05). As expected, NA treatment restored plasma NEFA (Table 1) and Hsl mRNA levels (3.75 ± 0.60 in HF-NA vs. 12.7 ± 0.56 in HF-S mice; P < 0.05) but maintained the gain in visceral obesity and BW in response to HF (Table 1). Consistently, plasma leptin levels remained elevated in HF-NA relative to RD-S mice (Table 1).
Histological analysis of hematoxylin and eosin (H and E)stained WAT sections (Fig. 1A) confirmed the maintenance of adipocytes' expansion to accommodate fat storage in WAT derived from HF-NA mice (adipocyte size of 2,078 ± 86 µm 2 in HF-NA vs. 1,848 ± 91 in HF-S and 576 ± 23 in RD-S).

NA reverses diet-induced fibrosis in WAT
Because visceral obesity is associated with increased fibrosis in WAT in rodents (32,33) and humans (34), we then examined whether NA treatment modulates HF-induced fibrosis in WAT. HF induced the mRNA levels of profibrogenic effectors: endotrophin/Col63, a protein that promotes metabolic derangement and fibrosis in adipose tissue (33), -smooth muscle actin (-Sma), and transforming growth factor- (Tgf-) (supplemental Table S2; HF-S vs. RD-S). In contrast, HF lowered the mRNA level of Smad7 that inhibits TGF- signaling (35, 36) (supplemental Table  S2; HF-S vs. RD-S). NA treatment reversed the effect of HF, with a lesser extent in Col63 mRNA relative to other markers (supplemental Table S2). This translated into limiting HF-induction of collagen deposition in WAT by NA, as indicated by Trichrome-C staining (Fig. 1B).

NA induces macrophage inflammation in WAT of HF-fed mice
qRT-PCR analysis in WAT revealed induction of mRNA of macrophages (CD68 and F4/80) and TNF- by HF, but without significantly changing the expression of other proinflammatory markers such as interleukin (IL)-1, Male mice (n = 6-10/feeding/treatment group) were fed RD or HF diet for 30 days. In the last 2 days of feeding, mice were treated with NA, and their body fat composition evaluated by nuclear magnetic resonance (NMR; Bruker Optics). Except for leptin and fed blood glucose levels, mice were fasted overnight for 18 h and their tissues extracted to assess visceral and brown fat mass relative to BW, and blood was drawn to determine plasma levels of NEFA, triacylglycerol, Ins, and C-peptide levels. Values are expressed as mean ± SEM. lowered plasma insulin levels (Table 1), in parallel to restoring insulin sensitivity ( Fig. 2A).
Supporting a direct effect of fatty acids on hepatic CEACAM1 expression, treating primary hepatocytes with 0.1 mM of a fatty acid mixture (FA) with a composition resembling that of the dietary fat reduced CEACAM1 protein level (Fig. 2C-ii). Consistent with our previous observations (12), the downregulatory effect of FA appears to be mediated by Ppar activation, as demonstrated by the ability of FA (0.1 mM) to reduce the transcriptional activity of Ceacam1 WT promoter (1,100), but not of the construct harboring a mutation of the sole active Ppar response element (PPRE/RXR) located between nucleotides -557 and -543 (Fig. 2D).

