|Home | About | Journals | Submit | Contact Us | Français|
Kinase Suppressors of Ras 1 and 2 (KSR1 and KSR2) function as molecular scaffolds to potently regulate the MAP kinases ERK1/2 and affect multiple cell fates. Here we show that KSR2 interacts with and modulates the activity of AMPK. KSR2 regulates AMPK-dependent glucose uptake and fatty acid oxidation in mouse embryo fibroblasts and glycolysis in a neuronal cell line. Disruption of KSR2 in vivo impairs AMPK-regulated processes affecting fatty acid oxidation and thermogenesis to cause obesity. Despite their increased adiposity, ksr2-/- mice are hypophagic and hyperactive, but expend less energy than wild type mice. In addition, hyperinsulinemic-euglycemic clamp studies reveal that ksr2-/- mice are profoundly insulin resistant. The expression of genes mediating oxidative phosphorylation is also down regulated in the adipose tissue of ksr2-/- mice. These data demonstrate that ksr2-/- mice are highly efficient in conserving energy, revealing a novel role for KSR2 in AMPK-mediated regulation of energy metabolism.
Molecular scaffolds, which coordinate the interaction of signaling molecules to affect efficient signal transduction (Burack and Shaw, 2000; Morrison and Davis, 2003), have the potential to serve as organizing nodes for multiple biological inputs. Kinase Suppressor of Ras 1 (KSR1) (Kortum and Lewis, 2004; Nguyen et al., 2002) serves as a scaffold for the coordination of signals through the Raf/MEK/ERK kinase cascade. Manipulation of KSR1 reveals its role in regulating the transforming potential of oncogenic Ras, neuronal and adipocyte differentiation, and replicative life span (Kortum et al., 2005; Kortum et al., 2006; Kortum and Lewis, 2004; Muller et al., 2000). A related protein, KSR2, has been detected in C. elegans, and humans (Channavajhala et al., 2003; Ohmachi et al., 2002). In C. elegans, KSR2 is required for germline meiotic progression and functions redundantly with KSR1 in excretion, vulva development, and spicule formation. Thus, KSR proteins appear to play critical roles in regulating multiple cell fates.
Cells must sense the nutritional status of the extracellular environment, monitor intracellular energy stores and integrate that information with intracellular pathways that drive cell fate. The trimeric AMP-activated protein kinase (AMPK) is a critical regulator of energy homeostasis that is activated when the nutritional environment is poor and intracellular ATP levels are low (Hardie, 2007). Under conditions of energy stress, ATP levels fall, and levels of the allosteric activator AMP rise, which promotes binding of the catalytic AMPK α-subunit to the γ-subunit, and protects against de-phosphorylation of a critical threonine in the activation loop of the α-subunit kinase domain (Sanders et al., 2007). ATP antagonizes the action of AMP on AMPK, making AMPK a sensor of cellular energy stores.
Upon activation, AMPK stimulates metabolic enzymes and induces gene expression programs to promote catabolic activity and inhibit anabolic activity. AMPK stimulates insulin-independent glucose uptake in muscle in response to exercise and hypoxia (Mu et al., 2001), and promotes the β-oxidation of long-chain fatty acids by phosphorylating and inhibiting acetyl-CoA carboxylase (ACC), the rate-limiting enzyme of malonyl-CoA synthesis (Ruderman et al., 2003). Malonyl-CoA is both a key substrate for fatty acid synthesis and an inhibitor of carnitine palmitoyltransferase 1 (CPT1), which mediates import of fatty acyl-CoA molecules into the mitochondria for oxidation. By inhibiting ACC, AMPK inhibits the synthesis of fatty acids and promotes their metabolism to generate ATP.
The mechanisms linking dietary nutrients and AMP-regulated energy metabolism to cell fate are incompletely understood. Our analysis of the scaffold KSR2, however, provides a missing link. Here we show that KSR2 interacts with AMPK in a functionally relevant manner in vitro and in vivo. In cultured cells AMPK-dependent effects on basal glucose uptake, fatty acid oxidation and glycolysis are enhanced by KSR2. Gene disruption of ksr2 reduces the glucose lowering capacity of an AMPK agonist in vivo. The white adipose tissue of ksr2-/- mice demonstrates defective phosphorylation of the AMPK substrate acetyl-CoA carboxylase, a key regulator of fatty acid synthesis and oxidation, and a down regulation of genes involved in oxidative phosphorylation. As a consequence of these impaired AMPK-dependent mechanisms, ksr2-/- mice are obese and insulin resistant. In contrast to most other animal models of obesity, ksr2-/- mice are hypophagic and more active. However, they expend less energy than wild type mice. These data reveal a novel role for KSR2 as a critical regulator of cellular energy balance affecting lipid and glucose metabolism.
