Hyperinsulinemia drives hepatic insulin resistance in male mice with liver-specific Ceacam1 deletion independently of lipolysis

Abstract

Background:

CEACAM1 regulates insulin sensitivity by promoting insulin clearance. Accordingly, global C57BL/6J.Cc1^−/− null mice display hyperinsulinemia due to impaired insulin clearance at 2 months of age, followed by insulin resistance, steatohepatitis, visceral obesity and leptin resistance at 6 months. The study aimed at investigating the primary role of hepatic CEACAM1 in insulin and lipid homeostasis independently of its metabolic effect in extra-hepatic tissues.

Methods:

Liver-specific C57BL/6J.AlbCre+Cc1^fl/fl mice were generated and their metabolic phenotype was characterized by comparison to that of their littermate controls at 2-9 months of age, using hyperinsulinemic-euglycemic clamp analysis and indirect calorimetry. The effect of hyperphagia on insulin resistance was assessed by pair-feeding experiments.

Results:

Liver-specific AlbCre+Cc1^fl/fl mutants exhibited impaired insulin clearance and hyperinsulinemia at 2 months, followed by hepatic insulin resistance (assessed by hyperinsulinemic-euglycemic clamp analysis) and steatohepatitis at ~ 7 months of age, at which point visceral obesity and hyperphagia developed, in parallel to hyperleptinemia and blunted hypothalamic STAT3 phosphorylation in response to an intraperitoneal injection of leptin. Hyperinsulinemia caused hypothalamic insulin resistance, followed by increased fatty acid synthase activity, which together with defective hypothalamic leptin signaling contributed to hyperphagia and reduced physical activity. Pair-feeding experiment showed that hyperphagia caused systemic insulin resistance, as manifested by blunted insulin signaling in white adipose tissue and lipolysis at 8-9 months of age.

Conclusion:

AlbCre+Cc1^fl/fl mutants provide an in vivo demonstration of the key role of impaired hepatic insulin clearance and hyperinsulinemia in the pathogenesis of secondary hepatic insulin resistance independently of lipolysis. They also reveal an important role for the liver-hypothalamic axis in the regulation of energy balance and subsequently, systemic insulin sensitivity.

1. Introduction

Insulin clearance occurs mostly in the liver and to a lower extent in kidneys and other peripheral tissues [1]. Upon its pulsatile release from pancreatic β-cells into the portal vein [2, 3], insulin crosses the sinusoidal endothelium to reach and activates its receptor on the hepatocytic surface membrane [4]. This phosphorylates other substrates and delivers insulin to early/late endosomes for degradation before the receptor recycles back to the surface membrane [5]. In this manner, excess insulin is rapidly removed and maintained at a physiologically higher concentration in the portal vein than in the systemic circulation [6]. Under hyperinsulinemic conditions, the receptor is diverted to lysosomal degradation [7, 8] to cause cellular insulin resistance.

Chronic hyperinsulinemia also induces hepatic de novo lipogenesis by activating SREBP1c, a master transcriptional regulator of lipogenic genes [9]. Thus, impaired hepatic insulin clearance leads to a concomitant increase in hepatic steatosis and hepatic insulin resistance.

Carcinoembryonic antigen-related cell adhesion molecule 1 (CEACAM1), a surface membrane substrate of the insulin receptor, promotes insulin clearance [10, 11] by taking part of the insulin-insulin receptor complex and increasing its rate of uptake and endosomal targeting [12, 13]. Subsequently, it binds to cytosolic fatty acid synthase (FASN), an event that detaches it from the complex to facilitate the dissociation of insulin from its receptor while mediating an inhibitory effect of insulin on FASN [14]. This restricts hepatic de novo lipogenesis and protects the liver against high portal insulin levels. Consistently, mice with null deletion of Ceacam1 gene (Cc1^−/−) manifested impaired insulin clearance and chronic hyperinsulinemia at 2 months, followed by insulin resistance, primarily hepatic, at ~ 6 months of age when propagated on C57BL/6J background [11]. Deletion of Ceacam1 gene induced triacylglycerol production and redistribution to white adipose tissue (WAT). This resulted in hepatic steatosis and visceral obesity at 2 months, followed by lipolysis and mobilization of non-esterified free fatty acids (NEFA) at ~6 months of age [11].

Because of the regulatory effect of CEACAM1 on lipid production (by preventing hyperinsulinemia or mediating insulin downregulation of FASN activity), it is possible that deleting Ceacam1 primarily increased FASN synthesis and hepatic de novo lipogenesis, independently of hyperinsulinemia, to lead to visceral obesity and NEFA mobilization, which causes hepatic insulin resistance (portal hypothesis) [15] that in turn induces a compensatory increase in insulin secretion and ensuing hyperinsulinemia. This model would be consistent with the commonly accepted paradigm of primary insulin resistance causing chronic hyperinsulinemia mainly by inducing a compensatory increase in insulin secretion. However, blocking lipolysis with nicotinic acid did not restore insulin clearance or sensitivity [16]. Together with intact β-cell mass [11], this indicates that hyperinsulinemia in Cc1^−/− mice resulted primarily from impaired insulin clearance rather than increased insulin secretion, and that it caused insulin resistance independently of lipolysis.

Cc1^−/− mice also developed leptin resistance concomitantly with insulin resistance, basal hyperphagia and decreased spontaneous locomotor activity [17]. Whereas hyperinsulinemia-driven insulin resistance in the hypothalamus caused energy imbalance, as shown by liver-specific rescuing of Ceacam1 in Cc1^−/− mice [16], deleting CEACAM1 from the arcuate nucleus region of the hypothalamus, in particular pro-opiomelanocortin (POMC)-expressing neurons, may contribute to the regulation of energy imbalance in Cc1^−/− mutants [17].

Given the multiple factors that could contribute to insulin resistance in Cc1^−/− mice, including impaired insulin clearance, lipolysis and leptin resistance, it became imperative to identify its primary underlying mechanism with the overarching goal to determine whether it caused or resulted from hyperinsulinemia. To this end, we generated a liver-specific C57BL/6J.Ceacam1 knockout mouse (AlbCre+Cc1^fl/fl) and characterized the hepatic and extra-hepatic mechanisms that could be implicated in its altered metabolic phenotype.

2. Methods

2.1. Generation of Liver-specific AlbCre+Cc1^fl/fl Mice

As in the conditional T cell-specific null mouse [18], the targeting construct inserted a loxP-neo cassette in intron 6 and a loxP fragment in intron 9, deleting a sequence that encodes the cytoplasmic domain [19]. We crossed Cc1^loxp/loxp mice with transgenic mice expressing Cre under the transcriptional control of the albumin gene promoter (AlbCre) on C57BL/6J background (Jackson Laboratories, Bar Harbor, ME) (Fig. S1). Heterozygotes were backcrossed >6x with C57BL/6J mice. Offsprings were genotyped by PCR analysis of ear DNA, using primers shown in Fig. S1. As controls, we used homozygotes with wild-type Ceacam1 allele with (AlbCre+Cc1^+/+) or without AlbCre (AlbCre–Cc1^+/+), and homozygotes with Ceacam1-floxed allele, without AlbCre (AlbCre–Cc1^fl/fl all from the same breeding to rule out potential confounding effects of floxing and introducing AlbCre.