NA restores lipid metabolism in the liver of mice fed an HF diet
HF increased fat accumulation in liver, as shown by H and E analysis (Fig. 3A, HF-S vs. RD-S) and the high hepatic triacylglycerol content (Table 1). While this could be attributed, at least in part, to reduced fatty acid flux from the adipose tissue to the liver, we investigated whether NA could also affect fatty acid synthesis and -oxidation in liver. Given the 3-fold increase in hepatic fatty acid oxidation in HF-S ( Fig. 3B; 3.31 ± 0.22 vs. 1.07 ± 0.07 nmol/g/ min in RD-S; P < 0.05), hepatic steatosis could in part, result from elevated Fasn protein levels ( Fig. 3C-ii), and activity ( Fig. 3C-i) in response to basal hyperinsulinemia ( Fig.  2B-i) (37). NA treatment significantly reversed fat accumulation in liver ( Fig. 3A; HF-NA vs. HF-S) and hepatic triacylglycerol levels ( Table 1; HF-NA vs. HF-S) in parallel to reversing basal hyperinsulinemia (Fig. 2B-i) and Fasn level (Fig. 3C-ii). Consistent with a previous report (11), the acute rise of insulin release at refeeding (RF) downregulates Fasn activity relative to fasting (F) in RD-S mice (Fig.  3C-i). This occurred in parallel to induced CEACAM1 phosphorylation as shown by immunoblotting with antiphospho-CEACAM1 (-pCC1) antibody ( Fig. 3C-ii), followed by its increased binding to Fasn, as shown by its detection in coimmunoprecipitation (Co-IP) with the Fasn immunopellet from RD-S mice (Fig. 3C-iii). In contrast, under chronic hyperinsulinemic conditions and in response to HF, insulin failed to phosphorylate CEACAM1 and induce its binding to Fasn to decrease its activity in HF-S. In HF-fed mice, NA restored refeeding-induced CEACAM1 phosphorylation (Fig. 3C-ii) and its binding to Fasn (Fig. 3C-iii) to downregulate its activity (Fig. 3C-i). Together with maintaining the rise in fatty acid -oxidation ( Fig. 3B; 4.39 ± 0.52 in HF-NA vs. 1.07 ± 0.07 nmol/g/min in RD-S; P < 0.05), NA's ability to restore the suppressive effect of insulin on Fasn activity could counter the effect of HF diet on fat deposition in liver. Fig. 1. Effect of NA on fat accumulation, fibrosis and inflammation in WAT. A: H and E staining of the WAT from mice fed ad libitum either with an RD or an HF diet for 30 days. In the last 2 days of feeding, RD-fed mice were administered a once daily intraperitoneal injection of saline (S) while HF-fed mice were treated with either saline or NA (n > 6 mice/feeding/treatment group). B: Trichrome C (Trich-C) stain was used to assess fibrosis in WAT. Black staining indicates nuclei. Cytoplasm and fibers are in red. Collagen is shown in blue. Representative images from three sections/ mouse are shown (20× magnification). C: Whole WAT was stained with BODIPY 558/568 to detect lipid (gray) for staining of adipocytes and with anti-F4/80 to detect macrophages (green). Nuclei from cells, including macrophages and adipocytes, were stained with DAPI (blue). All images were captured using laser scanning confocal microscopy with a 63× objective and are 2D projections of a 3D image z-stack. Increased DAPI staining indicating cellular inflammation is observed in HF-NA (iii) compared with RD-S (i) or HF-S (ii) treated animals. F4/80 staining was elevated in HF-NA (ix) compared with HF-S (viii) indicating increased inflammation in HF-NA treated mice, while there was no F4/80 staining in RD-S. The overlay Fig. 2. Effect of NA on metabolism and insulin signaling. A-i: RD-fed mice were injected with saline (RD-S, white square) and HF-fed mice with saline (HF-S, black square) or NA (HF-NA, gray circle) before intraperitoneal insulin tolerance test was carried out to measure glucose disposal at 0-180 min. ii: The AUC was calculated. n = 6-10 mice/feeding/treatment group. Values are expressed as mean ± SEM. * P < 0.05 versus RD-S (white bar); † P < 0.05 HF-NA (gray bar) versus HF-S (black bar). B: At the end of feeding/treatment period, mice were fasted overnight (F, white bar) and then refed (RF) for 7 h to measure plasma insulin level (i). Western analysis in liver (ii) and WAT lysates (iii) was carried out by immunoblotting (Ib) with -phospho-Akt (-pAkt) antibodies, followed by reimmunoblotting (re-Ib) with -Akt antibody to normalize per total Akt loaded. Gels represent 2 independent experiments on 2 mice/feeding/treatment group. C-i: Liver lysates were analyzed by Western Blotting, probed with -CEACAM1 (-CC1), followed by -Actin antibody. ii: Hepatocytes were isolated from WT mice and treated with either BSA or with BSA-coupled fatty acid mixture (FA) (0.1 mM), lysed, and analyzed by Western blot, using immunoblotting (Ib) with -CC1, followed by -GAPDH antibody for normalization. Gel represents three independent experiments. D: Constructs from the mouse promoter bearing the WT sequence from nt 1,100 to +30 or a block mutation of the only active PPRE/RXR site located between nt 557 and 543 were generated, subcloned into the pGL4.10 promoterless plasmid before their Luciferase activity in response to ethanol (Veh) (white bars) or FA (0.1 mM) (black bars) was determined in mouse Hepa1-6 cells. As a negative control, cells were transfected with the empty pGL4.10. The experiment was performed in quadruplet several times. Luciferase light units were expressed as mean ± SD in relative light units (RLU). The graph represents typical results from several separate experiments. * P < 0.05 versus Veh-treated cells (black vs. white bar per each construct).