We used mass spectrometry to identify novel regulators that may affect KSR1 and KSR2 function. Previously reported KSR1-interacting molecules (Muller et al., 2001; Ory et al., 2003; Ritt et al., 2007), including HSP90, HSP70, Cdc37, C-Tak1 and PP2A were detected in association with KSR2 as were peptidergic components of the α1, α2, β1 and γ1 subunits of AMP-activated protein kinase (AMPK) (Figure 1A). This interaction was verified by the co-precipitation of endogenous AMPK α1 subunit in immunoprecipitates of KSR2 but not in immunoprecipitates of KSR1 (Figure 1B). When both AMPKα and KSR1 were overexpressed, an interaction between the two proteins could be detected. However, when compared for their relative ability to precipitate endogenous or ectopic AMPKα, KSR2 consistently precipitated more AMPK than did KSR1 (Figure 1C). To determine the sites required for interaction with AMPK, full length and truncated versions of KSR1 and KSR2 were expressed in 293T or COS-7 cells and their ability to precipitate endogenous AMPK or the ectopic AMPKα subunit was tested. Deletion analysis indicated that the CA3 region contributes to the interaction of the AMPKα subunit with KSR1 (Figure 1D and E) and that the CA3 region and amino acids unique to KSR2 intervening between the CA2 and CA3 region (Dougherty et al., 2009) each contributed to the ability of KSR2 to interact with the AMPKα subunit (Figure 1F-G). The observation that KSR2 retains additional sequences contributing to its interaction with AMPK may explain its ability to precipitate a proportionally greater amount of AMPK.
We tested whether mutations that impaired the interaction of AMPK with KSR1 or KSR2 affected ERK activation. The ΔCA3 mutation severely affected maximal activation of ERK by KSR proteins as observed previously for KSR1 (Michaud et al., 1997) (Figures (Figures2A2A and S2A). However, KSR2Δ327-392 effects on ERK activation were equivalent to KSR2 (Figure 2A). These data indicate it is unlikely ERK activation plays a role in mediating the effects of KSR2 on AMPK signaling.
AMPK stimulates glucose uptake and the oxidation of fatty acids by mitochondria, thereby promoting ATP synthesis (Ruderman et al., 2003). To test whether KSR proteins affect these catabolic functions of AMPK, we generated ksr2-/- mice (Figure S1) and ksr2-/- MEFs. We measured glucose uptake in ksr2-/- and ksr1-/- mouse embryo fibroblasts (MEFs) and in null MEFs expressing their respective cognate ksr transgenes. Though MEFs do not express detectable KSR2 (not shown), expression of ectopic KSR2 in ksr2-/- MEFs increased basal glucose uptake 5-fold (Figure 2B) and expression of KSR1 in ksr1-/- MEFs increased basal glucose uptake 2-fold (Figure S2B). Deletion of the CA3 region (KSR2.ΔCA3) modestly impaired basal glucose uptake, but deletion of sequences between the CA2 and the CA3 regions (KSR2.Δ327-392) decreased KSR2-induced glucose uptake by 50% (Figure 2B). Deletion of the CA3 region (KSR1.ΔCA3) was sufficient to significantly disrupt the effects of KSR1 on basal glucose uptake (Figure S2B).
We then tested whether KSR2 or KSR1 affected cellular fatty acid metabolism. The rate of oxygen consumption (OCR) was used as an index of oxidative phosphorylation (Wu et al., 2007). OCR in ksr2-/- or ksr1-/- MEFs was compared to OCR in ksr2-/- or ksr1-/- MEFs expressing full length KSR2 or KSR1, respectively. In both ksr2-/- and ksr1-/- MEFs, palmitate had little or no ability to stimulate oxygen consumption. However, the reintroduction of KSR2 (Figure 2C) or KSR1 (Figure S2C) restored oxidation of palmitate. As with glucose uptake, disrupting the interaction of KSR2 or KSR1 with AMPK prevented restoration of fatty acid oxidation (Figures (Figures2C2C and S2C).
The neuroblastoma X glioma hybrid cell line NG108-15 expresses endogenous KSR2 (Dougherty et al., 2009). An shRNA was constructed to knock down KSR2 expression in these cells. In culture, neuronal cell lines in general, and NG108-15 cells in particular, synthesize ATP primarily from glycolysis (Ray et al., 1991). Since AMPK activation promotes glycolysis (Marsin et al., 2000), we examined the ability of KSR2 knockdown with and without the co-expression of a constitutively active AMPK construct to regulate glycolysis in NG108-15 cells. As an index of lactate production, extracellular acidification was markedly reduced in these cells when RNAi was used to inhibit KSR2 expression. However, expression of a constitutively active AMPK (Stein et al., 2000) restored glycolysis to levels seen in control cells (Figure 2D).
