The Krüppel-like factor KLF15 regulates the insulin-sensitive glucose transporter GLUT4.

Resistance to the stimulatory effects of insulin on glucose utilization is a key feature of type 2 diabetes, obesity, and the metabolic syndrome. Recent studies suggest that insulin resistance is primarily caused by a defect in glucose transport. GLUT4 is the main insulin-responsive glucose transporter and is expressed predominantly in muscle and adipose tissues. Whereas GLUT4 has been shown to play a critical role in maintaining systemic glucose homeostasis, the mechanisms regulating its expression are incompletely understood. We have cloned the murine homologue of KLF15, a member of the Krüppel-like family of transcription factors. KLF15 is highly expressed in adipocytes and myocytes in vivo and is induced when 3T3-L1 preadipocytes are differentiated into adipocytes. Overexpression of KLF15 in adipose and muscle cell lines potently induces GLUT4 expression. This effect is specific to KLF15 as overexpression of two other Krüppel-like factors, KLF2/LKLF and KLF4/GKLF, did not induce GLUT4 expression. Both basal (3.3-fold, p < 0.001) and insulin-stimulated (2.4-fold, p < 0.00001) glucose uptake are increased in KLF15-overexpressing adipocytes. In co-transfection assays, KLF15 and MEF2A, a known activator of GLUT4, synergistically activates the GLUT4 promoter. Promoter deletion and mutational analyses provide evidence that this activity requires an intact KLF15-binding site proximal to the MEF2A site. Finally, co-immunoprecipitation assays show that KLF15 specifically interacts with MEF2A. These studies indicate that KLF15 is an important regulator of GLUT4 in both adipose and muscle tissues.

Glucose uptake into cells is regulated by two families of cellular transporters, the sodium-linked glucose transporters (kidney, intestine) and the facilitated glucose transporters (GLUTs). With respect to the latter, GLUT4 is the main effector of insulin-stimulated glucose transport and is located primarily in muscle and adipose tissues (1).
Clinical and experimental observations suggest that insulinstimulated glucose transport via GLUT4 is critical in maintaining systemic glucose homeostasis. For example, heterozygous mice deficient in GLUT4 exhibit muscle insulin resistance and develop diabetes (2). Tissue-specific disruption of GLUT4 in adipose tissue and skeletal muscle results in the development of insulin resistance and glucose intolerance (3,4). In human type 2 diabetic patients, impairment of insulin-stimulated glucose transport is responsible for resistance to insulin-stimulated glycogen synthesis in muscle (5)(6)(7).
Studies in adipose and muscle tissues reveal that expression of the GLUT4 glucose transporter is controlled at the level of transcription (8,9). In vitro and in vivo promoter studies support a role for members of the MADS-box family of transcription factors termed MEF2 proteins in the regulation of the GLUT4 promoter. However, these studies suggest that MEF2 binding alone is not sufficient to fully support GLUT4 expression (10 -12).
The Krü ppel-like family of transcription factors are important regulators of cellular development, differentiation, and activation. The KLFs are a subfamily of Cys 2 /His 2 zinc finger DNAbinding proteins related to the Drosophila melanogaster segmentation gene product, Krü ppel. Previous studies demonstrate a critical role for this family in erythropoiesis (KLF1/EKLF) (13,14), T cell activation (KLF2/LKLF), vascular development (KLF2/LKLF) (15,16), lung development (17,18), and skin development (KLF4/GKLF/EZF) (19). We identified a novel member of this family termed KLF15 and provide evidence that it regulates GLUT4 expression in both adipose and muscle cells.

MATERIALS AND METHODS
Reagents-The heart cDNA library was obtained from Stratagene. Reagents for 3T3-L1 differentiation, 3-isobutyl-1-methylxanthine, dexamethasone, and insulin as well as nonradioactive 2-deoxy-D-glucose and cytochalasin B for glucose uptake assays were purchased from Sigma. The 3T3-L1, A10, NIH-3T3, and C2C12 cells were obtained from American Type Culture Collection. Neonatal rats and pups were obtained from Taconic Farms. Mice for aortic banding experiments were obtained from Charles River Laboratories. ES cells were obtained from Incyte. MEF2A expression construct was kindly provided by E. Olson (University of Texas, Southwestern).
