Hepatocyte nuclear factor-4 alpha regulates liver triglyceride metabolism in part through secreted phospholipase A2 GXIIB


  • Potential conflict of interest: Nothing to report.


Hepatocyte nuclear factor-4 alpha (HNF-4α) is an important transcription factor governing the expression of genes involved in multiple metabolic pathways. Secreted phospholipase A2 GXIIB (PLA2GXIIB) is an atypical member of a class of secreted phospholipases A2. We establish in this study that PLA2GXIIB is an HNF-4α target gene. We demonstrate that HNF-4α binds to a response element on the PLA2GXIIB promoter. Moreover, HNF-4α agonists induce PLA2GXIIB expression in human hepatocarcinoma cells. Importantly, PLA2GXIIB-null mice accumulate triglyceride, cholesterol, and fatty acids in the liver and develop severe hepatosteatosis resembling some of the phenotypes of liver-specific HNF-4α–null mice. These defects are in part due to compromised hepatic very low-density lipoprotein secretion. Finally, adenovirus-mediated overexpression of HNF-4α elevates serum triglyceride level in wild-type but not PLA2GXIIB-null mice. Conclusion: Collectively, these evidences suggest that HNF-4α is a key physiological PLA2GXIIB transcriptional regulator and that PLA2GXIIB is a novel mediator of triglyceride metabolism in the liver. (HEPATOLOGY 2011;53:458-466)

Hepatocyte nuclear factor-4 alpha (HNF-4α) was initially identified as a transcriptional factor required for liver-specific gene expression.1 Coenzyme A (CoA) derivatives of certain fatty acids can bind to the ligand-binding domain of HNF-4α and activate the receptor,2 suggesting that HNF-4α activity is subjected to modulation by metabolic and nutritional signals. HNF-4α regulates the expression of gluconeogenic genes, including phosphoenolpyruvate carboxykinase (PEPCK)3 and glucose-6-phosphatase (G6P),4 through binding to hormone response elements on their promoters and recruiting coactivators such as peroxisome proliferator-activated receptor γ coactivator-1α (PGC-1α) to their promoters.5 In addition, HNF-4α and PGC-1α regulate the expression of lipoproteins and packaging enzymes such as microsomal triglyceride transfer protein (MTP) that are involved in very low-density lipoprotein (VLDL) secretion. Significantly, liver-specific HNF-4α-null (HNF4αLivKO) mice suffer from severe defects in lipid homeostasis with hepatosteatosis and reduced serum triglyceride (TG) and cholesterol levels.6

Phospholipases A2 are enzymes that catalyze the hydrolysis of glycerophospholipids at the Sn2 position to release free fatty acids such as arachidonic acid and lysophospholipids that are precursors of signaling molecules.7 In particular, arachidonic acid and its metabolites leukotrienes and prostaglandins are key inflammatory regulators. On the basis of their protein structures and biochemical properties, the superfamily of PLA2 can be divided into five principal kinds of enzyme: cytosolic PLA2s (cPLA2s), Ca2+-independent PLA2s (iPLA2s), lysosomal PLA2s, platelet activating factor acetylhydrolases, and secreted PLA2s (sPLA2s).

Secreted PLA2s have relatively low molecular masses of 14-19 kDa, a large number of disulfides, and similar Ca2+-dependent catalytic mechanism. Mammalian sPLA2s include GIB, GIIA, GIIC, GIID, GIIE, GIIF, GIII, GV, GX, GXIA, GXIB, GXIIA, and GXIIB. Among them, PLA2GXIIB is considered an atypical member. Although mouse and human PLA2GXIIB share 90% identity between them, they share only 40% identity with their respective PLA2GXIIA counterparts, with the homology limited to the Ca2+-binding segment and active site.8 Importantly, a canonical histidine found in the active site of GXIIA and all other sPLA2s is not conserved in GXIIB at which a leucine is encoded instead. The lack of enzymatic activity of purified GXIIB expressed in Escherichia coli and poor phospholipid binding ability indicate that GXIIB is very likely catalytically inactive.8 Furthermore, mGXIIB is not an endogenous ligand for the mouse M-type receptor that binds to several other mouse sPLA2s.8 These evidences suggest that PLA2GXIIB may have unique function distinctive from other sPLA2s. However, the physiological role of this atypical member remains to be established.