Phosphorylation of Thr172 on the effector loop of AMPK promotes its activity (Stein et al., 2000). In MEFs the presence or absence of KSR2 did not affect AMPK Thr172 phosphorylation (not shown), suggesting that AMPK signaling may require KSR2 for proper spatial regulation similar to its effect on ERK (Dougherty et al., 2009). However, in NG108-15 cells, loss of KSR2 inhibits the phosphorylation of AMPK on Thr172 in response to AICAR and modestly impairs phosphorylation of the AMPK substrate ACC (Figure 2E). These data support the conclusion that KSR2 is an AMPK regulator. To determine the relative impact of KSR1 and KSR2 on AMPK function and fatty acid metabolism in vivo we examined the metabolic phenotype of ksr1-/- and ksr2-/- mice.
KSR2 protein is detectable by western blot only in brain (Figure S1C). Therefore, we determined the relative abundance of ksr2 mRNA in various tissues (Figure 3A). Consistent with western blot analysis, ksr2 mRNA is approximately 100-fold more abundant in brain than in skeletal muscle or liver. ksr2 mRNA is detectable at low levels in adipose tissue. In contrast, ksr1 mRNA is expressed strongly in skeletal muscle but present at approximately one third that level in brain and 5% of that level in adipose tissue. ksr1-/- mice have been previously reported to have enlarged adipocytes but normal amounts of all adipose tissues (Kortum et al., 2005). Further analysis revealed no altered response to an AMPK agonist (Figure S3) and no overt metabolic defects (Figure S4 and Table S1) in ksr1-/- mice.
To test the role of KSR2 in the regulation of AMPK-mediated signaling, we targeted exon 4 within the ksr2 locus for deletion in mice (Figure S1). The ksr2-null allele was transmitted through the germ line, and heterozygous intercrosses yielded all three genotypes in a ratio close to the expected Mendelian distribution (relative ratios ksr2+/+ 1, ksr2+/- 2.25, ksr2-/- 0.91; n = 441). During development in utero and at birth ksr2-/- mice were identical in size and weight to wild type and ksr2+/- mice (Figure 3B, left panel). However, while nursing, ksr2-/- mice grew at approximately 50% the rate observed in wild type and ksr2+/- mice (Figure 3B, center panel). Thirty two percent of ksr2-/- mice (31 of 98) failed to survive until weaning. Premature death was not due to the failure of ksr2-/- pups to nurse properly as all mice had milk in their stomachs upon necropsy. The addition of foster mothers did not improve survival. Furthermore, nutrient absorption was identical in wild type and ksr2-/- mice (not shown). We measured the growth rate of surviving ksr2-/- mice and observed that the ksr2-/- mice attained body weights similar to wild type and ksr2+/- mice 6-10 weeks after birth (Figure 3C). At 20-24 weeks of age, ksr2-/- mice exceeded the body weight of their wild type and ksr2+/- littermates and became obese (Figure 3B, right panel). Interestingly, disruption of ksr2-/- caused a doubling in fat mass and a 15% decrease in lean mass even before body weight differences became apparent (Figure 3D). All adipose depots from ksr2-/- mice were significantly increased in mass relative to wild type mice (Figure 3E). Histological analysis demonstrated that the cross-sectional area of adipocytes from white adipose tissue in ksr2-/- mice was increased in size relative to wild type mice (Figure 3F).
We tested whether an interaction between endogenous KSR2 and AMPK could be detected similar to that observed when KSR2 was overexpressed in COS7 cells (Figure 1). Brain lysates from wild type and ksr2-/- mice were precipitated with antibodies against the AMPKα subunit and probed for KSR2 on western blot. KSR2 was detected in anti-AMPKα immunoprecipitates from WT brain but not detected in anti-AMPKα immunoprecipitates from ksr2-/- brain or from wild type brain immunoprecipitated with a non-immune antibody (Figure 3G).
