Identification and characterization of basal and cyclic AMP response elements in the promoter of the rat hexokinase II gene.

Hexokinases catalyze the phosphorylation of glucose and initiate cellular glucose metabolism. Hexokinase II (HKII) is the principal hexokinase isoform in skeletal muscle, heart, and adipose tissue. Isoproterenol and exogenous cyclic AMP (cAMP) increase HKII gene transcription in L6 myotubes. Various segments of the HKII promoter that direct the expression of the chloramphenicol acetyltransferase reporter gene were transfected into L6 myotubes to identify basal and cAMP response elements. The 5′-flanking region that extends 90 base pairs upstream of the transcription start site includes a CCAAT box and a cAMP response element (CRE); both contribute to basal promoter activity and each provides an independent, maximal response to cAMP. An inverted CCAAT motif, or Y box, located just upstream of the CCAAT box, contributes to basal promoter activity but is not involved in the cAMP response. Homo- and heterodimers composed of the CRE-binding protein and activating transcription factor-1 bind specifically to the CRE. The Y box and the CCAAT box specifically bind the factor NF-Y (also known as CBF).

The four mammalian hexokinases (ATP:D-hexose-6-phosphotransferase, EC 2.7.1.1; designated HKI to HKIV) 1 convert glucose to glucose 6-phosphate (for review see Ref. 1). The hexokinases thus initiate glucose metabolism, and this maintains the downhill glucose concentration gradient that results in glucose entry into cells through facilitative glucose transporters (2,3). HKII is the principal hexokinase isoform in insulin-responsive tissues such as skeletal muscle, heart, and adipose tissue (4,5). The cloning of HKII cDNA permitted deduction of the primary amino acid sequence of the protein (6,7), provided insight into the complete structure of the HKII mRNA and gene (6), and led to studies of the regulation of the gene (6,8). The rate of HKII gene transcription is increased by insulin, catecholamines, and cyclic AMP (cAMP) in L6 myotubes (6,8), and this induction results in increased HKII mRNA, protein synthesis, and glucose phosphorylation. This is interesting in view of observations regarding the regulation of hexokinases in various physiologic circumstances. For example, skeletal muscle glucose utilization is increased by exercise (9), an effect that can be explained in part by the observation that a single bout of acute exercise increases HKII gene transcription, mRNA, and protein in rat skeletal muscle (10,71). Other physiologic stimuli, such as cold stress, swimming, or the administration of ␤-adrenergic agonists, increase total hexokinase activity in rat and mouse brown adipose tissue and muscle (11)(12)(13)(14). The mechanism of these effects is unknown but could involve cAMP.
The regulation of gene transcription by cAMP has been analyzed quite extensively (15)(16)(17)(18). Hormones such as ␤-adrenergic agonists that couple cell surface receptors to adenylate cyclase result in increased intracellular levels of cAMP. Cyclic AMP, in turn, activates protein kinase A, which enters the nucleus and phosphorylates cyclic AMP response elementbinding protein (CREB) on Ser 133 (19,20). Phosphorylated CREB, bound to the cyclic AMP response element (CRE), enhances transcription through linked promoters (21,22). Activating transcription factor-1 (ATF-1), a protein closely related to CREB, also binds CREs and mediates cAMP responsiveness in certain genes (23,24). A CCAAT box and an inverted CCAAT box mediate cAMP responsiveness in several promoters that lack typical CREs, but the factors that bind to these elements were not identified (25)(26)(27).
We initiated an analysis of the HKII promoter in L6 cells, a cell line representative of rat skeletal muscle, to further understand how hormones influence HKII gene expression. The basal promoter consists of about 160 bp of 5Ј-flanking sequence that includes an inverted CCAAT box (henceforth referred to as the Y box), a CCAAT box, and the CRE. The CCAAT box and the CRE are both involved in cAMP responsiveness and are functionally redundant in this regard. The Y box does not play a role in cAMP responsiveness but does contribute to basal promoter activity. Several known transcription factors bind to these sequences, notably CREB and ATF-1 to the CRE and NF-Y to both the Y and the CCAAT boxes.

