Patatin-like phospholipase domain-containing 3/adiponutrin deficiency in mice is not associated with fatty liver disease


  • Weiqin Chen,

    1. Diabetes and Endocrinology Research Center (DERC), Section of Diabetes and Endocrinology, Departments of Medicine, Molecular and Cellular Biology, and Biochemistry and Molecular Biology, Baylor College of Medicine, Houston, TX
    Search for more papers by this author
  • Benny Chang,

    1. Diabetes and Endocrinology Research Center (DERC), Section of Diabetes and Endocrinology, Departments of Medicine, Molecular and Cellular Biology, and Biochemistry and Molecular Biology, Baylor College of Medicine, Houston, TX
    Search for more papers by this author
  • Lan Li,

    1. Diabetes and Endocrinology Research Center (DERC), Section of Diabetes and Endocrinology, Departments of Medicine, Molecular and Cellular Biology, and Biochemistry and Molecular Biology, Baylor College of Medicine, Houston, TX
    Search for more papers by this author
  • Lawrence Chan

    Corresponding author
    1. Diabetes and Endocrinology Research Center (DERC), Section of Diabetes and Endocrinology, Departments of Medicine, Molecular and Cellular Biology, and Biochemistry and Molecular Biology, Baylor College of Medicine, Houston, TX
    2. St. Luke's Episcopal Hospital, Houston, TX
    • Diabetes and Endocrinology Research Center R614, Baylor College of Medicine, One Baylor Plaza, Houston, TX 77030
    Search for more papers by this author
    • fax: 713-798-8764

  • Potential conflict of interest: Nothing to report.

  • See Editorial on Page 818


PNPLA3 (adiponutrin), a novel patatin-like phospholipase domain-containing enzyme, is expressed at high level in fat, but also in other tissues including liver. Polymorphisms in PNPLA3 have been linked to obesity and insulin sensitivity. Notably, a nonsynonymous variant rs738409(G) allele of the PNPLA3 gene was found to be strongly associated with both nonalcoholic and alcoholic fatty liver disease. We have generated Pnpla3−/− mice by gene targeting. Loss of Pnpla3 has no effect on body weight or composition, adipose mass, or development, whether the mice were fed regular chow or high-fat diet or bred into the genetic obese Lepob/ob background. Plasma and liver triglyceride content and plasma aspartate aminotransferase and alanine aminotransferase levels were not different between Pnpla3+/+ and Pnpla3−/− mice while they were on regular chow, fed three different fatty liver-inducing diets, or after they were bred into Lepob/ob background. Hepatic Pnpla5 messenger RNA (mRNA) levels were similar in wild-type and Pnpla3−/− mice, although adipose Pnpla5 mRNA level was increased in Pnpla3−/− mice. A high-sucrose lipogenic diet stimulated hepatic Pnpla3 and Pnpla5 mRNA levels to a similar degree, but it did not affect adipose or liver triglyceride lipase (ATGL, known also as Pnpla2) mRNA in Pnpla3+/+ and Pnpla3−/− mice. Finally, Pnpla3+/+ and Pnpla3−/− mice displayed similar glucose tolerance and insulin tolerance tests while on regular chow or three different fatty liver–inducing diets. Conclusion: Loss of Pnpla3 does not cause fatty liver, liver enzyme elevation, or insulin resistance in mice. (HEPATOLOGY 2010)

Fatty liver disease, especially nonalcoholic fatty liver disease, is characterized by excessive triacylglycerol (TG) accumulation in the liver, and is the most common form of liver disease in western society.1, 2 Although alcohol abuse, obesity, insulin resistance, and diabetes are known to be associated with fatty liver disease, genetic factors causing predisposition to hepatic steatosis are not well understood. Recently, several genome-wide association studies have identified a common genetic variant (rs738409 G allele) in the PNPLA3 (patatin-like phospholipase domain-containing 3) gene to be strongly associated with increased liver fat content and susceptibility to nonalcoholic fatty liver disease.3-5 A subsequent study extends the association of this single-nucleotide polymorphism (SNP) also to alcoholic liver disease.6 PNPLA3 was recently identified as one of the genetic loci influencing plasma levels of liver enzymes,7 and the rs738409 G allele was also associated with elevated serum levels of aspartate aminotransferase (ALT) and alanine aminotransferase (AST).3, 5 A recent biochemical study showed that this nonsynonymous SNP is associated with an inactive PNPLA3, suggesting a possible direct relation between loss of function of PNPLA3 and an increase in hepatic TG content and inflammation.8

