Liver NF-κB-Inducing Kinase Promotes Liver Steatosis and Glucose Counterregulation in Male Mice With Obesity

Abstract

Obesity is associated with chronic inflammation and liver steatoses. Numerous proinflammatory cytokines have been reported to regulate liver glucose and lipid metabolism, thus contributing to the pathogenesis of liver steatosis and/or metabolic dysfunction. Nuclear factor-κB–inducing kinase (NIK) is stimulated by many cytokines and mediates activation of the noncanonical nuclear factor-κB pathway. We previously reported that liver NIK is aberrantly activated in obesity; inactivation of NIK by overexpressing dominant negative NIK(KA) suppresses hepatic glucose production. In the present study, we generated conditional NIK knockout mice using the Cre/loxp system. Mice with hepatocyte-specific or hematopoietic lineage-specific deletion of NIK were normal with either normal chow diet or high-fat diet (HFD) conditions. In contrast, deletion of NIK in the liver, including both hepatocytes and immune cells, protected against HFD-induced liver steatosis and attenuated hepatic glucose production. Mechanistically, deletion of liver NIK suppressed liver inflammation and lipogenic programs, thus contributing to protection against liver steatosis. Liver NIK also downregulated cyclic nucleotide phosphodiesterases, thereby augmenting the cyclic adenosine monophosphate/protein kinase A pathway and glucagon-stimulated hepatic glucose production. Together, our data suggest that NIK pathways in both hepatocytes and immune cells act in concert to promote liver steatosis and glucose production in the setting of obesity.

NIK was conditionally deleted in mice. Deletion of NIK in the liver but not hepatocytes or immune cells alone protects against HFD-induced liver steatosis and hepatic glucose production.

The liver is an essential metabolic organ responsible for endogenous glucose production. Liver glucose production is controlled by a balance between insulin and counterregulatory hormones. Insulin suppresses, and counterregulatory hormones (e.g., glucagon, catecholamines, glucocorticoids, and growth hormone) stimulate, hepatic glucose production. Obesity is associated with chronic inflammation, and numerous proinflammatory cytokines are able to promote liver steatosis and insulin resistance (1–3). However, intracellular mediators, which link liver inflammation to liver steatosis and hepatic glucose production, remain poorly understood.

Nuclear factor-κB–inducing kinase (NIK), also known as MAP3K14, is stimulated by a subset of cytokines and mediates activation of the noncanonical nuclear factor-κB (NF-κB) 2 pathways (4). NIK is ubiquitously expressed at low levels in quiescent cells owing to ubiquitin/proteasome-mediated degradation (4–7). Cytokine stimulation disrupts the association of NIK with TRAF3, an adaptor that recruits NIK to cIAP1/2 ubiquitin E3 ligase complexes, thereby increasing NIK stability and activity (8, 9). CHIP also mediates ubiquitination and degradation of NIK (10). NIK is well known for its ability to regulate immune responses (11–14); however, the metabolic function of NIK remains poorly understood. We recently reported that liver NIK is aberrantly activated in obese mice and enhances hepatic glucose production (15).

The liver is composed of heterogeneous populations of cells, including hepatocytes and nonparenchymal cells. Hematopoietic lineage cells (immune cells) and Kupffer cells (liver resident macrophages) are the main components of the nonparenchymal subpopulations (16). In the present study, we examined the metabolic role of NIK in both hepatocytes and immune cells, using conditional NIK knockout (KO) mice. We found that hepatocyte NIK and immune cell NIK act together to promote liver steatosis and hepatic glucose production in the setting of high-fat diet (HFD)–induced obesity.

