The Ca²⁺/Calmodulin-Dependent Protein Kinase Kinase, CaMKK2, Inhibits Preadipocyte Differentiation

Abstract

When fed a standard chow diet, CaMKK2 null mice have increased adiposity and larger adipocytes than do wild-type mice, whereas energy balance is unchanged. Here, we show that Ca²⁺/calmodulin-dependent protein kinase kinase 2 (CaMKK2) is expressed in preadipocytes, where it functions as an AMP-activated protein kinase (AMPK)α kinase. Acute inhibition or deletion of CaMKK2 in preadipocytes enhances their differentiation into mature adipocytes, which can be reversed by 5-aminoimidazole-4-carboxamide ribonucleotide-mediated activation of AMPK. During adipogenesis, CaMKK2 expression is markedly decreased and temporally accompanied by increases in mRNA encoding the early adipogenic genes CCAAT/enhancer binding protein (C/EBP) β and C/EBP δ. Preadipocyte factor 1 has been reported to inhibit adipogenesis by up-regulating sex determining region Y-box 9 (Sox9) expression in preadipocytes and Sox9 suppresses C/EBPβ and C/EBPδ transcription. We show that inhibition of the CaMKK2/AMPK signaling cascade in preadipocytes reduces preadipocyte factor 1 and Sox9 mRNA resulting in accelerated adipogenesis. We conclude that CaMKK2 and AMPK function in a signaling pathway that participates in the regulation of adiposity.

During the past several decades, the incidence of obesity and the metabolic syndrome has reached epidemic proportions worldwide (1–3). Obesity can arise due to excessive lipid accumulation in adipocytes resulting in increased cell size coupled with the recruitment of adipocyte progenitor cells (preadipocytes) to white adipose tissue (WAT) that differentiate into additional mature adipocytes. WAT is not only an energy storage depot that helps to maintain lipid homeostasis (4) but also functions as a dynamic endocrine organ. Adipocytes, which comprise the bulk of WAT mass, secrete hormones (adipokines) that play central roles in regulating energy balance, insulin sensitivity, immunological responses, and vascular disease (5). Because adipocytes play such a critical role, elucidating molecular pathways that control adipocyte differentiation may help to develop strategies for the prevention and treatment of obesity, insulin resistance, and diabetes.

Adipogenesis follows a highly ordered temporal sequence of transcriptional events that is regulated by well-defined transcription factors. Among the early events that follow proadipogenic hormonal stimulation is the expression of CCAAT/enhancer binding protein (C/EBP)β and C/EBPδ (6, 7). C/EBPβ and C/EBPδ then trigger the expression of peroxisome proliferator-activated receptor (PPAR)γ, which is a master regulator of adipogenesis (7). In turn, PPARγ can activate C/EBPα, which exerts positive feedback on PPARγ to maintain the differentiated state, which includes expression of adipocyte genes, such as adipocyte protein 2 (aP2), fatty acid synthase (Fas), and glucose transporter type 4 (Glut4) (8).

Additional studies have shown that adipogenesis is influenced by a variety of extrinsic factors and intracellular signaling pathways (9, 10). For example, preadipocyte factor 1 (Pref-1) (variants in other tissues include delta-like protein 1, fetal antigen-1, and human adrenal-specific mRNA pG2) activates ERK to maintain the expression of sex determining region Y-box 9 (Sox9), which inhibits adipogenesis by repressing C/EBPβ and C/EBPδ transcription (11–14). In addition, adipogenesis is regulated by calcium signaling pathways as increases of intracellular Ca²⁺ in preadipocytes during the early phase of differentiation can inhibit adipogenesis (15–17) as can exposure of preadipocytes to high levels of extracellular Ca²⁺ (18). Although the mechanism by which Ca²⁺ represses adipogenesis is not fully understood, a logical one to explore is the calmodulin (CaM) kinase cascade. In this cascade, a rise in Ca²⁺ results in activation of CaM, which, in turn, activates Ca²⁺/CaM-dependent protein kinase kinase 2 (CaMKK2). CaMKK2 then phosphorylates downstream kinases, including CaMKI, CaMKIV (19–25), and AMP-activated protein kinase (AMPK) (26–28). Of these three CaMKK2 substrates, AMPK seems to be the most relevant to adipogenesis for several reasons. First, it has been shown to inhibit fatty acid synthesis and increase fatty acid oxidation in WAT (29, 30). Second, there is evidence to suggest that AMPK is involved in preadipocyte differentiation, because 5-aminoimidazole-4-carboxamide ribonucleotide (AICAR), an activator of AMPK, inhibits adipogenesis in 3T3L1 or F442A preadipocytes (31–33). Third, metformin, a drug frequently used to treat type 2 diabetes, inhibits lipid accumulation in 3T3L1 cells via a pathway that involves activation of AMPK (34). Therefore, because the mechanistic details of how Ca²⁺ and AMPK participate in adipogenesis remain unresolved, we questioned whether CaMKK2 was involved.