PNPLA3, also called adiponutrin, belongs to a novel class of patatin-like phospholipase (PNPLA) domain family proteins that also contain a conserved lipase catalytic dyad (Gly-X-Ser-X-Gly and Asp-X-Gly/Ala).9 Similar to other PNPLA family proteins such as PNPLA2 (also called adipose triglyceride lipase [ATGL]) and PNPLA5, PNPLA3 possesses both lipase and transacylase activities in vitro.10, 11 Its expression is highest in fat, though it also occurs in other tissues, including liver, in mice and is markedly up-regulated during adipogenesis.12 In both rodents and humans, adipose PNPLA3 expression is highly up-regulated by glucose and insulin, but free fatty acids and a high-fat diet produce no effect.13-16 The regulation of PNPLA3 by insulin and glucose is reminiscent of the control of lipogenic enzymes, such as fatty acid synthase and acetyl-coenzyme A reductase,17 and is very different from that of lipolytic enzymes like ATGL.18 Genetic linkage analyses and clinical studies demonstrated that high adipose PNPLA3 expression occurs with obesity, although not all reports are consistent.13-15, 19 Indeed, Pnpla3 messenger RNA (mRNA) levels are increased in fa/fa obese Zucker rats that lack leptin receptor.12

To determine the role of PNPLA3 in fatty liver disease and obesity, we generated Pnpla3 knockout mice by gene targeting. Loss of Pnpla3 did not affect the liver TG content or serum AST or ALT levels, whether the mice were fed normal chow or three different fatty liver–inducing diets, or after they were bred into a genetic obese Lepob/ob background. Furthermore, Pnpla3−/− mice displayed normal body fat and composition and maintained normal glucose homeostasis and insulin sensitivity. We observed, however, an up-regulation of Pnpla5 in the fat depot but not liver of Pnpla3−/− mice. These data indicate that inactivation of Pnpla3 does not lead to hepatic TG accumulation or susceptibility to diet-induced hepatic steatosis, nor does it perturb glucose homeostasis, insulin sensitivity, or adipose development in mice.


ALT, aspartate aminotransferase; AST, alanine aminotransferase; ATGL, adipose triglyceride lipase; CHD, regular chow diet; FLD, fatty liver disease; GTT, glucose tolerance test; HFD, high-fat diet; HSD, high-sucrose very low-fat diet; IP, intraperitoneal; ITT, insulin tolerance test; MCD, methionine/choline-deficient diet; mRNA, messenger RNA; NAFLD, nonalcoholic fatty liver disease; PCR, polymerase chain reaction; PNPLA, patatin-like phospholipase domain-containing; SE, standard error; SNP, single-nucleotide polymorphism; TG, triacylglycerol; WAT, white adipose tissue.

Materials and Methods


Mice were maintained in a temperature-controlled facility with fixed 12-hour light and 12-hour dark cycles and free access to regular chow and water. Some experiments were done on animals fed with a high-fat diet (HFD; 42% of kilocalories from fat; Harlan Teklad TD88137), high-sucrose very low-fat diet (HSD; Harlan Teklad; TD03045) and methionine/choline-deficient diet (MCD; MP Biomedicals; 0296043910) for various periods of time as indicated. Animals of 8-22 weeks of age were used throughout this study unless otherwise indicated. All animal experiments were done using protocols approved by the Institutional Animal Care and Use Committee at Baylor College of Medicine.

Plasma Biochemistry.

Plasma nonesterified fatty acid (NEFA) (Wako), glycerol (Sigma), glucose (ThermoScientific), total cholesterol, total TG levels (Thermo DMA), and ALT and AST (Teco Diagnostics) were measured by enzymatic assay kits for determination of their concentrations (manufacturer names given in parentheses). Serum insulin was measured by enzyme-linked immunosorbent assay (Mercodia).

Tissue Lipid Analysis.

Tissues were homogenized in standard phosphate-buffered saline. Lipids were extracted according to Bligh and Dyer,20 and dissolved in 5% triton X-100 in phosphate-buffered saline. We measured TG concentration with an Infinity triacylglycerol assay kit (Thermo DMA) and normalized to sample weight.

Glucose Tolerance and Insulin Tolerance Tests.

