Liver Kinase B1 Is Required for White Adipose Tissue Growth and Differentiation

Antibodies to C/EBPα (2295S), PPARγ (2295S), phosphorylated (p-)AMPKα T172 (2535 L), Fabp4 (3544S), p-Akt Thr308 (9275S), p-Akt Ser473 (9271S), Akt (9272S), p-S6K1 T389 (9234S), S6K1 (2708S) and p-FoxO1 Ser256 (9461S) were from Cell Signaling Technology (Danvers, MA). Antibodies to LKB1 (sc-32245), GAPDH (sc-137179), and sterol regulatory element (SRE)–binding protein 1c (SREBP-1; sc-8984) were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Antibodies to uncoupling protein 1 (AB1426), p-IRS1 Tyr612 (07-846), and IRS1 (06-248) were purchased from Millipore (Temecula, CA). Anti-Fbw8 (ab85647) and p-IRS1 S636/639 (ab53038) were from Abcam (Cambridge, U.K.). Anti-fatty acid synthetase (NB400114) was from Novus Biologicals (Littleton, CO). Anti-CUL7 (13738–1-AP) was from Proteintech (Chicago, IL). Anti-FLAG (F7425) antibody was from Sigma-Aldrich (St. Louis, MO). The vectors expressing FLAG-tagged SREBP-1a, SREBP-1c, and SREBP-2 were purchased from Addgene (Cambridge, MA). FuGene HD transfection reagent was from Roche Applied Science (Indianapolis, IN). The Dual-Luciferase Assay Kit was purchased from Promega (Madison, WI).

To generate adipose tissue-specific disruption of the LKB1 gene (LKB1^ad−/−), LKB1^flox/flox mice were crossbred with transgenic mice expressing the Cre recombinase under the control of the Fabp4 promoter/enhancer. PCR-based genotyping of LKB1^ad−/− mice was performed with the following primers: Fabp4-Cre, 5′-TCTCACGTACTGACGGTGG-3′ and 5′-CCAGCTTGCATGATCTCC-3′; for the upstream 5′ loxP site, 5′-TCTAACAATGCGCTCATCGTCATCCTCGGC-3′; and for the downstream 3′ loxP site, 5′-GAGATGGGTACCAGGAGTTGGGGC-3′. The deletion of exons 3–6 of LKB1 was verified as described previously (22). The animals were housed in a controlled environment (20 ± 2°C, 12-h/12-h light/dark cycle) and had free access to water and standard chow diet. All experiments were conducted in accordance with the Animal Care and Use Committee of the University of Oklahoma Health Science Center.

Primary MEFs were isolated from embryos of LKB1^flox/flox mice at 13.5 days postcoitum. Cells were cultured at 37°C in high-glucose Dulbecco’s modified Eagle’s medium (GIBCO BRL, Carlsbad, CA) supplemented with 10% (volume for volume [v/v]) heat-inactivated FBS (Sigma-Aldrich) and 100 U/ml penicillin/streptomycin (GIBCO BRL). LKB1^flox/flox MEF cells were infected with adenovirus expressing lacZ (wild-type [WT]) or Cre recombinase (LKB1 KO). Two days later, the medium was replaced with the standard differentiation induction medium containing 0.5 mmol/L methylisobutylxanthine, 1 μmol/L dexamethasone, 10 μg/mL insulin, 10 μmol/L troglitazone (all from Sigma), and 10% (v/v) FBS. This medium was changed every other day. For the preparation of proteins for Western blot analysis, cells were harvested at 0, 2, 5, and 8 days of differentiation. AMPKα1- and -α2–deficient MEFs were prepared using the same procedure.

Oil Red O staining was performed at 8 days after cell differentiation. Briefly, cells were washed twice with PBS and fixed with 10% formalin in PBS for 1 h. They were washed three times with water and stained with Oil Red O (six parts of 0.6% Oil Red O dye in isopropanol and four parts of water) for 1 h. Excess stain was removed by washing with water, and then microscopic images were recorded.

Triglycerides (TGs) were extracted from cultured cells using hexane/isopropanol (3:2 [v/v]). The lipid extracts were dried under a gentle stream of nitrogen and dissolved in cold isopropanol and TG levels measured using a commercially available kit (Thermo Scientific, Waltham, MA).