Our early studies on CaMKK2 null mice resulted in the unexpected finding that they have more adipose tissue than wild-type (WT) mice when fed standard chow (5001). When mice were placed on a low-fat control diet (D12328; Research Diets, Inc., New Brunswick, NJ), which has slightly less fat content but significantly more carbohydrate than the standard 5001 chow, adiposity and adipocyte size significantly increased in WT mice, whereas adiposity and adipocyte size in the CaMKK2 null mice remained the same as in mice fed the 5001 chow. Additionally, when fed the corresponding high-fat diet (D12330; Research Diets), WT mice accumulated even more adiposity and larger adipocytes, but CaMKK2 null mice demonstrated a considerably smaller increase in adiposity and adipocyte size. Because of these findings, we were interested to evaluate a potential role for CaMKK2 in the development of adiposity. Here, we show that CaMKK2 is present in preadipocytes but significantly reduced in adipocytes. Deletion or inhibition of CaMKK2 results in enhanced adipogenesis when preadipocytes are exposed to medium that initiates their differentiation but does not affect cell proliferation. We further show that the effects of CaMKK2 on adipogenesis are due to its ability to phosphorylate AMPK (rather than either of its other two substrates, CaMKI or CaMKIV) in preadipocytes. Finally, in tracing the adipogenic pathway backwards, we show that loss of CaMKK2 or AMPK activity decreases Pref-1 mRNA. Our results suggest that a CaMKK2-AMPK signaling pathway functions to maintain expression of Pref-1 in preadipocytes, which functions to inhibit adipogenesis.

Materials and Methods

Animal care

All animals were bred and maintained in either the Duke University Levine Science Research Center or Medical Sciences Research Building II animal facilities under a 12-h light (0600–1800 h), 12-h dark (1800–0600) cycle. Food and water were provided ad libitum, and all care was given in compliance with National Institutes of Health and institutional guidelines on the care and use of laboratory and experimental animals.

Dual-energy x-ray absorptiometry (DEXA)

Mice were anesthetized (ketamine, 100 mg/kg; zylaxine, 10 mg/kg) and scanned four times (with no repositioning between scans) using a Lunar PIXImus II densitometer (software version 2.0; GE Lunar Corp, Madison, WI). All mice had access to food and water ad libitum before the DEXA measurements, which were typically performed between 0800 and 1130 h. Anesthetized mice were then placed on the imaging positioning tray in a prostrate fashion with the front and back legs extended away from the body. Heads were excluded from all analyses by placing an exclusion region of interest over the head. Thus, all body composition data exclude the head.

White fat analysis

White fat analysis from mice was performed on euthanized mice. After euthanasia, gonadal fat pads were removed and were placed in 10% neutral buffered formalin for histology. Histology was performed as previously described (35). Cell dimension measurements were performed on 5-μ hematoxylin and eosin-stained sections using a Zeiss AxioImager D1 microscope (Carl Zeiss, Thornwood, NJ) and Axiovision version 4.4 software.

Adipocyte number analysis

Adipocyte number was determined by the method described by Etherton et al. (36). Briefly, gonadal fat pads were washed in 0.9% NaCl and fixed in 2% osmium tetroxide (no. 19100; Electron Microscopy Sciences, Hatfield, PA) in collidine-HCl buffer (pH 7.4) (no. 11500; Electron Microscopy Sciences) at room temperature for 72 h. After a 24-h rinse in 0.9% NaCl, tissues were treated with 8 m urea in 0.9% NaCl at room temperature for 48 h. The mixture was filtered through a 230-μm tissue sieve, and free adipocytes in the filtrate were collected on a 25-μm tissue sieve. Adipocytes were resuspended in 0.9% NaCl with 0.01% TritonX-100 and counted by Coulter Counter (Z1; Beckman Coulter, Inc., Brea, CA).

Immunoblotting

Samples were fractionated by SDS-PAGE and transferred to Immobilon-P membranes (Millipore Corp., Bedford, MA). All blocking and secondary antibody incubation steps were in Tris-buffered saline (TBS) containing 5% nonfat milk, whereas primary antibody incubations were in TBS containing 5% BSA or milk. Washes were in TBS with 0.5% Tween 20. AMPKα antiphospho-Thr¹⁷² (no. 2535; Cell Signaling, Danvers, MA), anti-AMPKα (no. 2532; Cell Signaling), ACC antiphospho-S⁷⁹ (no. 3661; Cell Signaling), and anti-CaMKK (no. 610545; BD Biosciences, San Jose, CA) were used at a 1:1000 dilution, anti-β-actin (no. A5441; Sigma, St. Louis, MO) was used at 1:2000. Horseradish peroxidase-conjugated secondary antibodies were from Jackson ImmunoResearch Laboratories, Inc. (West Grove, PA) and were used at 1:5000 with the enhanced chemiluminescence kit from GE Healthcare (Piscataway, NJ). Protein amount is quantified with ImageJ software.

Preparation of preadipocytes, cell culture, and induction of adipogenesis

Isolation of preadipocytes was performed as previously described by Permana et al. (37) and Hausman et al. (38) with some modifications. Freshly isolated mouse gonadal fat pads were minced and digested with Collagenase P (no. 11249002001; Roche, Indianapolis, IN) dissolved in DMEM with 5% fat-free BSA at a concentration of 1 mg/ml. Tissues were incubated at 37 C for 45 min and filtered through a sterile 230-μm stainless steel tissue sieve (no. 1985-00069; BellCo Glass, Vineland, NJ). After centrifugation at 500 × g for 10 min, the pelleted cells containing preadipocytes were resuspended in DMEM with 10% fetal bovine serum (FBS) and strained through a sterile 25-μm stainless steel tissue sieve (no. 1985-00500; BellCo Glass). The filtrate was transferred to a 60-mm culture plate, cells were allowed to attach, and the next day, floating red blood cells were removed by aspiration. Medium was changed every other day. After 4 d, the cells were removed from the plate using trypsin/EDTA and counted using a Coulter Counter.