Hepatic adenoviral-redelivery of CEACAM1 reverses dietinduced metabolic abnormalities
We then tested whether reduced hepatic CEACAM1 is critical to diet-induced metabolic abnormalities. To this end, we rescued CEACAM1 in the liver using adenoviralmediated redelivery (Ad) and examined whether this reversed the abnormal diet-induced metabolic phenotype.
Mice fed HF for 21 days were injected with comparable titers of Ad-GFP control, adenovirus construct bearing WT rat CEACAM1 cDNA (Ad-rWT), and the adenovirus construct bearing the S503A phosphorylation-defective CEACAM1 mutant cDNA (Ad-rSA). Immunoblotting tissue lysates 3 days postadenoviral injection with -mouse CEACAM1 antibody detected endogenous CEACAM1 CEACAM1 expression was sustained for at least 21 days, as demonstrated by Western (Fig. 4A-ii) and qRT-PCR (Fig.  4A-iii) analyses of liver lysates. Per our earlier report (17), endogenous mouse Ceacam1 mRNA (Fig. 4A-iii) and CEACAM1 protein (Fig. 4A-ii) content were reduced by >60% by HF. In parallel, HF impaired insulin clearance, as assessed by steady-state C-peptide/insulin molar ratio, and hyperinsulinemia ( Table 2, HF-GFP vs. RD-GFP). This caused fed hyperglycemia (Table 2), and intolerance to insulin ( Fig. 4B-i, blue vs. black lanes and AUC bars) and glucose (Fig. 4B-ii), as assessed on day 13 and 16 postinjection, respectively. Ad-rWT delivery restored insulin clearance together with plasma insulin and C-peptide levels in HF-fed mice ( Table 2, HF-rWT vs. RD-GFP). It also restored insulin and glucose tolerance (Fig. 4B-i and ii, respectively, green vs. black lanes and AUC bars) and blood glucose levels ( Table 2). In contrast, injecting Ad-rSA CEACAM1 mutant (red) failed to restore these metabolic factors ( Table 2 and Fig. 4B).
Consistent with the positive effect of hyperinsulinemia on hepatic de novo lipogenesis (37), HF-GFP mice manifested higher mRNA levels of sterol regulatory elementbinding protein-1c, the master transcriptional regulator of lipogenic genes, than RD-GFP (supplemental Table S3). This increased Fasn mRNA (supplemental Table S3) and protein content (Fig. 5A-i) in the liver and subsequently, enzymatic activity (Fig. 5A-ii).