To determine whether ksr2-/- mice have defects in AMPK signaling, we tested the ability of an intraperitoneal injection of the AMPK agonist AICAR to lower blood glucose. Wild type mice showed a rapid and sustained reduction in blood glucose in response to the AMPK agonist AICAR that was not altered by deletion of ksr1 (Figure S3). However, disruption of ksr2 in mice delayed the onset of the drop in blood glucose and reduced the extent of the decrease (Figure 3H). Given the effect of AICAR in ksr2-/- mice, the phosphorylation state of AMPK and ACC was examined in white adipose tissue. While wild type mice showed strong phosphorylation of AMPK and ACC following AICAR injection, disruption of ksr2 impaired their phosphorylation in white adipose tissue (Figure 3I). Furthermore, when explants of subcutaneous adipose tissue were isolated and incubated ex vivo with AICAR, wild type adipose tissue showed robust phosphorylation of AMPK on Thr172 and of ACC on Ser79. However, AMPK and ACC from ksr2-/- adipose tissue remained unphosphorylated and ACC expression was markedly reduced (Figure 3J). These data suggest that, despite its low level of expression (Figure 3A), KSR2 functions in a cell autonomous manner in adipose tissue.
We next tested the hypothesis that due to the role of KSR2 in controlling AMPK function, global ksr2 gene disruption would result in impaired energy metabolism. Large lipid vesicles were detected in the brown adipose tissue (BAT) of ksr2-/- mice that were absent in wild type mice (Figure 4A). BAT is the major site of adaptive thermogenesis in rodents and adaptive thermogenesis protects mammals during cold exposure and regulates energy balance to compensate for alterations in nutrient intake (Lowell and Spiegelman, 2000). We next assessed whether this lipid accumulation was reflected in reduced heat generation. The rectal temperature of wild type and ksr2-/- mice was compared during distinct diurnal periods. Deletion of KSR2 lowered rectal temperature by as much as 1.5°C (Figure 4B) and modestly decreased UCP1 mRNA expression (Figure 4C, left panel) and protein levels (Figure 4C, right panel).
Hyperphagia is considered a hallmark of numerous well-characterized rodent models of obesity (Bray and York, 1979). However, obese ksr2-/- mice actually consumed less food than wild type mice in spite of increased adiposity (Figure 4D). Consistent with the observed increase in fat mass, serum leptin levels in the ksr2-/- mice were elevated seven-fold in females and 12-fold in males (Figure 4E). The morbid obesity and hyperleptinemia suggested massive leptin resistance. However, injection of leptin suppressed food consumption to the same degree in both wild type and ksr2-/- mice (Figure 4F). Key orexigenic hypothalamic neuropeptides agouti-related peptide (AgRP) and neuropeptide Y (NPY) (Ollmann et al., 1997; Schwartz et al., 1996), and anorexigenic neuropeptides proopiomelanocortin (POMC) and cocaine- and amphetamine-regulated transcript (CART) (Kristensen et al., 1998) were unchanged in ksr2-/- mice suggesting a novel compensatory mechanism independent from known major pathways of adipocyte-hypothalamus communication (Figure 4G).
Consistent with the ability of KSR2 to alter the rate of oxygen consumption in vitro (Figure 2D), young male ksr2-/- mice in total consume less oxygen than wild type mice and produce less CO2 (Figure 5A). In females, a similar trend is observable. These data support the conclusion that oxidative phosphorylation is impaired by the disruption of ksr2. Respiratory quotient (RQ) is lower in ksr2-/- mice during the dark cycle relative to wild type mice (Figures (Figures5A5A and S5A), indicating a preference for fatty acids as metabolic substrate of choice or a relative resistance to carbohydrate metabolism during active periods. General, ambulatory and stationary motor activity is increased in ksr2-/- mice (Figure (Figure5B5B and S5B-D), despite their obesity. The consequence of such energy balance physiology responses to the disruption of ksr2 would result in a decreased amount of energy stored as fat. However, overall ksr2-/- mice expend substantially less energy (Figures 5C) and therefore accrue larger fat deposits than wild type mice. In order to exclude a potential artifact resulting from relative cold-exposure of ksr2-/- and wild type mice while being studied at room temperature, we re-analyzed their metabolic phenotype at thermoneutrality (32°C). ksr2-/- mice exhibited significantly lower oxygen consumption under those conditions (Fig. 5D), consistent with an ambient temperature-independent and physiologically relevant deficit in basal metabolic rate. Thus, despite compensatory feeding and nutrient partitioning responses to their increased adiposity, ksr2-/- mice become obese due to low resting thermogenesis and high energy efficiency.
