Upregulation of the serine palmitoyltransferase subunit SPTLC2 by endoplasmic reticulum stress inhibits the hepatic insulin response

Endoplasmic reticulum (ER) stress is induced by various conditions, such as inflammation and the presence of excess nutrients. Abnormal accumulation of unfolded proteins leads to the activation of a collective signaling cascade, termed the unfolded protein response (UPR). ER stress is reported to perturb hepatic insulin response metabolism while promoting insulin resistance. Here, we report that ER stress regulates the de novo biosynthesis of sphingolipids via the activation of serine palmitoyltransferase (SPT), a rate-limiting enzyme involved in the de novo biosynthesis of ceramides. We found that the expression levels of Sptlc1 and Sptlc2, the major SPT subunits, were upregulated and that the cellular concentrations of ceramide and dihydroceramide were elevated by acute ER stress inducers in primary hepatocytes and HepG2 cells. Sptlc2 was upregulated and ceramide levels were elevated by tunicamycin in the livers of C57BL/6J wild-type mice. Analysis of the Sptlc2 promoter demonstrated that the transcriptional activation of Sptlc2 was mediated by the spliced form of X-box binding protein 1 (sXBP1). Liver-specific Sptlc2 transgenic mice exhibited increased ceramide levels in the liver and elevated fasting glucose levels. The insulin response was reduced by the inhibition of the phosphorylation of insulin receptor β (IRβ). Collectively, these results demonstrate that ER stress induces activation of the de novo biosynthesis of ceramide and contributes to the progression of hepatic insulin resistance via the reduced phosphorylation of IRβ in hepatocytes.


INTRODUCTION
Overnutrition-induced obesity is a risk factor for metabolic diseases, such as type II diabetes, hypertension, and coronary heart disease [1][2][3] . Elevated plasma free fatty acids (FFAs) lead to unregulated uptake by the liver, resulting in the development of nonalcoholic fatty liver disease 4 . Elevated levels of FFAs contribute to the induction of insulin resistance and diabetes by modulating glucose and lipid metabolism in skeletal muscle, liver, and adipose tissues 5 . Under obese conditions with excess FFAs in the circulation, saturated fatty acids (FAs) contribute to lipotoxicity via the synthesis of specific lipid metabolites that act as signaling molecules in response to saturated FFAs 6 . These lipid metabolites include diacylglycerol (DAG), ceramide or their metabolites that activate serine/threonine kinases or phosphatases that inhibit the insulin receptor substrate/AKT-dependent insulin signaling cascades in insulin-sensitive tissues 7 .
Ceramide is a sphingolipid metabolite and a major component of the plasma membrane and lipoproteins. De novo synthesis of ceramide starts from the condensation of serine and palmitoyl CoA. FAs are converted to ceramide via a series of reactions by serine palmitoyltransferase (SPT), 3-ketosphinganine reductase, ceramide synthase, and dihydroceramide desaturase. This de novo sphingolipid biosynthesis is activated by excess cellular palmitate and hyperlipidemic conditions in apoE-deficient livers 8,9 . Although this pathway is likely to be activated in the setting of obesity and diabetes, its mechanism of regulation remains uncertain.
SPT is a rate-limiting step in the de novo ceramide biosynthesis pathway and is composed of two main subunits, namely, Sptlc1 and Sptlc2, encoding 53-and 63-kDa proteins, respectively 10 . Deficiency of Sptlc1 or Sptlc2 is lethal in mice, and only heterozygous mice or tissue-specific null mice can survive 11,12 . Despite the presence of a third unit, Sptlc3, its function remains unknown 13 . Each subunit is stabilized by forming a dimer or multimer in the endoplasmic reticulum to produce ceramide 14 .
When cellular ceramide is increased by excess FFAs, protein phosphatase 2A is activated and dephosphorylates AKT, leading to inhibition of insulin action 15 . Phosphorylation of the AKT pleckstrin domain prevents binding to 3-phosphoinositides that activate AKT 16 . Taken together, cellular glucose uptake is reduced by ceramide-mediated inhibition of insulin signaling. In this manner, the formation of ceramide is associated with the development of metabolic dysfunctions, such as insulin resistance.
Pharmacological or genetic interventions to inhibit the biosynthesis of ceramide will help to alleviate various metabolic abnormalities, including atherosclerosis, cardiomyopathy, and diabetes [17][18][19] .