MATERIALS AND METHODS
The Rat HKII Promoter Sequence-The EcoRI-EcoRI fragment from gHK51 was inserted into pBluescript II SK Ϫ and named pgHK51.1 (6); this contains ϳ5.5 kb of 5Ј-flanking sequence of the rat HKII gene. The pgHK51.1 plasmid was used as the DNA template for dideoxy sequencing reactions performed using Sequenase 2.0 TM (U.S. Biochemical Corp.) and synthetic oligonucleotide primers.
Plasmid Constructs-The BamHI fragment of pgHK51.1, which contains the rat HKII promoter between Ϫ5.5 kb and ϩ147 bp (numbers relative to the transcription start site), was inserted into the BamHI  (28). This construct was named pHKIICAT(An)-1A, or more simply, 1A. A series of 5Ј deletion mutants were generated from the original 1A construct by restriction enzyme digestion at unique sites in the upstream polylinker and at the desired 5Ј termini of the HKII insert, and the construct was then religated. Incompatible termini were made blunt-ended with Klenow fragment before religation. Constructs that contain ϳ2600 (HindIII), 1176 (SalI), 487 (PstI), 157 (SacII), 78 (BssH II), and 41 (Nae I) bp of 5Ј-flanking sequence were named 2A, 3A, 4A, 6A, 7A, and 5A, respectively. Additional 5Ј deletion mutants were generated by the polymerase chain reaction (PCR) using upstream primers containing a HindIII site fused to the designed 5Ј termini. The upstream primers used were 5Ј-CCAAGCTTGCAGCCAATGAGC-GCGCCCA-3Ј for the 9A construct and 5Ј-CCAAGCTTTTGGGCGGC-CCAAAGAGCCG-3Ј for 10A; the downstream primer for both constructs was 5Ј-ACTTCGCCCACGGGATCCGTAAG-3Ј. The PCR reactions contained 10 ng of pgHK51.1 as a template, 0.2 mM of each dNTP, 1 M of each primer, and 1 unit of Taq polymerase (Perkin-Elmer) in 100 l of a solution that consisted of 50 mM KCl, 1.5 mM MgCl 2 , 0.01% gelatin, and 10 mM Tris-HCl (pH 8.3). The PCR conditions consisted of denaturation at 94°C for 4 min followed by 40 cycles of denaturation at 94°C for 1 min, annealing at 66°C for 2 min, and extension at 72°C for 1 min followed by a final extension period of 5 min at 72°C. Each PCR product was digested with HindIII and BamHI before insertion into HindIII/BglII-digested pCAT(An). These constructs contained 90 and 60 bp of 5Ј-flanking sequence and were named 9A and 10A, respectively. Mutations of 9A were made by introduction of synthetic oligonucleotides into unique HindIII, BssH II, or NaeI restriction sites in the 9A parent construct. The Y box, CCAAT box, and CRE mutations selected had previously been shown to be effective in eliminating protein binding to these elements or had been shown to reduce promoter activity (26,29,30). These mutants of 9A were named 9A CCAATmt-1, 9A CREmt-2, and 9A CCAATmt-1/CREmt-2, respectively ( Table I). The other 9A mutants contain an inverted CCAAT box region (between Ϫ90 and Ϫ80), or a 5-bp (5Ј-TGCAG-3Ј), or a 10-bp (5Ј-TGCAGTGCAG-3Ј) insertion between Ϫ79 and Ϫ78. These were named 9A-inv, 9Aϩ5, and 9Aϩ10, respectively. The 6A parent construct was employed for site-directed mutagenesis using PCR. The upstream primer used was 5Ј-CCAAGCTTGGGCTCTGGGCGCTGATGTGCTG-TGGACTGCGGGCG-3Ј, and the downstream primer was the same as that used for the 5Ј deletion mutants. The PCR reactions were performed using Vent DNA polymerase and a commercial buffer (New England Biolabs, Inc.; Beverly, MA) as described above except that 35 cycles were used, consisting of denaturation at 94°C for 30 s, annealing at 59°C for 30 s, and extension at 72°C for 30 s. The product was cut with HindIII and BssH II and was used to replace the HindIII and BssH II region of the 9A construct. This construct was named 6A Yboxmt-1. The identity of each construct was confirmed by restriction digestion and dideoxy DNA sequencing. All constructs were purified by CsCl gradient centrifugation before transfection, as described (31).