For the glucose tolerance test (GTT), mice were fasted 6 hours or overnight and then received an intraperitoneal (IP) injection of a bolus of 1.5 g glucose/kg body weight. Blood was collected before and at 15, 30, 60, and 120 minutes after injection. We measured glucose levels using a glucometer (FastTake; LifeScan, Inc.) and serum insulin level by an enzyme-linked immunosorbent assay (Mercodia). For the insulin tolerance test (ITT), mice were fasted for 6 hours and injected IP with humulin at a final concentration of 0.75 U/kg body weight. Blood was collected for glucose measurements before injection and at 15, 30, 60, and 120 minutes after injection.

Quantitative Reverse Transcription Polymerase Chain Reaction.

Total RNA was isolated from tissues or cultured cells with TRIzol (Invitrogen) and reverse-transcribed using Superscript II reverse transcriptase using random primers (Invitrogen). We performed real-time quantitative reverse transcription polymerase chain reaction (PCR) on the Stratagene MX3000 real-time detection system using iQ SYBR Green PCR reagent kit (Bio-Rad Laboratories). Primers used are shown in Supporting Table 2.

Statistical Analysis.

We applied the Student t test for statistical analysis. Differences were considered significant when P values were < 0.05. Results were expressed as means ± standard deviation or standard error (SE) as specified.


Absence of Pnpla3 in Mice Has no Effect on Body Fat Composition and Adipose Tissue Development.

The Pnpla3 gene was inactivated by replacing the first seven exons, including the translation initiation codon and the lipase consensus sequence motif (GXSXG, where G is Glycine, S is Serine, and X is any amino acid), with a neomycin selection cassette (Fig. 1A). Genomic PCR genotyping of tail DNA extracted from wild-type, heterozygous, and homozygous littermates are shown in Fig. 1B. Reverse transcription followed by PCR analyses confirmed that there was no Pnpla3 mRNA detectable in the white adipose tissue (WAT) of Pnpla3−/− mice (Fig. 1C).

Figure 1.

Generation of Pnpla3−/− mice. (A) We used a replacement vector that replaces the sequence encompassing exon 1 through exon 7 and part of intron 7 of the murine Pnpla3 gene with the neo gene driven by a phosphoglycerate kinase (PGK) promoter. Two thymidine kinase cassettes (TK) were ligated in tandem to the 5′ end of the targeting vector. H, HindIII; B, BamHI; E, EcoRI. (B) Genomic PCR genotyping produced bands of the predicted 522-base pair (p1 and p3) and 344-base pair (p1 and p2) nucleotides for mutant and wild-type alleles, respectively. (C) RNA from WAT of Pnpla3+/+ and Pnpla3−/− littermates was extracted and reverse-transcribed, followed by PCR analysis. No transcripts of Pnpla3 were detected in the RNA isolated from the WAT of Pnpla3−/− mice. ppia (also called cyclophilin A) transcripts were used as loading control.

Pnpla3−/− mice were born live with the expected Mendelian mode of inheritance and displayed no overtly abnormal phenotype. They were fertile and nursed their pups normally. The body weights of Pnpla3−/− and Pnpla3+/+ mice showed no difference either while they were fed a normal chow diet (CHD) (Fig. 2A) or HSD or HFD (Fig. 2B), indicating that Pnpla3 deficiency has no effect on body weight. There was also no difference in their 4-hour fasted blood glucose, plasma free fatty acids (NEFA), TG, total cholesterol or free glycerol, or overnight fasted plasma insulin level, between Pnpla3−/− mice and wild-type littermates while they were fed CHD or fed HSD or HFD for 10 weeks (Supporting Table 1). The adiposity index, determined by echo magnetic resonance imaging, of mice under the regular chow diet revealed similar body fat content with the fat making up ∼7.62% body weight in male wild-type compared with ∼8.18% of body weight in male Pnpla3−/− mice (Fig. 2C). Their lean mass was also similar (Fig. 2D). There was also no difference in the adiposity index in HSD-fed or HFD-fed Pnpla3+/+ and Pnpla3−/− mice (data not shown). In addition, we detected no difference in the weights of gonadal, subcutaneous, or brown adipose depots in Pnpla3+/+ and Pnpla3−/− mice (data not shown). Experiments on the in vitro differentiation of stromal vascular cells isolated from Pnpla3+/+ and Pnpla3−/− mice revealed no difference in the efficiency of adipocyte differentiation or TG accumulation between the two genotypes (Supporting Fig. 1A,B), which agrees with the fact that the adipose depot mass was similar in these mice. Furthermore, the basal and β-adrenergic agonist–stimulated lipolysis in fully differentiated stromal vascular cells in vitro (Supporting Fig. 1C,D) as well as in mice in vivo (Supporting Fig. 1E,F) was similar between wild-type and Pnpla3−/− cells or mice, indicating that, unlike ATGL and hormone-sensitive lipase, Pnpla3 does not contribute significantly to basal or β-adrenergic agonist–stimulated lipolysis.