PLA2GXIIB is highly expressed in the liver.8 Intriguingly, the expression level of PLA2GXIIB was induced by an adenovirus encoding HNF-4α and suppressed by small interfering RNA against HNF-4α in human hepatocarcinoma HepG2 cells.9 In this study, we found that PLA2GXIIB is transcriptionally regulated by HNF-4α. By means of generating PLA2GXIIB-null mice, we further established that PLA2GXIIB is important for hepatic VLDL secretion.


Apo, apolipoprotein; bp, base pair; CoA, coenzyme A; EMSA, electrophoresis mobility shift assay; G6P, glucose-6-phosphatase; HNF-4α, hepatocyte nuclear factor-4 alpha; kb, kilobase; mRNA, messenger RNA; MTP, microsomal triglyceride transfer protein; PCR, polymerase chain reaction; PEPCK, phosphoenolpyruvate carboxykinase; PGC-1α, peroxisome proliferator-activated receptor γ coactivator-1α; PLA2, phospholipase A2; TG, triglyceride; VLDL, very low-density lipoprotein.

Materials and Methods

Details on the materials and methods used are provided in the supporting online information.


Transcriptional Regulation of PLA2GXIIB by HNF-4α and PGC-1α.

Because modulation of HNF-4α activity altered PLA2GXIIB expression,9 we hypothesize that PLA2GXIIB is transcriptionally regulated by HNF-4α. Bioinformatics analysis of the mouse PLA2GXIIB promoter region spanning +1 to −1200 base pair (bp) upstream of the transcriptional start site revealed two putative HNF-4α–responsive elements (Fig. 1A). Referred to as sites A and B, these putative elements are located at positions −1070 to −1084 and −68 to −86, respectively (Fig. 1A). We first asked if this promoter region is sufficient to confer responsiveness to HNF-4α. We cloned the wild-type mouse PLA2GXIIB promoter into a luciferase reporter (pGL3-mPLA2GXIIB) and transiently transfected this reporter plasmid with expression vectors for HNF-4α and its coactivator PGC-1α into HeLa cells. Compared to a control pGL3-basic reporter, HNF-4α modestly increased pGL3-mPLA2GXIIB reporter expression (Fig. 1B). Although coactivator PGC-1α by itself was insufficient to modulate the reporter expression, it enhanced the HNF-4α–driven expression (Fig. 1B). To further identify the HNF-4α responsive element responsible for this regulation, we generated site A deleted (pGL3-mPLA2GXIIB ΔA), sites A and B deleted (pGL3-mPLA2GXIIB ΔAB), and site B mutated (pGL3-mPLA2GXIIB mut B) reporter plasmids (Fig. 1A). HNF-4α/PGC-1α promoted both the wild-type and ΔA reporter expression levels (Fig. 1C). On the other hand, this HNF-4α/PGC-1α–driven expression was lost by deleting or mutating site B (Fig. 1C). Similar results were obtained in human hepatocarcinoma HepG2 cells (Supporting Information Fig. 1).

Figure 1.

Transcriptional regulation of PLA2GXIIB-promoter luciferase by HNF-4α and PGC-1α. (A) Schematic representations of wild-type PLA2GXIIB promoter, deletion, and mutant constructs. (B) HeLa cells were transfected with pGL3-basic as a control or pGL3-mPLA2GXIIB-promoter luciferase reporter plasmid and an internal control Renilla luciferase together with expression vectors of HNF-4α and/or PGC-1α. (C) Effects of disrupting potential HNF-4α recognition sites on PLA2GXIIB-promoter-reporter. HeLa cells were cotransfected as in (B) except with pGL3-mPLA2GXIIB, pGL3-mPLA2GXIIB ΔA, pGL3-mPLA2GXIIB ΔAB, or pGL3-mPLA2GXIIB mutant B together with HNF-4α and PGC-1α expression vectors. (B,C) Luciferase activity was normalized to Renilla luciferase activity and expressed as fold stimulation over the control. The results shown are the means ± standard error (SE) of three different experiments. **P < 0.01; ***P < 0.001. (D) EMSA analysis was performed with nuclear extracts from Adeno-ΔE1E3 or Adeno-HNF-4α infected cells using biotin-labeled wild-type, unlabeled wild-type, or unlabeled mutant site B oligonucleotides with or without an anti–HNF-4α antibody. Free probe, HNF-4α specifically bound band, and antibody super-shifted band are indicated. (E) ChIP analysis of HNF-4α response element on G6P as a positive control, −214/−8 (spanning site B) and −1126/−940 (spanning site A) of PLA2GXIIB-promoter were performed using an anti–HNF-4α antibody. No antibody was used for mock, whereas a nonspecific antibody was used for IgG negative control.