All animals were housed in a 12 hr-dark/light cycle and fed standard chow (Harlan Teklad 2016; Harlan, Haslett, MI) ad libitum. All procedures and animal experiments were approved by the Animal Care and Utilization Committee of each institution.

2.2. Body Composition

Body composition was assessed by nuclear magnetic resonance technology (Bruker Minispec; Billerica, MA).

2.3. Indirect Calorimetry

Awake mice (n=4/genotype) were individually caged (CLAMS system, Columbus Instruments, Columbus, OH) over a 3 day-period after being acclimated for 2 days [17]. Mice had access to food and water ad libitum. Spontaneous physical activity was measured on the x axis (locomotor), y axis (ambulatory), and z axis (standing). Total activity was calculated as the average of x/y/z activities. Oxygen consumption (VO₂), CO₂ production (VCO₂), and heat generation were sampled every 20min and normalized to fat-free lean mass. The respiratory exchange rate was calculated as the VCO₂/VO₂ ratio. Data were represented as mean ± SEM of light (700–1900h) and dark (1900 to 700h) cycles.

2.4. Glucose and Insulin Tolerance Tests

Awake mice were fasted for 6–7 hr before being subjected to intraperitoneal dextrose and insulin injections, and blood glucose was measured from the tail at each time point at 0–180 min post-injection [16].

2.5. Hyperinsulinemic-euglycemic Clamp Analysis

A 2-hr hyperinsulinemic-euglycemic clamp was performed in awake overnight-fasted mice with primed and continuous infusion of human regular insulin (Humulin, Lilly, Indianapolis, IN) at a rate of 2.5mU×kg⁻¹×min⁻¹ [11]. Glucose metabolism was estimated with a continuous infusion of 0.05μCi/min of [3-³H] glucose (PerkinElmer and Analytical Sciences, Hopkinton, MS) and subsequently with 0.1μCi/min throughout the clamp.

2.6. Biochemical Parameters

Retro-orbital venous blood was drawn at 1100h from overnight fasted mice and plasma was analyzed by ELISA for insulin, C-peptide (ALPCO, Salem, NH), and adiponectin (Abcam, Cambridge, MA). Plasma NEFA and triacylglycerol were assayed by enzymatic colorimetric assays from Wako (Richmond, VA) and Pointe Scientific (Canton, MI), respectively, and hepatic triacylglycerol were assayed as previously described [11].

Mice were fasted 6-7 hr before leptin and glucagon were assayed by ELISA kits from ALPCO and R&D Systems (Minneapolis, MN), respectively.

2.7. Isolation of Primary Hepatocytes

Primary hepatocytes were isolated by perfusing livers (1ml/min) of anesthetized mice with Collagenase-Type II (1mg/ml) (Worthington, Lakewood, NJ), and plated in 6-well plates in complete Williams-E medium at 1×10⁶ cells/well [10].

2.8. Biotin Labeling

Primary cells were incubated in the absence or presence of 100nM insulin (Sigma-Aldrich, Saint Louis, MO) at 37°C for 5min, followed by incubation on ice with biotin (1mg/ml) (Pierce, Rockford, IL) in PBS for 30min [16]. Cells were lysed in 1%Triton-X and proteins immunoprecipitated with streptavidin (Fisher Scientific, Waltham, MA), and immunoblotted with 1:1000 of insulin receptor alpha (IRα) antibody (N-20, Santa Cruz) or a custom-made mouse Ab3759 polyclonal antibody against CEACAM1 [20], followed by horseradish peroxidase-conjugated mouse anti-rabbit IgG antibody (Jackson Immunoresearch, West Grove, PA) and enhanced chemiluminescence (ECL, Amersham Pharmacia, Sunnyvale, CA).

2.9. Insulin Internalization

Human [¹²⁵I]Insulin (PerkinElmer Life Sciences, Akron, OH) (30,000cpm) was allowed to bind to primary hepatocytes at 4°C for 5hr in Krebs-Ringer phosphate (KRP) buffer [16], before unbound insulin was removed, and surface-bound insulin was collected in acidic PBS (pH 3.5). Cells were washed and lysed with 0.4N NaOH to account for cell-associated internalized insulin. Internalized insulin was calculated as percent cell-associated per specifically bound ligand (the sum of surface-bound plus cell-associated ligand).

2.10. Ex-vivo Palmitate Oxidation

Liver homogenate was added to 1ml of solution A [0.2mM of [^1-14C]palmitate (0.5mCi/ml) (American Radiolabeled Chemicals Inc, St Louis, MO)-2mM ATP] and left at 30°C for 45min in a sealed beaker [21]. Benzothonium hydroxide (Sigma-Aldrich) was added to a basket attached to the beaker and the reaction was terminated with perchloric acid to recover trapped CO₂ radioactivity and the partial oxidation products to be measured by liquid scintillation in CytoCint (MP Biomedicals, Solon, OH). The oxidation rate was expressed as the sum of total and partial fatty acid oxidation in nmoles/g/min.

2.11. Fatty Acid Synthase Activity

Livers were homogenized, centrifuged and the supernatant added to a reaction mix containing 0.1 μCi [¹⁴C]malonyl-CoA (Perkin-Elmer) and 25nmol malonyl-CoA in the absence or presence of 500μM NADPH (Sigma-Aldrich) [14]. The reaction was stopped with 1:1 chloroform:methanol solution and samples were centrifuged, butanol-extracted and counted. FASN enzymatic activity was calculated as cpm of [¹⁴C]-incorporated Bq/μg cell lysates.

2.12. Daily Food Intake and Pair-feeding

The average daily food intake was assessed in individually caged mice over a 5-day-period. Based on the average daily food intake of individually caged 8-month-old mice over a 5-day-period, mutants were subjected to a pair-feeding regimen: 2.5g food/day (0.5g less than the ad libitum-fed mutants) for the first week [22]. Because this feeding program prevented weight gain, the amount of food was increased to 3.0g/day as opposed to 5.0g of food/day to the ad libitum-fed group in the second week, and insulin tolerance was assessed.