Figure 2.

Pnpla3 deficiency has no effect on body weight or body fat. Growth curves of wild-type and Pnpla3−/− mice fed (A) a regular chow (CHD) and (B) high-fat (HFD) diets starting at 6 weeks, or high-sucrose very low-fat diet (HSD) starting at 4 weeks of age for the indicated number of weeks. (C,D) Echo magnetic resonance imaging analysis of 13-week-old male wild-type and Pnpla3−/− mice fed chow diet. (C) Total fat and (D) lean mass were normalized to body weight. Each group contains 10 mice. No significant differences were found.

Loss of Pnpla3 Does Not Alter Liver TG Content or Serum ALT and AST Levels.

The nonsynonymous rs738409 SNP in PNPLA3 was predicted to cause the loss of PNPLA3 enzymatic activity, a consequence functionally similar to the targeted inactivation of Pnpla3 in our mouse model.3-6 Microscopic examination of Pnpla3−/− mouse liver sections revealed normal histology (data not shown). We analyzed liver TG content in wild-type and Pnpla3−/− mice fed regular chow, and after they had been placed on three different fatty liver–inducing diets. As shown in Table 1, mice in C57BL/6 background fed the different fatty liver–inducing diets (including HSD, HFD, and MCD diets) displayed varying degrees of increased liver TG content compared with mice fed CHD. However, there was no significant difference in the degree of hepatic TG accumulation between wild-type and Pnpla3−/− mice under each type of diet, indicating that loss of Pnpla3 had no direct impact on liver TG accumulation.

Table 1. Liver TG Content and Serum ALT and AST Levels in Wild-Type and Pnpla3−/− Mice Fed Regular Chow or Different Fatty Liver–Inducing Diets or After They Were Bred into Genetic Obese Lepob/ob Background
DietsLiver TGs (mg/g Tissue)ALT (U/L)AST (U/L)
  1. All mice were sacrificed after 4 hours fasting. Mice under regular chow diets were sacrificed at 18 weeks old (C57BL/6J) and 24 weeks old (Lepob/ob background). Wild-type and Pnpla3−/− mice were fed from 6 weeks old with HFD for 20 weeks and HSD for 15 weeks, respectively, before sacrifice; 18-week-old mice were fed with MCD diet for 2 weeks before sacrifice. Each group contains at least 6-8 mice. Each value represents the mean ± SE. No significant differences were observed in the liver TGs, ALT, and AST levels between the wild-type and Pnpla3−/− mice.

Regular chow diets, two genetic backgrounds
 C57BL/6J0.087 ± 0.030.071 ± 0.0219.4 ± 421.6 ± 4.354.3 ± 1.255.5 ± 12.1
 Lepob/ob0.42 ± 0.050.41 ± 0.07162 ± 6142 ± 12159 ± 15154 ± 24
Fatty liver–inducing diets, C57BL/6J background
 HSD0.127 ± 0.030.127 ± 0.0219.3 ± 4.221.3 ± 12.534.1 ± 3.436.3 ± 13.3
 HFD0.27 ± 0.030.25 ± 0.0579 ± 3971 ± 2353 ± 1946 ± 22
 MCD0.26 ± 0.050.28 ± 0.03183 ± 34169 ± 20151 ± 31162 ± 45

Genetic variations at PNPLA3 have been reported to be associated with increased serum levels of liver enzymes in human populations.3, 5 We found that serum ALT and AST levels varied with the diet conditions (Table 1), being highest in mice fed an MCD diet, which may be related to the significant liver damage and inflammation induced by this diet. However, no difference in ALT or AST level was observed between the two genotypes, suggesting that lack of Pnpla3 in mice does not cause an elevated aminotransferase response in liver either under CHD or after the mice were fed the different fatty liver–inducing diets.