Materials and Methods

Animals

Animal experiments were conducted following the protocols approved by the University of Michigan Institutional Animal Care and Use Committee. To generate conditional NIK KO mice, two loxp sites were inserted, through gene targeting in mouse embryonic stem cells, into 2 NIK introns that flank 5 exons, including the exon encoding the ATG start codon [Fig. 1(a)]. To delete NIK in hepatocytes and hematopoietic cells (NIK^∆liver), NIK^flox/flox mice were crossed with Mx1-Cre drivers (Jackson Laboratory, Bar Harbor, ME). NIK^flox/flox;Mx1-Cre progenies were treated with polyinosinic:polycytidylic acid [poly(I:C); 250 μg/mouse intraperitoneally (IP), three times every other day; InvivoGen] to induce Cre expression specifically in hepatocytes and in hematopoietic lineage cells, thereby deleting NIK specifically in these cells. To generate hepatocyte-specific (NIK^∆hep) or myeloid cell-specific (NIK^∆myeloid) NIK KO mice, NIK^flox/flox mice were crossed with albumin-Cre or lysM-Cre drivers (Jackson Laboratory), respectively. Mice were backcrossed with C57BL/6 mice for more than six generations (except for those used to measure liver cyclic adenosine monophosphate [cAMP] levels). The mice were housed with a 12-hour light-dark cycle in the Unit for Laboratory Animal Medicine at the University of Michigan and fed ad libitum either a normal chow diet (calories 9% fat; TestDiet, St. Louis, MO) or a HFD (calories 60% fat; Research Diets, New Brunswick, NJ). The mice were transduced with adenoviral vectors (10¹¹ viral particles/mouse) via tail vein injection (100 µL/mouse).

Figure 1.

Deletion of liver NIK protects against HFD-induced liver steatosis. (a) A schematic representation of conditional deletion of NIK. NIK^∆liver and control (Con) male mice (8 weeks old) were fed a HFD for 10 weeks. (b) Liver, white adipose tissue (WAT), and skeletal muscle extracts were immunoblotted with antibodies against NF-κB2 or tubulin. (c) Growth curves (n = 11). (d) Plasma ALT activity in mice fed a HFD for 10 weeks (n = 9). (e) Plasma TAG levels in NIK^∆liver (n = 11) and control (n = 11) mice fed a HFD for 10 weeks. (f) NIK^∆liver (n = 11) and control (n = 11) mice were fed a HFD for 10 weeks. Liver TAG levels were measured and normalized to liver weight. (g) Hematoxylin and eosin (H&E) and Nile red staining of liver sections. (h) NIK^flox/flox male mice were injected with poly(I:C) (n = 4) or PBS (n = 4) and fed a HFD for 10 weeks. Liver TAG levels were measured and normalized to liver weight. *P < 0.05.

Glucose, insulin, pyruvate, and glucagon tolerance tests

The mice were fasted overnight for the glucose tolerance test (GTT) or 4 to 6 hours for the insulin tolerance test (ITT), glucagon tolerance test (GlucgTT), and pyruvate tolerance test (PTT), and intraperitoneally (IP) injected with human insulin, glucose, pyruvate, or glucagon (the doses are described in the figure legends that appear later), and blood glucose was measured 0, 15, 30, 60, and 120 minutes after injection. Blood samples were collected from tail veins, and plasma insulin, glucagon, and alanine aminotransferase (ALT) were measured using mouse insulin enzyme-linked immunosorbent assay kits (Crystal Chem, Downers Grove, IL), the glucagon radioimmunoassay kit (EMD Millipore, Billerica, MA), and the ALT reagents (Pointe Scientific, Canton, MI), respectively.

Adoptive bone marrow transfers

NIK^flox/flox recipient males (5 weeks old) were treated with GdCl₃ (10 mg/kg body weight IP two times at a 4-day interval) and then with lethal irradiation (2 × 6 Gy, 3 hours apart). They received donor bone marrow cells (2 × 10⁶ cells/mouse) via tail vein injection 6 hours after irradiation. Donor bone marrow cells were harvested from the femurs and tibias of wild-type (WT) or whole body NIK KO mice (5 weeks old) and depleted of red blood cells using a red blood cell lysis buffer (NH₄Cl, 155 mM; KHCO₃, 10 mM; EDTA, 0.1 mM; pH 7.3). The recipients drank acid water (pH 2.6) during GdCl₃ treatment and for an additional 2 weeks (supplemented with 0.1 mg/mL neomycin) after bone marrow transplantation.