2.13. Hypothalamic Leptin Signaling

Mice (n=3–5/treatment/genotype) were injected intraperitoneally with vehicle (Veh, open or solid bars) or leptin (Lep, grey- and black-striped bars) 45min prior to tissue isolation [17]. Coronal sections from the medial hypothalamus were subjected to immunohistochemical analysis with phospho-STAT3 antibody (rabbit polyclonal antiserum, Cell Signaling Technologies, Beverly, MA). Images of stained hypothalamic neurons of the medial hypothalamus were counted using Open CFU [23].

2.14. Western Blot Analysis

Tissue lysates were analyzed by immunoprobing with custom-made rabbit polyclonal Ab3759 against the mouse CEACAM1 extracellular domain [24], and phosphorylated CEACAM1 (α-pCEACAM1) (Bethyl Laboratories, Montgomery, TX) [25, 26]. Antibodies against phospho-insulin receptor β (pIRβ) (phospho-Y1361), IRβ (C18C4) (Abcam), FASN, phospho-Akt^Ser473 and Akt (Cell Signaling) were also used, α-Tubulin monoclonal antibody (Cell Signaling) was used for normalization. Blots were incubated with horseradish peroxidase-conjugated donkey anti-rabbit IgG antibody or anti-mouse IgG antibody (GE Healthcare Life Sciences, Amersham), and detected by ECL.

2.15. Quantitative Real-time-PCR (qRT-PCR)

Total RNA was isolated with PerfectPure RNA Tissue Kit (Fisher Scientific, Waltham, MA). cDNA was synthesized by iScript cDNA Synthesis Kit (Bio-Rad), using 1μg of total RNA and oligodT primers (Table S1). cDNA was evaluated with qRT-PCR (StepOne Plus, Applied Biosystems, Foster City, CA), and mRNA was normalized to Gapdh [21].

2.16. Statistical Analysis

Data were analyzed using one-way ANOVA analysis with Bonferroni correction or two-tailed Student-t-test using GraphPad Prism6 software. Data were presented as mean ± SEM. P<0.05 was considered statistically significant.

3. Results

3.1. Liver-specific Deletion of CEACAM1

qRT-PCR analysis demonstrated that mouse Ceacam1 mRNA (mCc1) was absent in primary hepatocytes from AlbCre+Cc1^fl/fl mice (Alb+Cc1^fl/fl for simplicity) (Fig. 1A). Immunoblotting with mouse α-CEACAM1 (Ib:α-mCc1) detected CEACAM1 protein in the livers of controls (Alb–Cc1^+/+, Alb+Cc1^+/+ and Alb–Cc1^fl/fl), but not Alb+Cc1^fl/fl mutants (Fig. 1Bi). CEACAM1 protein content was intact in other tissues from mutant mice, including kidney, hypothalamus and heart (Fig. 1B.i-ii).

Fig. 1.

Tissue-specific expression of mouse CEACAM1 (mCC1). (A) Primary hepatocytes from Alb–Cc1^+/+ (white bar), Alb+Cc1^+/+ (light grey bar), Alb–Cc1^fl/fl (dark grey bar) and Alb+Cc1^fl/fl (black bar) were isolated and analyzed by qRT-PCR in triplicate to assess mouse Ceacam1 mRNA level. Values are expressed as mean ± SEM; *P<0.05 vs Alb–Cc1^+/+, Alb+Cc1^+/+ and Alb–Cc1^fl/fl. (B) Western blot analysis of mouse (mCC1) CEACAM1 protein content in (i) liver and kidney, and (ii) hypothalamus and heart was performed by immunoblotting (Ib) the upper portion of the blot with a polyclonal antibody against CEACAM1 (α-CC1) and the lower portion with α-Tubulin to normalize for protein loading. Gels represent more than two experiments (different mice/genotype).

3.2. Early Onset of Impaired Insulin Clearance and Hyperinsulinemia in Alb+Cc1^fl/fl Mice

Like Cc1^−/− [11], Alb+Cc1^fl/fl exhibited hyperinsulinemia at 2 months of age (Fig. 2A.i) in parallel to impaired insulin clearance, as indicated by steady-state C-peptide/insulin molar ratio (Fig. 2A.iii). Consistent with increased rate of receptor-mediated insulin endocytosis by phosphorylated CEACAM1 [12], [¹²⁵I]-insulin internalization was significantly reduced (P <0.05) in primary hepatocytes isolated from Alb+Cc1^fl/fl (black circles) by comparison to hepatocytes from Alb–Cc1^fl/fl controls (dark grey circles) (Fig. 2B). In addition, immunoblotting the biotin-streptavidin immunopellet (Ip) with polyclonal antibodies against IR_α, the extracellular subunit of the insulin receptor (IR), and mouse CEACAM1 (Fig. 2C) revealed insulin-induced IR_α and CEACAM1 internalization, measured by the loss of biotin-labeled surface membrane proteins in insulin (Ins)- versus buffer-treated hepatocytes [10] in control groups but not in Alb+Cc1^fl/fl mice. Lower insulin (Fig. 2B) and biotin-labelled IR_a (Fig. 2C) internalization in Alb+Cc1^fl/fl hepatocytes demonstrated defective targeting of the insulin-insulin receptor complex to the degradation process in the absence of CEACAM1, an event that could impair insulin clearance.

Fig. 2.

Analysis of insulin clearance. (A) Alb–Cc1^+/+ (white bar), Alb+Cc1^+/+ (light grey bar), Alb–Cc1^fl/fl (dark grey bar) and Alb+Cc1^fl/fl (black bar) (n=5/genotype; 2-month-old) were fasted overnight and retro-orbital blood was drawn to assess plasma insulin (i) and C-peptide (ii) levels to calculate steady-state C-peptide/insulin molar ratio (iii) as a measure of insulin clearance. Assays were performed in triplicate. Values are expressed as mean ± SEM; *P<0.05 vs Alb–Cc1^+/+, Alb+Cc1^+/+ and Alb–Cc1^fl/fl. (B) [¹²⁵I]insulin internalization was measured in triplicate as percent of specifically-bound ligand in primary hepatocytes from Alb–Cc1^fl/fl (dark grey circles) and Alb+Cc1^fl/fl (black circles). n=6 mice/genotype. *P<0.05 vs Alb+Cc1^fl/fl. (C) Cell surface proteins in primary hepatocytes treated with buffer (−) or insulin (Ins) and then labelled with biotin. Proteins were immunoprecipitated (Ip) with α-streptavidin beads prior to analysis by 7% SDS-PAGE and immunoblotting (Ib) with antibodies against IRα and mouse CEACAM1. Total lysates were also analyzed by immunoblotting with mouse Actin antibody (α-mActin) to normalize against total loaded proteins. Gels represent more than two experiments (different mice/genotype/experiment).