Additionally, HF induced mRNA levels of genes implicated in fatty acid transport, such as Fatp-1 and Fatp-4 (supplemental Table S3, HF-GFP vs. RD-GFP). In contrast, HF lowered carnitine palmitoyltransferase-1 (Cpt-1) mRNA levels (supplemental Table S3). Consistent with the role of CPT-1 in transporting fatty acids to mitochondria for oxidation, hepatic mRNA expression (supplemental Table  S3) and plasma levels of FGF21 (Table 2) were 2-fold lower in HF-GFP than RD-GFP mice. Injecting Ad-rWT, but not Ad-rSA, normalized the level of these genes (supplemental Table S3) as well as plasma FGF21 in HF-rWT, but not HF-rSA mice (Table 2), further supporting restored fatty acid -oxidation by Ad-rWT redelivery. Together, this yielded limited fat accumulation in the liver of HF-rWT relative to HF-GFP and HF-rSA mice, as shown by histological analysis of the H and E stain of the liver (Fig. 5B), and by hepatic triacylglycerol content ( Table 2). Consistent with redistribution of substrates to WAT, adipocytes' expansion was also restricted in HF-rWT by comparison to HF-GFP mice (Fig. 5C), as opposed to mice injected with Ad-rSA mutant in which adipocytes' expansion was comparable to that in Ad-GFP controls (Fig. 5C).
Consistent with enhanced inflammation by increased fat accumulation (38), qRT-PCR analysis showed induced proinflammatory (F4/80 and Tnf-), and profibrotic genes (-Sma and Col63) in the liver (supplemental Table S3) of HF-GFP versus RD-GFP mice, all of which were reversed by Ad-rWT, but not Ad-rSA injection. Of note, qRT-PCR analysis showed comparable mRNA levels of F4/80 and TNF- in the liver of HF-fed mice 3 days postinjection of Ad-rWT and Ad-rSA CEACAM1 (supplemental Table S4). This indicates that the changes in the inflammatory signals Fig. 3. Effect of NA on lipid metabolism in liver. A: Liver histology was assessed by H and E stained sections (n = 5 mice/feeding/treatment group). HF-S mice exhibit microvesicular lipid infiltration alternating with normal liver parenchyma. HF-NA manifested a normal histology like that of RD-S. Representative images from three sections/mouse are shown (20× magnification). B: Hepatic fatty acid oxidation was determined in 4 mice/feeding/treatment group. Experiments were repeated at least three times. Values are expressed as mean ± SEM. * P < 0.05 versus RD-S. C-i: Fasn enzymatic activity was measured in the liver by [ 14 C]malonyl-CoA incorporation on mice that had been fasted overnight (F) and refed for 7 h (RF). n = 5/feeding/treatment group. Each assay was performed in triplicate. Values are expressed as mean ± SEM. * P < 0.05 versus F/each treatment group; † P < 0.05 versus F or RF in other treatment groups. ii: Western analysis of Fasn (normalized to GAPDH) and phosphorylated CEACAM1 normalized to total CEACAM1 in liver lysates. iii: Co-IP analysis of Fasn binding to CEACAM1 was carried out by immunoprecipitating (IP) with -Fasn antibody followed by immunoblotting (Ib) with -CC1 antibody. Gels represent more than two separate experiments. protein expression in liver, with insignificant expression in WAT and skeletal muscle (Fig. 4A-i, mCC1), as expected from the limited expression of CEACAM1 in these tissues (9). In contrast, immunoblotting with -rat CEACAM1 antibody revealed restricted expression of rat CEACAM1 (rCC1) to the liver but not WAT or muscle of mice injected with Ad-rWT or Ad-rSA, but not Ad-GFP. Transgenic rat to RD-GFP mice (Fig. 6A). Indirect calorimetry analysis over a 24 h period revealed lower energy expenditure (Fig.  6B, heat generation), O 2 consumption (VO 2 ) (not shown), CO 2 production (VCO 2 ) (not shown), calculated RER (Fig. 6C), and spontaneous locomotor activity along the combined X, Y, Z axes (Fig. 6D) in HF-GFP relative to RD-GFP mice. Injecting mice with Ad-rWT, but not Ad-rSA CEACAM1, reversed the negative effect of HF on energy expenditure and the spontaneous locomotor activity. Preserved energy expenditure and physical activity could contribute to lower fat mass and BW in HF-rWT than HF-rGFP and HF-rSA mice ( Table 2).