Cloning of KLF15-The C-terminal zinc finger region of KLF1/EKLF was used as a probe to screen a mouse embryonic cDNA library under low-stringency conditions. A partial clone of KLF15 was obtained and used to screen a mouse heart cDNA library to obtain the full-length clone. Both strands of the full-length clone were sequenced by the dideoxy chain termination method.
Cell Culture-3T3-L1 differentiation was performed as previously described (20). Briefly, 3T3-L1 cells were grown in 10% FCS 1 /DMEM and media was changed to 10% FCS/DMEM supplemented with 3-isobutyl-1-methylxanthine, dexamethasone, and insulin (differentiation medium) at 2 days post-confluence ("day 0"). After 48 h, media was changed to 10% FCS/DMEM supplemented with 1 ⁄4 the concentration of insulin used at day 0. After 48 additional hours, cells were maintained in 10% FCS/DMEM. C2C12 myoblasts were cultured in 10% FBS/ DMEM. Differentiation was induced post-confluence using 2% horse serum/DMEM that was changed daily. A10 cells were cultured in 20% FBS/DMEM; NIH-3T3 fibroblasts were cultured with differentiation medium to day 11 as described for 3T3-L1 cells. For retroviral studies, the indicated cDNA was cloned into the retroviral vector green fluorescent protein-RV (gift K. Murphy) and retrovirus were generated as described (21). For infection of target cells, retroviral supernatant and culture medium (10% FCS/DMEM ϩ 4 g/ml Polybrene) were mixed at a 1:1 ratio and added to preconfluent cells. Within 24 -48 h nearly 100% infectivity was noted by assessment for green fluorescent protein. 2-Deoxy-D-glucose uptake assays in 3T3-L1 cells were performed as described previously (22).
Electrophoretic Mobility Shift Assays-The zinc finger region of KLF15 bearing an N-terminal FLAG tag was generated by PCR and cloned into the expression vector pCDNA3 (Invitrogen). Gel-shift studies were performed as previously described (23).
Cardiomyocyte Isolation and Hypertrophy Studies-Primary neonatal rat ventricular cardiomyocytes were isolated as previously described (24). Cardiomyocyte quiescence was induced in DMEM supplemented with insulin, transferrin, and selenium. Cardiac hypertrophy was induced by aortic banding of the proximal aorta as previously described (25). Ventricles were harvested at 3 weeks after banding.
Antibody Generation, Immunohistochemistry, and ␤-Galactosidase Staining-An anti-peptide antibody to the C-terminal 14 amino acids was generated by Zymed Laboratories. Formalin-fixed and paraffinembedded tissues were stained using a 1:250 dilution of the primary antibody with horseradish peroxidase-linked secondary antibody. For ␤-galactosidase staining, tissues were removed and fixed in 1.25% PBS/ glutaraldehyde at room temperature for 10 min, rinsed several times in PBS, and stained overnight at 37°C in the dark. Tissues were rinsed in PBS and dehydrated in ethanol before sectioning.
Generation of KLF15 ϩ/Ϫ LacZ Mice-The murine KLF15 genomic clone was obtained by hybridization of a mouse 129SVJ library with an EcoRI-BamHI cDNA fragment from the mouse KLF15 coding region. The targeting vector was generated by inserting a 3-kb SmaI-EcoRI genomic fragment from the first intron into the BamHI (blunt) and EcoRI sites of pPNT (26) followed by simultaneous insertion between pPNT NotI and XhoI sites of a 3.6-kb HindIII-XhoI fragment containing a nuclear lacZ gene (27) and a 3.5-kb genomic fragment extending from the genomic NotI site to an engineered HindIII site at the KLF15 ATG. The resulting targeting construct was linearized with NotI before electroporation into 129SVJ ES cells. Neo-resistant transfectants were selected by growth in G418 (200 g/ml) and ganciclovir (1 M). DNA from ES cell clones was digested with BamHI, Southern blotted, and hybridized with a 1.5-kb genomic fragment located 3Ј of the knockout construct. ES cells from two independently derived KLF15 ϩ/Ϫ clones were microinjected into C57BL/6 donor blastocysts that were implanted into pseudopregnant females. The resulting male chimeras were mated with C57BL/6 females and the agouti offspring were genotyped using Southern analysis.