All fat tissues were fixed in 10% neutral buffered formalin and embedded in paraffin. Five-micrometer sections were cut, stained with hematoxylin and eosin (H&E), and photographed under a light microscope.

Blood was taken from anesthetized mice. The levels of leptin, adiponectin, IGF-1, and insulin were measured by an ELISA kit (R&D Systems, Minneapolis, MN). Serum TG and cholesterol levels were measured using the corresponding kit from Thermo Scientific.

Fresh fat tissues were collected and immediately frozen in liquid nitrogen. Proteins from whole-cell lysates were extracted in lysis buffer with sonication and analyzed using Western blot. The intensities (density three area) of individual bands were measured by densitometry (Model GS-700, Imaging Densitometer; Bio-Rad, Hercules, CA).

Total RNA was extracted from frozen tissues (kept at −80°C) using the Tri Reagent (Ambion, Austin, TX), and 1 μg of RNA was reverse-transcribed into cDNA using iScript cDNA Synthesis Kits (Bio-Rad). PCR amplification was performed using the SYBR PCR mix (Bio-Rad). The oligonucleotides used in PCR analysis were as follows: LKB1, 5′-AGCTGCGCAGGATCCCCAAT-3′ and 5′-TGGCACACAGGGAAGCGCTT-3′; IRS1, 5′-ACTATGCCAGCATCAGCTTC-3′ and 5′-AAGGAGGATTTGCTGAGGTC-3′; Fbw8, 5′-GCCACGGAGCCCGAGCCGCTGG-3′ and 5′-GCACCTCGTCTTCAGCAATCAC-3′; and GAPDH, 5′-TTGTCAAGCTCATTTCCTGGTATG-3′ and 5′-GCCATGTAGGCCATGAGGTC-3′.

Chromatin immunoprecipitation (ChIP) analyses of MEF cells were performed using the Magna ChIP A/G Chromatin Immunoprecipitation Kit (Millipore), following the manufacturer's recommendations. LKB1^flox/flox MEF cells were first infected with adenovirus expressing lacZ (WT) or Cre recombinase (LKB1 KO) for 48 h and then cross-linked using formaldehyde; DNA was then sheared by sonication and incubated with 1 μg of normal rabbit IgG (sc-2027) or anti–SREBP-1 (sc-8984). Immunoprecipitation was performed with the magnetic beads included in the kit. For PCR, 2 μl of the 50 μl total immunoprecipitated DNA was analyzed with the following oligonucleotides: SRE1, 5′-CTCAGAGAGGTCTTCATGACAGCT-3′ and 5′-TGGCGCACGCCTTTAATCCCAGCA-3′; SRE2, 5′-TTGCCTACATCAGCACCCTGTTCT-3′ and 5′-CTACTTGATATTTCCACCGATGTA-3′; and SRE3, 5′-CTGATGAAGGCTCAGGCCTCTGAC-3′ and 5′-TGATGTAGGCAAGGGTGGCTAAGG-3′.

The DNA fragment from the mouse Fbw8 promoter cloned into pGL3Basic (Promega, Madison, WI) to generate the WT-luc construct was amplified by PCR using the following oligonucleotides: 5′-GCTAGCCACAAAGGCATGTACACAGGTATA-3′ and 5′-AAGCTTACTGTCCATCCGCCGGAAGTGTCC-3′. Mutant constructs of the predicted SRE sites in the Fbw8 promoter were generated by site-directed mutagenesis (Promega) using the following primers: SRE1, 5′-CCTGGCTGTCCTGGAACCAAATCCATAGACCAGGCTGG-3′ and 5′-CCAGCCTGGTCTATGGATTTGGTTCCAGGACAGCCAGG-3′; SRE2, 5′-TCTGGGAACATGGACTTAGGCAAACCCAATCTCACGTCCCATTG-3′ and 5′-CAATGGGACGTGAGATTGGGTTTGCCTAAGTCCATGTTCCCAGA-3′; and SRE3, 5′-ACAGATGACATCACTGTTTACTTTTGGAAACCCAGGACAGCCAGCGACAGC-3′ and 5′-GCTGTCGCTGGCTGTCCTGGGTTTCCAAAAGTAAACAGTGATGTCATCTGT-3′. For the luciferase assay, the luciferase reporter plasmids along with the SREBP-1c plasmid were transfected into MEF cells in 24-well plates using FuGene HD reagent (Roche Applied Science). The pRL-TK plasmid carrying the Renilla luciferase under the control of the thymidine kinase promoter was also cotransfected as an internal control for transfection efficiency. Cells were harvested 48 h after transfection, and luciferase activities were then analyzed using the Dual-Luciferase Assay Kit (Promega).