3.3. Alb+Cc1^fl/fl Mice Developed Secondary Hepatic Insulin Resistance at 6-7 Months of Age

Up to 5-6 months of age, Alb+Cc1^fl/fl mutants displayed normal glucose clearance in response to intraperitoneal injections of insulin (Fig. 3Ai, black circles, black bars) and glucose (Fig. 3Aii) relative to their three controls [Alb–Cc1^+/+ (white circles and bars), Alb+Cc1^+/+ (light grey triangles and bars) and Alb–Cc1^fl/fl (dark grey squares and bars)]. At 6-7 months; however, Alb+Cc1^fl/fl mutants exhibited intolerance to exogenous insulin (Fig. 3Bi) and glucose (Fig. 3Bii), as supported by the ~2-fold higher area under the curve (AUC) (graphs in Figs. 3Bi’ and 3Bii’). They also exhibited fed hyperglycemia starting at 6-7 months of age (Tables 1 and S2). Like global Cc1^−/− mice [11], Alb+Cc1^fl/fl mutants maintained fasting euglycemia up to 8-9 months of age (Table S3), reflecting intact β-cell function and ruling out defect in insulin secretion. This was supported by elevated C-peptide levels (Fig. 2A.ii), normal plasma glucagon levels (Tables 1, S2 and S3) and normal pancreatic proinsulin and proglucagon mRNA levels as compared to the controls (Table S4).

Fig. 3.

Insulin and glucose tolerance tests. 7 hr-fasted mice at (A) 5-6 and (B) 6-7 months of age (n=7-8/genotype) were injected intraperitoneally with insulin or glucose to assess insulin (i) and glucose (ii) tolerance, respectively. Accompanying graphs represent the area under the curve (AUC) (i’ and ii’). Alb–Cc1^+/+ (white circles and bars), Alb+Cc1^+/+ (light grey triangles and bars), Alb–Cc1^fl/fl (dark grey squares and bars) and Alb+Cc1^fl/fl (black circles and bars). Values are expressed as mean ± SEM. *P<0.05 vs Alb–Cc1^+/+, Alb+Cc1^+/+ and Alb–Cc1^fl/fl.

Table 1.

Plasma and tissue biochemistry in mice at 6-7 months of age

	Alb–Cc1+/+	Alb+Cc1+/+	Alb– Cc1fl/fl	Alb+Cc1fl/fl
Body Weight (g)	30.1 ± 0.5	31.2 ± 0.5	31.4 ± 0.5	36.9 ± 0.9*†‡
% Fat Mass	6.2 ± 0.5	6.1 ± 0.2	6.5 ± 0.3	9.7 ± 0.3 *†‡
% Lean Mass	51.4 ± 1.2	52.7 ± 1.5	52.8 ± 1.5	46.4 ± 1.2 *†‡
% WAT/BW	2.2 ± 0.1	2.2 ± 0.2	2.0 ± 0.4	3.5 ± 0.4 *†‡
NEFA (mEq/L)	0.4 ± 0.0	0.7 ± 0.0	0.7 ± 0.0	0.8 ± 0.1
Insulin (pmol/L)	32.3 ± 1.4	34.1 ± 1.5	36.7 ± 1.5	56.1 ± 1.7 *†‡
C-peptide (pmol/L)	242. ± 34.	271. ± 42.	255. ± 39.	575. ± 50.*†‡
C/I molar ratio	9.8 ± 1.4	9.5 ± 1.2	9.8 ± 1.7	6.5 ± 1.9 *†‡
Glucagon (pg/mL)	65.7 ± 5.4	67.1 ± 4.9	63.8 ± 5.2	69.8 ± 6.1
Triacylglycerol (TG, mg/dL)	51.1 ± 2.7	49.1 ± 4.2	48.2 ± 5.1	52.7 ± 4.5
Hepatic TG (μg/mg protein)	66.2 ± 5.4	64.1 ± 4.3	62.3 ± 2.9	88.2 ± 3.4 *†‡
Fast blood glucose (mg/dL)	78. ± 2.	80. ± 2.	79. ± 2.	85. ± 2.
Fed blood glucose (mg/dL)	131. ± 2.	126. ± 2.	129. ± 3.	158. ± 3. *†‡
Leptin (ng/mL)	3.1 ± 0.1	3.1 ± 0.2	3.2 ± 0.1	10.4 ± 0.3 *†‡
Adiponectin (ng/mL)	33.6 ± 0.9	31.8 ± 1.2	32.4 ± 1.3	26.2 ± 0.8 *†‡

Male mice (6-7 months of age, n≥6/genotype) were fasted overnight (except for leptin and glucagon when mice were fasted for 6-7 hr) before blood was drawn and tissues were excised. Values refer to plasma levels, unless otherwise mentioned. Values are expressed as mean ± SEM.

^*P<0.05 vs AlbCre–Cc1^+/+,

^†P<0.05 vs AlbCre+Cc1^+/+,

^‡P<0.05 vs AlbCre–Cc1^fl/fl..

WAT: White adipose tissue; BW: Body weight; C/I: C-peptide/Insulin molar ratio as measure of insulin clearance; NEFA: Non-esterified fatty acid; TG: Triacylglycerol.

To further assess the effect of losing hepatic CEACAM1 on insulin action in vivo, a 2-hr hyperinsulinemic-euglycemic clamp analysis was performed on overnight-fasted, awake 7-month-old Alb+Cc1^fl/fl mice and all littermate controls (Fig. 4). The glucose infusion rate required to maintain euglycemia was lower in Alb+Cc1^fl/fl mice (Fig. 4A), indicating insulin resistance in these mutants.

Fig. 4.

Hyperinsulinemic–euglycemic clamp analysis performed on 7-month-old awake overnight-fasted mice. Measurements under clamp conditions with primed and continuous infusion of insulin are shown for Alb–Cc1^+/+ (white bar), Alb+Cc1^+/+ (light grey bar), Alb–Cc1^fl/fl (dark grey bar) and Alb+Cc1^fl/fl (black bar) mice. In (C), measurements for the basal (b) and clamp (c) conditions are also shown. The data are cumulative of 2 sets of experiments performed on different sets of mice at the same age. (n=5 for Alb–Cc1^+/+, n=12 for Alb+Cc1^+/+ and Alb–Cc1^fl/fl and n=20 for Alb+Cc1^fl/fl). Values are expressed as mean ± SEM; *P<0.05 vs Alb–Cc1^+/+, Alb+Cc1^+/+ and Alb–Cc1^fl/fl; †P<0.05 vs basal.