The endoplasmic reticulum (ER) is the key organelle responsible for the folding and modification of proteins and protein transport to other cellular compartments 20,21 . ER stress is induced by inflammation and nutrient overload that disrupts ER homeostasis due to the accumulation of unfolded or misfolded proteins in the ER lumen 21,22 . To manage this organelle dysfunction, cells activate a collective signaling cascade referred to as the unfolded protein response (UPR) to relieve ER-associated stress 23 . The conditions that elicit ER stress include glucose or nutrient deprivation, viral infections, disturbance of calcium homeostasis, and excess lipid accumulation 24 . Obesity induces ER stress and plays an important role in the development of insulin resistance and diabetes by activating c-Jun N-terminal kinase (JNK) via activation of inositolrequiring enzyme 1 alpha (IRE1α) and the X-box binding pathway and inhibition of insulin receptor substrate 25,26 .
In this study, we investigated whether ER stress could induce the de novo biosynthesis of ceramide and inhibit the insulin response. Sptlc2, a subunit of SPT, was induced by ER stress, and the in vivo effect of ceramide was investigated using liver-specific Sptlc2 transgenic mice. Collectively, the "sphingolipid rheostat," the balance of ceramide and sphingosine 1-phosphate (S1P) that is necessary for cell survival, is independently regulated by the UPR pathway and contributes to the regulation of metabolic function.

Primary hepatocyte culture
Primary hepatocytes were isolated from 6-8-week-old C57BL/6 mice (Charles River Laboratories, Wilmington, MA, USA) as described previously 27 . Briefly, after the abdomen was opened, the intestines were displaced to the left, and the portal vein was exposed. The chest was opened, and the hepatic vein was cannulated in a retrograde fashion via an incision in the right atrium with a 24-gauge needle connected to a variable-speed pump that delivered perfusion solution 1 (142 mM NaCl, 6.7 mM KCl, and 10 mM HEPES) at a rate of 8 mL/min. The hepatic portal vein was cut to allow the perfusion fluid to escape. After 2 min, perfusion solution 1 was switched to perfusion solution 2 (66.7 mM NaCl, 6.7 mM KCl, 100 mM HEPES, and 4.8 mM CaCl 2 ·2H 2 O) containing 25 mg of collagenase type 4 (Sigma) and 2% bovine serum albumin (BSA) in 50 mL, and the perfusion was continued for another 2-3 min at the same rate. Perfused liver was then excised and transferred to a 60-mm culture dish containing the primary cell culture medium. The liver was broken apart gently with forceps. Cells were counted and plated in medium 199 (Sigma) supplemented with 10% fetal bovine serum (FBS), 1 U/mL penicillin, 1 µg/mL streptomycin, 10 nM dexamethasone, and 23 mM HEPES. The cells were incubated at 37°C in a 5% carbon dioxide (CO 2 ) atmosphere.

Cell culture
Human hepatocellular carcinoma HepG2 cells were obtained from the American Type Culture Collection (ATCC) (Manassas, VA, USA) and cultured in Dulbecco's modified Eagle's medium (DMEM; Welgene, Korea) supplemented with 10% heat-inactivated FBS, 1 U/mL penicillin, and 1 μg/mL streptomycin. The cells were incubated at 37°C in a 5% CO 2 atmosphere. Human embryonic kidney cells (HEK-293 cells) were used to generate the recombinant adenovirus and to determine the virus titers.
Generation of liver-specific Sptlc2 transgenic (lSptlc2-Tg) mice and animal experiments A 1.7-kb fragment of human Sptlc2 cDNA was inserted into the MluI and XhoI sites of the pLIV-7 plasmid. A linearized fragment of the construct containing the promoter, first exon, first intron, and part of the second exon of the human apoE gene, the human Sptlc2 cDNA, the adenylation sequence and the hepatic control region of the apoE/C-I gene was isolated and microinjected into C57BL/6 J mouse embryos. These embryos were implanted into pseudopregnant C57BL/6J foster mothers. Primers used in the generation of Sptlc2 cDNA with restriction sites were as follows: 5′-AAAAACGCGTATGCGGCCGGAGCCCGGA-3′ and 5′-AAAACTC-GAGTCAGTCTTCTGTTTCTTC-3′. Founder animals confirmed by Sptlc2 genotyping were propagated by breeding with C57BL/6J wild-type (WT) mice. Male C57BL/6J mice (8-10 weeks old; Charles River Laboratories) were fed a normal chow diet (NCD; LabDiet, PICO Lab Rodent Diet 20), a 60% high-fat diet (HFD, D12492, Research Diet Inc., NJ, USA), or a highfructose diet (60% fructose; Central Lab Animal Inc., Korea). ER stress was induced by intraperitoneal injection of tunicamycin (2.5 μg/g body wt) for the indicated times in WT C57BL/6J mice (Charles River Laboratories). At the indicated times, the mice were killed by CO 2 inhalation, and the livers were harvested to measure sphingolipids and gene expression. To examine the insulin response, an insulin tolerance test (ITT) and glucose tolerance test (GTT) were performed after 9 weeks of feeding. For the GTT, mice were fasted for 16 h, and glucose (2 g/kg body wt) was injected intraperitoneally. For the ITT, mice were starved for 4 h, and insulin (0.75 U/ kg body wt, Eli Lilly, Toronto, Canada) was injected intraperitoneally. All procedures were approved by the Gachon University of Medicine Institutional Animal Care and Use Committee (IACUC-R2019011).