Mean ± SEM values were analyzed using GraphPad Prism (version 3; GraphPad Software Inc., San Diego, CA). Statistical comparisons between groups were performed using the Student t test. Curves were compared using two-way ANOVA and Bonferroni posttests. Groups were considered significantly different at P values ≤0.05.

We first investigated the mRNA and protein levels of LKB1 in WAT and BAT from WT mice. As depicted in Fig. 1A, the mRNA levels of LKB1 in WAT were fourfold higher than those in BAT. Compared with the weak staining seen for LKB1 in BAT, higher expression of LKB1 proteins was found in WAT (Fig. 1B), suggesting that LKB1 is mainly present in WAT.

To further elucidate the role of LKB1, we generated mice containing an adipose tissue–specific LKB1 deletion using mice containing the floxed LKB1 allele (22). Because Fabp4, like LKB1, is mainly expressed in WAT, we crossed LKB1^flox/flox with Fabp4-Cre mice to generate WAT-specific LKB1 KO (LKB1^ad−/−) mice. The LKB1^flox/flox/Cre⁻ mice were used as WT controls. As expected, LKB1 was not detectable in the WAT and remained low in the BAT, of LKB1^ad−/− mice (Fig. 1C). Consistent with this, the level of phospho-AMPKα (Thr172), one of the most important substrates of LKB1, was barely detectable in the WAT of LKB1^ad−/− mice, in contrast to the unaltered levels of p-AMPK seen in the BAT of LKB1^ad−/− mice (Fig. 1C and D). Taken together, our results suggest that LKB1 is essential for maintaining AMPK Thr172 phosphorylation in WAT in vivo.

Compared with their WT counterparts (Fig. 1E), LKB1^ad−/− mice exhibited severe postnatal growth retardation, starting at day 4 after birth. The body weights of LKB1^ad−/− mice at birth and for the first 3 days after birth were indistinguishable from their littermates (Fig. 1F). However, from day 5 after birth, all LKB1^ad−/− mice had significantly reduced body weights relative to those of their WT counterparts (Fig. 1F). At day 17, male LKB1^ad−/− mice weighed 49.5% less than their WT littermates (5.15 ± 0.45 vs. 10.20 ± 0.52 g, n = 5; P < 0.001). Importantly, LKB1^ad−/− mice began to die at day 3 after birth. None of the LKB1^ad−/− mice survived beyond postnatal day 19 (Fig. 1G), whereas all WT pups survived. The blood glucose levels of LKB1^ad−/− mice measured at days 3, 9, and 18 were markedly lower than those of WT mice (Fig. 1H).

Next, we examined the effect of LKB1 deletion on WAT and BAT. Necropsy analysis revealed that the amount of WAT, including subcutaneous, visceral, or gonadal fat tissue, was severely reduced in LKB1^ad−/− mice compared with their age-matched littermates. Fig. 2A shows epididymal fat from 18-day-old WT and LKB1^ad−/− mice. The 18-day-old male WT mice had a mean value of 9.5% white fat of body weight, whereas LKB1^ad−/− mice had 1.27% white fat of body weight (Fig. 2B). The amount of BAT was also smaller in LKB1^ad−/− mice than in WT (Fig. 2A); however, the ratio of BAT to body weight was not lower in the LKB1^ad−/− mice (Fig. 2B). Histological analyses of WAT and BAT using H&E staining showed that white adipocytes were smaller in LKB1^ad−/− mice than in WT mice (Fig. 2C and D). In contrast, there was no difference in the morphology of BAT between WT and LKB1^ad−/− mice, further confirming that deletion of LKB1 affects only WAT but not BAT. Interestingly, the ratios of liver, kidney, and heart tissue to body weight were similar between WT and LKB1^ad−/− mice (Fig. 2B). Histological examination confirmed normal structures in both heart and liver in LKB1^ad−/− mice (Supplementary Fig. 1).