DISCUSSION
HF diet causes a progressive decline in hepatic CEACAM1 levels reaching 60% after 3 weeks, at which point, it impairs insulin clearance to cause hyperinsulinemia with attendant insulin resistance and hepatic steatosis (17). Conversely, protecting CEACAM1 by fat-inducible liverspecific CEACAM1 overexpression restricted the metabolic derangement caused by HF diet, including visceral obesity and disturbed energy balance (17). While these studies 21 days postinjection are not related to the acute inflammatory response to adenoviral injection, but rather to the metabolic effect of WT CEACAM1, which appears to require longer than 3 days.

Adenoviral-redelivery of CEACAM1 rescues energy expenditure in HF-fed mice
Ad-rWT, but not Ad-rSA CEACAM1 totally reversed BW gain, visceral adiposity, and NEFA plasma levels ( Table 2). It also restored the mRNA levels of UCP-1, a marker of brown adipogenesis, in WAT (supplemental Table S5). Additionally, it reduced, but not fully restored total fat mass ( Table 2, HF-rWT vs. RD-GFP). This provided impetus to investigate potential changes in energy balance by indirect calorimetry 21 days postinjection. As expected (17), HF feeding reduced daily food intake in all mouse groups relative Fig. 4. Effect of adenoviral-mediated redelivery of CEACAM1 on glucose and insulin intolerance caused by HF diet. A-i: Western analysis of rat and mouse CEACAM1 expression (rCC1 and mCC1, respectively) in lysates from liver, WAT and skeletal muscle of 2-month-old RD-fed WT mice 3 days postinjection with Ad-GFP, Ad-rWT, and Ad-rSA. Membranes were reprobed with -GAPDH antibody for normalization. ii: Western analysis (n = 2/each group) on liver lysates from mice fed with RD or HF for 41 days and injected with adenoviral particles in the last 21 days of feeding. iii: qRT-PCR analysis of liver lysates to measure rat and mouse Ceacam1 mRNA levels normalized to Gapdh (n = 5/each group in duplicate). B: Intraperitoneal insulin tolerance (i) and glucose tolerance (ii) tests were carried out 13 and 16 days, respectively, after being injected with Ad-GFP [black (RD) and blue (HF)], Ad-rWT (green), and Ad-rSA (red) (n = 6-8/per group). The AUC was measured and presented in graphs on the right. Values are expressed as means ± SEM. * P < 0.05 versus RD-fed; † P < 0.05 versus HF-GFP.
Acute rise in fatty acids can also activate protein kinase Cmediated pathways to impair insulin signaling in liver (24), which would in turn, suppress insulin-mediated Ceacam1 transcription (42). We have shown that repression of CEACAM1 does not translate into systemic insulin resistance until about 3 weeks of HF feeding (43) when hepatic CEACAM1 content is reduced by >60% (17). Complete reversal of these metabolic abnormalities by rescuing WT CEACAM1 in the liver of mice kept on a HF diet demonstrates that the decrease in hepatic CEACAM1 plays a causative role in sustaining systemic insulin resistance and hepatic steatosis in response to HF diet. The physiological consequence of reduced CEACAM1 expression by HF diet appears to maintain hepatic fatty acids -oxidation at times of excessive energy supply. We have reported that CEACAM1 downregulates Fasn activity to prevent steatosis in liver (11). We have also shown that PPAR reduces Ceacam1 transcription to regulate fatty acid -oxidation during the fasting-refeeding transition (12) and in response to fenofibrate treatment (44). Because fatty acids are the endogenous ligands of PPAR (45), it is conceivable that adipose tissue-derived fatty acids reduce CEACAM1 levels by a PPAR-mediated mechanism as shown by failure of fatty acids to downregulate the transcriptional activity of a promoter construct bearing a mutation on the active PPRE/RXR site in Ceacam1 promoter, as they did to the WT promoter. As recently shown (12), suppressing CEACAM1 would provide a positive feedback mechanism on -oxidation as it is expected to alleviate the negative effect of CEACAM1 on Fasn activity (11) and subsequently, reduce malonyl-CoA level to relieve its inhibitory effect on fatty acids translocation to the mitochondria to undergo -oxidation. This presents a novel mechanistic underpinning for the regulation of lipid oxidation by plasma fatty acids (46). Adenoviral-redelivery of WT CEACAM1 in the liver restored insulin clearance and, subsequently, insulin sensitivity and underscored the importance of CEACAM1-dependent hepatic insulin clearance in promoting insulin sensitivity, they did not fully identify the primary factors that cause diet-induced CEACAM1 repression or the cause-effect relationship between impaired insulin clearance and insulin resistance in response to HF diet. The current studies showed that redistribution of WAT-derived fatty acids to the liver during lipolysis mediated the suppressive effect of HF diet on hepatic CEACAM1 expression, and that restoring CEACAM1 function by adenoviral redelivery completely reversed diet-induced metabolic abnormalities.