We next investigated if HNF-4α can bind to site B response element. We prepared nuclear extracts from cells infected with a control adenovirus (Adeno-ΔE1E3) or an adenovirus encoding HNF-4α (Adeno-HNF-4α) to perform electrophoresis mobility shift assays (EMSAs). Using a biotin-labeled oligonucleotide based on site B sequence as a probe, we found that Adeno-ΔE1E3 nuclear extract bound to this probe, generating several nonspecific bands that were not competed by a nonlabeled site B probe (Fig. 1D). In contrast, Adeno-HNF-4α nuclear extract specifically bound to the probe generating a specific band that was competed by a wild-type but not by a mutant site B nonlabeled probe (Fig. 1D). In addition, this specific band was super-shifted by an antibody against HNF-4α (Fig. 1D), indicating that HNF-4α binds to site B probe specifically in vitro. We then used a chromatin immunoprecipitation (ChIP) analysis to confirm that liver HNF-4α binds to site B in vivo. An antibody against HNF-4α immunoprecipitated a chromatin fragment containing site B; whereas, a nonspecific antibody (immunoglobulin G [IgG]) or mock control did not (Fig. 1E). On the other hand, none of the antibodies or the mock control immunoprecipitated a chromatin fragment spanning site A (Fig. 1E). As a positive control, we found that a chromatin fragment spanning the published HNF-4α binding site on G6P was specifically pulled down by the HNF-4α antibody (Fig. 1E). Collectively, our analysis indicated that HNF-4α and PGC-1α control the expression of PLA2GXIIB through a bona fide HNF-4α response element located at site B.

PLA2GXIIB Expression Regulated by HNF-4α.

To further address if modulating HNF-4α activity alters PLA2GXIIB expression, we first used the HNF-4α agonistic modulators propionate and butyrate to enhance HNF-4α activity.2, 10 HeLa cells were transfected with either the wild-type or site B mutant reporter plasmid together with HNF-4α and PGC-1α expression vectors. We found that the HNF-4α/PGC-1α–driven pGL3-mPLA2GXIIB reporter expression was further enhanced by both propionate and butyrate (Fig. 2A); whereas, these treatments had no effect on the site B mutant reporter (Fig. 2A). In addition, we found that the messenger RNA (mRNA) expression level of PLA2GXIIB was induced by propionate and butyrate in HepG2 cells, which express HNF-4α endogenously (Fig. 2B). In contrast, HNF-4α antagonists such as polyunsaturated fatty acyl-CoAs4 and acyl-CoA thioesters of hypolipidemic peroxisome proliferators11 down-regulated the HNF-4α/PGC-1α–driven pGL3-mPLA2GXIIB reporter expression (Supporting Information Fig. 2). Because hepatic activities of HNF-4α and PGC-1α are elevated by fasting,12 we examined and found that hepatic PLA2GXIIB expression is induced by fasting similar to other HNF-4α target genes PEPCK, G6P, and MTP (Fig. 2C), suggesting that HNF-4α regulates PLA2GXIIB expression in vivo.

Figure 2.