In addition to its primary function as an energy reservoir, WAT is an endocrine organ, secreting adipocytokines that regulate glucose and lipid metabolism (23). We therefore examined whether the defective adipogenesis seen in LKB1^ad−/− mice affected the secretion of critical hormones such as adiponectin, leptin, and IGF-1. Serum adiponectin levels were lower in LKB1^ad−/− mice (3.88 ± 1.17 μg/mL) than in WT mice (6.08 ± 1.28 μg/mL) (Fig. 3A). Circulating levels of leptin were barely detectable in 18-day-old LKB1^ad−/− mice (Fig. 3B). Similarly, serum IGF-1 levels were reduced by 89.6% in LKB1^ad−/− mice (Fig. 3C). However, insulin levels were similar between WT and LKB1^ad−/− mice (Fig. 3D).

Next, we detected whether reduced WAT altered the levels of TG in LKB1^ad−/− mice. As shown in Fig. 3E, LKB1 deletion in adipose tissue resulted in decreased serum TG, indicating that LKB1 deficiency leads to impaired adipogenesis and lipogenesis in mice. However, LKB1 deficiency in WAT did not change serum cholesterol levels (Fig. 3F) or the mRNA level of hydroxymethylglutaryl (HMG) CoA reductase, the rate-limiting enzyme in cholesterol synthesis, in the livers (Supplementary Fig. 2). Our results support the notion that liver but not WAT is the central organ in maintaining whole-body cholesterol homeostasis (24), and WAT is poor at de novo cholesterol biosynthesis (25).

We further investigated how LKB1 deficiency affects adipogenesis. We first monitored the levels of C/EBPα and PPARγ, two known transcription factors for adipogenesis. As shown in Fig. 4A, the levels of C/EBPα and PPARγ were significantly lower in LKB1-deficient WAT, but not in BAT, compared with WT. As expected, the levels of Fabp4 were significantly reduced in LKB1-deficient WAT, but not in BAT. The levels of uncoupling protein 1 (UCP1) were reduced in LKB1-deficient WAT but not in BAT (Fig. 4A).

Insulin/Akt signaling is known to be essential in the process of adipogenesis; therefore, we determined whether LKB1 deficiency altered the insulin/Akt pathway. As depicted in Fig. 4B, total IRS1 levels were lower in LKB1-deficient WAT. Consistent with this, the levels of phospho-Akt (T308 and S473) were also lower upon LKB1 deletion (Fig. 4B and C).

To elucidate how LKB1 deletion causes a decrease in IRS1, we detected IRS1 mRNA levels using RT-PCR. As shown in Fig. 4D, IRS1 mRNA levels were similar between WT and LKB1^ad−/− adipose tissues, suggesting that LKB1 deletion may lower IRS1 levels by increasing its degradation. Since IRS1 was reported to be degraded by the E3 ubiquitin complex containing CUL7 and the Fbw8-substrate-targeting subunit (14), we tested whether LKB1 deletion altered the levels of CUL7 and Fbw8. The level of CUL7, which is the scaffolding protein in the complex, remained unchanged in LKB1-deficient WAT, whereas the level of Fbw8, the IRS1-target subunit, was higher (Fig. 4E). Consistent with this, Fbw8 mRNA levels were eightfold higher in LKB1-deficient adipose tissue than in WT adipose tissue (Fig. 4F).

We further researched how LKB1 altered adipogenesis in cultured cells. We cultured LKB1^flox/flox MEFs, and infection with Ad-Cre virus led to LKB1 deletion in these cultured cells. LKB1^flox/flox MEFs were first infected with the lacZ or Cre virus and then induced to differentiate into adipocytes for 8 days. Oil Red O staining showed that the number of adipocytes obtained from LKB1-deficient MEFs was significantly lower than that obtained from WT (Fig. 5A). As expected, the TG levels were also lower upon LKB1 deletion (Fig. 5B). Western blot results indicated that the levels of the adipogenesis markers Fabp4 and fatty acid synthetase (FAS), as well as those of C/EBPα and PPARγ, were significantly decreased in LKB1-deficient cells (Fig. 5C), consistent with the results from LKB-deficient WAT (Fig. 4A).