Mouse 3T3L1 cells from the American Type Culture Collection (Manassas, VA) were cultured in DMEM (17-205-CV; Cellgro, Herndon, VA) with 10% bovine serum (16170-078; Life Technologies, Inc., Carlsbad, CA). C3H10T1/2 cells from the American Type Culture Collection were culture in α-MEM (no. 12561-056; Life Technologies, Inc.) with 10% heat-inactivated FBS (no. S11150; Atlanta Biologicals, Lawrenceville, GA). Mouse embryonic fibroblasts (MEF) were cultured in DMEM with 10% heat-inactivated FBS. On d 0 (2 d after cell reaches confluence), adipogenesis was induced by 1 μg/ml (0.2 μg/ml for 3T3L1) insulin (no. 12585-014; Life Technologies, Inc.), 0.5 mm isobutylmethylxanthine (no. I7018; Sigma), and 1 μm dexamethasone (no. D4902; Sigma) in DMEM with 10% FBS. Forty-eight hours later, cells were changed to a medium containing 1 μg/ml insulin (0.2 μg/ml for 3T3L1) for 4 d and then DMEM with 10% FBS for 4 d. After differentiation, cells were fixed in 10% formalin and stained with Oil Red O. Oil Red O was extracted by isopropyl alcohol, and its optical density was monitored spectrophotometrically at 520 nm. To test different compounds in adipogenesis, 2 μm STO-609 (Tocris Bioscience, Ellisville, MO), 0.25 mm or 0.5 mm AICAR (Calbiochem, San Diego, CA), or 0.5 μm KN-93 (Calbiochem) were added into medium 2 d before confluence (day −4) and refreshed every other day. Compound C was purchased from Calbiochem.

Short hairpin RNA (shRNA) and small interfering RNA (siRNA) transfection

shRNA against CaMKK2 was obtained from Open Biosystems (Huntsville, AL) and transfected into 293T cells with lentiviral packaging plasmids to make viral particles. 3T3L1 preadipocytes were infected using polybrene according to the manufacturer's recommendations; cells were screened by puromycin.

siRNA against AMPKα1 and AMPKα2 were obtained from Integrated DNA Technologies (Coralville, IA) and transfected into WT MEF using Lipofectamine 2000 reagent (no. 11668-019; Invitrogen, Carlsbad, CA) according to the manufacturer's recommendations. Protein and RNA were harvested 4 d after transfection.

Ad-CMV-GFP and Ad-Cre-IRES-GFP viruses were obtained from Vector Biolabs (Philadelphia, PA). Flox-CaMKK2 MEF were plated in six-wells plates and infected with 4 × 10⁹ plaque formation unit/well viruses by polybrene. Protein and RNA were exacted 6 d after infection

RNA isolation, RT, and RT-PCR

Total RNA from cells and tissues was extracted using TRIzol (15596-026; Invitrogen) and quantified spectrophotometrically. Single-stranded cDNA was synthesized using SuperScript II Reverse Transcriptase (no. 18064-022; Invitrogen) according to the manufacturer's directions. Real-time PCR was carried out using iCycler (Bio-Rad, Hercules, CA) with the iQ STBR Green Supermix (no. 170-8882; Bio-Rad) and the following primers: CaMKK2 5′-CATGAATGGACGCTGC-3′ (forward) and 5′-TGACAACGCCATAGGAGCC-3′ (reverse), 18S ribosomal protein 5′-AGGGTTCGATTCCGGAGAGG-3′ (forward) and 5′-CAACTTTAATATACGCTATTGG-3′ (reverse), C/EBPβ 5′-CGCCTTTAGACCCATGGAAG-3′ (forward) and 5′-CCCGTAGGCCAGGCAGT-3′ (reverse), C/EBPδ 5′-GACTCCTGCCATGTACGA-3′ (forward) and 5′-GGTTGCTGTTGAAGAGGT-3′ (reverse), C/EBPα 5′-TATAGACATCAGCGCCTACATCGA-3′ (forward) and 5′-GTCGGCTGTGCTGGAAGAG-3′ (reverse), PPARγ 5′-GGCCATCCGAATTTTTCAAG-3′ (forward) and 5′-GGGATATTTTTGGCATACTCTGTGA-3′ (reverse), aP2 5′-AAGAAGTGGGAGTGGGCTTTG-3′ (forward) and 5′-CTCTTCACCTTCCTGTCGTCTG-3′ (reverse), Glut4 5′-CATTCCCTGGTTCATTGTGG-3′ (forward) and 5′-GAAGACGTAAGGACCCATAGC-3′ (reverse), Fas 5′-CCGAAGCCACAAAGCT-3′ (forward) and 5′-GTCAAGGTTCAGGGTGC-3′ (reverse), Pref-1 5′-AATAGACGTTCGGGCTTGCA-3′ (forward) and 5′-TCCAGGTCCACGCAAGTTCCATTGTT-3′ (reverse), and Sox9 5′-AGGTTTCAGATGCAGTGAGGAGCA-3′ (forward) and 5′-ACATACAGTCCAGGCAGACCCAAA-3′ (reverse). After deriving the relative amount of each transcript from a standard curve, transcript levels were normalized to 18S ribosomal RNA.