Quantitative reverse-transcription polymerase chain reaction (PCR)
Total RNA from either mouse primary hepatocytes or HepG2 cells was extracted using an easy-spin (DNA free) Total RNA Extraction Kit (iNtRON Biotechnology, Korea), and the mouse livers were homogenized in TRIzol reagent (Invitrogen, Carlsbad, CA, USA). cDNA was synthesized by the iScript cDNA synthesis Kit (Bio-Rad, CA, USA). Real-time PCR analysis was performed in ABI7300 equipment (Applied Biosystems Inc., Carlsbad, CA, USA) using SYBR Green Master Mix (Toyobo, Japan). The mRNA expression levels were expressed as a ratio normalized to those of β-actin 28

SPT activity analysis
HepG2 cells were grown to 80-90% confluency and treated with 1.25 μg/mL tunicamycin or 0.5 mmol/L thapsigargin for 24 h. Harvested cells were homogenized in 100 mM HEPES (pH 8.0), 0.5 mM EDTA (pH 8.0), protease inhibitors, and 10% (w/v) sucrose monolaurate. SPT activity was measured with L-serine and palmitoyl CoA as substrates as described previously by Rutty et al. 29 . Briefly, the SPT reaction was initiated by adding 50 µM palmitoyl CoA, 5 mM L-serine, and 20 µM pyridoxal 5′-phosphate to the cell extracts containing 200 µg of protein. After incubation for 60 min at 37°C, the reactions were terminated by adding 50 μL of NaBH 4 (5 mg/mL) and reacted for 5 min at room temperature. Lipid fractions were extracted by the addition of methanol/KOH:CHCl 3 (4:1, v/v), CHCl 3 , alkaline water, and 2 N NH 4 OH, and the lower organic layer was isolated and dried under N 2 gas. The residues were redissolved in 150 µL of methanol:ethanol:H 2 O (85:47.5:17.5). Then, the samples were analyzed by LC-MS/MS for sphinganine.

Luciferase reporter assay
HepG2 cells (2 × 10 5 cells/mL) were cultured in 24-well culture plates to 80% confluency for transfection experiments. Cells were transiently cotransfected with pSptlc2-pGL3-Luc and pcDNA3 vector containing ATF4 or sXBP1, respectively, using Lipidofect-P reagent (Lipidomia, Sungnam, Korea) according to the manufacturer's instructions. An expression vector for Renilla luciferase, pTK-RL, was used to normalize the transfection efficiency. After incubation for 24 h, the cells were washed with PBS once and lysed in passive lysis buffer, and luciferase and Renilla activities were measured using the Dual-Luciferase Assay System (Promega, Madison, WI, USA) according to the manufacturer's instructions 28 .

Chromatin immunoprecipitation (ChIP) assay
HepG2 cells were treated with tunicamycin (1.25 µg/ml) for 2 h. Cells were isolated, and ChIP assays were performed according to the manufacturer's protocol (Abcam, Cambridge, UK). HepG2 cells (3 × 10 6 ) were fixed with 1.1% formaldehyde for 10 min at room temperature and stopped with glycine reagent. Chromatin samples, collected by SDS lysis buffer with protease inhibitor, were sheared with sonication for 7 min on ice. The sheared chromatin samples were immunoprecipitated with anti-sXBP1 antibody (Cell Signaling, Danvers, MA, USA) or IgG as a control overnight with rotation at 4°C. The chromatin/antibody samples were bound to protein A beads, centrifuged to remove insoluble material, and the chromatin/antibody/protein A beads were collected. The DNA was purified with proteinase K. PCR was carried out on 1 µl of DNA using primers for the sXBP1 promoter. The PCR conditions were 95°C for 5 min; followed by 40 cycles of 95°C for 30 s, 61°C for 30 s and 72°C for 1 min; and a final extension at 72°C for 7 min. Gels were run on a 1% agarose gel and analyzed with an SML-01 LED Transilluminator (MAESTROGEN, Taiwan).

Statistical analysis
Data are generally expressed as the mean ± standard error of the mean (SEM). The difference between the 2 groups was analyzed by Student's t-test. A value of less than 0.05 was considered significant.