Co-immunoprecipitation Studies-293T cells were transfected with the indicated expression vectors. 48 h post-transfection the cells were rinsed with PBS and harvested in 300 l of RIPA buffer per 10-cm dish. Cellular debris was pelletted and the lysates were subjected to immunoprecipitation with either 4 g of ␣-FLAG M2 monoclonal antibody or 4 g of IgG 1 control antibody (Sigma) for 2 h at 4°C after which protein A/G-Sepharose was added (O/N, 4°C). The beads were washed in RIPA buffer followed by PBS and proteins were separated by SDS-PAGE.

RESULTS
Identification and Characterization of KLF15-A partial cDNA fragment of KLF15 was identified through a low stringency homology screen of a mouse embryonic cDNA library using the zinc finger region of EKLF as a probe. The full-length cDNA was subsequently obtained by screening of a heart cDNA library. Analysis of the open reading frame revealed that KLF15 is a 415-amino acid protein with three Cys 2 /His 2 zinc fingers at the C terminus. The N terminus was notable for a discrete glutamic acid-rich region (Fig. 1A, underlined). A putative nuclear localization signal was present at amino acid 369 (-RHRR-). A GenBank TM search revealed that our factor is the mouse homolog of a recently identified rat cDNA termed KKLF (28). Based upon the recommendation of the Mouse and Human Nomenclature Committee we refer to this factor as KLF15.
Northern analysis of mouse tissues showed KLF15 expression levels to be the highest in adipose, muscle tissues (heart, skeletal muscle, and aorta), kidney, and liver ( Fig. 1B and data not shown). To define the cellular expression of the KLF15 protein, an anti-peptide antibody was generated to the C-terminal 14 amino acids (Zymed Laboratories). This antibody was able to detect a single band using a KLF15 in vitro transcribed . Immunohistochemical studies on white adipose tissue were performed using a 1:250 dilution of the primary antibody with horseradish peroxidase-linked secondary antibody. Counterstaining was performed with eosin. Magnification ϫ1000. Arrows indicate nuclear staining. KLF15lacZ ϩ/Ϫ mice were generated as described under ''Materials and Methods.'' 5-Bromo-4-chloro-3-indolyl-␤-D-galactopyranoside (X-gal) staining reveals expression in adipocytes, skeletal myocytes, cardiomyocytes, and smooth muscle cells in a small blood vessel. Counterstaining was performed using eosin. Magnification was ϫ1000 for all tissues except blood vessel (ϫ400). and translated product (data not shown). Immunohistochemical studies in adipose tissue reveal a punctate pattern of nuclear staining in adipocytes from white fat (Fig. 1C, left panel). To further assess its expression pattern and function we initiated efforts to generate mice deficient in KLF15 by homologous recombination. The knockout strategy included insertion of a nuclear lacZ gene at the KLF15 ATG. Assessment of ␤-galactosidase activity in KLF15 LacZϩ/Ϫ heterozygous mice demonstrated staining in adipocytes, cardiomyocytes, skeletal myocytes, and smooth muscle cells (Fig. 1C, right panels).