Statistical analysis

Statistical significance was tested with unpaired Student's t test. P values of less than 0.05 were considered significant.

Results

CaMKK2 null mice have increased adiposity characterized by larger adipocytes

DEXA scanning, used to evaluate body composition, revealed increased body fat percentage in adult CaMKK2 null mice fed standard chow relative to WT mice, and this difference was not age dependent, because it was also evident in 3-wk-old mice at the time of weaning (Fig. 1C). hematoxylin and eosin staining of gonadal fat pad sections revealed that the size of adipocytes in adult CaMKK2 null mice was much larger than WT controls (Fig. 1A). Quantitative analysis confirmed the visual results and revealed a striking shift in adipocytes in WAT of CaMKK2 null mice toward a significantly larger size (Fig. 1B). Dissection of epididymal fat, retroperitoneal fat, and scapular brown adipose tissue, as well as regional DEXA scanning, showed that every fat pad examined was larger in CaMKK2 null mice (Supplemental data, published on The Endocrine Society's Journals Online web site at http://endo.endojournals.org). We also found that gonadal fat pads from CaMKK2 null mice contained more adipocytes relative to those of WT mice (Supplemental data). The increased adiposity was not correlated with changes in energy balance, because null mice consumed similar amounts of chow and did not show differences in energy expenditure compared with WT controls (data from comprehensive lab animal monitoring system analysis not shown). Thus, these data infer that the increased adiposity of CaMKK2 null mice could be due to altered fat accumulation instead of energy balance. Because CaMKK2 expression in WAT has not been reported, we tested for protein expression by immunoblot analysis and found it to be undetectable in WAT (Fig. 1D). However, it was clearly expressed in WT primary preadipocytes freshly isolated from WAT and MEF but absent in similar preparations of cells from CaMKK2 null mice (Fig. 1D).

Fig. 1.

CaMKK2 null mice have more adipose tissue and larger adipocytes compared with WT mice. A, Adipocytes are larger in CaMKK2 null mice. Paraffin-embedded sections of gonadal WAT from 3-month-old mice were stained with hematoxylin and eosin. B, Quantitative distribution of adipocyte size in WT and CaMKK2 null (KO) mice. C, DEXA scans of CaMKK2 null mice indicate a significant increase in adiposity compared with WT animals after 3 months on standard 5001 chow. n = 10; *, P < 0.005. There is also a significant increase at weaning (3 wk). n = 10; *, P < 0.02. Data are presented as mean ± sem. D, CaMKK2 protein is expressed in preadipocytes. Protein extract from WAT, primary cultured preadipocytes, and MEF were Western blotted with a panCaMKK antibody. To help distinguish between CaMKK1 and CaMKK2, WT and CaMKK2 null mouse brain extracts were included as controls. CaMKK2 protein is present in WT preadipocytes and MEF but is undetectable in WAT.

CaMKK2 expression decreases after adipogenesis

The fact that CaMKK2 is expressed in primary preadipocytes but barely detectable in WAT suggested to us that CaMKK2 expression may decrease during adipogenesis. To address this we used real-time PCR to compare CaMKK2 mRNA before and after adipogenesis with aP2 as a diagnostic marker for mature adipocytes. Primary preadipocytes had high amounts of CaMKK2 mRNA and protein but little aP2 mRNA; however, this was reversed in mature adipocytes as aP2 mRNA greatly increased, whereas CaMKK2 mRNA and protein greatly decreased (Fig. 2, A and B). This reciprocal relationship between CaMKK2 and aP2 was also observed in three other cell lines that can be induced to differentiate into adipocytes, namely, 3T3L1, MEF, and C3H10T1/2 cells (Fig. 2, A and B). These data show that CaMKK2 is present in four types of preadipocytes, but mature adipocytes differentiated from these cells contain markedly decreased amounts of CaMKK2 mRNA and protein. These data suggested to us that CaMKK2 might be inhibitory to adipogenesis.

Fig. 2.

CaMKK2 mRNA and protein expression decreases after adipogenesis. A, CaMKK2 mRNA decreases after adipogenesis. Total mRNA was isolated from primary-cultured preadipocytes, adipocytes, and from undifferentiated (UD) (d 0) or differentiated (FD) (d 10) MEF, 3T3L1, and C3H10T1/2 cells. CaMKK2 mRNA level was quantified by RT-PCR (top panel). aP2 is included as a marker of adipogenesis (bottom panel). Data are presented as mean ± sem. n = 3; *, P < 0.05. B, CaMKK2 protein decreases after adipogenesis. Protein was extracted from primary-cultured preadipocytes, adipocytes, and from undifferentiated (d 0) or differentiated (d 10) MEF, 3T3L1, and C3H10T1/2 cells. Samples were Western blotted for panCaMKK. One representative blot (upper) and quantification (lower) of three independent experiments are shown. The amount of CaMKK2 protein in each type of preadipocyte is normalized to 1.