Activation of SPT by ER stress elevates the levels of sphingolipids in HepG2 cells
To determine whether SPT is activated by ER stress in primary hepatocytes and HepG2 cells, we treated primary mouse hepatocytes and human hepatoma HepG2 cells with tunicamycin, an acute ER stress inducer. Tunicamycin upregulated the expression levels of Sptlc1, Sptlc2 and UPR genes, such as ATF4, CHOP, GRP78, the spliced form of XBP1 (sXBP1) and the unspliced form of XBP1 (uXBP1), in primary mouse hepatocytes. We also observed upregulation of the expression levels of Sptlc1 and Sptlc2 in HepG2 cells (Fig. 1a, b). Furthermore, we measured SPT activity in HepG2 cells treated with tunicamycin and found that SPT activity was increased (Fig. 1c). Next, we examined the expression of Sptlc1 and Sptlc2 in HepG2 cells treated with tunicamycin by immunoblotting (Fig. 1d). The expression of Sptlc1 protein was increased at 24 h, and that of Sptlc2 protein was increased at 12 h. These data suggest that the subunits of SPT, namely, Sptlc1 and Sptlc2, were upregulated by ER stress in hepatocytes.
To assess whether ER stress regulated the metabolism of sphingolipids via the upregulation of SPT, we measured sphingolipid metabolites by LC-MS/MS in HepG2 cells after treatment with tunicamycin or thapsigargin, which are ER stress inducers. Ceramide, dihydroceramide, and sphingomyelin (SM) levels in tunicamycin-or thapsigargin-treated HepG2 cells were significantly elevated (Fig. 1e-g). However, sphinganine (SA) and sphingosine (SO) levels were only elevated in thapsigargintreated HepG2 cells and were not altered in tunicamycin-treated HepG2 cells (Fig. 1h, i). These results indicated that ER stress upregulates the transcription of SPT subunits, leading to the activation of de novo sphingolipid biosynthesis in hepatocytes.

SPT subunits are upregulated by ER stress in vivo
To determine whether SPT is upregulated by ER stress in vivo, we injected tunicamycin intraperitoneally into C57BL/6 J WT mice, and the livers were isolated at various times postinjection. All ER stress markers, that is, ATF4, ATF6, uXBP1, sXBP1, CHOP, and GRP78, were upregulated by tunicamycin in a time-dependent manner ( Fig. 2a-f). In addition, the mRNA expression levels of Sptlc1 and Sptlc2 were upregulated by tunicamycin treatment in the livers of mice treated with tunicamycin in a time-dependent manner (Fig. 2g, h).
To assess whether ER stress regulates sphingolipid metabolism, we measured sphingolipid metabolites in mouse livers treated with tunicamycin by LC-MS/MS. Total ceramide, dihydroceramide, SM, SA, and SO levels in the livers of tunicamycin-treated mice were elevated in a time-dependent manner (Fig. 3a-e). These results suggest that the expression of SPT subunits is upregulated and that de novo sphingolipid biosynthesis is activated by ER stress. sXBP1 upregulates the expression of Sptlc2 in HepG2 cells HFD consumption induces metabolic dysfunction and insulin resistance via the activation of ER stress [31][32][33][34] . To examine whether ER stress induced by hyperlipidemic conditions regulates Sptlc2, 8-week-old C57BL/6J WT male mice were fed a normal chow diet (NCD) or a HFD (60% of kcal from fat) for 4 weeks. Mouse livers were isolated, and mRNA expression was measured. Under hyperlipidemic conditions, the expression of Sptlc1 and Sptlc2 was upregulated. HFD feeding increased hepatic sXBP1 activation but decreased ATF4, and no transcriptional changes in other UPR genes were found. Consistent with the mRNA results, the protein expression levels of Sptlc1, Sptlc2, and sXBP1 were increased in the livers from mice fed a HFD compared to levels in the livers from control animals (Fig. 4a, b).