KLF15 Induces GLUT4 Expression and Glucose Uptake in 3T3-L1 Cells-KLF15 was highly expressed in adipose tissue. Given the availability of cell lines that faithfully recapitulate adipogenesis in vitro, we assessed the expression of KLF15 during 3T3-L1 differentiation. Stimulation of these cells with an empirically derived prodifferentiation mixture leads to a characteristic pattern of induction for various transcription factors and target genes involved in adipogenesis and lipogenesis (29). As shown in Fig. 2A, KLF15 mRNA was first detected by Northern analysis at day 3 of differentiation subsequent to the induction of peroxisome proliferator-activated receptor (PPAR) ␥, a critical regulator of adipogenesis. Comparison of KLF15 expression to a number of other genes induced during 3T3-L1 differentiation demonstrates an expression pattern most similar to that of GLUT4. To further investigate the function of KLF15, we retrovirally infected 3T3-L1 cells with either fulllength KLF15 (KLF15), empty vector (EV), or deletion mutants lacking either the N-terminal 56 or 200 amino acids of KLF15 (KLF15⌬56 and KLF15⌬200) and then differentiated the cells for 4 days. In comparison to EV-infected cells, we noted a marked induction of GLUT4 mRNA in KLF15 overexpressing cells. No effect on GLUT4 expression was seen in the KLF15⌬56 mutant (data not shown). In contrast, KLF15⌬200 cells exhibited a reduction in GLUT4 mRNA consistent with a dominant negative effect (Fig. 2B). A modest effect was also noted on fatty acid synthase and CCAAT/enhancer-binding protein family (C/EBP␣) expression along with minimal effects on the expression of PPAR␥ and fatty acid-binding protein.
To determine whether the induction of GLUT4 mRNA was a specific result of KLF15 overexpression, we tested the ability of two additional members of the Krü ppel-like family, KLF2/ LKLF and KLF4/GKLF, to affect GLUT4 levels. 3T3-L1 cells were retrovirally infected with EV, KLF2, KLF4, or KLF15, differentiated to day 4, and then harvested for total RNA. Northern blot analysis using a GLUT4 probe demonstrated that only KLF15 was able to induce GLUT4 mRNA above the level found in EV-infected cells. In contrast, both KLF2 and KLF4 reduced GLUT4 expression (Fig. 2C).

FIG. 2. KLF15 expression in 3T3-L1 cells and effects on GLUT4 expression and glucose uptake.
A, KLF15 mRNA is induced during 3T3-L1 differentiation. 3T3-L1 cells were induced to differentiate using hormonal agents as described under ''Materials and Methods.'' Cells were harvested at the indicated number of days post-induction, and total RNA was isolated and subjected to Northern analysis using the indicated probes. B, overexpression of KLF15 induces GLUT4. 3T3-L1 cells were retrovirally infected with either full-length cDNA (KLF15), EV, or a mutant lacking the N-terminal 200 amino acids (KLF15⌬200) and differentiated to day 4. Total RNA was isolated and subjected to Northern analysis using the indicated probes. C, induction of GLUT4 is specific to KLF15. 3T3-L1 cells were retrovirally infected with KLF4, KLF2, EV, or KLF15 and differentiated to day 4. Northern analysis was performed using a GLUT4 probe. D, KLF15 induces glucose uptake. 3T3L1 cells were infected with KLF15⌬200, EV, or KLF15 retrovirus and differentiated to day 6. The cells were cultured without serum for 5 h before addition of 100 nM insulin (shaded bars, ϩI) for 15 min. The cells were washed and glucose transport assays were performed as described under ''Materials and Methods.'' Results are shown after subtracting counts taken in the presence of the transport inhibitor cytochalasin B and dividing by the total protein concentrations. Results are normalized to the EV value in the absence of insulin (n ϭ 4). * ϭ p Ͻ 0.007; ** ϭ p Ͻ 0.001, compared with the (Ϫ)insulin EV value); † ϭ p Ͻ 0.008; † † ϭ p Ͻ 0.00001, compared with the (ϩ)insulin EV value).