Obesity is associated with elevated serum lipids and predisposes rodents and humans to impaired glucose homeostasis (Lazar, 2005). We find that the absence of KSR2 replicates both conditions since ksr2-/- mice show elevated lipolysis, free fatty acids, triglycerides (Figure 6A) and fasting insulin levels (Figure 6B). Adiponectin, resistin, PAI-1, IGF-1, and thyroxin were not significantly altered by disruption of ksr2. MCP-1 was significantly elevated in male but not female ksr2-/- mice (Figure S6). These data suggested that elevated endogenous insulin levels might be compensating for peripheral insulin resistance caused by the disruption of ksr2. We, therefore, performed hyperinsulinemic euglycemic clamps on wild type, ksr1-/-, and obese ksr2-/- mice. Glucose homeostasis in ksr1-/- mice mirrored that of wild type mice (Table S1). In contrast, ksr2-/- mice maintained comparable blood glucose concentrations only at glucose infusion rates that ranged between 20-50% of that observed in wild type mice (Figure 6C). Though hepatic glucose production (HGP) was reduced in ksr2-/- mice in comparison to wild type mice, the ability of insulin to suppress HGP in the null mice was markedly suppressed (Figure 6D). Similarly, whole body glucose turnover, glycolysis and glycogen production were reduced in knockout mice (Figure 6E). These data are consistent with markedly reduced glycogen content in the livers and glycolytic muscle of ksr2-/- mice (Figure 6F). To estimate insulin-stimulated glucose uptake in individual tissues, 2-deoxy-D-[1-14C] glucose was administered as a bolus (10 mCi) 75 min after the start of clamp. In comparison to wild type mice, ksr2-/- mice showed profound insulin resistance in skeletal muscle (gastrocnemius) and both white (epididymal) and brown adipose tissue (Figure 6G). In combination with the effect on HGP, these data reveal that disruption of ksr2 impairs glucose uptake at the major sites of insulin action.
To test whether the effect of KSR2 disruption had a cell autonomous effect on insulin-stimulated glucose uptake, we examined glucose uptake in isolated EDL muscles from wild type and ksr2-/- mice. Insulin significantly stimulated glucose uptake in the isolated muscle from either genotype to a similar degree (Figure 6H). In contrast, the AMPK agonist AICAR had a small but significant impact on glucose uptake in wild type EDL muscle but no effect on EDL muscle from ksr2-/- mice. These data suggest that KSR2 has a direct effect on AMPK-stimulated glucose uptake in EDL muscle. However, the insulin resistance observed in skeletal muscle of ksr2-/- mice appears to be a non-cell autonomous effect, perhaps secondary to the obesity and impaired lipid metabolism caused by KSR2 disruption (Bergman et al., 2006).
The observation that KSR2 expression regulated oxidative metabolism in vitro (Figure 2C), led us to examine gene expression in the white adipose tissue of ksr2-/- and wild type mice. Gene Set Enrichment Analysis (GSEA) (Subramanian et al., 2005) revealed that gene sets previously identified as down regulated in obese mouse models and in some humans with abnormal glucose tolerance are regulated by ksr2 (Figure S7A, Supplemental Table S2). Strongest among these was a gene set previously shown to be down regulated in ob/ob mice (Nadler et al., 2000). A notable number of sets that included genes for oxidative metabolism were also down regulated in ksr2-/- mice. A subset of oxidative phosphorylation genes co-regulated across different mouse tissues (OXPHOS-CR) was identified previously (Mootha et al., 2003). This gene set corresponds to two thirds of the genes encoding the oxidative phosphorylation biochemical pathway. The OXPHOS-CR gene set is reported to be down regulated in skeletal muscle of some humans with impaired glucose tolerance and correlated with changes in total body metabolism. We examined the expression of a custom gene set corresponding to the mouse orthologs (mOXPHOS-CR) and determined that this set was significantly down regulated (p<.001, NES = -2.39, FDR<.001) in the white adipose tissue of ksr2-/- mice. 83% (20 of 24) of the genes demonstrated lower expression in ksr2-/- mice in comparison to wild type mice (Figure S7). The transcriptional co-regulator PGC1α is a potent regulator of OXPHOS-CR genes (Mootha et al., 2003). Microarray analysis revealed that PGC1α is also markedly decreased in the white adipose tissue of ksr2-/- mice (Figure S7). These data demonstrate that the AMPK regulator KSR2 plays a potent role in controlling the expression of PGC1α-dependent gene programs whose alteration may contribute to impaired glucose tolerance in KSR2-/- mice.
In this study, we show that the molecular scaffold KSR2 is an essential regulator of AMPK activity controlling cellular thermogenesis, fat oxidation, and glucose metabolism. These data suggest a model (Figure 7) whereby a decrease in AMPK function impairs fatty acid oxidation and increases lipid storage, contributing to obesity and insulin resistance in ksr2-/- mice. In wild type mice, KSR2 interacts with AMPK to negatively regulate ACC and promote the expression of PGC1α-dependent OXPHOS genes. Inhibition of ACC prevents the synthesis of the CPT1 allosteric negative regulator malonyl CoA. Together with OXPHOS gene expression, elevated CPT1 activity enhances the oxidation of fatty acids and reduces their storage as triglycerides (TG). In the absence of KSR2 AMPK function is impaired, OXPHOS gene expression declines, ACC activity is deregulated causing decreased long chain fatty acid uptake into mitochondria via CPT1 and enhanced triglyceride synthesis, which promotes obesity. These findings reveal a novel pathway regulating energy expenditure and glucose metabolism, the elucidation of which may facilitate therapeutic intervention in obesity.