Nile red staining, liver triacylglycerol levels, and liver phosphodiesterase activity

For Nile red staining, liver frozen sections were fixed with 4% paraformaldehyde for 20 minutes, washed twice with phosphate-buffered saline (PBS), stained with Nile red (1 µg/mL in PBS) for ∼30 minutes, washed twice with PBS, and visualized using a fluorescent microscope. To measure liver triacylglycerol (TAG) levels, liver samples were homogenized in 1% acetic acid, and liver lipids were extracted using 80% chloroform/methanol (2:1). The organic fractions were dried in a chemical hood, resuspended in 3 M KOH, incubated at 70°C for 1 hour, mixed with MgCl₂ (0.75 M), and centrifuged. The aqueous fractions were used to measure the TAG levels using Free Glycerol Reagent (Sigma-Aldrich, St. Louis, MO). To assess PDE activity, liver samples were homogenized in lysis buffer (20 mM MOPS, 1% Triton X-100, 5 mM EGTA, and 2 mM EDTA with proteasome inhibitors, 1 mM phenylmethylsulfonyl fluoride, 10 mg/mL aprotinin, and 10 mg/mL leupeptin). Liver PDE activity was measured using a PDE activity assay kit following the manufacturer’s protocols (Abcam, Cambridge, UK) and presented as the rates of converting cAMP to 5′-AMP.

Immunoblotting

Tissues were homogenized in a lysis buffer (50 mM Tris HCl [pH 7.5], 1.0% NP-40, 150 mM NaCl, 2 mM EGTA, 1 mM Na₃VO₄, 100 mM NaF, 10 mM Na₄P₂O₇, 1 mM phenylmethylsulfonyl fluoride, 10 mg/mL aprotinin, and 10 mg/mL leupeptin). Tissue or cell extracts were immunoblotted with the indicated antibodies listed in Table 1.

Table 1.

Antibodies Used

Peptide/Protein Target	Antigen Sequence (If Known)	Name of Antibody	Manufacturer, Catalog No. or Name of Source	Species Raised in, Monoclonal or Polyclonal	Dilution Used	RRID
NF-κB2		NF-κB2 p100/p52 antibody	Cell Signaling, #4882	Rabbit, polyclonal	1:1000, 5% BSA in TBST	AB_10828354
Tubulin		α-Tubulin antibody (B-7)	Santa Cruz, sc-5286	Mouse monoclonal	1:5000, 5% BSA in TBST	AB_628411
Phosphorylated CREB (pSer133)		Phospho-CREB (Ser133) (87G3) Rabbit mAb	Cell Signaling, #9198	Rabbit, monoclonal	1:1000, 5% BSA in TBST	AB_2561044
CREB		CREB (48H2) rabbit mAb	Cell Signaling, #9197	Rabbit, monoclonal	1:1000, 5% BSA in TBST	AB_331277
Fasn		Fatty acid synthase (C20G5) rabbit mAb	Cell Signaling, #3180	Rabbit, monoclonal	1:5000, 5% BSA in TBST	AB_2100796
ACC		Acetyl-CoA carboxylase (C83B10) rabbit mAb	Cell Signaling, #3676	Rabbit, monoclonal	1:5000, 5% BSA in TBST	AB_2219397
SREBP1		SREBP-1 antibody (2A4)	Santa Cruz, sc-13551	Mouse, monoclonal	1:1000, 5% BSA in TBST	AB_628282

Abbreviations: BSA, bovine serum albumin; mAB, monoclonal antibody; RRID, Research Resource Identification; TBST, Tris-buffered saline and Tween 20.