Insulin suppresses hepatic glucose production by inhibiting gluconeogenesis and stimulating net hepatic glucose uptake and subsequent glycogen synthesis [27]. Whole-body glycogen synthesis was intact in Alb+Cc1^fl/fl mutants (Fig. 4B). Whereas preclamped basal (b) hepatic glucose production, a measurement of the appearance rate (Ra), was normal in Alb+Cc1^fl/fl mutants, insulin’s ability to suppress hepatic glucose production (HGP) was compromised, as shown by higher hepatic glucose production during clamp (Fig. 4C) and reduced hepatic insulin action (Fig. 4D) in mutants relative to controls. In contrast, insulin-stimulated whole-body glucose turnover (R_d) was intact in 7-month-old Alb+Cc1^fl/fl mice (Fig. 4E). This is consistent with intact glucose uptake in the gastrocnemius muscle (Fig. 4F), WAT (Fig. 4G) and brown adipose tissue (BAT) (Fig. 4H). Together, this suggests that Alb+Cc1^fl/fl mice developed primarily hepatic insulin resistance at about 7 months of age.

3.4. Alb+Cc1^fl/fl Mice Displayed Hepatic Fat Accumulation and Inflammation at 6-7 Months of Age

Flepatic triacylglycerol content was higher in Alb+Cc1^fl/fl mice compared to the three controls starting at 6-7 months of age (Tables 1 and S2). Consistently, histological evaluation of H&E-stained liver sections revealed that unlike the three controls, Alb+Cc1^fl/fl mice displayed diffused fat infiltration with both micro- and macro-steatosis at this age (Fig. 5A).

Fig. 5.

Lipid metabolism in liver. (A) H&E staining in the liver of 6-7 month-old (i), Alb–Cc1^+/+, (ii), Alb+Cc1^+/+ (iii), Alb–Cc1^fl/fl, and (iV) Alb+Cc1^fl/fl mice (n=5/genotype). The circle in panel (iV) points to foci of inflammatory cell infiltrates. (B) Mice (6-7 months of age) were fasted overnight (F) and refed for 7 hr (RF) and retro-orbital blood was removed to analyze (i) plasma insulin levels (n=6/genotype/feeding state). Values are expressed as mean ± SEM. *P<0.05 refed vs fasted per genotype, †P0.05 Alb+Cc1^fl/fl vs other genotypes at fasting, ‡P<0.05 Alb+Cc1^fl/fl vs other genotypes at refeeding. (ii) Western blot analysis of liver lysates was carried out to assess insulin receptor protein level (α-IRβ) and phosphorylation (α-pIRβ). Immunoblotting with α-tubulin was used for normalization. (iii) Some aliquots were subjected to immunoprecipitation (Ip) with α-FASN antibody followed by immunoblotting (Ib) with α-pCEACAM1 antibody (α-pCC1). Gels represent two separate experiments performed on different mice/genotype/feeding state, (iv) Fasn mRNA expression relative to Gapdh was analyzed in triplicate by qRT-PCR (n=5/genotype/feeding state; performed in triplicate). Values are expressed as mean ± SEM. †P<0.05 Alb+Cc1^fl/fl vs other genotypes at fasting, ‡P<0.05 Alb+Cc1^fl/fl vs other genotypes at refeeding, (v) FASN activity was measured in triplicate by [¹⁴C]malonyl-CoA incorporation (n=5/genotype/feeding state). Values are expressed as mean ± SEM. †P<0.05 Alb+Cc1^fl/fl vs other genotypes at fasting, ‡P<0.05 Alb+Cc1^fl/fl vs other genotypes at refeeding. (C) Hepatic FAO (palmitate) in fasted 6-7 months of age mice, Alb–Cc1^+/+, (white bar), Alb+Cc1^+/+ (light grey bar), Alb–Cc1^fl/fl (dark grey bar) and Alb+Cc1^fl/fl (black bar) mice (n=5/genotype). Assays were performed in triplicate. Values are expressed as mean ± SEM. *P<0.05 vs Alb–Cc1^+/+, Alb+Cc1^+/+ and Alb–Cc1^fl/fl.

Hepatic steatosis could be due to high de novo lipogenesis and low fatty acid β-oxidation (FAO). The former was supported by elevated mRNA levels of hepatic genes involved in de novo lipogenesis (Srebp1c and Fasn) in mutant mice (Table S5). Because acute release of insulin represses FASN activity by inducing CEACAM1 phosphorylation and its binding to FASN [14], we refed mice (RF) for 7 hours following an overnight fast (F) and assayed their FASN activity. Consistent with increased transcription of lipogenic genes by activated SREBP-1c under chronic hyperinsulinemia [9], basal (F) hepatic Fasn mRNA (Fig. 5B.iv and Table S5) and protein (Fig. 5B.iii) levels were higher in liver lysates of Alb+Cc1^fl/fl relative to their three normo-insulinemic controls. Consequently, fasting FASN activity was higher in mutant than control mice (Fig. 5B.v). As expected from the acute negative effect of insulin on FASN activity under normo-insulinemic conditions [14], acute insulin release suppressed FASN activity in all controls (Fig. 5B.v; RF vs F). This was mediated by the ability of insulin to induce phosphorylation of the β-subunit of IR (pIR_β) (Fig. 5B.ii) and subsequently, CEACAM1 phosphorylation (pCc1) and its binding to FASN, as demonstrated by its detection in the α-FASN immunopellet (Fig. 5B.iii). In contrast, hyperinsulinemia drove insulin receptor downregulation in Alb+Cc1^fl/fl mice, as shown by lower IRβ protein level than controls in the immunoblot of liver lysates with α-IRβ antibody (Fig. 5B.ii). This translated into restricted ability of insulin to induce IR_P phosphorylation (Fig. 5B.ii). Consistent with dependence on CEACAM1 phosphorylation for insulin to trigger CEACAM1/FASN interaction [14], pCC1 was not detected in the α-FASN immunopellet in mutant mice (Fig. 5B.iii). Consequently, insulin release failed to suppress FASN activity in the liver of refed relative to fasted Alb+Cc1^fl/fl, leading to increased hepatic lipogenesis in mutant mice.

Additionally, hepatic mRNA level of Cptlα a protein that transports fatty acids into mitochondria for FAO, was lower in Alb+Cc1^fl/fl livers (Table S5). Consistently, FAO was reduced in Alb+Cc1^fl/fl as shown by lower partial and total conversion of [^1-14C]palmitate to CO₂ in mutant mice relative to controls (Fig. 5C).

H&E-stained sections of Alb+Cc1^fl/fl mice showed multiple foci of inflammatory cell infiltrates in the hepatic lobules (Fig. 5A). This was supported by increased mRNA of F4/80 (a marker of macrophage pool) and of other inflammatory markers such as hepatic tumor necrosis factor-α (Tnfα) and interleukin 6 (Il-6) (Table S5).