Tissue Culture-L6 cells purchased from the American Type Culture Collection (ATCC; Rockville, MD) were grown in 10-cm plates to near confluence and were differentiated as described previously (6). Following differentiation, myotubes were maintained in Dulbecco's modified Eagle's medium containing 2% horse serum for 18 h before transfection or nuclear extract preparation. Cells were uniformly transfected or harvested for nuclear extract preparation within a low passage number of receipt from the ATCC.
Transfection and CAT Assay-Transient transfection was performed using cationic poly-L-ornithine coupled with dimethyl sulfoxide shock (32,33). Monolayers of L6 myotubes were incubated in 3 ml of Dulbecco's modified Eagle's medium containing 2% horse serum, 10 -15 g of plasmid DNA, and 30 g of poly-L-ornithine⅐HBr for 6 h at 37°C before a 5-min 25% dimethyl sulfoxide shock at room temperature. Plates were subsequently maintained in serum-free Dulbecco's modified Eagle's medium until they were harvested 40 h later. Cell extract preparation and assays for CAT activity were performed as described by Nordeen et al. (34). CAT activity in the extracts was normalized for protein content, as measured using Bio-Rad reagent. Co-transfection of cells with the expression plasmid pRSVCat␣ (a generous gift of Dr. R. A. Maurer, Oregon Health Sciences University), which encodes the cAMP-dependent protein kinase ␣-catalytic subunit (PKA) (35), was used to functionally map the CRE because this method provided more consistent results than did cAMP treatment, which sometimes affected cell viability upon prolonged incubation (Ͼ24 h). Plasmids used were the pHKIICAT(An) series (10 g) as a reporter and either 5 g of pRSVCat␣ as the effector or pRSVNeo (36) as a control for DNA amount when no effector was used. The plasmid pRSV-luciferase (2.5 g) was also transfected as an internal control to ensure that changes in promoter activity were not due to differences in transfection efficiency (data not shown). Luciferase activity was assayed in cell extracts as described (37).
Electrophoretic Mobility Shift Assay (EMSA)-L6 cell nuclear extracts were prepared as described previously (38) with the following modifications: phenylmethylsulfonyl fluoride (1 mM) and benzamidine (1 mM) were added to the hypotonic buffer, the nuclear resuspension buffer, and the dialysis buffer. The ammonium sulfate precipitation was omitted, and the nuclear extract was dialyzed twice for 90 min at 4°C. Cytoplasmic fractions were prepared as described (39). The synthetic oligonucleotides used as probes or competitors for EMSAs are listed in Table I. Oligonucleotides that contain the rat somatostatin CRE (40), a high affinity CCAAT/enhancer-binding protein (C/EBP) binding site (41), and the murine E␣ Y box (42) have been employed previously. Mutated sequences were identical to those employed in transfection experiments. Oligonucleotide pairs were gel-isolated and annealed prior to labeling with [␥-32 P]dATP by T4 polynucleotide kinase. Binding conditions were as described (43) with the following modifications: each probe was incubated with 10 g of L6 cell nuclear extract at room temperature for 15 min except for the CRE probes, which were incubated for 30 min at 4°C. Poly(dI-dC)⅐(dI-dC) (50 ng/l), Nonidet P-40 (0.5%), and a 100-fold molar excess of the single-stranded sense oligonucleotide of each probe were added to reduce nonspecific binding. Competition experiments were performed in the presence of a 100-fold molar excess of unlabeled oligonucleotide pairs and titration experiments were performed to determine the affinity of each DNA element for its binding protein(s). Polyclonal anti-CREB antibody (␣244) was the generous gift of Dr. M. Montminy (Salk Institute; La Jolla, CA) (44); the monoclonal anti-NF-YA (monoclonal antibody YA7) and polyclonal anti-NF-YB antibodies were the generous gifts of Dr. R. Mantovani (University of Milan) (45), and the monoclonal anti-ATF-1 antibody (sc-243 X) was purchased from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). Antibodies (anti-CREB, 1:400; anti-NF-Y, 1:100; or anti-ATF-1, 1:20 dilution) when used were added to each EMSA as indicated, and incubation conditions were as described for the binding assays. Reaction samples were electrophoretically separated on 6% nondenaturing polyacrylamide gels at room temperature or at 4°C when the CRE probes were employed. The relevant bands were visualized by autoradiography and analyzed by scanning densitometry.