Blocking lipolysis with NA caused adipocytes' expansion to accommodate fat retention in WAT. This tissue remodeling was likely facilitated by a parallel decrease in fibrosis, a cellular event commonly found in the white adipose depot of insulin-resistant, obese rodents (32,33) and humans (34). The anti-fibrotic effect of NA could be mediated by the 2-fold rise in IFN- (40) that could counter the profibrogenic effect of Col63 (33) and IL-6 (40), and by the combined effect of the rise in WAT-derived TNF- and plasma leptin levels (41).
While NA treatment caused fat accumulation, triggered more inflammation, and blunted insulin signaling in WAT of mice fed a HF diet for 30 days, it protected hepatic insulin signaling together with systemic insulin response. In parallel, it protected hepatic CEACAM1 levels against HF diet, demonstrating that lipolysis mediated the decline in hepatic CEACAM1 expression. This is consistent with findings of compromised insulin clearance and insulin signaling in parallel to reduced CEACAM1 protein content in rats receiving an intralipid-heparin infusion (24).
Lipolysis could occur shortly after initiating HF intake due to dysregulated hypothalamic control in the absence of insulin resistance in WAT (22). Released fatty acids could then suppress hepatic CEACAM1 content, as suggested by their downregulatory effect on Ceacam1 promoter activity and on its protein level in primary hepatocytes. Male mice (n = 6-10/feeding/treatment group) were fed RD or HF diet for 20 days before being injected through the tail vein with comparable amounts of adenoviral particles of Ad-GFP (as control), Ad-rWT, and Ad-rSA while they were maintained on RD or HF diet. At the end of the experiments (21 days postinjection), mice were euthanized, and tissues and blood removed to carry out the same analyses described in the legend to Table 1. Values are expressed as mean ± SEM.  followed by systemic insulin resistance and increased lipid production in liver and redistribution to adipose tissue to cause visceral obesity with increased lipolysis and FFA output (13). Consistent with downregulated insulin receptor by chronically elevated insulin levels (52,53), L-SACC1 mice manifested reduced insulin receptor number and compromised insulin-induced signaling in hepatocytes (13) in addition to reduced ability of insulin to suppress hepatic glucose production, as demonstrated by hyperinsulinemic-euglycemic clamp analysis (54). Additionally, L-SACC1 mice developed insulin resistance including reduced glucose transport in muscle and adipose and increased lipid accumulation in liver and peripheral tissues (54). Hyperinsulinemia can cause insulin resistance in adipose tissue by reducing Glut4-mediated glucose transport (55). Moreover, chronic hyperinsulinemia can cause insulin resistance to the suppression of plasma FFA levels and increasing de novo lipogenesis but not with regard to hepatic gluconeogenesis (56). Hepatic insulin resistance and increased gluconeogenesis appear to be regulated by increased mobilization of FFA and adipokines from WAT (57,58). Consistently, inhibiting lipolysis and inducing fatty acid oxidation restored insulin action in L-SACC1 transgenics (59), pointing to the role of altered fat metabolism in their sustained insulin resistance that was secondary normal lipid metabolism. It also restored locomotor activity and energy expenditure in mice fed a HF diet in parallel to reversing the gain in BW and visceral adiposity. Given that CEACAM1 is not produced by adipose tissue (9), it is likely that adenoviral delivery of CEACAM1 drives the expression of a set of factors that mediate this positive effect on energy expenditure and adipose tissue biology (reversal of adipocytes' expansion and limited fibrosis and inflammation). One of these factors might be the rise in plasma FGF21 (47,48), which induces locomotor activity (49) to elevate energy dissipation (50,51). Recapitulating the protective effect of forced expression of hepatic WT CEACAM1 on metabolism (17) and on adipose tissue biology and energy expenditure (30) in response to HF diet, the restorative metabolic effect caused by hepatic adenoviral redelivery of WT CEACAM1 on adipocytes supports the critical role of hepatocytic CEACAM1 in regulating insulin action and metabolism in other tissues.