PLA2GXIIB expression regulated by HNF-4α modulators. (A) HeLa cells were transfected with pGL3-mPLA2GXIIB or pGL3-mPLA2GXIIB mutant B with expression vectors of HNF-4α and PGC-1α. Water used as a vehicle control, 2.5 mM propionate, or butyrate was added for another 24 hours before luciferase assays. Luciferase activity was normalized to Renilla luciferase activity and expressed as fold stimulation over the control. (B) HepG2 cells were treated with vehicle, 2.5 mM propionate, or butyrate was added for 10 hours before assessing the mRNA expression levels of PLA2GXIIB. (A,B) Results shown are the means ± SE of three different experiments. **P < 0.01 for propionate or butyrate compared to vehicle. (C) The 12-week-old male C57BL/6J mice (n = 3 per group) were either not fasted (Fed) or fasted for 8 or 24 hours. Hepatic mRNA expression levels of PLA2GXIIB, MTP, PEPCK1, and G6P were measured by quantitative reverse transcription PCR. Expression levels were normalized to β-actin and the relative levels of fed mice were set at 1. Fast versus fed mice: *P < 0.05; **P < 0.01; ***P < 0.001.

Altered Lipid Metabolism in PLA2GXIIB-Null Mice Partially Resembled That of HNF4αLivKO Mice.

Because HNF-4α governs the expressions of PEPCK, G6P, and MTP to regulate gluconeogenesis and VLDL-TG secretion we next examined if PLA2GXIIB participates in these processes. To establish a physiology function of PLA2GXIIB, we generated PLA2GXIIB-null mice. Two loxP sites flanking a neo cassette were introduced into exon 1 of the PLA2GXIIB gene with a sequence-replacement gene-targeting vector (Fig. 3A). Deletion of exon 1 results in a frame-shift mutation. Chimeric mice were selected based on deletion of the neo cassette (generating a PLA2GXIIB allele) and crossed to wild-type mice to generate lines carrying the PLA2GXIIB allele. Southern blot analysis of tail genomic DNA detected a 13.2-kilobase (kb) fragment from wild-type mice whereas two fragments at 8.2 and 5 kb were detected in null mice (Fig. 3B). Polymerase chain reaction (PCR) analysis of tail genomic DNA detected 1200 and 327 bp fragments for wild-type and null alleles respectively (Fig. 3C). Additionally, we confirmed using an antibody directed against PLA2GXIIB that its expression level was below detection limit in PLA2GXIIB-null mice (Fig. 3D). Intercrossing of the heterozygous mice (PLA2GXIIB+/−) indicated that the transmission of the PLA2GXIIB allele was close to Mendelian inheritance ratio (PLA2GXIIB+/+, 23%; PLA2GXIIB−/−, 21%; PLA2GXIIB+/−, 56%).

Figure 3.

Generation of PLA2GXIIB-null mice. (A) Schematic of the sequence-replacement gene-targeting strategy. Wild-type PLA2GXIIB allele (PLA2GXIIB+) spans the promoter (sequences upstream from exon 1) and 4 exons. The locations of the 326-bp 5′ flanking probe P1 and 329 bp 3′ flanking probe P2 are indicated. Recombination of the gene-targeting vector with the cognate sequences in the chromosomal DNA produces a mutant PLA2GXIIB allele, PLA2GXIIBflox, in which exon 1 of the PLA2GXIIB allele are flanked by loxP sites (filled triangles). EIIaCre-mediated excision of the neo fragment produces a null PLA2GXIIB allele (designated PLA2GXIIB). (B) Southern blot using probe P1 and P2 illustrating KpnI-digested genomic DNA from PLA2GXIIB+/+, PLA2GXIIB+/−, and PLA2GXIIB−/− mice are indicated. (C) PCR genotyping of mice. Tail genomic DNA was amplified using primers flanking the loxP site in the PLA2GXIIB null allele, yielding respectively products of 1200 and 327 bp for the wild-type and null alleles. (D) An antibody against PLA2GXIIB was used in a Western analysis comparing liver samples from PLA2GXIIB+/+ and PLA2GXIIB−/− mice with bacterially expressed PLA2GXIIB as a positive control.