3.5. Increased Visceral Obesity in Alb+Cc1^fl/fl Mice at 6-7 Months of Age

Consistent with increased redistribution of VLDL-triacylglycerol from liver to WAT, Alb+Cc1^fl/fl mice developed visceral adiposity starting at 6-7 months of age (Tables 1 and S2). Despite loss of adiponectin (Table 1), this did not translate into insulin resistance in WAT at this age, as shown by intact glucose transport under hyperinsulinemic clamp conditions (Fig. 4G), intact insulin-stimulated Akt phosphorylation in refed (RF) relative to overnight-fasted (F) mice (Fig 6A.i), and absence of lipolysis (as indicated by normal plasma NEFA (Table 1 and Fig. 6B.i) and mRNA levels of Hsl (Fig. 6C.i), the gene that encodes hormone-sensitive lipase, a key enzyme in lipolysis). In contrast, insulin resistance developed in WAT of mutant mice at 8-9 months of age, as supported by a blunted ability of insulin to induce Akt phosphorylation (Fig. 6A.ii), and the ~2-fold higher plasma NEFA (Fig. 6B.ii) and Hsl mRNA levels (Fig. 6C.ii). Thus, WAT of Alb+Cc1^fl/fl mutants became insulin resistant at 8-9 months of age, releasing NEFA and contributing to sustained systemic insulin resistance [26, 28].

Fig. 6.

Insulin signaling in white adipose tissue. (A) WAT was isolated from (i) 6-7 and (ii) 8-9 month-old fasted (F) and refed (RF) mice. Western analysis was performed by immunoblotting (Ib) with antibodies against phospho-Akt (α-pAkt), followed by α-Akt for normalization. Gels represent more than two separate experiments performed on different mice/genotype/treatment group. (B) Plasma NEFA levels were assayed in (i) 6-7 and (ii) 8-9 month-old overnight fasted mice [n>6/each of Alb–Cc1^+/+ (white bar), Alb+Cc1^+/+ (light grey bar), Alb–Cc1^fl/fl (dark grey bar) and Alb+Cc1^fl/fl (black bar)]. (C) mRNA level of Hsl was analysed by qRT-PCR relative to Gapdh in triplicate (n=5/genotype). in (B and C), values are expressed as mean ± SEM. *p<0.05 vs Alb–Cc1^+/+, Alb+Cc1^+/+ and Alb–Cc1^fl/fl.

3.6. Alb+Cc1^fl/fl Mice Exhibited Energy Imbalance at 6-7 Months of Age

In addition to visceral adiposity, Alb+Cc1^fl/fl mutants exhibited an increase in total fat mass with a reciprocal decrease in lean mass and consequently, increased total body weight starting at 6-7 months of age (Tables 1, S2 and S3).

To investigate whether altered fat distribution was associated with energy imbalance, we assessed daily food intake and subjected mice to indirect calorimetry. Relative to the three controls, Alb+Cc1^fl/fl mutants displayed hyperphagia starting at 6-7 months of age (Figs. 7A and S2A). They also developed a marked decrease in total spontaneous physical activity at this age (Figs. 7B.v and S2B.v), including lower locomotor, ambulatory and standing activity (not shown). Like Cc1^−/− mice [17], VO₂ consumption (Fig. 7B.i), CO₂ production (Fig. 7B.ii) and heat generation (Fig. 7B.iv) were not altered in Alb+Cc1^fl/fl mutants.

Fig. 7.

Daily food intake and energy balance. (A) Daily food intake over 5 consecutive days was assessed on 6-7 months of age Alb–Cc1^+/+ (white bar), Alb+Cc1^+/+ (light grey bar), Alb–Cc1^fl/fl (dark grey bar) and Alb+Cc1^fl/fl (black bar) (n=6/genotype). Values are expressed as mean ± SEM. *P<0.05 vs Alb–Cc1^+/+, Alb+Cc1^+/+ and Alb–Cc1^fl/fl. (B) 6-7 month-old mice were individually caged (n=4/genotype) and analyzed by indirect calorimetry (CLAMS system) for 5 days to measure (i) VO₂ consumption, (ii) VCO₂ production, (iii) respiratory exchange ratio, (iV) heat production, (v) calculated total activity every 20 min at a flow rate of 0.5 L/min for 24 hr. Values are expressed as mean ± SEM of each time interval in the last 3 days in the light (07:00h to 19:00h) and dark (shaded; 19:00h to 07:00h) cycle. *P<0.05 vs Alb–Cc1^+/+, Alb+Cc1^+/+ and Alb–Cc1^fl/fl.

3.7. Role of Hyperphagia in Systemic Insulin Resistance in Alb+Cc1^fl/fl Mice

To test whether hyperphagia contributed to delayed systemic insulin resistance in Alb+Cc1^fl/fl, we subjected mutants at 7-8 months of age to a pair-feeding regimen to decrease their food intake and subsequently, lower their body mass [Fig. 8A.i; calorie restricted (CR) versus ad libitum (AL)-fed Alb+Cc1^fl/fl] to that of their AL-fed controls (Fig. 8A i). As Fig. 8A.ii shows, caloric restriction restored tolerance to exogenous insulin in (CR)Alb+Cc1^fl/fl mice. This indicates that hyperphagia contributed to systemic insulin resistance in older mutants.

Fig. 8.

Hyperphagia and systemic insulin resistance. (A) Pair-feeding experiments were performed on 7-8 month-old mice with some (i) Alb+Cc1^fl/fl being fed ad libitum (AL) and others being subjected to caloric restriction (CR) for 2 weeks to decrease their body mass to the level of Ad libitum fed controls (n=6/genotype/feeding group). (ii) At the end of the feeding period insulin tolerance was determined. (AL)Alb–Cc1^+/+ (white circle), (AL)Alb+Cc1^+/+ (light grey triangle), (AL)Alb–Cc1^fl/fl (dark grey square), (AL)Alb+Cc1^fl/fl (black circle) and (CR)Alb+Cc1^fl/fl (hatched circle). Values are expressed as mean ± SEM at each time point. *P<0.05 vs (AL)Alb–Cc1^+/+, (AL)Alb+Cc1^+/+, (AL)Alb–Cc1^fl/fl and (CR)Alb+Cc1^fl/fl. (B) Ceacam1 (Cc1) mRNA content was analysed by qRT-PCR in triplicate in liver (i) and hypothalamus (ii) of Alb–Cc1^fl/fl (grey bar) and Alb+Cc1^fl/fl (black bar) mice aged 2–9 months (n=5/genotype/age group). (iii) Cc2 mRNA was analysed in the hypothalamus of Alb–Cc1^fl/fl (grey bar) and Alb+Cc1^fl/fl (black bar) mice aged 2–9 months, as in (ii). Values are expressed as mean ± SEM. *P<0.05 vs Alb–Cc1^fl/fl of the same age group. †P≤0.05 vs mice at the earliest age examined.