Contrary to WT CEACAM1, adenoviral delivery of the Ad-rSA phosphorylation-defective CEACAM1 mutant failed to reverse the negative metabolic effects of HF diet. Using the L-SACC1 transgenic mouse with liver-specific dominant-negative overexpression of this SA phosphorylation-defective mutant, we demonstrated that impairment of insulin clearance causes chronic hyperinsulinemia Western analysis of Fasn levels in the liver lysates from mice fed with RD or HF for 41 days and injected with adenoviral particles in the last 21 days of feeding. ii: Fasn activity was assayed in liver lysates of RD-GFP (black), HF-GFP (blue), HF-rWT (green), or HF-rSA (red) (n = 6-8/per group). Values are expressed as mean ± SEM. * P < 0.05 versus RD-GFP; † P < 0.05 versus. HF-GFP. B: Liver histology was assessed by H and E stained sections (n = 5/each group). Whereas HF-GFP and HF-rSA exhibited microvesicular lipid infiltration alternating with normal liver parenchyma, RD-GFP and HF-rWT exhibited normal histology. Representative images from three sections/mouse are shown (40× magnification). C: H and E stain on WAT sections (n = 5/each group). HF diet caused enlarged adipocytes relative to RDfed mice (HF-GFP vs. RD-GFP), as shown by the adipocyte size in the accompanying bar graph. This was reversed by Ad-rWT, but not Ad-rSA, injection. Representative images from three sections/mouse are shown (20× magnification), and values of adipocyte size are expressed as mean ± SEM. * P < 0.05 versus RD-GFP; † P < 0.05 versus HF-GFP.
In summary, the current studies provide a novel role for hepatic CEACAM1-dependent pathways in regulating fatty acid metabolism along the adipocyte-hepatocyte axis in response to excess energy intake. Given that hepatic CEACAM1 content is markedly reduced in the liver of insulin-resistant obese subjects with fatty liver disease (20), the current findings propose that inducing CEACAM1 could constitute a critical therapeutic target that serves to mitigate diet-induced metabolic abnormalities, including obesity and fatty liver disease.
The authors thank Ms. M. Kopfman at the Najjar Laboratory for her technical assistance in the generation, genotyping, and maintenance of mice, in addition to carrying out routine RNA and DNA analyses.
to impaired insulin clearance. Given that HF diet increases apoA1 (17), it is conceivable that elevated plasma FFA in L-SACC1 mice maintain the elevated level of the dominant-negative rat transgene expression driven by apoA1 promoter, while repressing that of the mouse endogenous gene by a PPAR-dependent mechanism (44). This could contribute to the suppression of endogenous CEACAM1 activity on hepatic insulin clearance and lipid metabolism by the S503A phosphorylation-defective Ceacam1 mutant transgene. Fig. 6. Effect of adenoviral-mediated redelivery of CEACAM1 on diet-induced energy imbalance. Mice were individually caged (n = 4 per group), fed ad libitum, and subjected to indirect calorimetry analysis in a 24 h period for 5 days to measure: daily food intake (A); heat production (energy expenditure) (kcal/h/kg lean mass) (B); VCO 2 production and VO 2 consumption (mg/h/kg lean mass) to calculate RER as VCO 2 /VO 2 ratio (C), and spontaneous locomotor activity along the X, Y, and Z axes (counts/day) (D). Values are expressed as mean ± SEM of each time interval in the last 3 days. * P < 0.05 versus RD-GFP per cycle; † P < 0.05 versus HF-GFP per cycle.