CaMKK2 inhibits adipogenesis via phosphorylation of AMPK

Because aP2 is a late marker of adipogenesis, we questioned whether depletion or inhibition of CaMKK2 resulted in accelerated fat accumulation when preadipocytes are exposed to agents that stimulate adipogenesis. As shown in Fig. 3A, MEF from CaMKK2 null mice form more triglyceride (based on Oil Red O staining) containing cells than do WT MEF. Similarly, the presence of the CaMKK2 selective inhibitor STO-609 accelerated adipogenesis of 3T3L1 and C3H10T1/2 cells (Fig. 3, B and C). Finally, the depletion of CaMKK2 from 3T3L1 cells using a shRNA strategy similarly accelerated adipogenesis (Fig. 3D). These data support our contention that CaMKK2 functions in a signaling pathway that prevents differentiation of preadipocytes in several cell models of adipogenesis. However, CaM kinase pathways have been shown to support cell proliferation (39). Because cells must exit from the proliferative cycle to differentiate, we questioned whether the absence of CaMKK2 altered the doubling time of preadipocytes isolated from CaMKK2 null mice or MEF in culture. We found that the absence of CaMKK2 did not alter cell cycle progression of preadipocytes or MEF (Supplemental Fig. 1, A and B), providing additional support that CaMKK2 primarily functioned to inhibit adipogenesis.

Fig. 3.

Loss of CaMKK2 activity enhances adipogenesis. A, WT and CaMKK2 null (KO) MEF were induced to differentiate as outlined in Materials and Methods. After 10 d of differentiation, intracellular lipid was stained with Oil Red O (left). To quantify the amount of Oil Red O, the dye was eluted with isopropyl alcohol, and its optical density was monitored spectrophotometrically at 520 nm (right). B and C, Inhibition of CaMKK2 in 3T3L1 or C3H10T1/2 preadipocytes enhances adipogenesis. 3T3L1 or C3H10T1/2 preadipocytes were induced to differentiate in the presence of vehicle or 2 μm STO-609. Amount of adipogenesis was determined by Oil Red O staining (left), which was quantified as in A shown on the right. D, Deletion of CaMKK2 in 3T3L1 preadipocytes enhances adipogenesis. 3T3L1 preadipocytes were infected with control or one of two different shRNA viruses against CaMKK2 (3143, 3145). The efficiency of CaMKK2 knockdown was determined by Western blot analysis. Cells were induced to differentiate and analyzed as in A. Quantification of Oil Red O is presented as mean ± sem; *, P < 0.05. Data shown are representative of at least three independent experiments, and each was done in triplicate.

CaMKK2 has been shown to have three direct substrates, CaMKI, CaMKIV, and AMPK (19–28), so we next questioned which of these substrates may lie downstream of CaMKK2 in the signaling pathway that prevents adipogenesis. Because we had previously generated mice null for CaMKIV, we compared the ability of CaMKIV null MEF with WT and CaMKK2 null MEF in the triglyceride accumulation assay. The absence of CaMKIV did not accelerate adipogenesis (Fig. 4A). CaMKI is essential for cell proliferation and functions predominantly in G₁ of the cell cycle (39, 40). Because the absence of CaMKK2 did not slow cell proliferation, we surmised that CaMKI was not likely to be the relevant substrate supporting that antiadipogenic role for CaMKK2. To confirm this idea, we cultured 3T3L1 cells with KN-93, which inhibits with similar potency CaMKI and CaMKIV (and CaMKII) (41–44). Because KN-93 did not accelerate adipogenesis (Fig. 4B), we reasoned that CaMKI (or CaMKIV or CaMKII) was not the relevant target of CaMKK2. To test the third substrate, AMPK, we cultured 3T3L1 cells from which CaMKK2 had been depleted by shRNA and CaMKK2 null MEF with the AMPK activator, AICAR. AICAR inhibited the ability of CaMKK2-depleted cells to support adipogenesis in a dose-dependent manner, suggesting that activation of AMPK was negatively correlated with adipogenesis and could be the relevant substrate by which CaMKK2 inhibits adipogenesis (Fig. 4, C and D).

Fig. 4.

CaMKK2 works through AMPK to inhibit adipogenesis. A, WT, CaMKIV null, and CaMKK2 null MEF were induced to differentiate. Loss of CaMKIV did not alter adipogenesis in MEF as determined by Oil Red O staining. B, 3T3L1 preadipocytes were induced to differentiate in the presence of vehicle or 5 μm KN-93; adipogenesis was not affected as determined by Oil Red O staining. C, 3T3L1 preadipocytes infected with CaMKK2 shRNA (3145) were induced to differentiate in the presence of 0, 0.25, or 0.5 mm AICAR. Adipogenesis was inhibited by AICAR as determined by Oil Red O staining. *, P < 0.05 compared with 0 mm AICAR. D, CaMKK2 null MEF (KO) were induced to differentiate in the presence of 0, 0.25, or 0.5 mm AICAR. Adipogenesis was inhibited by AICAR as determined by Oil Red O staining. *, P < 0.05 compared with 0 mm AICAR. Quantification of Oil Red O is presented as mean ± sem. Data shown are representative of three independent experiments, and each was done in triplicate.

We wanted to confirm that CaMKK2 is a functional AMPK kinase in preadipocytes. As has been shown in the hypothalamus (45, 46), Western blot analyses of primary preadipocyte extracts showed that basal AMPK phosphorylation was markedly decreased in the absence of CaMKK2 (Fig. 5A). In WT MEF, ionomycin, which increases intracellular Ca²⁺ leading to CaMKK2 activation, up-regulated AMPK phosphorylation, and its effect was blocked by preincubation of cells with the CaMKK2 inhibitor, STO-609. On the other hand, not only was basal AMPK phosphorylation reduced in CaMKK2 null MEF, but it failed to respond to ionomycin or STO-609 (Fig. 5B). Additionally, deletion of CaMKK2 from 3T3L1 preadipocytes by shRNA resulted in a large decrease in basal AMPK phosphorylation (Fig. 5C). As found with WT MEF, treatment of normal 3T3L1 preadipocytes with ionomycin increased AMPK phosphorylation and was blocked by STO-609 (Fig. 5D). Ionomycin and STO-609 had similar effects in normal C3H10T1/2 preadipocytes to those in MEF (Fig. 5E). We conclude from these data that CaMKK2 functions as an AMPK kinase in several types of preadipocytes.