The 12-week-old PLA2GXIIB-null mice developed normally and were fertile with similar body, liver, and epididymal fat pad weights compared to age-matched PLA2GXIIB+/+ littermates (Table 1). However, the livers from 16-week-old PLA2GXIIB−/− mice were heavier compared to age-matched PLA2GXIIB+/+ mice (data not shown) and were pale in color due to massive accumulations of lipids (Fig. 4A). Histological analysis of the liver from a 16-week-old PLA2GXIIB−/− mouse showed that hepatocytes were filled with fat droplets compared to a PLA2GXIIB+/+ mouse liver (Fig. 4A). Liver cholesterol content was elevated by two-fold and TG content was elevated by up to three-fold (Fig. 4B). Furthermore, hepatic free fatty acids concentration was elevated by 67%, whereas concentration of phospholipids remained unchanged (Fig. 4B). Even in 12-week-old PLA2GXIIB−/− mice, total serum cholesterol as well as HDL-cholesterol and LDL-cholesterol levels of the PLA2GXIIB−/− mice were dramatically lowered compared to age-matched wild-type mice (Table 1), whereas serum alanine aminotransferase, asparatate aminotransferase, and glucose levels were normal (Table 1). In addition, serum TGs, free fatty acids, and phospholipid levels fell by 79%, 63%, and 75%, respectively (Table 1). Remarkably, many of these phenotypes bear close resemblances to those of the HNF4αLivKO mice which also display severe hepatosteatosis with reduced serum TG and cholesterol levels,6 suggesting that PLA2GXIIB functions downstream of HNF-4α to control lipid metabolism.

Figure 4.

PLA2GXIIB-null mice display fatty liver phenotype. (A) Histological analysis of the livers of 16-week-old wild-type and PLA2GXIIB−/− female mice. (a) Normal liver of a wild-type mouse. (b) Fatty liver of a PLA2GXIIB−/− mouse. Hematoxylin and eosin stains of (c) wild-type and (d) PLA2GXIIB−/− livers. Magnification, ×400. Liver of PLA2GXIIB−/− mouse showed marked vacuolation. (e,f) Frozen sections of normal liver from a wild-type mouse and fatty liver from a PLA2GXIIB−/− mouse were stained by Oil red O. Magnification, ×400. (e) Wild-type liver contained some small fat droplets in hepatocytes throughout the liver. (f) PLA2GXIIB−/− liver contained abundant amount of large fat droplets (red) in hepatocytes. (B) Lipid analysis of livers from 16-week-old control and PLA2GXIIB−/− female mice. Concentrations of liver cholesterol, TGs, free fatty acids, and phospholipids were measured on male PLA2GXIIB+/+ and PLA2GXIIB−/− mice. Each bar represents the mean ± SE of data from at least four mice. *P < 0.05; **P < 0.01; ***P < 0.001 for null compared to wild-type mice.

Table 1. Phenotypic Comparison Between PLA2GXIIB+/+ and PLA2GXIIB−/− Mice
  • The 12-week-old male mice were fed a chow diet prior to study and fasted for 5 hours before sample collection. Each value represents the mean ± SD of at least four mice.

  • *

    P < 0.05.

  • **

    P < 0.01.

  • ***

    P < 0.001 indicate statistical significance (Student t test) between PLA2GXIIB+/+ and PLA2GXIIB−/− mice. ALT, alanine aminotransferase; AST, asparate aminotransferase; HDL, high-density lipoprotein; LDL, low-density lipoprotein.

Body weight (g)30.18 ± 1.5628.83 ± 1.56
Liver weight (g)1.24 ± 0.181.32 ± 0.14
Epididymal fat pad weight (g)0.30 ± 0.110.31 ± 0.11
Serum chemistry  
 ALT (U/L)57.0 ± 14.254.8 ± 5.4
 AST (U/L)197.5 ± 53.7205.1 ± 43.9
 Glucose (mM)6.7 ± 0.46.6 ± 0.6
 Total bile acid (μM)2.5 ± 0.82.2 ± 0.4
 Total cholesterol (mM)5.69 ± 0.660.48 ± 0.15***
 HDL-cholesterol (mM)4.29 ± 0.640.34 ± 0.13***
 LDL-cholesterol (mM)0.57 ± 0.250.15 ± 0.05*
 Triglycerides (mM)1.74 ± 0.270.37 ± 0.09***
 Free fatty acids (mM)1.75 ± 0.540.64 ± 0.33**
 Phospholipids (mM)2.63 ± 0.330.67 ± 0.09***

VLDL-TG Secretion Is Compromised in PLA2GXIIB-Null Mice.