To confirm the presence of impaired IRS1/Akt signaling in LKB1-deficient cells, total IRS1 was detected by Western blot. As seen in Fig. 5D, the IRS1 level was significantly decreased in LKB1-deficient cells. A phospho-specific antibody directed at Tyr612 of IRS1 (a putative binding domain for the p85 subunit of phosphoinositide-3-kinase) was used to examine the insulin-stimulated activation of IRS1 (26,27). As expected, the level of p-IRS1 (Tyr612) was also decreased in LKB1-deficient cells (Fig. 5D). Consistent with this, the levels of phospho-Akt (Thr308 and Ser473) and p-FoxO1 (Ser256) were also decreased (Fig. 5D).

To investigate if Akt reduction contributed to reduced adipogenesis by LKB1, LKB1-deficient cells were infected with adenoviruses encoded with constitutively active (CA)-Akt or GFP. As shown in Fig. 5E, adenoviral overexpression of CA-Akt but not GFP restored the adipogenesis in LKB1-deficient cells. Consistently, overexpression of CA-Akt but not GFP normalized the expression of Fabp4, PPARγ, and C/EBPα in LKB1-deficient cells (Fig. 5F).

To further confirm impaired insulin/Akt signaling in LKB1-deficient cells, WT or LKB1-deficient cells were treated with insulin (100 nmol/L) for 15 min, and Akt phosphorylation was detected by Western blots. As shown in Fig. 5G, both phospho-Akt S473 and T308 were decreased in LKB1-deficient cells, suggesting impaired Akt phosphorylation in LKB1-deficient cells. Since IRS1 was reported to be phosphorylated by S6K1 with consequent Akt inhibition (28), we further probed for p-S6K1 in LKB1^−/− adipocytes. As shown in Fig. 5H, p-S6K1 (T389) was increased in LKB1-deficient WAT whereas the levels of p-IRS1 (Ser636/639) in WAT from LKB1^−/− were reduced when compared with WT, suggesting that increased Akt inhibition in LKB1-deficient WAT was not due to increased phosphorylation of IRS1 by S6K1. Consistently, the ratios of p-IRS1 (Ser636/639) to total IRS1 were not altered in LKB1-deficient WAT (data not shown). Since the total IRS1 was significantly lower in LKB1-deficient WAT than WT (Fig. 4B), our data suggest that LKB1 deletion results in decreased IRS1 levels with consequent Akt reduction, thereby leading to impaired adipogenesis.

Our results above indicated that the expression of Fbw8 is increased in LKB1-deficient WAT. To further confirm that LKB1 deletion leads to Fbw8 overexpression, LKB1^flox/flox MEFs were infected with virus encoding the Cre recombinase for different times, and Western blotting was performed to detect the Fbw8 level. As shown in Fig. 6A, Cre-mediated LKB1 loss led to decreased p-AMPK, increased Fbw8, and decreased IRS1 levels. Fig. 6B depicts the time course of the relationships among LKB1, p-AMPK, Fbw8, and IRS1. LKB1 levels decreased within 12 h, the phosphorylation of AMPK was significantly decreased and the Fbw8 level elevated at 24 h, and the IRS1 level began to decrease at 36 h. These data suggest that LKB1 loss decreased AMPK phosphorylation.

To establish a causative role for Fbw8-induced IRS1 degradation, Fbw8 small interfering RNA (siRNA) was used to knockdown Fbw8, and then the Cre virus was used to delete LKB1. LKB1 deletion increased IRS1 degradation (Fig. 6C), and as expected, knockdown of Fbw8 blocked LKB1 deletion-induced IRS1 degradation. Taken together, these data indicate that LKB1 deletion increases IRS1 degradation via Fbw8 overexpression.

Because the level of p-AMPKα was significantly decreased in LKB1-deficient WAT (Fig. 1C), we next determined whether LKB1 loss-induced IRS1 degradation was AMPK-mediated. Overexpression of a CA-AMPK viral construct increased AMPK activity, as evidenced by the increased phosphorylation of acetyl-CoA carboxylase (ACC), one of the downstream enzymes of AMPK (Fig. 6D). As shown in Fig. 6D, transfection with CA-AMPK ablated both Fbw8 expression and the IRS1 reduction caused by LKB1 deletion. These data indicate that LKB1 regulates Fbw8 expression in an AMPK-dependent manner.