Statistical Analysis-All data are presented as the mean Ϯ S.E. Statistical comparisons were made by single-factor factorial analysis of variance using STATVIEW II TM software for Apple Macintosh computers (Abacus Concepts; Berkeley, CA).

Effect of 5Ј Deletion Mutations on the Basal Promoter Activity
and PKA Inducibility-We cloned about 5.5 kb of 5Ј-flanking sequence of the rat HKII gene, sequenced the proximal 1176 bp (6), and constructed a series of 5Ј deletion mutants that consist of the HKII promoter fused to the CAT reporter gene. These constructs were transiently transfected into differentiated L6 cells to localize functional cis-acting DNA elements responsible for mediating basal promoter activity and cAMP responsiveness. The PKA expression vector was co-transfected with the HKII promoter constructs when cAMP responsiveness was evaluated.
The basal promoter activities of all constructs that contained at least 90 bp of 5Ј-flanking sequence were significantly greater than those of their shorter counterparts (Fig. 1). Construct 9A, with 90 bp of flanking sequence, was thus taken as the standard and assigned a relative basal activity of 100%. Construct 6A, which contains the additional sequence between Ϫ157 and Ϫ90, showed enhanced basal activity compared with 9A (190 Ϯ 10%; p Ͻ 0.05). The largest drop-off of basal activity occurred when the region between Ϫ90 and Ϫ78 was deleted (100% for 9A versus 27 Ϯ 1% for 7A, p Ͻ 0.05), and another drop-off of basal activity occurred when the region between Ϫ78 and Ϫ60 was deleted (27 Ϯ 1% for 7A versus 14 Ϯ 4% for 10A; p Ͻ 0.05). The low promoter activity observed for 10A (Ϫ60 bp) was also found for the shorter construct 5A (Ϫ41 bp).
Regions involved in cAMP (PKA) responsiveness were sought using the same 5Ј deletion constructs (Fig. 1). A 2-4-fold increase of CAT expression in response to PKA (PKA inducibility is described as the ratio of PKA-treated to basal value) was noted in constructs that contained at least 78 bp of 5Ј-flanking sequence; the lower total PKA response of 7A as compared with 9A is due to a reduction of the basal promoter strength in the former construct. Complete loss of PKA inducibility occurred when the sequence upstream from Ϫ60 was removed (0.8 Ϯ 0.1-fold induction). We conclude from these findings that the sequence within Ϫ157 is important for basal promoter activity and that the sequence within Ϫ78 is required for cAMP (PKA) responsiveness.