Quantitative real-time reverse transcription polymerase chain reaction

Total RNA was extracted using TRIzol reagent (Invitrogen Life Technologies, Carlsbad, CA). First-strand complementary DNA was synthesized using random primers and murine Moloney murine leukemia virus reverse transcription (Promega, Madison, WI). Quantitative polymerase chain reaction (qPCR) was performed using the Absolute QPCR SYBR Mix (Thermo Fisher Scientific Life Sciences, Waltham, MA) and Mx3000P real time PCR system (StrataGene, La Jolla, CA). qPCR primers are listed in Supplemental Table 1 ^{(399.9KB, pdf)} .

cAMP levels

Male mice (9 weeks old; 129Sv/C57BL background) were fasted for 5 hours and stimulated with glucagon (30 µg/kg IP), and the liver was harvested 5 minutes later. Liver samples were lysed with 0.1 M HCl, and cAMP levels were measured using the EIA kit following the manufacturer’s instructions (Cayman Chemicals, Ann Arbor, MI).

Statistical analysis

Data are presented as the mean ± standard error of the mean. Differences between the two groups were analyzed using two-tailed Student’s t tests. The longitudinal data (growth curves, GTT, ITT, PTT, and GlucgTT) were analyzed using analysis of variance and followed by the Bonferroni post-test using Prism 6 (GraphPad Software, La Jolla, CA). P < 0.05 was considered statistically significant.

Results

Liver-specific deletion of NIK attenuates HFD-induced liver steatosis

To examine the function of liver NIK, we generated liver-specific NIK KO (NIK^∆liver) mice by crossing NIK^flox/flox mice with Mx1-Cre drivers. NIK^flox/flox mice were generated by inserting loxp sites that flank 5 exons [Fig. 1(a)]. Mx1-Cre drivers are commonly used to delete floxed genes in hepatocytes, hematopoietic lineage cells, and Kupffer cells on poly(I:C) stimulation (17–19). To verify deletion of liver NIK, NIK^flox/flox;Mx1-Cre mice were intraperitoneally injected with poly(I:C) (NIK^∆liver) or PBS vehicle (controls). NIK messenger RNA (mRNA) levels were substantially lower in the NIK^∆liver than in the control littermates [Supplemental Fig. 1(A) ^{(399.9KB, pdf)} ]. To further confirm liver-specific deletion of NIK, NIK^flox/flox;Mx1-Cre mice (8 weeks old) were fed a HFD for 6 weeks to increase liver NIK levels (15) and then treated with either poly(I:C) or PBS vehicle. Liver NIK mRNA levels were significantly lower in NIK^∆liver than in control littermates [Supplemental Fig. 2(A) ^{(399.9KB, pdf)} ]; notably, deletion efficiency was lower in the HFD-fed than in the chow-fed NIK^∆liver mice. We did not detect NIK proteins in control mice using commercial anti-NIK antibodies (data not shown), likely because the liver NIK levels were below the detection thresholds of the antibodies. Therefore, we analyzed NIK-dependent proteolytic cleavages of NF-κB2 precursor p100. Generation of mature p52 from p100 was blocked in the liver but not in adipose tissue or skeletal muscle of NIK^∆liver mice [Fig. 1(b)], confirming that poly(I:C) treatment induces liver-specific deletion of NIK in NIK^flox/flox;Mx1-Cre mice.

NIK^∆liver mice had a normal body weight with a standard chow diet [Supplemental Fig. 1(B) ^{(399.9KB, pdf)} ]. To assess glucose metabolism, we performed GTT, ITT, PTT, and GlucgTT and did not detect substantial differences between NIK^∆liver and control littermates [Supplemental Fig. 1(C) ^{(399.9KB, pdf)} ]. The liver weight and TAG levels were also comparable between these two groups [Supplemental Fig. 1(D) ^{(399.9KB, pdf)} ]. Next, we placed NIK^flox/flox;Mx1-Cre mice on a HFD for 6 weeks and then injected them with poly(I:C) (NIK^∆liver) or PBS vehicle (control). The body weight [Fig. 1(c)], plasma ALT activity [Fig. 1(d)], and plasma TAG levels [Fig. 1(e)] were similar between NIK^∆liver and control mice. To assess the role of NIK in hepatic lipid metabolism, we measured hepatic TAG levels. Liver TAG content was significantly lower in NIK^∆liver mice relative to control littermates [Fig. 1(f)]. To corroborate these findings, we analyzed hematoxylin and eosin or Nile red (neutral lipid dye) staining of liver sections. Lipid droplets were less in number and smaller in size in NIK^∆liver than in control littermates [Fig. 1(g)]. To exclude the possibility that poly(I:C) treatment might nonspecifically mitigate liver steatosis, NIK^flox/flox mice were fed a HFD and treated with poly(I:C). Poly(I:C) treatment did not decrease liver TAG content in NIK^flox/flox mice [Fig. 1(h)]. These data indicate that deletion of liver NIK protects against HFD-induced liver steatosis in mice.