Elevated levels of cholesterol and fatty acid synthesis may be responsible for the accumulation of lipids in the liver. The rate-limiting enzymes for cholesterol synthesis are HMG-CoA synthase 1 (HMGCS1) and HMG-CoA reductase (HMGCR). We found that the hepatic mRNA expression levels of these enzymes were in fact down-regulated in PLA2GXIIB−/− compared to PLA2GXIIB+/+ mice (Supporting Information Fig. 3A). Additionally, the mRNA expression levels of fatty acid synthase (FASN) and stearoyl CoA desaturase-1 (SCD1), two key enzymes for fatty acid synthesis, were also down-regulated in the liver of PLA2GXIIB−/− (Supporting Information Fig. 3A). The reduced expression levels of these genes are likely to be a result of a negative feedback mechanism in response to elevated levels of hepatic lipids. Consistently, the expression levels of several fatty acid transport proteins (FATP) were also reduced (Supporting Information Fig. 3B). In contrast, the expression levels of several genes involved in fatty acid β-oxidation were not significantly altered (Supporting Information Fig. 3C). Therefore, alterations in lipid synthesis, transport, and catabolism are unlikely to be responsible for the fatty liver phenotype in PLA2GXIIB-null mice.

Hepatic secretion of TGs and cholesterol in the form of VLDL-TG is responsible for transporting endogenous lipids into the serum. We next investigated if defects in this pathway are responsible for the accumulation of lipids in the liver. After an intravenous injection of Triton WR1339, which blocks the catabolism of VLDL, the rate of serum TG accumulation was measured by determining TG levels in the serum at appropriate time points (Fig. 5A). The TG accumulation rate calculated for each individual mouse was normalized to its body weight. The secretion rate was calculated from the slope of the individual lines and expressed as millimole/kilogram/hour (Fig. 5B). PLA2GXIIB−/− mice showed an approximately 50% reduction in VLDL-TG secretion (Figs. 5A and 6B). Because each VLDL particle contains one molecule of apolipoprotein B (ApoB), using an antibody that recognizes both ApoB100 and its processed form ApoB48, we found that ApoB100 within the VLDL fraction was significantly reduced whereas ApoB48 was selectively less affected in PLA2GXIIB−/− mice by western blot analysis (Fig. 5C). Consistently, total serum ApoB level was reduced in PLA2GXIIB−/− mice by an ELISA assay (Fig. 5D); whereas, liver total ApoB level was actually modestly increased (Fig. 5D), strongly suggesting that the secretion of VLDL-TG is defective.

Figure 5.

Reduced liver VLDL-TG secretion rate in PLA2GXIIB-null mice. (A) Serum TG concentrations in 13-week-old PLA2GXIIB+/+ (•) and PLA2GXIIB−/− (○) female mice after injections of Triton WR-1339. (B) VLDL-TG secretion rates (mmol/kg/hour) were calculated from the serum TG versus time curve. (A,B) Each value represents the mean ± standard deviation (SD) of five mice. The lines represent the mean TG rates. (C) Western blot of apolipoprotein B (ApoB) contents in nascent VLDL particles of PLA2GXIIB+/+ and PLA2GXIIB−/− mice in (B). (D) Serum (left panel) and hepatic (right panel) ApoB concentration in 16-week-old PLA2GXIIB+/+ and PLA2GXIIB−/− male mice. Each value represents the mean ± SD of seven mice. **P < 0.01; ***P < 0.001 for null compared to wild-type mice.

Figure 6.

Adeno-PLA2GXIIB and HNF-4α on serum TG levels in wild-type and PLA2GXIIB-null mice. (A) Changes in serum TG level after injections of Triton WR-1339 in 13-week-old PLA2GXIIB−/− male mice infected with Adeno-ΔE1E3(○) or Adeno-PLA2GXIIB(•). (B) VLDL-TG secretion rates of mice in (A). (C) Serum TG levels of mice in (A). (D) Serum TG levels of 13-week-old wild-type and PLA2GXIIB−/− male mice 3 days after injection of a control adenovirus or an adenovirus encoding HNF-4α. (A-D) Each value represents the mean ± SD of five mice.