To further analyze whether loss of Pnpla3 affects fatty liver development associated with a genetic form of obesity, we intercrossed the Lepob/+ mice with Pnpla3−/− mice to produce Lepob/ob/Pnpla3+/+ and Lepob/ob/Pnpla3−/− mice. The obesity phenotype was unchanged in these mice. The Lepob/ob/Pnpla3−/− mice gained similar weight with time as compared to their Lepob/ob/Pnpla3+/+ counterparts (data not shown). Both Lepob/ob/Pnpla3−/− mice and Lepob/ob/Pnpla3+/+ littermates displayed fatty liver, and there was no difference in the hepatic TG content between the two genotypes (Table 1). Moreover, their serum ALT and AST levels were also not different (Table 1). These data indicate that loss of Pnpla3 in mice has no impact on fatty liver development under basal conditions, after they are on different fatty liver–inducing diets, or bred into a genetic model associated with obesity and fatty liver.

Loss of PNPLA3 Does Not Cause Glucose Intolerance or Insulin Resistance.

Lack of PNPLA3 has been postulated to perturb glucose homeostasis and insulin sensitivity in vivo.13, 21 As such, we measured the rate of glucose disposal and insulin sensitivity in wild-type and Pnpla3−/− mice by GTT and ITT. After administration of an exogenous glucose load, Pnpla3−/− mice and their wild-type littermates showed similar basal and stimulated blood glucose and insulin levels, indicating a normal glucose disposal rate and insulin secretory response to hyperglycemia in the absence of Pnpla3 (Fig. 3A,B). Pnpla3−/− mice and wild-type littermates on a normal chow diet also displayed a similar blood glucose during ITT (Fig. 3C), indicating no significant insulin resistance associated with loss of Pnpla3. We fed these mice an HFD for 15 weeks, and found that the blood glucose levels during the GTT in HFD-fed Pnpla3−/− mice was minimally lower than those in wild-type littermates (Fig. 3D). The plasma insulin was, however, similar during the GTT (Fig. 3E), as was the blood glucose response during an ITT (Fig. 3F) in the two HFD-fed groups. Further examination of a cohort fed an HSD for 12 weeks also revealed no difference in either blood glucose or insulin levels during GTT (Supporting Fig. 2A,B), or blood glucose levels during an ITT (Supporting Fig. 2C), between Pnpla3−/− mice and Pnpla3+/+ mice. Finally, we examined the role of Pnpla3 in mice with the genetic obese Lepob/ob background and found that Lepob/ob/Pnpla3−/− and Lepob/ob/Pnpla3+/+ mice displayed similar blood glucose and insulin levels during GTT (Supporting Fig. 2D,E) and similar blood glucose levels during ITT (Supporting Fig. 2F). Therefore, not only did the absence of Pnpla3 not affect hepatic TG content, it also did not impact the glucose intolerance and insulin resistance that often accompany hepatic steatosis.

Figure 3.

Glucose tolerance test (GTT) and insulin tolerance test (ITT) in Pnpla3−/− and Pnpla3+/+ mice fed regular chow or high-fat diet. GTT was performed on 12-week-old male wild-type and Pnpla3−/− mice after 15-hour fasting, using glucose given at 1.5 g/kg body weight by IP injection. (A) Plasma glucose levels and (B) insulin levels during the GTT test were presented as means ± SE. ITT was performed on 6-hour fasted 17-week-old to 18-week-old male wild-type and Pnpla3−/− mice with IP injection of humulin at 0.75 U/kg body weight. (C) Plasma glucose levels were presented as percent of change from glucose level at time 0. GTT was performed on wild-type and Pnpla3−/− mice fed HFD for 15 weeks as above. (D) Plasma glucose levels and (E) plasma insulin levels during the GTT test were presented as means ± SE. (F) ITT analyses on 6-hour fasted mice that had been fed HFD for 20 weeks after IP injection of humulin at 2.0 U/kg body weight. Plasma glucose levels were presented as percent of glucose change by mean ± SE.

Up-Regulated Pnpla5 mRNA Level in the WAT but not Liver of Pnpla3−/− Mice.