Prolonged feeding of a HFD and a high-sucrose diet are known to increase de novo lipid synthesis in the liver via the activation of sXBP1 33,34 . We examined whether XBP1 activation regulates Sptlc2 by giving the mice a fructose diet (FD). WT mice were fed a FD for 2 d or 7 d, and the liver was isolated to measure Sptlc2 expression. While no change was found in the mRNA expression of the SPT subunits and ER stress markers after FD feeding (Fig. 4c), Sptlc2 protein was increased after 2 days of FD feeding (Fig. 4d). As reported previously, the FD induced sXBP1 protein expression after 2 days of FD feeding (Fig. 4d). However, 7-day fructose feeding had no effect on Sptlc1 or Sptlc2. These results indicate that sXBP1 upregulates Sptlc2 under FD-induced physiological ER stress conditions. Next, we examined which UPR pathways are involved in the regulation of Sptlc2 expression and the de novo biosynthesis of sphingolipids. To investigate which UPR gene is involved in Sptlc2 upregulation, the promoter of Sptlc2 was cloned into the pGL3-luciferase plasmid to make a reporter construct (pSptlc2-luc) to measure the transcriptional activity of Sptlc2. The pSptlc2-luc reporter plasmid was cotransfected with the pcDNA3.1 plasmid containing ATF4 or sXBP1 into HepG2 cells. Although ATF4 cotransfection had no effect on the expression of Sptlc2, the transcriptional activity of pSptlc2 was enhanced by sXBP1 cotransfection compared to that of the pcDNA3.1 control Fig. 1 Upregulation of serine palmitoyltransferase long chain base subunit (Sptlc)-1 and Sptlc2 by ER stress inducers in primary hepatocytes and HepG2 cells. Mouse primary hepatocytes or HepG2 cells were treated with dimethyl sulfoxide (DMSO) or tunicamycin (1.25 µg/mL) for 24 h. Isolated mRNA was analyzed by quantitative reverse-transcription polymerase chain reaction (qRT-PCR) (a, b), and enzyme activity was measured using the cell extracts (c). HepG2 cells were treated with tunicamycin for 24 h, and the Sptlc1 and Sptlc2 proteins were analyzed by immunoblotting (d). Total ceramides (Cer) (e), dhCer (f), sphingomyelin (SM) (g), sphinganine (SA) (h) and sphingosine (SO) (i) were analyzed by liquid chromatography-tandem mass spectrometry (LC-MS/MS). *p < 0.05 vs. control. n = 3. vector (Fig. 4e). Then, HepG2 cells were infected with adenoviruses (AdGFP, AdATF4, or AdsXBP1) to activate UPR genes. Adenoviral overexpression of sXBP1 increased Sptlc2 protein levels compared with those of ATF4 overexpression or control GFP adenovirus (Fig. 4f). These results suggest that sXBP1 upregulates the expression of Sptlc2. To confirm whether Sptlc2 is a downstream gene of sXBP1, we suppressed the expression of XBP1 by transfecting XBP1-specific siRNA. Mouse primary hepatocytes were transfected with siRNA control and XBP1-specific siRNA. After 48 h, tunicamycin was administered for 6 h to induce ER stress, and then the cells were harvested to examine mRNA and protein expression levels. We found that XBP1 knockdown suppressed the expression levels of Sptlc1 and Sptlc2 (Fig. 4g, h). We performed a ChIP assay to examine whether Sptlc2 is the direct target of sXBP1. When ER stress was activated by tunicamycin, the Sptlc2 promoter region was immunoprecipitated by sXBP1 antibody (Fig. 4i). In the absence of tunicamycin, no band was detected by sXBP1 antibody-mediated ChIP assay. Therefore, these results suggest that sXBP1 is a direct transcriptional regulator of Sptlc2. Fig. 2 Regulation of the ER stress markers Sptlc1 and Sptlc2 by tunicamycin in wild-type (WT) mice. Tunicamycin (2.5 µg/g body wt) was injected intraperitoneally into C57BL/6 J WT mice. The liver was isolated for mRNA expression analysis at 0, 1, 2, 4, 6 and 24 h after injection. The expression levels of ER stress markers, including ATF4 (a), ATF6 (b), unspliced XBP1 (uXBP1) (c), spliced XBP1 (d), CHOP (e), GRP78 (f), Sptlc1 (g), and Sptlc2 (h), were measured by qRT-PCR. *p < 0.05 vs. 0 h. n = 3.

Liver-specific Sptlc2 transgenic mice fed a HFD exhibit increased insulin resistance
To investigate the role of Sptlc2 in hepatic glucose/lipid metabolism, we constructed liver-specific Sptlc2 transgenic (lSptlc2-Tg) mice. Three lines of transgenic mice were produced by embryonic injection of pLiv.7 vectors containing human Sptlc2 cDNA (gene ID: BC005123.2). Among them, one line of transgenic mice (F06) was selected for further studies. In the selected line of transgenic mice (lSptlc2-Tg F06), the breeding ability was normal, and the enzyme activity was higher than that in the other lines (lSptlc2-Tg F03 and F08). Hepatic SPT enzyme activity was significantly increased by 2-fold compared with that in the WT (Fig. 5a). Increased Sptlc2 expression was confirmed by immunoblotting analysis (Fig. 5b). We examined whether the Sptlc2 transgene is specific to the liver. We found that the liver showed increased expression of Sptlc2 and that nonhepatic organs showed no change in Sptlc2 expression compared with the expression in WT mice (Fig. 5c).