3.8. Mechanisms of Energy Imbalance in Alb+Cc1^fl/fl Mice

Consistent with increased visceral adiposity, hyperleptinemia was initiated at 6-7 months of age in Alb+Cc1^fl/fl mice (Tables 1 and S2) in parallel to increased daily food intake (Fig. 7A). Flyperphagia could be caused at least partly, by blunted leptin signaling as shown by failure of intraperitoneally injected leptin to induce STAT3 phosphorylation (pSTAT3) in 7-month-old Alb+Cc1^fl/fl like it did in their age-matched Alb–Cc1^fl/fl controls (Fig. 9A).

Fig. 9.

Hypothalamic leptin signaling. (A) 7-month-old Alb–Cc1^fl/fl and Alb+Cc1^fl/fl mice (n=5/treatment group) were injected intraperitoneally with vehicle (Veh, solid grey or black bars) or leptin (Lep, grey-and black-striped bars) 45min prior to tissue isolation. Coronal sections from the medial hypothalamus were subjected to immunohistochemical analysis with phospho-STAT3 (pSTAT3) antibody. Values are presented as mean ± SEM of stained hypothalamic neurons of the medial hypothalamus. *P<0.05 vs vehicle/genotype. (B) Mice (7 months of age) were fasted overnight (F) and refed for 7 hr (RF). (i) Western blot analysis investigating hypothalamic insulin receptor protein level (α-IRβ) and phosphorylation (α-pIRβ). α-tubulin was used to normalize IRβ against total loaded proteins. (ii) Some hypothalamic aliquots were used for immunoprecipitation (Ip) with α-FASN followed by immunoblotting (Ib) with α-pCEACAM1 antibody (α-pCC1). Gels represent two separate experiments performed on different mice/genotype/feeding state. (iii) Fasn mRNA expression was analyzed by qRT-PCR in triplicate relative to Gapdh (n=5/genotype/feeding state). Values are expressed as mean ± SEM. †p<0.05 Alb+Cc1^fl/fl vs other genotypes at fasting, ‡p<0.05 Alb+Cc1^fl/fl vs other genotypes at refeeding. (iv) FASN activity was measured in triplicate by [¹⁴C]malonyl-CoA incorporation (n=5/genotype/feeding state). Values are expressed as mean ± SEM. *P<0.05 refed vs fasted/genotype, †p<0.05 Alb+Cc1^fl/fl vs other genotypes at fasting, ‡p<0.05 Alb+Cc1^fl/fl vs other genotypes at refeeding.

Because hypothalamic FASN activation causes hyperphagia independently of leptin [29-31], we then examined FASN activity in the hypothalamus of 7-month-old mice. As in the liver (Fig. 5B), Western blot analysis revealed lower IR_β level in hypothalamic lysates of Alb+Cc1^fl/fl relative to age-matched controls (Fig. 9B.i). Consistently, insulin release during refeeding (RF) failed to activate the insulin receptor in Alb+Cc1^fl/fl hypothalami as it did in control mice [assessed by immunoblotting with phospho-IR_β antibody (α-pIR_β)] (Fig. 9B.i, RF vs F). Insulin induced CEACAM1 phosphorylation (pCC1) and detection in the FASN immunopellet in controls but not Alb+Cc1^fl/fl mice (Fig. 9B.ii, RF vs F). This mediated a lower FASN activity in refed normo-insulinemic controls, but not in mutants in which chronic hyperinsulinemia drove higher basal mRNA (Fig. 9B.iii) and protein (Fig. 9B.ii) hypothalamic FASN levels and activity (Fig. 9B.iV), in parallel to a blunted ability of insulin to suppress FASN activity in refed mice (Fig. 9B.iV, RF vs F). Together with suppressed hypothalamic leptin signaling, hyperinsulinemia-driven defect in hypothalamic insulin signaling could contribute to hyperphagia to maintain progression of systemic insulin resistance in Alb+Cc1^fl/fl mice.

CEACAM1 is expressed in the anorexigenic POMC neurons in the arcuate nucleus as well as in other hypothalamic neuronal populations [17]. Because insulin induces Ceacam1 transcription [32], we then tested whether older Alb+Cc1^fl/fl mutants exhibited lower hypothalamic Ceacam1 levels resulting from systemic insulin resistance. As we have shown [33], Ceacam1 mRNA level progressively decreased with age in wild-type livers (Fig. 8B.i). This cause a reduction in hepatic insulin clearance to compensate for age-related decrease in insulin secretion [34] and maintain insulin sensitivity in wild-type mice (Fig. 8B.i). Similarly, hypothalamic Ceacam1 mRNA levels progressively declined, reaching ~30% loss in wild-type mice at 9 months, a level that does not compromise the regulatory effect of CEACAM1 on metabolism (Fig. 8B.ii). In the insulin-resistant mutants; however, the loss in hypothalamic Ceacam1 mRNA levels reached at 9 months (Fig. 8B.ii) the 60% threshold that causes metabolic abnormalities [21]. In contrast, hypothalamic mRNA level of Ceacam2, a close relative to Ceacam1 that is detected in neuropeptide Y-expressing neurons of the dorsomedial hypothalamus [35] and is involved in food intake suppression [22, 36], was not altered with age (Fig. 8B.iii). The marked loss of CEACAM1 in hypothalamic POMC neurons and other hypothalamic neuronal populations could contribute to sustained energy imbalance and systemic insulin resistance in older Alb+Cc1^fl/fl mutants.

4. Discussion

The cause-effect relationship between insulin resistance and hyperinsulinemia remains elusive [37]. Whereas primary insulin resistance causes hyperinsulinemia mainly by inducing a compensatory increase in insulin secretion [38], evidence in support of the causative role of chronic hyperinsulinemia in insulin resistance is mounting [39, 40], in particular when hyperinsulinemia results from impaired insulin clearance [41]. This is underlined by several mechanisms, including downregulation of the insulin receptor under hyperinsulinemic conditions [7, 8]. This paradigm is bolstered by our previous findings that global deletion of Ceacam1 gene impaired receptor-mediated insulin uptake and degradation to cause chronic hyperinsulinemia at the early age of 2 months, followed by downregulation of the insulin receptor and compromised insulin signaling in insulin target tissues to ultimately translate into insulin resistance, primarily hepatic, at about 6 months of age when the mutation was propagated onto the C57BL/6J background, as opposed to a mixed FVB background onto which insulin resistance developed concomitantly with hyperinsulinemia [11]. At 6 months of age, C57BL/6J.Cc1^−/− mice also developed leptin resistance [17] and elevated lipolysis [11], but without any increase in β-cell mass [11], pointing to the relatively minor role of CEACAM1 in insulin secretion and ruling out a significant contribution by insulin secretion to hyperinsulinemia in global mutants. The current studies showed that liver-specific AlbCre+Cc1^fl/fl null mice exhibited a stepwise progression of the pathogenesis of insulin resistance and altered lipid homeostasis. Liver-specific ablation of Ceacam1 gene caused hyperinsulinemia at 2 months of age together with impaired insulin clearance when mice were propagated onto C57BL/6J background. This was not associated with any other metabolic derangement until 6-7 months of age, when insulin resistance, particularly hepatic, arose concomitantly with hepatic steatosis, visceral obesity, and compromised leptin signaling in association with energy imbalance (hyperphagia and reduced spontaneous physical activity). Hyperphagia caused progression to systemic insulin resistance, which became more pronounced at 8-9 months of age, as shown by impaired insulin signaling in WAT and elevated lipolysis. Absence of lipolysis until after hepatic insulin resistance developed in AlbCre+Cc1^fl/fl nulls suggests that elevated plasma NEFA arose secondarily to chronic hyperinsulinemia which could cause insulin resistance in adipose tissue by reducing Glut4-mediated glucose transport [42], as supported by compromised insulin-stimulated Akt phosphorylation in WAT of the older AlbCre+Cc1^fl/fl mutants. In addition, chronic hyperinsulinemia causes insulin resistance to the suppression of plasma NEFA levels and increasing de novo lipogenesis [43]. Given its lipotoxicity effect [28], elevated plasma NEFA could contribute to sustained systemic insulin resistance in older AlbCre+Cc1^fl/fl mice [26]. This is consistent with the ability of L-Carnitine to ameliorate systemic insulin resistance in parallel to restoring plasma NEFA levels without directly affecting insulin degradation in LSACc1 mice with liver-specific inactivation of Ceacam1 [44].