It remains unclear to what extent specific tissues contribute to the metabolic defects caused by KSR2 disruption. Ex vivo experiments suggest a role for KSR2 in adipose tissue and muscle. However, KSR2 is detectable by western blot only in the brain, and mRNA profiling detects much lower levels of KSR2 in muscle, liver, and adipose tissue. These data suggest that important actions of KSR2 on AMPK are probably exerted within the central nervous system.
Resting metabolic rate is the major contributor to obligatory energy expenditure and is strongly associated with fat-free mass (Cunningham, 1991). However, fat-free mass accounts for only 70% to 80% of the variability in resting metabolic rate (Sparti et al., 1997). Genetically determined differences in the ability of organisms to consume oxygen to make ATP have been proposed as an explanation for at least some of the remaining variability (Harper et al., 2008). The reduced oxygen consumption and energy efficiency of ksr2-/- mice identify ksr2 as a novel genetic determinant of metabolic rate. To some extent, decreased energy expenditure in ksr2-/- mice could be ascribed to the decrease in UCP1 expression, which might contribute to the 1.5°C drop in rectal temperature of ksr2-/- mice. The reduced rectal temperature of ksr2-/- mice therefore suggests that ksr2 is essential for the physiological regulation of resting thermogenesis.
Through its phosphorylation of ACC, AMPK plays a critical role in determining whether fatty acids are oxidized in mitochondria to generate ATP or stored as triglycerides. Our in vivo and ex vivo data demonstrate that deletion of KSR2 in white adipose tissue impairs the ability of AMPK to phosphorylate ACC, which should inhibit fatty acid oxidation and promote triglyceride storage (Ruderman et al., 2003). This defect provides a simple explanation for the increased fat mass of ksr2-/- mice (Brommage et al., 2008). Accordingly, ampkα2-/- mice share traits with ksr2-/- mice. On a high fat diet, ampkα2-/- mice demonstrate elevated body weight, morbidly increased adipose mass and adipocyte hypertrophy without an increase in food intake relative to control mice (Villena et al., 2004). Similar to ksr2-/- mice, deletion of the AMPK α2 subunit causes insulin resistance and AICAR intolerance (Viollet et al., 2003). Also consistent with our findings, loss of AMPK signaling that results from hepatic disruption of the AMPK-kinase, LKB1, causes hyperglycemia and glucose intolerance, while not impairing insulin action (Shaw et al., 2005). Thus, KSR2 may affect cell autonomous energy homeostasis in multiple tissues by directly modulating lipid and glucose metabolism via AMPK, while behavioral phenotypes of KSR-deficient mice such as hypophagia and hyperactivity are likely compensatory responses for body setpoint defense.
We performed microarray analysis on mRNA isolated from wild type and ksr2-/- adipose tissue to identify genetic pathways affected by the deletion of KSR2. GSEA of microarray data identified curated gene sets that are significantly regulated by the presence or absence of KSR2 (Figure S7 and Table S2). Grouping the highest ranking (p = 0.000; FDR ≤ 0.001; FWER ≤ 0.045) gene sets according to overlapping genes identifies three general cellular functions regulated by KSR2. In white adipose tissue, KSR2 appears to most potently affect genes regulating adipocyte differentiation, genes involved in oxidative phosphorylation, and genes affecting the metabolism of branched-chain amino acids and short chain lipids. KSR2-dependent regulation of genes controlling adipocyte differentiation (e.g., Nadler Obesity Down, IDX TSA Up Cluster 6, and TNFα Down gene sets) plays a role in the increased adipocyte cell number observed in ksr2-/- mice. Decreased expression of genes involved in oxidative phosphorylation (e.g., Electron Transport, Mootha VoxPhos, and Mitochondria gene sets) likely contributes to the decreased ability of ksr2-/- mice to metabolize lipid and carbohydrate. Gene sets for branched chain amino acid catabolism, and propionate, butanoate and pyruvate metabolism represent key pathways that generate key substrates (e.g., acetyl-CoA, succinyl-CoA) to the TCA cycle and fatty acid synthesis. The regulation of these pathways by KSR2 suggests a role for the scaffold in controlling both the availability of these key substrates as well as the enzymes necessary for their metabolism. However, future experiments will be required to determine whether these changes in gene expression are mediated by the action of KSR2 in adipose tissue or in another tissue with prominent expression, like the brain.