The correlation between sphingolipid metabolites and insulin response has been reported 35 . Ceramide, a typical sphingolipid metabolite, inhibits serine phosphorylation of the AKT/PKB pathway via the activation of protein phosphatase 2 A (PP2A) 22,23,26 . To test whether hepatic Sptlc2 overexpression decreases the hepatic insulin response, we fed a HFD to mice for 8 weeks. We confirmed Sptlc1 and Sptlc2 expression at the protein level (Fig. 5d). To investigate insulin signaling in the liver, insulin was injected intraperitoneally (50 U/kg body weight) for 10 min before collection of liver tissue. As a result, serine phosphorylation of AKT/PKB was decreased slightly through inhibition of tyrosine phosphorylation of IRβ in HFD-fed lSptlc2-Tg mice compared with that in HFD-fed WT mice (Fig. 5e). In contrast, the expression of gluconeogenic proteins, such as glucose 6-phosphatase (G6Pase) and phosphoenolpyruvate carboxykinase (PEPCK), was not altered. These results suggest that insulin signaling was inhibited by the reduced phosphorylation of IRβ and AKT, but gluconeogenesis was not altered in the lSptlc2-Tg livers.
Hepatic Sptlc2 overexpression elevates the synthesis of ceramide We measured selected plasma parameters to determine whether the liver-specific SPTLC2 transgene alters the metabolic state in the plasma. While ALT, AST, total cholesterol, triglycerides, HDL, and LDL were not altered, fasting plasma glucose levels were drastically elevated in lSptlc2-Tg mice (Table 1).
To further examine the effects of lSpltlc2-Tg on insulin sensitivity and glucose metabolism, glucose tolerance and insulin tolerance tests were performed. The responses to glucose and insulin were not altered in NCD-fed WT and lSptlc2-Tg mice. In contrast, the glucose intolerance and insulin responsiveness were aggravated in lSptlc2-Tg mice fed a HFD compared with those in WT mice fed a HFD (Fig. 6a, b). These results suggest that lSptlc2-Tg mice fed a HFD have attenuated insulin responses via inhibition of the serine phosphorylation of AKT.
To examine whether hepatic Sptlc2 overexpression affects the synthesis of sphingolipid metabolites, we measured sphingolipid metabolites, including ceramide, dihydroceramide, SM, SO, and SA, in the liver tissues from WT and lSptlc2-Tg mice by LC-MS/MS. We found that total ceramide was increased significantly, but dihydroceramide, SM, SO, and SA were not altered compared with levels in the livers of WT mice (Fig. 6c-f). These findings suggest that hepatic Sptlc2 overexpression elevates ceramide levels in the liver and contributes to the dysregulation of glucose metabolism in the liver.

DISCUSSION
The endoplasmic reticulum (ER) is an organelle that is responsible for protein folding, maturation, and trafficking. ER stress occurs due to the accumulation of unfolded proteins, which further leads to insulin resistance and metabolic dysfunction 25 . Overnutrition and infection-mediated inflammatory responses are physiological conditions causing ER stress that triggers the UPR to restore ER homeostasis. Under continuous ER stress, lipid homeostasis is perturbed, and cellular messengers, such as DAG and ceramide, lead to the development of insulin resistance in the liver. We found the following: (1) SPT-mediated de novo ceramide biosynthesis was activated by ER stress via the XBP1 pathway, (2) ER stress-mediated upregulation of Sptlc2 was responsible for increased ceramide in the liver, and (3) increased ceramide inhibited insulin signaling by decreasing the phosphorylation of IRβ and AKT.
The ER is the major site of protein synthesis, as well as the location for the proper maturation and transportation of correctly Fig. 5 The liver-specific Sptlc2 transgene (lSptlc2-Tg) inhibits insulin signaling. Liver-specific Sptlc2 transgenic mice were constructed, and three germ lines were generated. The hepatic SPT activity of the three germ lines (F03, F06, and F08) was measured (a). n = 3-4, mean ± SEM. *p < 0.05 vs. wild-type (WT) mice. The protein expression of Sptlc2 in F06 cells was analyzed by western blotting (b). The overexpression of hepatic Sptlc2 protein was confirmed (b), and there was no overexpression in other organs by immunoblotting (c). WT mice and lSptlc2 mice were fed either a normal chow diet (NCD) or a high-fat diet (HFD) for 4 weeks and their livers were isolated to measure the protein expression levels of Sptlc1 and Sptlc2 (d). The protein intermediates in the insulin signaling pathway (IR, insulin receptor; AKT) and gluconeogenic proteins (G6Pase, glucose 6-phosphatase; PEPCK, phosphoenolpyruvate carboxykinase) were measured 10 min postinjection of insulin (1 U/ kg) by western blotting (e). Fig. 4 An obesogenic diet upregulates the expression of Sptlc2 via the activation of sXBP1. C57BL/6 J wild-type (WT) mice were fed a normal chow diet (NCD) or a high-fat diet (HFD, 60% of kcal from fat) for 4 weeks, and their livers were collected. Hepatic mRNAs were isolated and analyzed by qRT-PCR (a). Protein expression levels of Sptlc1, Sptlc2, sXBP1, and ATF4 were measured by western blotting (b). n = 5-6, mean ± standard error of the mean (SEM). *p < 0.05 vs. NCD. WT mice were fed a high fructose diet (FD, 60% of kcal from fructose) for 2 and 7 d, and livers were isolated for mRNA (c) and protein (d) analysis. n = 5, mean ± SEM. *p < 0.05 vs. NCD. HepG2 cells were cotransfected with pSptlc2-pGL3-Luc and pcDNA3 containing either ATF4 or sXBP1. Sptlc2 promoter activity was measured by the luciferase assay (e). n = 5, mean ± SEM. *p < 0.05 vs. pcDNA. HepG2 cells were infected with adenovirus containing ATF4 or sXBP1. Twenty-four hours postinfection, the protein expression levels of Sptlc1 and Sptlc2 were determined by western blotting (f). Control and XBP1 siRNA were transfected into mouse primary hepatocytes. After incubating for 48 h, cells were treated with tunicamycin (1.25 µg/mL) for 6 h. mRNA (g) and protein (h) levels of Sptlc1 and Sptlc2 were analyzed by qRT-PCR and western blotting, respectively. n = 3, mean ± SEM. *p < 0.05 vs. siRNA control (CON). HepG2 cells were treated with tunicamycin (1.25 µg/ml) for 2 h. ChIP assays were performed with harvested cells as described in the Materials and Methods and according to the manufacturer's protocol (i).