Adeno-HNF-4α Elevates Serum TG levels in Wild-Type But Not PLA2GXIIB-Null Mice.

To confirm the phenotypes of PLA2GXIIB−/− mice were due to loss of PLA2GXIIB function, we generated an adenovirus encoding PLA2GXIIB. Compared to the control Adeno-ΔE1E3, this Adeno-PLA2GXIIB virus was capable of overexpressing PLA2GXIIB in HepG2 cells (Supporting Information Fig. 4A). Importantly, this Adeno-PLA2GXIIB virus but not the control virus elevated the rate of hepatic VLDL secretion in PLA2GXIIB−/− mice close to that of the wild-type level (Fig. 6A,B) and restored the decrease in serum TG level in PLA2GXIIB−/− mice (Fig. 6C), strongly indicating that PLA2GXIIB functions to regulate lipid metabolism. Finally, to confirm that PLA2GXIIB functions down-stream of HNF-4α to control lipid metabolism, we injected into wild-type and PLA2GXIIB−/− mice the control Adeno-ΔE1E3 or Adeno-HNF-4α and measured the changes in serum TG levels. We established that Adeno-HNF-4α was effective in overexpressing HNF-4α and inducing PEPCK, MTP, and PLA2GXIIB mRNA expressions in HepG2 cells (Supporting Information Fig. 4B,C). Although Adeno-HNF-4α elevated the serum TG level in wild-type mice compared to the control adenovirus (Fig. 6D), it failed to elevate serum TG level in PLA2GXIIB−/− mice (Fig. 6D). In all, our analysis strongly suggested that PLA2GXIIB is an important target of HNF-4α necessary for controlling lipid metabolism.


We demonstrated in this study that PLA2GXIIB is an HNF-4α target gene. First, close to its transcriptional start site at positions −68 to −86, PLA2GXIIB promoter contains an HNF-4α response element composed of 5′-AGAGGACAAAGGTGAAAC-3′, representing a direct repeat with a 1 base pair spacer (DR1) of an imperfect nuclear hormone receptor consensus binding sequence AGGTCA. Second, HNF-4α bound to this response element by EMSA analysis and that an anti-HNF-4α antibody immunoprecipitated a chromatin fragment spanning this response element from mouse liver. Third, HNF-4α modulators regulated PLA2GXIIB expression in HepG2 cells and fasting induces hepatic PLA2GXIIB expression similar to other HNF-4α target genes. Noticeably, HNF-4α overexpression by adenovirus or knockdown by small interfering RNA also regulated PLA2GXIIB expression.9 Moreover, PLA2GXIIB expression is strongly reduced in HNF4αLivKO mice.6 Importantly, PLA2GXIIB-null mice accumulated TG, cholesterol, and fatty acids in the liver and developed severe hepatosteatosis despite reduced serum TG and cholesterol levels, closely resembling some of the phenotypes of HNF4αLivKO mice.6 Because cholesterols, TGs, and phospholipids are first exported from the liver via VLDL-TG particles which then serve as key precursors for LDL and HDL cholesterol,13 we found that PLA2GXIIB-null mice are defective in hepatic VLDL-TG secretion, which is likely responsible for the hepatosteatosis and reduced serum total TG, cholesterol, and phospholipids levels observed. Critically, an adenovirus encoding HNF-4α failed to elevate serum TG levels in PLA2GXIIB-null mice.

Collectively, these evidences suggest that HNF-4α is a key physiological PLA2GXIIB transcriptional regulator in the liver and that loss of PLA2GXIIB expression as in HNF4αLivKO and PLA2GXIIB-null mice lower serum TG levels due to reduced hepatic VLDL-TG secretion rates (Fig. 7). Incidentally, PLA2GXIIB is expressed in the small intestine where HNF-4α is functionally active14; therefore, HNF-4α also likely drives PLA2GXIIB expression in the small intestine.

Figure 7.