To assess the functional consequence of KLF15-mediated GLUT4 induction, we performed [ 3 H]2-deoxyglucose uptake assays in 3T3-L1 in the presence and absence of insulin. As shown in Fig. 2D, compared with EV-infected cells, KLF15 overexpressing cells exhibited a 3.3-fold increase in basal glucose uptake (p Ͻ 0.001), whereas KLF15⌬200 expressing cells showed a 25% reduction in glucose uptake (p Ͻ 0.007). Following insulin stimulation, we observed a 2.5-fold increase in glucose uptake in KLF15 overexpressing cells compared with EV-infected cells (p Ͻ 0.00001), whereas KLF15⌬200 expressing cells showed a 40% reduction in glucose uptake (p Ͻ 0.008). The fact that KLF15⌬200-, EV-, and KLF15-infected cells all showed a ϳ3-fold increase in the amount of glucose taken up in response to insulin (Fig. 2D, compare the shaded bar to white bar for each construct) suggests that KLF15 does not sensitize cells to insulin. However, by augmenting the level of GLUT4 in 3T3-L1 cells, KLF15 overexpression does result in an increase in both basal and insulin-stimulated glucose uptake.
KLF15 Induces GLUT4 in Muscle Cell Lines-In addition to adipose tissue, GLUT4 expression is also seen in all muscle cells, both striated (skeletal and cardiac muscle) and smooth muscle (blood vessel). To assess the role of KLF15 in skeletal muscle cells, we used C2C12 cells as a model system. Interestingly, it is recognized that virtually all muscle cell lines are deficient in GLUT4 expression (12). Indeed, we were unable to detect either GLUT4 or KLF15 mRNA by Northern analyses using 20 g of total RNA (data not shown). To assess whether the deficiency of KLF15 in C2C12 may account for the absence of GLUT4 in this cell line, we retrovirally overexpressed KLF15 in C2C12 cells. As shown in Fig. 3A, overexpression of KLF15 in C2C12 cells robustly induced GLUT4 mRNA. This effect was specific as another glucose transporter, GLUT1, was not induced. In addition, KLF15 overexpression induced GLUT4 mRNA in NIH-3T3 fibroblasts and A10 cells (a smooth muscle cell line; Fig. 3B).
Glucose is an important source of energy for the heart at rest. The heart becomes increasingly dependent on glucose under certain physiologic and pathologic states such as exercise, hyperthyroidism, and hypertrophy (30). GLUT4 expression in the rodent heart is induced during the first few weeks after birth and is accompanied by a concomitant decrease in GLUT1 expression (9,31). We compared the expression pattern of KLF15 and GLUT4 in the postnatal mouse heart as well as after the induction of cardiac hypertrophy. Total RNA from mouse ventricles was harvested at various time points after birth and subjected to Northern analysis (Fig. 3C). We observed very low levels of KLF15 mRNA at postnatal days 3 and 10 that increased by day 15 and reached near adult levels by day 20. This pattern of expression is similar to that of GLUT4 and correlates inversely with cyclin A, a marker of cellular growth known to decrease during the postnatal period (32). Isolation of neonatal cardiomyocytes and culture under quiescent conditions confirmed both KLF15 and GLUT4 expression in cardiomyocytes (Fig. 3D). Finally, previous studies show that GLUT4 levels are reduced following the induction of cardiac hypertrophy (33). To assess KLF15 expression under these conditions, we used aortic banding to induce pressure-overload hypertrophy in 8-weekold C57BL/6 mice and harvested the total ventricle RNA 3 weeks after banding. In contrast to sham operated animals, the banded animals showed a reduction in the levels of both KLF15 and GLUT4 mRNA (Fig. 3E). Taken together these data sup-

FIG. 3. Role of KLF15 in muscle cells. A-E show Northern analyses with the probes indicated at the left of each panel.