folded proteins, along with the Golgi apparatus 25 . In addition, the ER is an essential apparatus in the coordination of metabolism by controlling the synthetic and catabolic pathways of various nutrients. Monitoring of ER status and signaling through the UPR are mediated by three ER membrane-associated proteins, protein kinase RNA-like ER-associated kinase (PERK), IRE1α, and ATF6. In the normal state of the ER, these three membrane proteins bind to BiP/GRP78 and inactivate these signaling proteins 36,37 . The accumulation of improperly folded proteins in the ER results in the release of BiP/GRP78 from these UPR sensors and the activation of PERK and IRE1α. PERK is responsible for the phosphorylation and inactivation of eIF2α, while it activates ATF4, which induces the translation of certain UPR gene clusters 38 . IRE1α is an endoribonuclease that cleaves and generates sXBP1, producing a basic leucine zipper transcriptional activator for the induction of UPR-response genes 39,40 . The UPR also activates ATF6 by promoting its proteolytic cleavage and release from the ER lumen. Liberated ATF6 moves to the nucleus to stimulate a subset of genes, facilitating clearance of misfolded proteins from the ER 41,42 . Improper ER function and activation of the UPR disrupt lipid homeostasis and activate LIPIN2 expression in chronic and acute ER stress conditions via ATF4 activation 32 . DAG, a product of LIPIN2, activates PKCε and inhibits insulin signaling via the inactivation of insulin receptors 35 .
Sphingolipids are basic cellular structural and signaling molecules. Among them, elevation of ceramide by obesity is widely believed to have deleterious effects 16,[43][44][45] and is associated with toxicity in several tissues, including the liver 46 . De novo ceramide biosynthesis has been known to cause the inhibition of insulin signaling. Pharmacological and genetic inhibition of de novo ceramide biosynthesis improved glucose intolerance in diet-induced obese (DIO) or heterozygous Sptlc2-deficient mice 47,48 . This finding implied the independent involvement of ceramide in obesity-induced hepatic dysfunction via Sptlc2, a catalytic subunit of SPT. In the liver, sXBP1 regulates the transcription of genes involved in fatty acid synthesis, including stearoyl CoA desaturase-1 (SCD1), acetyl-CoA carboxylase 2 (ACC2), and diacylglycerol acyltransferase 2 (DGAT2), for the accumulation of TG 33 . HFD feeding for 4 weeks caused upregulation of only sXBP1 and the accumulation of lipid droplets in hepatocytes 31 . Early adaptation to ER stress might be initiated by the upregulation of de novo ceramide biosynthesis. Induction of acute ER stress by tunicamycin upregulated Sptlc1 and Sptlc2, the subunits of SPT, leading to increased ceramide production. HFD feeding upregulated both SPTLC1 and SPTLC2, while the fructose diet only elevated SPTLC2 (Fig. 4b, d). Even in lSPTLC2-Tg mice, SPTLC1 was upregulated (Fig. 5b, d). The stoichiometry of the SPTLC1 and SPTLC2 subunits is not known exactly, but imbalanced expression of the subunits may be sensed, and the binding partner SPTLC1 seems to be upregulated to restore proper stoichiometry and enzyme activity. These results suggest that the expression of the respective SPTLC subunits is regulated by different insults and that the stoichiometry of the SPTLC1 and SPTLC2 subunits should be maintained to ensure proper enzyme activity response.