Liver-specific deletion of NIK suppresses hepatic lipogenic program

We next sought to identify the underlying mechanism whereby NIK deficiency protects against liver steatosis. We measured the expression of key genes that control fatty acid uptake (e.g., CD36 and FATP5), oxidation (e.g., PGC1α, PPARα, and CPT1α), synthesis (e.g., SREBP1c, LXRα, ChREBP, ChREBPβ, Fasn, SCD1, and PPARγ), lipolysis (e.g., ATGL and LPL), and very-low-density lipoprotein secretion (MTTP and ApoE) and did not detect differences between control and NIK^∆liver mice fed a normal chow diet [Supplemental Fig. 1(A) ^{(399.9KB, pdf)} ]. Next, we placed NIK^∆liver and control littermates on a HFD and observed that the mRNA abundance of liver Fasn, SREBP1c, and LXRα was significantly lower in NIK^∆liver than in control mice [Fig. 2(a)]. To validate these findings, we measured key lipogenic regulator protein levels by immunoblotting liver extracts. Fasn, ACC1, and SREBP1c levels were significantly lower in NIK^∆liver than in control mice [Fig. 2(b)]. In agreement with these results, the mRNA and protein abundance of liver Fasn, ACC1, and SREBP1c was lower in poly(I:C)-treated NIK^flox/flox;Mx1-Cre mice than in poly(I:C)-treated NIK^flox/flox mice fed a HFD [Supplemental Fig. 3(A) and 3(B) ^{(399.9KB, pdf)} ]. These data suggest that deletion of liver NIK suppresses the hepatic lipogenic program, thereby attenuating obesity-associated liver steatosis.

Figure 2.

Deletion of liver NIK inhibits the hepatic lipogenic program. NIK^∆liver and control (Con) male mice were fed a HFD for 10 weeks and euthanized under nonfasted conditions. (a) Gene expression in the liver was quantified by qPCR and normalized to 36B4 expression. (NIK^∆liver, n = 4; control, n = 4). (b) Liver extracts were immunoblotted with the indicated antibodies. ACC1, Fasn, and SREBP1 protein levels were normalized to tubulin levels. *P < 0.05.

Liver-specific deletion of NIK attenuates hepatic glucose production in obese mice

We previously reported that adenoviral-mediated overexpression of dominant negative NIK(KA) inhibits hepatic gluconeogenesis in obese mice (15). To determine whether poly(I:C)-induced deletion of liver NIK similarly decreases hepatic glucose production, NIK^flox/flox;Mx1-Cre male mice were fed a HFD for 6 weeks and then treated with poly(I:C) (NIK^∆liver) or PBS vehicle (control). PTT and GlucgTT were performed 3 weeks after poly(I:C) treatment. Blood glucose levels (by analysis of variance) and areas under the curve were significantly lower in NIK^∆liver mice relative to control mice [Fig. 3(a) and 3(b)]. As an additional control, poly(I:C) administration did not alter PTT and GlucgTT in HFD-fed NIK^flox/flox mice [Fig. 3(c)]. Notably, glucagon resistance was more severe in NIK(KA) adenoviral-transduced mice than in NIK^∆liver mice (Supplemental Fig. 4 ^{(399.9KB, pdf)} ). Moreover, NIK(KA) (Supplemental Fig. 4 ^{(399.9KB, pdf)} ) but not NIK^∆liver [Fig. 3(d)] mice displayed glucose intolerance. Because poly(I:C) induces incomplete deletion of liver NIK [Supplemental Fig. 2(A) ^{(399.9KB, pdf)} ], the residual NIK activity in the liver is likely to be greater in NIK^∆liver mice than in NIK(KA)-overexpressing mice, which might explain the discrepancy between these two models. To test this idea, NIK^∆liver mice were fed a HFD and transduced with NIK(KA) or β-galactosidase adenoviral vectors. Overexpression of NIK(KA) still attenuated the hyperglycemic responses to glucagon and pyruvate in NIK^∆liver mice (Supplemental Fig. 4 ^{(399.9KB, pdf)} ).