A, KLF15 induces GLUT4 mRNA in C2C12 cells. C2C12 cells were infected with EV or KLF15 (KLF) retrovirus and maintained in high serum (10% FCS) until confluent. The cells were then either harvested (myoblast) or switched to differentiation medium (2% horse serum) and cultured for an additional 5 days (myotube). B, KLF15 induces GLUT4 in smooth muscle and fibroblast cell lines. Preconfluent NIH-3T3 fibroblasts and A10 smooth muscle cells were infected with EV or KLF15 retrovirus. The fibroblasts were treated with hormonal inducing agents and harvested 11 days after induction. The A10 cells were harvested 72 h after infection. C, KLF15 mRNA is induced in the postnatal mouse heart. Total RNA was harvested from ventricles of C57BL/6 mice on the indicated postnatal day and subjected to Northern analysis. Ribosomal hybridization with 28 S is shown to indicate loading. D, KLF15 is expressed in neonatal cardiomyocytes. Cardiomyocytes were harvested from 2-day-old rat neonatal pups and cultured under growing (G; 10% FCS) or quiescent conditions (Q; 0% FCS ϩ insulin, transferrin, and selenium). E, KLF15 mRNA levels are reduced with cardiac hypertrophy. Eight-week-old mice were subjected to aortic banding or sham operation. Ventricles were harvested 3 weeks after banding and total was RNA isolated. Each number represents an individual animal. port a role for KLF15-mediated regulation of GLUT4 in muscle tissues.
KLF15 Binds DNA and Transactivates the GLUT4 Promoter-Members of the Krü ppel-like family bind to specific DNA elements (5Ј-CNCCC-3Ј) to exact their function. To assess the ability of KLF15 to bind DNA we performed electrophoretic mobility shift assays using FLAG-tagged KLF15 and a probe containing the consensus sequence (5Ј-CACCC-3Ј). As shown in Fig. 4A, a single DNA-protein complex was seen. This complex was specific as it can be competed by an identical but not a nonspecific (5Ј-CATGTG-3Ј) or mutated (5Ј-CACCG-3Ј) oligomer. Finally this complex can be supershifted with an anti-FLAG antibody but not with an unrelated antibody (IgG). Thus, KLF15 was able to bind the KLF consensus sequence.
To define the mechanism(s) underlying the ability of KLF15 to induce GLUT4 expression, we performed transient transfection studies using deletion constructs of the GLUT4 promoter. Previous studies in transgenic mice have demonstrated that the proximal 895 bp of the GLUT4 promoter was sufficient to recapitulate endogenous expression and that a proximal regulatory region (Ϫ412 Ͼ Ϫ526) is essential. Within this region lies a consensus binding site for myocyte enhancer factors (MEFs). Mutation of the MEF site results in loss of GLUT4 promoter activity in transgenic animals (Fig. 4B) (10,11,34).
Analysis of the 895-bp GLUT4 promoter revealed several potential KLF15-binding sites (shaded boxes; Fig. 4B). One of these sites (Ϫ499 3 Ϫ503) lies in close proximity to the MEFbinding element (Ϫ454 3 Ϫ464). We observed an ϳ8 -9-fold (p Ͻ 0.0001) induction of both the 2.2-kb and 813-bp GLUT4 promoter constructs by KLF15 (Fig. 4C). Approximately twothirds of this induction was eliminated with the Ϫ500-bp GLUT4 promoter (ϳ3.5-fold induction; p Ͻ 0.001); the remainder of the activity was completely eliminated with the Ϫ149-bp promoter fragment (ϳ1.5-fold induction; p Ͻ 0.1). To determine which potential KLF-binding site between base pairs Ϫ500 3 Ϫ813 was responsible for the loss of transactivation, we introduced a point mutation at the Ϫ499-bp position (CACCC3 CACCG) because this site was closest to the MEF element and is contained within the proximal regulatory region. This single base pair mutation in the Ϫ813-bp promoter resulted in a loss of transactivation similar to that of the Ϫ500-bp promoter (Fig. 4C). These data suggest that KLF15 can transactivate the GLUT4 promoter by binding near the MEF2 consensus site.
KLF15 and MEF2A Synergistically Transactivate the GLUT4 Promoter-The proximity of the KLF15-and MEFbinding sites raised the possibility that these two factors may function in a coordinated manner to induce the GLUT4 promoter. To assess this possibility, we performed co-transfection studies. As shown in Fig. 5A, we observed that the combination of KLF15 and MEF2A resulted in a synergistic activation of the 813-bp GLUT4 promoter (ϳ13-fold; p Ͻ 0.0001 by comparison to KLF15 alone). This synergistic effect was not seen in the 500-bp GLUT4 promoter that lacks the KLF15-binding site (Fig. 5A, right panel). These studies suggest that KLF15 can induce the GLUT4 promoter and this may occur through a coordinated effort with MEF2A.