Concomitantly with hepatic insulin resistance, AlbCre+Cc1^fl/fl mice developed hepatic steatosis resulting from increased de novo lipogenesis and reduced fatty acid oxidation. Because the effect of CEACAM1 on FASN activity depends on the prior insulinemic state, it is likely that increased de novo lipogenesis primarily resulted from the hyperinsulinemic state caused by the loss of Ceacam1 in liver. Nonetheless, hepatic steatosis was followed by redistribution of lipid substrates, preferentially to WAT, as expected from the C57BL/6J background [45], to cause visceral obesity. In addition to releasing NEFA, increase in visceral obesity caused a decrease in plasma adiponectin level with a reciprocal increase in leptin release. The former could contribute to sustained systemic insulin resistance in older AlbCre+Cc1^fl/fl mice [46]. The latter could contribute to reduced hypothalamic STAT3 signaling in response to intraperitoneally injected leptin and hyperphagia [47], which in turn, caused systemic insulin resistance, as demonstrated by pair-feeding experiments.

Hyperphagia and reduced locomotor activity could be caused by hyperinsulinemia-driven systemic factors. By downregulating insulin receptors, hyperinsulinemia caused hypothalamic insulin resistance and restricted the ability of insulin to induce CEACAM1 phosphorylation and subsequently, suppress FASN activity. Because high hypothalamic FASN activity causes hyperphagia independently of leptin resistance [29, 30], it is likely that hyperinsulinemia-driven hypothalamic insulin resistance caused hyperphagia, which in turn led to systemic insulin resistance in older AlbCre+Cc1^fl/fl mice [48, 49]. Consistent with hypothalamic insulin resistance being a main determinant of lipolysis [50], it preceded adipocytic insulin resistance and the rise in plasma NEFA in AlbCre+Cc1^fl/fl mice. Thus, in addition to hepatic insulin resistance, chronic hyperinsulinemia drove hypothalamic insulin resistance, which could contribute to progressive systemic insulin resistance by causing hyperphagia and energy imbalance [49]. This central effect of chronic hyperinsulinemia agrees with our previous findings showing restoration of energy balance and all of the metabolic abnormalities by curbing hyperinsulinemia upon liver-specific rescuing of CEACAM1 [16].

Consistent with reduced insulin clearance with aging [4], the current as well as our previous studies [33], showed that hepatic CEACAM1 expression progressively decreased with age in wild-type mice, likely to compensate for the age-related decline in insulin secretion [34] in order to maintain physiologic insulin homeostasis and action. Similarly, hypothalamic Ceacam1 mRNA levels progressively decreased with age in wild-type mice, but by ~35%, which would not cause a significant metabolic abnormality, as expected from the normal phenotype in heterozygous global Cc1^+/− mice [11]. In AlbCre+Cc1^fl/fl mice; however, the age-related loss of hypothalamic Ceacam1 mRNA reached ≥60% at 9 months of age, likely resulting from the parallel progression of systemic insulin resistance and resultant loss of transcriptional upregulation by insulin [32]. Because CEACAM1 is expressed in POMC and other hypothalamic neurons [17] that control energy balance, its age-related significant reduction in this neuronal population could contribute to energy imbalance, and subsequently, to sustained insulin resistance by inducing lipolysis in older AlbCre+Cc1^fl/fl mutants [50]. This hypothesis must await further studies, including deleting Ceacam1 gene in POMC neurons, to be tested.

In summary, our data demonstrated that liver-specific deletion of Ceacam1 primarily caused chronic hyperinsulinemia resulting from impaired insulin clearance, and that this led to hepatic insulin resistance and steatosis, and to hypothalamic insulin resistance, which by triggering energy imbalance, mediated the progression of systemic insulin resistance and altered hypothalamic control of lipolysis that in turn, contributed to sustained systemic insulin resistance in older mice. The earlier onset of hyperinsulinemia and impaired insulin clearance than hepatic insulin resistance followed by lipolysis in this liver-specific loss-of-function AlbCre+Cc1^fl/fl model provided an in vivo demonstration that chronic hyperinsulinemia resulting primarily from impaired insulin clearance can cause secondary hepatic insulin resistance independently of lipolysis. Moreover, by causing hypothalamic insulin resistance, hyperinsulinemia can lead to energy imbalance and progressive systemic insulin resistance.

Highlights.

• Mice with liver-specific deletion of Ceacam1 gene (AlbCre+Cc1^fl/fl) manifest impaired insulin clearance and hyperinsulinemia at a very early age.

• Hyperinsulinemia emerging from impaired insulin clearance causes hepatic insulin resistance, independently of lipolysis

• Hyperinsulinemia drives hypothalamic insulin resistance and subsequently, energy imbalance (hyperphagia and decreased spontaneous locomotor activity)

• Energy imbalance drives progression of systemic insulin resistance

Abbreviations

AlbCre+Cc1^fl/fl

Mice with liver-specific deletion of Ceacam1 gene

AlbCre–Cc1^+/+

Wild-type controls

AlbCre+Cc1^+/+

Albumin-Cre controls

AlbCre–Cc1^fl/fl

Ceacam1-floxed controls

FASN

Fatty acid synthase

IR_β

β subunit of the insulin receptor

NEFA

Non-esterified fatty acids

POMC

Pro-opiomelanocortin

STAT3

Signal transducer and activator of transcription 3

WAT

White adipose tissue