Figure 3.

Deletion of liver NIK suppresses glucose counterregulation. (a and b) NIK^∆liver (n = 9) and control (Con) (n = 9) male mice were fed a HFD for 9 weeks. PTT (sodium pyruvate, 1 g/kg body weight) and GlucgTT (glucagon, 15 µg/kg) were performed and areas under the curve (AUCs) calculated. (c) NIK^flox/flox male mice (8 weeks old) were injected with poly(I:C) (n = 4) or PBS (n = 4) and fed a HFD for 10 weeks. PTT and GlucgTT were performed. (d–f) NIK^∆liver (n = 9) and control (n = 9) male mice (8 to 9 weeks old) were fed a HFD for 10 weeks. GTT (glucose, 1 g/kg), ITT (insulin, 1 U/kg), blood glucose, and plasma insulin and glucagon were analyzed. (g) Gene expression in the liver was quantified by qPCR and normalized to 36B4 expression (NIK^∆liver, n = 4; control, n = 4; *P < 0.05). Abbreviations: IL, interleukin; MCP1, monocyte chemoattractant protein-1; TNF, tumor necrosis factor.

We did not detect GTT and ITT differences between HFD-fed NIK^∆liver and control mice [Fig. 3(d) and 3(e)]. Overnight fasting blood glucose, insulin, and glucagon levels were also similar between these two groups [Fig. 3(f)]. These data are consistent with our previous findings that hepatocyte NIK does not regulate insulin sensitivity. To determine whether liver inflammation is involved in NIK action, we measured the expression of proinflammatory cytokines using qPCR. Deletion of liver NIK markedly reduced the expression of liver interleukin-1β, tumor necrosis factor-α, monocyte chemoattractant protein-1, and F4/80 in HFD-fed NIK^∆liver mice relative to control littermates [Fig. 3(g)].

Deletion of liver NIK increases hepatic cyclic nucleotide PDE activity

To explore the underlying mechanism by which NIK enhances glucagon-stimulated hepatic glucose production, we assessed phosphorylation of liver cAMP response element binding protein (CREB), a key gluconeogenic transcription factor. Phosphorylation of CREB (pSer133), which is mediated by the cAMP/protein kinase A (PKA) pathway, was lower in HFD-fed NIK^∆liver mice relative to control mice [Fig. 4(a)]. As an additional control, CREB phosphorylation was lower in poly(I:C)-treated NIK^flox/flox;Mx1-Cre than in poly(I:C)-treated NIK^flox/flox mice [Supplemental Fig. 3(B) ^{(399.9KB, pdf)} ]. To further analyze the cAMP/PKA pathway, WT and NIK KO male mice (9 weeks old) were stimulated with glucagon or PBS vehicle for 5 minutes. Liver cAMP levels were significantly lower in KO relative to WT mice after glucagon stimulation [Fig. 4(b)].

Figure 4.