Thus far, our data indicate that there was no evident change in hepatic TG content or whole-body glucose homeostasis between Pnpla3−/− and Pnpla3+/+ mice under four dietary conditions (CHD, HFD, HSD, and MCD) and two genotypes (C57BL/6J Lepob/ob and C57BL/6J Lep+/+). Because the PNPLA gene family encompasses three paralogous gene products in mice, we next examined the dynamics of the three paralogs in the liver under various dietary conditions. We found that the hepatic Pnpla3 mRNA in wild-type mice was markedly up-regulated (∼32-fold) by HSD feeding but only moderately by HFD (∼5-fold) (Fig. 4A, left panel). Interestingly, HSD and HFD both stimulated hepatic Pnpla5 mRNA expression to the same degree in Pnpla3−/− and Pnpla3+/+ mice (Fig. 4A, middle panel). In contrast, as compared with regular chow (CHD), none of the fatty liver–inducing diets (HSD, HFD, and MCD) affected the level of ATGL mRNA expression (Fig. 4A, right panel). It is noteworthy that although MCD diet induced the largest TG accumulation in the liver compared with feeding with other diets (Table 1), it did not have any effect on the mRNA expression of the three different patatin-like family members (Fig. 4A). In any case, there was no evidence of compensatory adjustment in hepatic Pnpla5 or ATGL expression in the absence of Pnpla3 in the liver (Fig. 4A, middle and right panels).

Figure 4.

Quantitative analysis of mRNA expression of PNPLA family members in the liver and WAT of Pnpla3−/− and Pnpla3+/+ mice fed different diets. Total RNA was extracted and reverse transcribed from (A) liver and (B) WAT of wild-type and Pnpla3−/− mice fed CHD (18 weeks), HSD (15 weeks), HFD (20 weeks), or MCD (2 weeks, only for liver analysis) diets. We performed real-time PCR to analyze the mRNA level of Pnpla family proteins, including Pnpla3, Pnpla5, and ATGL (Pnpla2). Data were normalized to cyclophilin A and compared with CHD-fed wild-type mice. Data were presented as means ± SE. Each group contains 5-7 mice. *P < 0.05; **P < 0.005 versus CHD-fed wild-type mice. #P < 0.05; ##P < 0.005 versus wild-type mice fed the same diet.

We next examined the mRNA expression of PNPLA family genes in perigonadal WAT in wild-type and Pnpla3−/− mice. As reported previously22 and confirmed by us, Pnpla3 expression in the WAT of wild-type mice was significantly induced by HSD diet (∼2.5-fold) and slightly up-regulated by HFD diet (∼1.5-fold, not significant; Fig. 4B, left panel). Under the same conditions, the expression of Pnpla5 was not significantly affected (Fig. 4B, open bars, middle panel). The mRNA expression of ATGL was not altered under the different diets in the wild-type WAT; furthermore, the diets did not affect ATGL mRNA in WAT in the two genotypes (Fig. 4B, right panel). Interestingly, the mRNA level of Pnpla5, normally expressed in WAT at very low level compared with the other two paralogs (Lake et al.23 and our own data), was up-regulated by ∼5-fold in Pnpla3−/− mice fed regular chow (CHD). This up-regulation of Pnpla5 was also observed in the gonadal fat of Pnpla3−/− mice fed HSD or HFD, although a little less in the HFD group (Fig. 4B, solid bars, middle panel). It thus appears that increased mRNA expression of another patatin-like family member, Pnpla5, may partly compensate for the loss of Pnpla3 in mice, specifically in WAT, but not in liver.


Genome-wide association studies have identified the Pnpla3/adiponutrin gene to be associated with obesity and insulin sensitivity,13, 21, 24 and more recently with nonalcoholic,3-5 as well as alcoholic, fatty liver disease6 and elevated AST and ALT,3, 5 implicating PNPLA3 in the control of body fat, liver fat, and whole-body glucose and lipid homeostasis. However, to our surprise, we found that loss of Pnpla3 in mice does not have any effect on body weight, adiposity, or plasma lipid or glucose levels (Fig. 1 and Supporting Table 1), nor does it cause detectable alterations in hepatic TG content or serum ALT and AST levels (Table 1). Furthermore, the whole-body glucose homeostasis and insulin sensitivity remained normal. These were evident whether the Pnpla3-null mice were fed CHD, HFD, HSD, or MCD regimens or in mice bred into a genetic obesity Lepob/ob background. We conclude that Pnpla3 appears dispensable for liver TG metabolism and normal adipose development in mice.