To further confirm the role of AMPK in adipogenesis, primary AMPKα1- and -α2–deficient and WT MEF cells were cultured for 25 days. Both AMPKα1- and -α2–deficient MEFs exhibited dramatically impaired adipocyte differentiation compared with WT MEFs (Fig. 6E). Western blot results showed that the levels of PPARγ and Fabp4 were decreased in differentiated AMPKα1- and -α2–deficient cells (Fig. 6F). Thus, LKB1 deletion has a phenotype similar to that of AMPKα loss in adipogenesis, which further confirms the importance of the LKB1–AMPK–Fbw8–IRS1 axis in adipogenesis.

We next dissected how LKB1/AMPK regulates Fbw8 expression. Promoter analysis of Fbw8 was performed using the Transcription Factor Database (http://www.gene-regulation.com) internet-based transcription factor binding site program. Three SRE sites were found in the Fbw8 promoter (Fig. 7A), suggesting that SREBP may bind to the Fbw8 promoter to regulate its expression.

There are three SREBP isoforms, and they have different distributions and target genes. To determine which SREBP isoform regulated Fbw8 expression, flag-tagged active SREBP isoforms were transfected into cells, and Western blots were performed to detect Fbw8 expression. As shown in Fig. 7B, SREBP1c, but not SREBP1a or SREBP2, induced the overexpression of Fbw8.

AMPK is reported to phosphorylate and inhibit SREBP-1c activity (29). Next, we investigated whether LKB1 deletion induced the maturation of SREBP-1c and its translocation into the nucleus. Cell fractionation was performed to separate cytoplasmic and nuclear proteins. As shown in Fig. 7C, LKB1 deletion not only increased the conversion of the SREBP-1 precursor (125 kDa) into active SREBP-1 (65 kDa) but also promoted the translocation of SREBP-1 into the nucleus.

To further confirm that SREBP-1c can bind to the SRE sites in the Fbw8 promoter, different primers were designed for a DNA ChIP experiment. Deletion of LKB1 induced binding of SREBP-1c to all of the three SRE sites in the Fbw8 promoter (Fig. 7D). To further analyze the functional role of SREBP-1c binding sites on the activity of the Fbw8 promoter, we mutated the core SREBP-1c binding sites in three SREs and then inserted these into luciferase plasmids. As shown in Fig. 7E, SREBP-1c transfection significantly increased luciferase activity from the WT Fbw8 promoter. In contrast, the SRE mutants showed a reduction in SREBP-1c–mediated Fbw8 promoter activation.

Finally, SREBP-1c siRNA was used to knockdown SREBP-1c, and then the Cre virus was used to delete LKB1. As shown in Fig. 7F, LKB1 deletion resulted in Fbw8 overexpression, as observed in an earlier experiment. Moreover, SREBP-1c knockdown ablated LKB1 deletion-induced Fbw8 overexpression (Fig. 7F).

In this study, we have demonstrated that the LKB1-IRS1 pathway is indispensable for adipogenesis. The impaired adipogenesis seen in mice lacking LKB1 in adipose tissue (LKB1^ad−/− mice) results in reductions in WAT and early death. Thus, our results establish that LKB1 is essential for the development and metabolism of adipose tissue.

LKB1^ad−/− mice were significantly smaller and more slender than the control littermates and typically died within 3 weeks after birth. Some studies have reported viable mouse models with reduced adipose deposition (30,31), indicating that the early lethality seen in LKB1^ad−/− mice may be unrelated to the inability of these mice to form WAT. Blood glucose levels are much lower in LKB1^ad−/− mice than in WT mice. Some reports showed that hypoglycemia in Sirt6-KO mice or C/EBPα-KO mice caused their early death (32,33). Thus, the early lethality in LKB1^ad−/− mice may result from the hypoglycemic phenotype.