NIK regulates the PDE/cAMP/PKA pathway. (a, c, and d) NIK^∆liver and control (Con) male mice were fed a HFD for 10 weeks and euthanized under nonfasted conditions. (a) Liver extracts were immunoblotted with antibodies against phosphorylated CREB (pCREB; pSer133) or CREB. pCREB levels were normalized to total CREB levels. (b) WT and NIK KO male mice were stimulated with glucagon (30 µg/kg) for 5 minutes. Liver cAMP levels were normalized to liver weight (n = 3). (c) Liver PDE activity (normalized to liver protein levels) in NIK^∆liver (n = 8) and control (n = 8) male mice. (d) Liver gene expression was measured in NIK^∆liver (n = 4) and control (n = 4) male mice by qPCR and normalized to 36B4 levels. *P < 0.05.

Intracellular cAMP levels are negatively regulated by PDEs (20). We observed that liver PDE activity was significantly higher in HFD-fed NIK^∆liver mice relative to control mice [Fig. 4(c)]. Consistently, PDE3B expression was significantly higher in NIK^∆liver than in control mice [Fig. 4(d)]. In line with these findings, PDE3B expression was also higher in poly(I:C)-treated NIK^flox/flox;Mx1-Cre mice than in poly(I:C)-treated NIK^flox/flox mice [Supplemental Fig. 3(A) ^{(399.9KB, pdf)} ]. These results suggest that NIK deficiency in the liver increases hepatic PDE3B expression and PDE activity, thereby inhibiting the cAMP/PKA pathway, glucose counterregulation, and hepatic glucose production.

Deletion of NIK in hepatocytes or hematopoietic lineage cells alone is insufficient to protect against HFD-induced liver steatosis

To determine the role of hepatocyte NIK, we generated hepatocyte-specific NIK KO (NIK^∆hep) mice by crossing NIK^flox/flox mice with albumin-Cre drivers. NIK^∆hep and NIK^flox/flox male mice were fed a HFD for 10 weeks. The body weight and overnight fasting plasma insulin and glucagon levels were similar between NIK^∆hep and NIK^flox/flox mice [Fig. 5(a) and 5(b)]. ITT, GTT, PTT, GlucgTT, and liver TAG levels were comparable between these two groups [Fig. 5(c) and 5(d)]. The expression of cytokines and lipogenic genes in the liver was also similar between NIK^∆hep and NIK^flox/flox littermates [Fig. 5(e)].

Figure 5.

Hepatocyte-specific deletion of NIK does not alter liver glucose and lipid metabolism. NIK^∆hep (n = 6) and NIK^flox/flox (n = 6) male mice (7 to 8 weeks old) were fed a HFD for 16 weeks. (a) Growth curves. (b) Overnight fasting plasma insulin and glucagon levels (HFD for 10 weeks). (c) GTT (glucose, 1 g/kg; HFD for 9 weeks), ITT (insulin, 0.7 U/kg; HFD for 9 weeks), PTT (pyruvate, 1 g/kg; HFD for 10 weeks), and GlucgTT (glucagon, 15 µg/kg; HFD for 10 weeks). (d) Liver TAG levels (normalized to liver weight; HFD for 10 weeks) under nonfasted conditions. (e) Liver gene expression was measured in NIK^∆hep (n = 4) and NIK^flox/flox (n = 4) male mice by qPCR and normalized to 36B4 levels. Abbreviations: IL, interleukin; MCP1, monocyte chemoattractant protein-1; PPAR, peroxisome proliferator-activated receptor; TNF, tumor necrosis factor.

In addition to hepatocytes, the liver contains Kupffer cells, hematopoietic lineage cells, and several other cell types. To assess the role of NIK in immune cells, we generated myeloid cell-specific NIK KO (NIK^∆myeloid) mice by crossing NIK^flox/flox mice with lysM-Cre drivers. NIK^∆myeloid and NIK^flox/flox males were fed a HFD for 15 weeks. The body weight and GTT and ITT results were comparable between NIK^∆myeloid and NIK^flox/flox mice (data not shown). Also, the liver TAG levels were similar between these two groups [Fig. 6(a)]. To assess the role of NIK in all hematopoietic lineage cells, we prepared bone marrow from WT or whole body NIK KO mice. NIK^flox/flox recipients were pretreated with lethal radiation to destroy endogenous hematopoietic lineage cells and then reconstituted with WT or KO donor bone marrow. The chimeric mice were fed a HFD for 12 weeks. Liver TAG levels were indistinguishable between WT and KO bone marrow recipients [Fig. 6(b)].