Fig. 5.

CaMKK2 phosphorylates AMPK in preadipocytes. A, Basal AMPK phosphorylation is down-regulated in CaMKK2 null (KO) preadipocytes. Protein extract from WT and CaMKK2 null primary cultured preadipocytes were immunoblotted for T172 phosphorylated AMPK (pAMPK) and total AMPK. *, P < 0.05 compared with WT. B, WT and CaMKK2 null (KO) MEF were incubated for 1 h with vehicle or 2 μm STO-609, followed by stimulation with 1 μm ionomycin for 5 min. Cell extracts were immunoblotted for pAMPK and total AMPK. Ionomycin increased AMPK phosphorylation and was blocked by STO-609 in WT but not CaMKK2 null MEF. *, P < 0.05 compared with control WT MEF. C, CaMKK2 was knocked down by shRNA (3143, 3145) in 3T3L1 preadipocytes. Basal AMPK phosphorylation was down-regulated in CaMKK2 null cells. *, P < 0.05 compared with control 3T3L1 preadipocytes. D and E, Ionomycin increased AMPK phosphorylation and was blocked by STO-609 in 3T3L1 or C3H10T1/2 preadipocytes. Ionomycin increased acetyl-CoA carboxylase phosphorylation (pACC) and was blocked by STO-609 in 3T3L1 preadipocytes. *, P < 0.05 compared with control. For A–E, one representative blot and quantification of three independent experiments are shown.

Loss of CaMKK2 activity increases mRNA of adipogenic transcription factors and markers

The temporal sequence of transcriptional events that lead to adipogenesis is well known. In response to agents that stimulate adipogenesis, a sequence of events occurs. Early in the process C/EBPβ and C/EBPδ are transcriptionally induced, and in turn, these transcriptional activators trigger transcription from the PPARγ and C/EBPα genes. PPARγ and C/EBPα then induce transcription of genes that are required in adipocytes, such as aP2, Fas, and Glut4. To determine whether CaMKK2 acts at any of these steps that lead from preadipocyte to adipocyte, we used real-time PCR to quantify the various adipogenic gene transcripts in WT and CaMKK2 null MEF after exposure to adipogenic stimulation. We found that adipocytes derived from CaMKK2 null MEF express much more mRNA encoding all seven adipogenic genes than adipocytes derived from WT MEF (Fig. 6A). This was also true for 3T3L1 adipocytes derived from cells treated with CaMKK2 shRNA or cultured in STO-609 relative to those cells subjected to a nonsense shRNA or cultured in the absence of the inhibitor, respectively (Fig. 6, B and C). The presence of STO-609 during adipogenic stimulation of C3H10T1/2 cells also resulted in enhanced expression of all seven adipogenic mRNA examined (Fig. 6D). These results suggest that CaMKK2 must function very early in adipogenesis or in a pathway that inhibits adipogenesis in preadipocytes.

Fig. 6.

Loss of CaMKK2 activity increases adipogenic gene transcripts during adipogenesis. A, mRNA levels of C/EBPβ, C/EBPδ, PPARγ, C/EBPα, aP2, Fas, and Glut4 are increased in CaMKK2 null (KO) MEF adipocytes compared with WT. Total RNA was isolated from WT or CaMKK2 null MEF after adipogenesis. mRNA levels of adipogenic genes were determined by RT-PCR. B, 3T3L1 preadipocytes infected with control or CaMKK2 shRNA were induced to differentiate, and total RNA was isolated after adipogenesis. RT-PCR indicated mRNA levels of PPARr, C/EBPa, aP2, Fas, and Glut4 are increased in 3T3L1 adipocytes with CaMKK2 knocked down. C and D, mRNA levels of C/EBPβ, C/EBPδ, PPARγ, C/EBPα, aP2, Fas, and Glut4 are increased in 3T3L1 or C3H10T1/2 adipocytes treated with 2 μm STO-609 during differentiation. For A–D, data are presented as means ± sem; *, P < 0.05. Data shown are representative of three independent experiments, and each was done in triplicate.