A model for HNF-4α regulation of VLDL secretion. HNF-4α independently regulates the expressions of ApoB, MTP, and PLA2GXIIB to direct hepatic VLDL secretion.

Although both HNF4αLivKO and PLA2GXIIB-null mice share many common phenotypes such as fatty liver and reduced serum lipid levels, they have other unique characteristics. HNF-4α regulates PEPCK to guide gluconeogenesis, short-heterodimer partner (SHP) to govern bile acid homeostasis, and ornithine transcarbamylase (OTC) to regulate ureagenesis; not surprisingly, the serum glucose and urea levels of HNF4αLivKO-null mice are lowered whereas bile acids and ammonia levels are elevated compared to their wild-type counterparts.6 However, these serum biochemical parameters were not significantly altered in PLA2GXIIB-null mice (Table 1; Supporting Information Fig. 5; data not shown). On the other hand, the serum free fatty acids level was significantly lowered in PLA2GXIIB-null but not in HNF4αLivKO mice (Table 1).6 Because PLA2GXIIB is a secreted protein, its action may extend to tissues other than the liver to affect the homeostasis of fatty acids.

Although hepatic VLDL-TG secretion is inhibited by PLA2GXIIB deficiency, the mechanistic connection is still an open question. Intriguingly, MTP-null mice also have lowered serum TG, cholesterol, and phospholipids levels as in PLA2GXIIB-null mice (Table 1) and develop mild hepatosteatosis.15 Nevertheless, the mRNA expression level of MTP, which is an HNF-4α target gene, remained normal in PLA2GXIIB-null mice (Supporting Information Fig. 6A). Beside, the expression levels of two other HNF-4α target genes PEPCK and G6P were not altered (Supporting Information Fig. 6B), implying that HNF-4α activity remains intact in PLA2GXIIB-null mice. Hepatic VLDL-TG secretion not only depends on the function of MTP but also plasma phospholipid transfer protein (PLTP).16 We found that the liver mRNA expression level of PLTP was not significantly altered in PLA2GXIIB-null mice (Supporting Information Fig. 6). Bile acids can regulate both gluconeogenesis and VLDL-TG secretion through suppressing HNF-4α activity. Our preliminary analysis indicated that the amounts of hepatic, urinary, fecal, and gallbladder bile acids did not significantly differ between wild-type and PLA2GXIIB-null mice (Supporting Information Fig. 5).

As demonstrated by their respective knockout mice, the functions of HNF-4α and its target genes MTP and PLA2GXIIB are indispensable for VLDL-TG secretion. In a complementary analysis, overexpression of MTP by adenovirus elevated serum TG levels and the rate of hepatic VLDL-TG secretion.17 Remarkably, we showed that overexpression of PLA2GXIIB by adenovirus also affected these parameters (Fig. 6). Based on these observations, we propose that HNF-4α acts upstream to control MTP- and PLA2GXIIB-dependent pathways that are independent but acting in parallel to drive hepatic VLDL-TG secretion (Fig. 7).

The phenotypes of PLA2GXIIB−/− mice bear some resemblances to a human disease familial hypobetalipoproteinemia (FHBL) characterized by low plasma levels of LDL cholesterol and FHBL subjects often have fatty liver.18 Approximately half of the FHBL subjects are carriers of mutations in the ApoB gene, whereas the causes for other FHBL patients are not known.19 Intriguingly, PLA2GXIIB−/− mice display compromised ApoB-VLDL secretion and develop severe fatty liver partially resembling those displayed by FHBL patients. It is therefore reasonable to speculate that some of those FHBL patients without mutations in the ApoB gene may have aberrant expression or activity levels of HNF-4α, MTP, and PLA2GXIIB.


We thank Ms. Xuehua Zheng, Mr. Yichu Liu, Dr. Hui Zhi, Dr. Zhaoyu Lin, and the staff at the Animal Center of GIBH for assistance throughout the project. This study was supported by the Knowledge Innovation Program of the Chinese Academy of Sciences (No. KSCX1-YW-10), National Key Science and Technology Specific Projects of China (2008ZX10001-001), and National Science Fund for Distinguished Young Scholars of China (No.30688004).