These data also raised the possibility that KLF15 and MEF2A may interact directly to transactivate the GLUT4 promoter. To investigate this hypothesis, we cotransfected 293T cells with expression vectors for MEF2A and full-length FLAG-tagged KLF15, FLAG-KLF4, or empty vector (pCDNA3). An ␣-FLAG antibody immunoprecipitated MEF2A protein from the lysate of cells transfected with FlagKLF15 and MEF2A, but not with empty vector ϩ MEF2A or FlagKLF4 ϩ MEF2A (Fig. 5B, left panel). This indicates that KLF15 directly interacts with MEF2A. The specificity of this interaction was further demonstrated by the fact that an IgG 1 isotype control antibody was unable to immunoprecipitate MEF2A (Fig. 5B, right panel). FIG. 4. KLF15 transactivates the GLUT4 promoter. A, KLF15 binds the CACCC element. In vitro translated, FLAG-tagged KLF15 was incubated with a 32 P-labeled oligomer containing the wild type CACCC-binding site or plus one of the following cold competitor probes: nonspecific, CACCC mutant, or increasing amounts of wild type CACCC. A single dominant DNA-protein complex was seen. Competition and supershift (*) studies verified that the dominant complex was specific for KLF15-CACCC binding. Electrophoretic mobility shift assays were performed as described under ''Materials and Methods.'' B, schematic diagram of the GLUT4 promoter. Potential KLF-binding sites are represented as shaded rectangles. C, KLF15 induces the GLUT4 promoter. NIH-3T3 cells were transfected with the indicated GLUT4 promoter and either the EV or KLF15 expression construct. The molar ratio of reporter to expression plasmid was 1:2.5. Luciferase and ␤-galactosidase assays were performed. Results are expressed as -fold induction compared with vector alone (n ϭ 6 -9 per group). * ϭ p Ͻ 0.0001 compared with the respective empty vector; † ϭ p Ͻ 0.001 compared with the empty vector value on the 500-bp promoter; ** ϭ p Ͻ 0.001 compared with the KLF15 vector value on the 813-bp wild type promoter.

DISCUSSION
A number of disease states such as diabetes and obesity are characterized by glucose intolerance and insulin resistance. Clinical and experimental observations suggest that a defect in glucose transport contributes to these conditions (5). Whereas decreased GLUT4 expression per se is not the cause of insulin resistance in these diseases, previous studies show that enhanced GLUT4 levels can augment insulin responsiveness and glucose tolerance (1,(35)(36)(37)(38). Indeed, glucose tolerance and insulin responsiveness are increased by overproduction of GLUT4 in muscle and/or adipose tissue in both normal and diabetic mice (35)(36)(37)(38)(39). These findings suggest that a better understanding of the mechanisms regulating the insulin-sensitive glucose transporter GLUT4 may ultimately lead to novel strategies for the treatment of various insulin-resistance states. We provide in this report evidence for KLF15 as an important regulator of GLUT4 gene expression.
Insulin-sensitive glucose transport is a late event in the process of differentiation that characterizes a mature fat cell (40). Current models of the transcriptional basis for adipocyte differentiation highlight an interplay between members of two major families, the C/EBP and PPAR families (41). Studies to date indicate that C/EBP␤ and C/EBP␦ induce PPAR␥, an essential regulator of adipogenesis and insulin-sensitive glucose uptake. The uptake of glucose in response to insulin re-quires expression of components of the insulin signaling pathway as well as GLUT4. PPAR␥ induces C/EBP␣, which augments the expression and phosphorylation of the insulin receptor and insulin receptor substrate-1 (42). PPAR␥ is also essential for GLUT4 expression (22,43). It has been hypothesized that an additional factor downstream of PPAR␥ is also involved in the induction of GLUT4 (22).