CaMKK2 and AMPK maintain Pref-1 signaling

Pref-1 is expressed in preadipocytes, where it inhibits adipogenesis by regulating a signaling pathway that leads to the phosphorylation and activation of ERK and subsequently to the transcriptional up-regulation of Sox9, which functions to transcriptionally repress the C/EBPβ and C/EBPδ genes (11–14). To examine whether CaMKK2 might function in the Pref-1 signaling pathway, we quantified Pref-1 and Sox9 mRNA in primary preadipocytes and MEF from WT and CaMKK2 null mice. These data revealed that the absence of CaMKK2 resulted in considerably reduced amounts of Pref-1 and Sox9 mRNA in both cell types (Fig. 7, A and B). We also used immunoblot analysis to quantify the phosphorylation of ERK1/2 in MEF prepared from WT mice compared with MEF isolated from mice heterozygous (Het) or null (KO) for CaMKK2. As shown in Fig. 7C, ERK phosphorylation was decreased in both heterozygous and KO cells relative to WT. To confirm that CaMKK2 played a role in regulating the Pref-1 pathway, we cultured WT MEF in the presence of STO-609 and found that the CaMKK2 inhibitor also reduced the amount of Pref-1 and Sox9 mRNA (Fig. 7D). To determine if reduction of Pref-1 and Sox9 resulted from acute depletion of CaMKK2 from MEF, we examined the effect of Cre-mediated deletion of CaMKK2 isolated from CaMKK2 loxP mice and found indeed that both mRNA were reduced (Fig. 7E). Finally, shRNA-mediated depletion of CaMKK2 from 3T3L1 cells also markedly reduced Pref-1 and Sox9 mRNA (Fig. 7F). Therefore, CaMKK2 is important to maintain the level of Pref-1 and consequently the Pref-1 signaling pathway in multiple preadipocyte cell types.

Fig. 7.

CaMKK2 maintains Pref-1 mRNA. A and B, Pref-1 and Sox9 mRNA levels are decreased in CaMKK2 null primary preadipocytes and MEF. Total RNA was isolated from WT and CaMKK2 null (KO) primary preadipocytes or MEF; Pref-1 and Sox9 mRNA levels were quantified by RT-PCR. C, Protein extracts from WT, CaMKK2−/+, and CaMKK2 null MEF were immunoblotted for phosphorylated ERK (pERK) and total ERK (left). Phosphorylation of ERK is reduced in CaMKK2 null MEF as shown in quantification of blot (right). Each lane represents MEF derived from an individual embryo. D, Total RNA was isolated from WT MEF incubated with vehicle or 2 μm STO-609 for 2 h. Pref-1 and Sox9 mRNA levels were determined by RT-PCR. E, MEF from CaMKK2 loxP mice were infected with Ad-CMV-GFP (GFP) or Ad-Cre-IRES-GFP (CRE) viruses. Cells were harvested 6 d after infection. Protein extract was immunoblotted for CaMKK2 and pAMPK (left). Ad-Cre-IRES-GFP virus effectively depleted CaMKK2 and decreased AMPK phosphorylation. Pref-1 and Sox9 mRNA levels were reduced as determined by RT-PCR (right). F, Total RNA was extracted from 3T3L1 preadipocytes infected with control or CaMKK2 shRNA. Pref-1 and Sox9 mRNA levels were decreased in shRNA infected cells compared with control, as determined by RT-PCR. Data are presented as means ± sem. *, P < 0.05. Data shown in A, B, and D–F are representative of at least two independent experiments, and each was done in triplicate. Het, Heterozygous.

We have presented evidence that AMPK serves as the primary CaMKK2 substrate to regulated the pathway by which CaMKK2 inhibits adipogenesis (Fig. 5). It has been reported that AICAR, a compound which can activate AMPK, also reduces adipogenic gene transcripts and restores Pref-1 mRNA during the differentiation of 3T3L1 cells (32). Thus, we questioned whether AMPK was involved in regulating Pref-1 in MEF. Knocking down the AMPKα subunits by siRNA decreased Pref-1 and Sox9 mRNA in WT MEF (Fig. 8A). Furthermore, Compound C, an inhibitor of AMPK, reduced Pref-1 and Sox9 mRNA in WT MEF (Fig. 8B), but treatment with KN-93, a compound that inhibits both CaMKI and CaMKIV, did not (data not shown). Finally, AICAR, an activator of AMPK, increased Pref-1 and Sox9 mRNA in CaMKK2 null MEF (Fig. 8C). These data reveal that CaMKK2 functions in preadipocytes to activate AMPK, which results in up-regulation of Pref-1 mRNA, leading to maintenance of the Pref-1 signaling pathway and therefore inhibition of adipogenesis (Fig. 8D).

Fig. 8.

AMPK maintains Pref-1 mRNA. A, WT MEF were transfected with control or AMPKα siRNA. Protein and total RNA were extracted 4 d after transfection. Both phosphorylated AMPK (pAMPK) and total AMPK were effectively decreased, as determined by immunoblot (left). Pref-1 and Sox9 mRNA levels were decreased, as determined by RT-PCR (right). *, P < 0.05. Data shown are from one experiment that is representative of three independent experiments, which were each done in triplicate. B, Total RNA was extracted from 3T3L1 preadipocytes incubated with vehicle or 2 μm Compound C (Calbiochem) for 1 h. Pref-1 and Sox9 mRNA levels were decreased after Compound C treatment as determined by RT-PCR. *, P < 0.05. Data shown are from one experiment that is representative of three independent experiments, which were each done in triplicate. C, Total RNA was extracted from CaMKK2 null (KO) MEF treated with vehicle or 0.25 mm AICAR for 1 h. Pref-1 and Sox9 mRNA levels were increased in AICAR-treated MEF as determined by RT-PCR. *, P < 0.05, n = 6. Data shown are from one experiment that is representative of two independent experiments. D, CaMKK2 inhibits adipogenesis by maintaining Pref-1 mRNA via AMPK. In response to increased intracellular Ca²⁺ concentration, CaMKK2 phosphorylates and activates AMPK. CaMKK2 and AMPK maintain mRNA level of Pref-1 through an unknown mechanism. Pref-1 stimulates phosphorylation and activation of ERK1/2, which maintains expression of Sox9. Sox9 inhibits transcription of the early adipogenic transcription factors C/EBPβ and C/EBPδ, which reduces expression of PPARγ and C/EBPα and, in turn, adipocyte-specific genes, such as aP2, Glut4, and Fas.