Figure 6.

NIK deficiency in hematopoietic lineage cells does not protect against HFD-induced liver steatosis. (a) NIK^∆myeloid (n = 4) and NIK^flox/flox (n = 6) male mice (7 to 8 weeks old) were fed a HFD for 15 weeks, and liver TAG levels (under nonfasted conditions) were measured and normalized to liver weight. (b) WT (n = 9) or NIK KO (n = 8) donor bone marrow was transferred into NIK^flox/flox male recipients. The chimeric mice were fed a HFD for 12 weeks. Liver TAG levels were measured under nonfasted conditions. (c) A model of NIK action. Obesogenic factors activate NIK in both hepatocytes and liver immune cells. NIK directly stimulates hepatocytes to release mediators that activate immune cells; reciprocally, NIK also directly stimulates immune cells to secrete factors that regulate hepatocyte metabolism. This hepatocyte–immune cell crosstalk promotes liver steatosis and glucose counterregulation in obesity.

Discussion

Liver-specific overexpression of dominant negative NIK(KA) suppresses liver glucose production in obese mice (15). In agreement with these findings, we found that liver-specific deletion of NIK attenuated glucagon and pyruvate-stimulated hepatic glucose production. Importantly, we uncovered a previously unknown mechanism by which liver NIK promotes glucose counterregulation. Liver NIK decreases hepatic PDE3B expression and PDE activity, thus increasing activation of the cAMP/PKA pathway and hepatic glucose production. In line with our findings, NF-κB1, which is activated by NIK, was reported to suppress PDE3B transcription (21).

We observed that adenoviral-mediated overexpression of dominant negative NIK(KA) suppressed hepatic glucose production to a greater extent than poly(I:C)-induced deletion of liver NIK in NIK^∆liver mice. Furthermore, NIK(KA) further reduced hepatic glucose production in NIK^∆liver mice. These observations suggest that poly(I:C)-induced deletion of liver NIK is incomplete in NIK^∆liver mice. Liver NIK mRNA levels were reduced by ∼50% in NIK^∆liver mice relative to control mice. Hence, the metabolic function of liver NIK was underestimated under our experimental conditions using NIK^∆liver mouse models.

We found that deletion of liver NIK substantially attenuated HFD-induced liver steatosis in NIK^∆liver mice. Expression of key lipogenic genes, but not the genes that control fatty acid uptake, β-oxidation, or very-low-density lipoprotein secretion, was considerably lower in NIK^∆liver mice. These data suggest that liver NIK promotes liver steatosis by stimulating the hepatic lipogenic program. Notably, body weight and insulin sensitivity were similar between HFD-fed NIK^∆liver and control mice. Expression of proinflammatory cytokines in the liver was lower in NIK^∆liver mice, suggesting that liver NIK promotes hepatic lipogenesis, at least in part, through augmenting liver inflammation.

Deletion of NIK in either hepatocytes or immune cells alone was unable to protect against HFD-induced liver steatosis. Hepatocyte glucose production was also similar between HFD-fed NIK^∆liver (or NIK^∆myeloid) and NIK^flox/flox mice. Hence, NIK pathways in multiple liver cell types, including hepatocytes, Kupffer cells, and other immune cells, might act in concert to promote liver steatosis and/or glucose counterregulation in the setting of obesity [Fig. 6(c)]. Activation of hepatocyte NIK is known to induce secretion of mediators that activate Kupffer cells/macrophages (22). Activation of NIK pathways in Kupffer cells and/or other immune cells might regulate hepatocyte metabolism by a paracrine mechanism. Thus, NIK is likely to regulate hepatocyte-immune cell crosstalk, thereby promoting liver steatosis and hepatic glucose production in obesity.