Recently, interest in PNPLA3 gene was kindled by genetic epidemiologic analyses by multiple groups showing a strong association between a nonsynonymous variant (rs738409) in the PNPLA3 gene and hepatic fat content, suggesting the involvement of PNPLA3 in the pathogenesis of hepatosteatosis.3-6, 25 This rs738409 variant leads to an Ile148Met substitution which has been shown to inactivate the enzyme by blocking substrate access to the catalytic site.8 If PNPLA3 mediates lipolysis in the liver, one would expect to see increased TG content in the liver of Pnpla3-deficient mice. However, there was no excess hepatic TG in Pnpla3−/− mice (Table 1). We next fed mice with different fatty liver–inducing diets to test whether loss of Pnpla3 engenders increased susceptibility to fatty liver development under dietary stress. However, none of the diets tested produced a detectable difference in hepatic TG content in Pnpla3−/− mice as compared with the wild-type counterparts in any of the cohorts (Table 1). We also did not detect any significant association of serum AST or ALT in Pnpla3−/− mice fed regular chow or three different fatty liver–inducing diets, or after they were bred into a genetic obesity Lepob/ob background, although the PNPLA3 locus, and specifically the enzymatically inactive rs738409(G) genetic variant, has been shown to be associated with elevated serum liver enzymes in humans.7, 21, 24

It is unclear whether the lack of phenotype in lipid accumulation in the liver of Pnpla3−/− mice can be explained by an indirect effect resulting from an adipose-specific up-regulation of Pnpla5, a paralogous PNPLA family protein, expression of which remains very low and unchanged in the liver of Pnpla3−/− mice. PNPLA5 is highly conserved among different species and is located immediately upstream of the PNPLA3 gene in the human, rat, and mouse genome. PNPLA5 exhibits both lipase and transacylase activities in vitro, and its mRNA in WAT is regulated by changes in energy balance that is not too dissimilar from that seen with Pnpla3.23 It should be noted that Pnpla3 and Pnpla5 mRNA levels in liver are much lower than those in WAT in mice. Pnpla5 mRNA especially is barely detectable in mouse liver (Lake et al.23 and our data). On the other hand, we showed that a lipogenic high-sucrose diet treatment leads to a marked stimulation of the transcripts of both Pnpla3 and Pnpla5 in the liver (Fig. 4A). An HFD also up-regulated these mRNAs but MCD failed to do so, a finding that may be related to the different mechanism whereby MCD induces hepatic TG accumulation.26 Although the regulatory pattern of Pnpla3 by lipogenic diets could suggest an involvement in an anabolic process, whether Pnpla3 normally plays a direct role in hepatic lipogenesis is unclear. It is interesting that its ablation in mice does not affect hepatic TG accumulation under multiple dietary conditions.

He et al. recently reported that adenovirus-mediated overexpression of a human PNPLA3 Ile148Met mutant in liver increased hepatic TG content, whereas overexpression of the normal human PNPLA3 failed to lower TG.8 The seemingly paradoxical observation suggests that Pnpla3 may not be a primary liver TG-metabolizing enzyme and it is possible that the Ile148Met mutant has a dominant negative action on other liver TG hydrolases. In our study, the up-regulation of Pnpla5 mRNA only happens in the WAT but not in the liver of Pnpla3−/− mice. Such adipose tissue–specific up-regulation of Pnpla5 mRNA was consistently seen among the different cohorts receiving the different dietary manipulations (Fig. 4B), implicating a dynamic interaction between Pnpla3 and Pnpla5 in WAT. On the other hand, we cannot rule out interspecies differences of PNPLA3 or PNPLA5 action or expression between humans and mice. For example, one very recent study suggested that PNPLA3 expression is higher in the liver than in the WAT of humans,27 in contrast to mice where its expression is significantly higher in adipose tissue than in liver (our data and Lake et al.23).

In conclusion, our study constitutes the initial study demonstrating that loss of Pnpla3 in mice has no effect on hepatic TG accumulation. The observation of the up-regulation of Pnpla5 specifically in fat but not liver in Pnpla3−/− mice is intriguing. It is tempting to speculate that up-regulated adipose Pnpla5 expression may be a confounding factor that underlies, or possibly modulates, the association between the rs738409(G) allele and the presence or absence of fatty liver at the individual level.


We are indebted to our coworkers in the Chan laboratory, especially Dr. Vijay Yechoor, Dr. Minako Imamura, and Dr. Yisheng Yang, for suggestions and discussions. We are also grateful to Dr. Saul J. Karpen for a critical review of the manuscript.