Currently, it is not clear whether this ER stress-mediated SPTLC2 induction process is a defensive mechanism to relieve chronic ER stress or a contributing factor for insulin resistance by excess lipids. HFD feeding for 4 weeks increased hepatic Sptlc1 and Sptlc2 in mice. Consistently, short-term feeding with a fructose diet induced Sptlc2 protein expression. A carbohydrate-rich fructose diet activates XBP1 to induce hepatic lipogenesis 33 . We found that sXBP1 upregulated Sptlc2 but that ATF4 did not. Accordingly, only sXBP1 is associated with the regulation of SPT activity and increased ceramide production in hepatocytes. According to a previous report that ATF4 upregulates sphingosine kinase 2 31 , "the sphingolipid rheostat," which is the balance between ceramide and sphingosine 1-phosphate (S1P), is regulated differently by distinct UPR pathways. The physiological relevance of the synthesis of ceramide or S1P via UPR pathways needs further elucidation.
Overexpression of Sptlc2 had different results in vitro and in vivo. During ER stress, JNK and JNK-mediated serine phosphorylation of IRS-1 via the IRE1α-sXBP1 pathway lead to insulin resistance 49 . In vitro, increased ceramide by adenoviral overexpression of Sptlc2 activated JNK, which phosphorylated IRS-1 at Ser 30750 . While insulin signaling was inhibited by Ser phosphorylation of IRS1 in vitro, phosphorylation of IRβ was reduced in the livers of lSptlc2-Tg mice. This result indicated that inhibition of insulin signaling by ER stress is due to ceramide production and decreased AKT phosphorylation. The increase in the levels of ceramide may have decreased the phosphorylation of AKT via protein phosphatase 2A (PP2A) 43 . When adenovirus bearing Sptlc2 was injected into mice fed a NCD, lipid droplets were reduced, and insulin sensitivity and glucose intolerance were improved 50 . This conflicting result is due to the diet (NCD) and nonphysiologically high expression of Sptlc2 in the liver. Hepatic Sptlc2 expression activated VLDL secretion and reduced lipid droplets in the liver. Indeed, the fasting glucose levels were even lower than physiological levels in mice injected with Sptlc2 adenovirus. Under normal conditions, elevated ceramide causes stress in the liver and reduces lipid droplets by increasing VLDL secretion. These results suggest that hepatic glucose/FA metabolism is regulated differently depending on diet condition and the degree of Sptlc2 overexpression.
We observed hyperglycemia in lSptlc2-Tg mice fed a HFD. Decreased hepatic ceramide levels by whole animal or tissuespecific ablation of dihydroceramide desaturase 1 in the liver or adipose tissue ameliorates hepatic steatosis and insulin resistance caused by obesogenic diets 51 . The major cause for the hyperglycemic conditions could be elevated hepatic ceramide levels, but steatosis was not found in lSptlc2-Tg livers (data not shown) since we did not find any changes in the expression of gluconeogenic genes. Interestingly, hyperglycemia was found only in lSptlc2-Tg mice fed a HFD, while no change was found in fasting glucose levels in mice fed a NCD. As reported previously, HFD feeding increases hepatic ceramide levels and causes elevation of fasting glucose levels via the inhibition of insulin signaling 52,53 . In our genetic study, hepatic FA uptake under a HFD supplied substrates for ceramide synthesis and aggravated the dysfunction of glucose metabolism in lSptlc2-Tg livers. We found that only ceramide was elevated in the liver and contributed to the development of hyperglycemia and insulin resistance. Fig. 6 The hepatic Sptlc2 transgene elevates the level of ceramide and induces hyperglycemia. Eight-week-old wild-type (WT) and lSptlc2-Tg mice were fed a normal chow diet (NCD) or a high-fat diet (HFD, 60% kcal fat) for 4 weeks. After 16 h of fasting, glucose (2 g/kg) was injected intraperitoneally into the mice. Blood glucose levels were measured at various times postinjection (a). n = 5, mean ± SEM. *p < 0.05 vs. WT-HFD. After 4 h of fasting, insulin (0.75 U/kg) was injected intraperitoneally into the mice. Blood glucose levels were measured at various times postinjection (b). n = 5, mean ± SEM. *p < 0.05 vs. WT-HFD. The levels of ceramide (c), dihydroceramide (d), SM (e), and SO and SA (f) were analyzed by LC-MS/MS in the livers from WT and lSptlc2-Tg mice fed a HFD for 4 weeks. n = 5, mean ± SEM. *p < 0.05 vs. WT.
In conclusion, we demonstrated that ER stress-mediated activation of Sptlc2 was regulated by sXBP1. The activation of Sptlc2 increased ceramide levels, inhibiting insulin signaling both in vitro and in vivo. In addition, elevated ceramide decreased pAKT and inhibited the insulin response. The reason for this kind of regulation via sphingolipids might be mainly to reduce the stress on the liver. Taken together, these results indicated that ER stress upregulates Sptlc2 and increases the levels of ceramide, partly contributing to the development of hepatic insulin resistance.