KSR1 and KSR2 are best known for their function as scaffolds for the Raf/MEK/ERK signaling cassette, facilitating the activation of Raf and MEK (Dougherty et al., 2009; Kortum and Lewis, 2004; Nguyen et al., 2002). The interaction of AMPK with these molecular scaffolds raises the intriguing possibility that KSR proteins function not only as scaffolds, but also as components of an energy and nutrient sensor that couples information about the nutritional environment and intracellular energy status of a cell to a kinase cascade with potent effects on cell proliferation, differentiation, and survival. The discovery that C-TAK1/MARK3/Par1a, a member of the AMPK kinase family, phosphorylates KSR1 and inhibits its ability to promote the activation of MEK by Raf (Muller et al., 2001), supports this concept. In complex with KSR2, AMPK might provide similar means to restrict energy-intensive proliferative stimuli emanating from activated ERK when ATP is limited. ksr2-/- mice might thus develop impaired metabolic homeostasis due to an inability to respond appropriately to energy deficits and curb ERK signaling when the nutritional environment is inadequate.
Consistent with that paradigm, our results demonstrate that disruption of ksr2 causes obesity through a reduction in cellular energy consumption despite hypophagia. Therefore, our observations reveal ksr2-/- mice to be a novel model of obesity with potential relevance to obesity-related dysregulation of glucose metabolism. That molecular scaffolds regulating the activation of Raf, MEK, and ERK can have a profound effect on fat accumulation suggests that factors affecting energy balance may have previously unappreciated roles on MAP kinase signaling. In particular, future studies focusing on cell type-specific mechanisms of KSR2/AMPK interactions may provide important insight into novel mechanisms regulating physiological control of energy storage and expenditure with implications for glucose homeostasis.
ksr1-/- mice were described previously (Kortum et al., 2005; Nguyen et al., 2002). Standard gene-targeting techniques and homologous recombination were used to generate ksr2-/- mutant mice. The Institutional Animal Care and Use Committee (University of Nebraska Medical Center, Omaha, NE) approved all studies. Animals were maintained on a 12-hour light/dark schedule (light on at 0600) and had free access to laboratory chow (Harlan Teklad LM 485) and water. All in vivo analyses were performed on mice of 3-7 months of age.
KSR1-/- and KSR2-/- MEFs were generated from day 13.5 embryos and immortalized by 3T9 protocol as described (Kortum and Lewis, 2004) or by expression of Sv40 Large T antigen. Expression of KSR1, KSR2 and corresponding mutants in MEFs was also measured as described previously (Kortum and Lewis, 2004). NG108-15 cells, COS-7 and 293T cells were obtained from ATCC.
Immunoprecipitation were performed on post-nuclear membranes with antibodies to the FLAG, Pyo, and Myc epitope tags as described previously (Kortum and Lewis, 2004; Ritt et al., 2007). Antibodies for AMPKα, phospho-Thr172 AMPK, ACC, and phospho-Ser79 ACC, UCP1, and GAPDH were from Cell Signaling. Anti-KSR2 antibody 1G4 was from Abnova. Anti-α-tubulin antibodies were from Santa Cruz.
Glucose uptake was measured with 2-deoxy-D[2,6-3H]glucose in the presence or absence of 20 μM cytochalasin B and 200 μM phloretin as described previously (Chaika et al., 1999). ERK phosphorylation was quantified on the Odyssey system (LI-COR) with anti-phospho-ERK1/2 (Cell Signaling No. 9106) and anti-ERK1 (Santa Cruz Biotechnology, sc-93) primary antibodies and goat anti-mouse Alexa Fluor 680 (Invitrogen) and goat anti-rabbit IRDye 800 (Rockland) as secondary antibodies as described (Kortum and Lewis, 2004).
Fatty acid oxidation (FAO) and glycolysis was determined by measuring oxygen consumption rate (OCR) and extracellular acidification rate, respectively, in cultured cells using the XF24 Analyzer (Seahorse Bioscience) (Wu et al., 2007).
A short hairpin targeting the nucleotides for amino acids 868-874 of mouse KSR2 was cloned into the lentiviral MISSION® pLKO.1-puro vector. Puromycin resistant cells were selected using 2 μg/ml puromycin (Sigma).