In this study, we report for the first time, based on ChIP and luciferase assays, that SREBP-1c binds to the Fbw8 promoter and regulates its expression. Thus, LKB1 deletion led to SREBP-1c activation and Fbw8 overexpression, resulting in decreased IRS1 levels and impaired adipogenesis. Our findings clarify the function of SREBP-1c in adipose tissue and further explain the phenotype of transgenic mice with adipose tissue–specific expression of SREBP-1c. SREBP-1s are considered to be profoundly involved in the transcriptional regulation of lipogenic enzymes (34), and the predominant SREBP-1 isoform in liver and adipose tissue is SREBP-1c rather than SREBP-1a (35). However, the involvement of SREBP-1c in fatty acid synthesis differs fundamentally between the liver and adipose tissue. Several lines of evidence, especially from transgenic and KO mouse models (36,37), have proved that mature hepatic SREBP-1c regulates lipogenic gene expression in the liver. In contrast, targeted disruption of the SREBP-1 gene scarcely affected the dynamic changes in lipogenic gene expression in the adipose tissue (37). Moreover, SREBP-1c overexpression in adipocytes disrupted the differentiation processes of adipocytes, resulting in lipodystrophy in transgenic mice (38). Hence, the role of SREBP-1c in lipogenic gene regulation in adipocytes is not well-understood.

Another important finding in this study is that LKB1 regulates SREBP-1c activity through AMPK. This observation is consistent with a previous report on the relationship between AMPK and SREBP-1c (29). However, the role of AMPK in adipogenesis is still unclear and controversial. Some studies have reported that treatment of 3T3-L1 cells with AMPK agonists leads to inhibition of clonal expansion (39,40); therefore, AMPK is thought to be antiadipogenic. In our study, AMPK activity was decreased in LKB1-deficient WAT, and this was accompanied by impaired adipogenesis. This suggests that AMPK is adipogenic. Besides, AMPKα-deficient MEFs could not differentiate into adipocytes. However, whole-body AMPKα1 or -α2 KO mice do not exhibit the lipodystrophic or obese phenotype. It is therefore possible that AMPKα1 and -α2 have compensatory effects in mice bearing KOs of the other gene. For example, AMPKα2 activity in BAT, WAT, and liver was significantly greater in AMPKα1^−/− than in WT mice (41). Besides, AMPK was reported to be required for brown adipocyte differentiation (42). Thus, it is necessary to generate adipose tissue–specific AMPKα1 and -α2 double-KO mice to further confirm the role of AMPK in adipogenesis.

LKB1^ad−/− mice exhibit severely decreased WAT mass, which is linked to decreased adipocyte size. This model is useful for studying the differentiation process of WAT. However, the role of LKB1 in BAT is not clear. Because Fabp4 was found mainly in WAT, Fabp4-mediated LKB1 deletion mainly occurred in WAT (Figs. 1A and B and 4A). In addition, we found that LKB1 is normally expressed in BAT and that LKB1^ad−/− mice have normal BAT (Figs. 1A and B and 4A). Thus, our study has not addressed the importance of LKB1 in BAT, and further studies on the role of LKB1 in BAT are warranted. As expected based on the reduced fat storage, LKB1^ad−/− mice also show reduced levels of circulating adiponectin, leptin, TGs, and IGF-I. IGF-1 plays an important role in skeletal development and normal postnatal growth (43). IGF-1 deficiency in mice results in marked skeletal dysplasia and growth retardation (44). Thus, the lower level of serum IGF-1 in LKB1^ad−/− mice may contribute to their small skeleton size and growth retardation.

In summary, we report the identification of LKB1 as an essential player in adipocyte differentiation. Our work provides a new mouse model that exhibits impaired adipogenesis, which will prove very useful in studies of the mechanism of adipogenesis. As preadipocytes from elderly humans retain the capacity to differentiate in vitro (45), it is possible that adipogenesis occurs throughout life at a low rate until energy storage demands promote further differentiation. Our study opens up new avenues of research into human lipid metabolism. Our future studies will investigate the relationship between LKB1 and obesity.

This study was supported by National Institutes of Health grants HL-079584, HL-074399, HL-080499, HL-089920, HL-096032, HL-110488, HL-105157, and HL-110488 and by funds from the Warren Chair in Diabetes Research of the University of Oklahoma Health Science Center (all to M.-H.Z.). Part of this work was also supported by an American Heart Association Scientist Development Grant (11SDG5560036 to P.S.). M.-H.Z. is a recipient of the National Established Investigator Award from the American Heart Association.

No potential conflicts of interest relevant to this article were reported.

W.Z. designed and performed the experiments, analyzed data, and prepared the manuscript. Q.W. and P.S. researched the data. M.-H.Z. conceived the projects, designed the experiments, reviewed the data, and wrote the manuscript. M.-H.Z. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.