C/EBP␣ has also been implicated in the regulation of GLUT4 expression. For example, El-Jack and colleagues (40) found that reconstitution of insulin-sensitive glucose transport in fibroblasts requires expression of both PPAR␥ and C/EBP␣. These authors found that in NIH-3T3 cells, the expression of C/EBP␣ was required for GLUT4 expression (40). In contrast, several lines of evidence suggest that C/EBP␣ expression is not requisite for GLUT4 expression. For example, C/EBP␣ null mice do not exhibit any reduction in GLUT4 expression (44). More recently, Wu and colleagues (43) found that reconstitution of C/EBP␣ null cells with PPAR␥ resulted in GLUT4 expression. Taken together, these data suggest that the role of C/EBP␣ remains controversial. Our data suggest that KLF15 may serve as a downstream effector as it is expressed subsequent to PPAR␥ and is able to induce GLUT4. These data raise the possibility that during the process of adipogenic differentiation both C/EBP␣ and KLF15 may function in a coordinated manner to confer full insulin responsiveness to the adipocyte. C/EBP␣ induces components of the insulin signaling pathway, whereas KLF15 induces expression of GLUT4 (42,43).
In addition to adipose, GLUT4 is highly expressed in muscle tissues where it plays critical roles in glucose homeostasis and tissue function. For example, loss of GLUT4 expression in skeletal muscle leads to the development of hyperglycemia and insulin resistance (3). In the heart, GLUT4 deficiency leads to death of the animal secondary to the development of a cardiomyopathy (30). Finally, GLUT4 is expressed in smooth muscle cells where it may be involved in the regulation of smooth muscle cell contraction (45,46). Given these important functions for GLUT4 in muscle, we assessed the expression of KLF15 in these tissues. We observed KLF15 expression in all muscle cells in vivo (Fig. 1C). KLF15 overexpression induced GLUT4 expression in both skeletal and smooth muscle cell lines (C2C12 and A10). Finally, KLF15 expression in cardiomyocytes paralleled that of GLUT4 during the postnatal period and in a disease model of left ventricular hypertrophy. It is noteworthy that the effects of KLF15 bear similarity to those observed for the transcriptional coactivator PGC-1 in muscle cell lines. Like KLF15, PGC-1 induces GLUT4 expression in C2C12 cells, augments basal glucose uptake, and functions in a cooperative manner with MEFs (12). Whether a direct relationship exists between PGC-1 and KLF15 is currently under investigation.
GLUT4 mRNA and protein expression are subject to complex regulation by a number of hormonal/metabolic influences and physiologic states. A major form of regulation involves the translocation of GLUT4 protein from the interior of cells to the plasma membrane in response to insulin stimulation (reviewed in Ref. 1). However, GLUT4 mRNA levels are also regulated in conditions such as experimental diabetes and fasting (47,48). In addition, Santalucia and co-workers (9) recently demonstrated that perinatal expression of cardiac GLUT4 is controlled directly at the level of gene transcription. Thus, an understanding of the mechanisms governing GLUT4 gene expression has been of considerable interest. A series of elegant promoter deletion analyses using transgenic approaches demonstrate that the proximal 895 bp of the GLUT4 promoter contains the necessary elements to recapitulate endogenous expression of GLUT4 (10). These studies identify a proximal regulatory region that contains a critical MEF-binding element (10,34). This MEF site is necessary but not sufficient to support expression of the GLUT4 gene, suggesting that other factors likely participate (10). We found that KLF15 is able to strongly induce the GLUT4 promoter and that the majority of this activity was mediated by a binding site that lies in proximity to the MEF site. The functional importance of this was verified by cotransfection studies that revealed a synergistic activation of the GLUT4 promoter by KLF15 and MEF2A. Finally, a direct interaction between KLF15 and MEF2A is supported by co-immunoprecipitation studies. Thus, our data add to the current understanding of the transcriptional control of GLUT4 expression and support a coordinated effort by KLF15 and MEF2A. The generation of KLF15 null mice or KLF15 and MEF2A double knockout mice will provide further insights into the role of these factors in GLUT4 regulation. These studies are currently in progress.