Total RNA was isolated from selected mouse tissues using Tri-Reagent (Molecular Research Center, Inc). ksr1, ksr2, GusB, and Tbp were simultaneously quantified from 5 μg of total RNA using QuantiGene 2.0 Plex gene sets (Panomics), following the manufacturer’s recommendations. Total RNA was hybridized to custom probe set 21121 for 24 hours at 54 °C while shaking at 900 RPM with an orbital diameter of 3 mm before signal amplification and quantification using a Luminex 200 instrument. The median fluorescent intensity of at least 50 beads was used to determine the average gene expression for tissue samples from three mice performed in duplicate.
RNA was extracted from brown adipose tissue (BAT) and hypothalamus using TRizol reagent (Invitrogen), according to the manufacturers instructions. After subsequent DNase treatment, reverse transcriptions were performed using SuperScript III (Invitrogen) and Oligo-dT20 primers (Invitrogen). Real-time PCR for UCP1 in BAT and neuropeptides in hypothalamus, and the ribosomal house keeping gene L32, were performed on a Biorad iCycler using iQ SybrGreen Supermix (Biorad). Relative quantification of the target transcript in comparison to a reference transcript was calculated from the real-time PCR efficiencies and the crossing point deviation of the target sample versus its control (Pfaffl, 2001).
Body composition was measured using Nuclear Magnetic Resonance technology (NMR, EchoMRI, Quantitative Magnetic Resonance Body Composition Analyzer, Echo Medical Systems, LLC, Houston, TX, USA). Total adipose tissue from each depot was excised and the wet weight was determined. Abdominal, subcutaneous, and brown adipose tissue was fixed in Bouin’s fixative, sectioned in a microtome and stained with hematoxylin and eosin. Adipocyte cross-sectional area was determined from photomicrographs of epididymal fat pads using IPLab software (Scanalytics Inc., Fairfax, VA)(Kortum et al., 2005).
Food intake was measured daily manually over five consecutive days in freely feeding mice. Total energy expenditure, locomotor activity, and respiratory quotient (relative rates of carbohydrate versus fat oxidation) of mice were determined by indirect calorimetry using a customized 32-cage indirect calorimetry system (TSE Systems Midland, MI). The mice were placed in the calorimetry system cages for up to 6 days and nights, with at least 24h for adaptation before data recording. For oxygen consumption measurements at thermoneutral conditions, the environmental temperature requiring the least energy for organismal heating or cooling processes was determined using indirect calorimetry within a climate control system. Thermoneutral zone for ksr2-/- and wild type mice was determined. Oxygen consumption analysis was then performed to ascertain that the thermogenetic phenotype was not an artifact resulting from relative cold exposure of ksr2-/- and wild type mice.
Blood glucose was measured with an Ascensia Glucometer Elite (Fisher Scientific). Plasma insulin was measured with the Mouse Insulin Elisa Kit (ChrystalChem, Chicago, IL) using mouse standards. Serum free fatty acids were measured colorimetrically (Roche). Plasma triglycerides and glycerol were measured using the GPO-Trinder colorimetric assay kit (Sigma). Plasma leptin was measured using the Rat Leptin RIA kit (Linco Research, St Louis, MO). Glycogen content was analyzed with the Glucose HK assay (Sigma). For measurement of lipolysis, mice were fasted overnight for 12 h. Subcutaneous fat was excised, minced in Krebs-Ringers bicarbonate buffer at 37 °C for 3 h, and a sample of media was assayed for glycerol content using Free Glycerol Reagent (Sigma).
Subcutaneous fat pads were removed from 7-month old wild type and ksr2-/- mice, prepared and treated as described (Gaidhu et al., 2009).
Glucose uptake was measured in isolated EDL muscle from wild type and ksr2-/- mice using 2-deoxy-D[2,6-3H]glucose and [14C]mannitol as described previously (Sakamoto et al., 2005).
Data are expressed as mean ± sem. Differences between two groups were assessed using the unpaired two-tailed t-test and among more than two groups by analysis of variance (ANOVA).
We thank the members of the Lewis laboratory for comments and criticism. M. Birnbaum (U. Pennsylvania) is thanked for his gift of the constitutively active AMPK construct. This work was supported by NIH grants to R.E.L (DK52809) and M.H.T. (DK69987, DK59630 and DK56863) and by support from the Nebraska Research Initiative to R.E.L. M.R.F was supported by the Skala Fellowship and M.B. was supported by the Bukey Fellowship from UNMC. K.F. was supported by a Physician/Scientist Training award from the American Diabetes Association. The UNMC Microarray Core Facility receives partial support from NIH grant P20 RR016469. For those studies conducted at USARIEM, the opinions or assertions contained herein are the private views of the author(s) and are not to be construed as official or reflecting the views of the Army or the Department of Defense. Any citations of commercial organizations and trade names in this report do not constitute an official Department of the Army endorsement of approval of the products or services of these organizations.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.