Discussion

CaMKK2 null mice have been used to demonstrate a number of roles for this protein kinase in the brain (45, 47–49). However, these mice have not been employed to identify cell autonomous effects of CaMKK2 in peripheral cells. Here, we show CaMKK2 to be expressed in preadipocytes and that germ-line depletion of this gene results in increased mass of WAT coupled with a decreased number of preadipocytes that can be isolated from WAT (Supplemental Fig. 2) when mice are fed standard lab chow (5001). When exposed to adipogenic stimuli primary preadipocytes (in vivo) or MEF (in culture) from CaMKK2 null mice differentiate into adipocytes to a greater extent than do cells isolated from WT mice, and the increase in molecular markers of the mature fat cell is inversely correlated with the disappearance of CaMKK2. That this is cell autonomous rather than mediated via the nervous system or developmentally is revealed by the fact that virtually identical changes result when CaMKK2 activity is blocked from by pharmacological inhibition or shRNA-mediated knockdown in 3T3L1 and C3H10T1/2 cells in culture. We demonstrate that the inhibitory effect of CaMKK2 on adipogenesis is due to a role of CaMKK2 and its substrate AMPK in maintaining the level of Pref-1 mRNA. The Pref-1 protein functions via the activation of ERK to enhance the amount of SOX9, which, in turn, acts as a transcriptional repressor to prevent the expression of C/EBPβ and C/EBPδ, two early genes that positively regulate the differentiation of preadipocytes to adipocytes (11–14). Thus, in response to initiating a series of events that lead to adipocyte formation and fat production by these cells, adipogenic stimulation also results, by a yet to be discovered mechanism, in down-regulation of CaMKK2 mRNA and protein. In this context, CaMKK2 and AMPK participate in a signaling pathway to maintain the preadipocyte rather than to stimulate the production of mature adipocytes from these progenitor cells (Fig. 8D).

We have previously reported that CaMKK2 null mice accumulate less body weight when fed a high-fat diet, which is primarily due to reduced expansion of WAT (45). As shown in Fig. 1, the average size of mature adipocytes is greater in WAT isolated from CaMKK2 null than it is in WT mice fed standard lab chow (5001; 28.5% protein, 13.5% fat, and 58% carbohydrate). In the study by Anderson et al. (45), the pregnant females were also maintained on lab chow and remained on this diet while nursing their pups. At the time of weaning, the pups were placed on solid food for the first time and fed either a high-fat diet (D12230; 16.4% protein, 58% fat, and 25.5% carbohydrate) or its “control” diet (D12328; 16.4% protein, 10.5% fat, and 73.1% carbohydrate) for 30 wk. We carried out histology of the WAT and found that the size of adipocytes of WT mice increased markedly between chow and control diets and increased again, in a statistically significant manner, between WT mice fed control and high-fat diet (Supplemental Fig. 3). However, although adipocyte size was greater in WAT from CaMKK2 null mice that WT mice fed chow, there was no increase in adipocyte size in WAT from the CaMKK2 null mice between chow and control diets and only a very small increase in size between control and high-fat diet. These data are consistent with our data showing depletion of preadipocytes in CaMKK2 null mice fed a chow diet and suggest that preadipocyte depletion is likely an important factor contributing to the resistance to diet-induced obesity exhibited by the CaMKK2 null mice. It is possible that diet-induced fat accumulation is also reduced in CaMKK2 null mature adipocytes, but if that occurs, it is due to a different and yet to be explored mechanism.

It came as a considerable surprise to us that preadipocytes were depleted in mice with a germ-line deletion of the Camkk2 gene, because CaMKK2 had not previously been shown to be expressed in these cells. Our novel data reveal that AMPK serves as the primary CaMKK2 substrate in the signaling pathway by which CaMKK2 delays (or prevents) adipogenesis and that a relevant consequence of this pathway is regulation of Pref-1 mRNA. Intriguingly, forced expression of Pref-1 in 3T3L1 cells makes them resistant to differentiation (11, 13), and reduced expression of Pref-1 is required for differentiation of these cells into adipocytes. The same is true for activation or inhibition of CaMKK2 or AMPK in 3T3L1 cells. This sequence of events suggests that the down-regulation of CaMKK2 in response to adipogenic stimuli may be an even earlier event than a reduction in Pref-1 and be required to down-regulate Pref-1 mRNA. Regulation of Pref-1 transcription is poorly understood, but studies have reported that hypoxia-inducible factor 1α, hypoxia-inducible factor 2α, and forkhead box A2 promote Pref-1 transcription, whereas SMAD family member 1 and 4 and hairy and enhancer of split 1 inhibit it (50–53). This information may help to inform future efforts designed to identify the primary AMPK target in the CaMKK2-mediated pathway that inhibits adipogenesis. Elucidating the relevant stimuli and resultant signaling events both upstream and downstream of CaMKK2/AMPK may lead to a more complete understanding of how adipogenesis is gated physiologically. It would also support CaMKK2 as a novel and potentially attractive target to develop therapeutics to combat diet-induced obesity.