Analysis of the Rat HKII Promoter Sequence-Examination of the sequence between nucleotides Ϫ157 and Ϫ60 relative to the transcription initiation site, a region important for basal promoter activity and cAMP (PKA) responsiveness, allowed us to identify several putative eukaryotic transcription factor binding motifs including a Y box, a CCAAT box, and a CRE ( Fig. 2A) (16,46). A 14-bp Y box containing sequence Ϫ148 to Ϫ135 is identical at 13 positions with the reverse complement of the region identified as a CCAAT box, which is located between nucleotides Ϫ89 and Ϫ76 (see Fig. 2, A and B). Interestingly, this CCAAT box is closer to the Y box consensus sequence than it is to a C/EBP consensus sequence. In this sense, the HKII CCAAT box may be more properly termed a "proximal inverted Y box," but to avoid confusion with nomenclature, we use the term "CCAAT box" for this element. The putative CRE, located between nucleotides Ϫ73 and Ϫ66, shares a 6 of 8 bp identity with a consensus CRE core sequence (Fig. 2, A and B). These three elements are all located upstream of the putative TATA box, which is located at nucleotides Ϫ30 and Ϫ26 (Fig. 2A).
Effect of Mutations of the Y Box, CCAAT Box, and CRE on Basal Promoter Activity and PKA Inducibility-Site-directed mutagenesis was employed to determine if the putative Y box, CCAAT box, and CRE are involved in basal and cAMP-induced expression of the rat HKII gene (Fig. 3). The 6A construct showed enhanced basal promoter activity when compared with FIG. 2. Analysis of the rat HKII gene promoter sequence. A, the sequence of the HKII promoter between Ϫ157 and ϩ147 is shown to illustrate the location of a Y box, a CCAAT box, a CRE, and a TATA box. Arrows indicate distal and proximal CCAAT motifs. The distal, inverted CCAAT sequence is defined as the Y box, and the proximal CCAAT sequence is defined as the CCAAT box. The 5Ј boundaries of the various promoter segments used are shown as 6A, 9A, etc. B, comparison of the rat HKII Y box, CCAAT box, or CRE to Y box, C/EBP, or CRE consensus sequences (16,46). Vertical lines indicate identical bases.
FIG. 1. Effect of 5 deletion mutations on basal activity and PKA inducibility of the rat HKII gene promoter. Each HKIICAT plasmid construct (10 g) was transiently transfected into L6 myotubes with 5 g of pRSVCat␣ (PKA-treated) or pRSVNeo (basal), and the CAT activity of each plate of cells was measured as described under "Materials and Methods." In order to make meaningful comparisons between experiments performed at different times, CAT activity was normalized to the basal activity of the 9A construct for each experiment. The average basal CAT activity of 9A was 23,500 Ϯ 2,300 dpm/mg⅐h (n ϭ 12). The basal and PKA-treated CAT activity of each 5Ј deletion construct is shown as mean ϩ S.E. (as the percentage of 9A basal), determined in at least three independent experiments with three replicate plates for each construct. The PKA inducibility (PKA-treated/basal) of each construct is shown as the -fold induction Ϯ S.E. 9A (190 Ϯ 10%, p Ͻ 0.05, as described above and illustrated in Fig. 3), and the ratio of PKA-treated to basal was lower for 6A than 9A (1.7 Ϯ 0.2 versus 3.5 Ϯ 0.4-fold induction; p Ͻ 0.05; Fig.  3) because of this. The Y box, located between the 5Ј end points of the 6A and 9A constructs, was mutated to generate the 6AYboxmt-1 construct. The basal activity of this construct, like that of 9A, is lower than that of 6A. Accordingly, since the total PKA responses are similar, the PKA-treated/basal ratio of both is higher than that of 6A. The effect of this mutation was identical to that seen when the region between Ϫ157 and Ϫ90 was deleted, and it effectively converted the promoter activity of 6A to that observed for 9A.
The putative CCAAT box was mutated within 9A to generate 9A CCAATmt-1 (Fig. 3). This mutation resulted in a reduction of basal activity identical to that seen when the region between Ϫ90 and Ϫ78 was deleted (compare with construct 7A, Fig. 1). No change in PKA inducibility (3.2 Ϯ 0.7-fold) was detected because PKA-treated CAT activity was proportionally decreased along with the reduction of the basal activity (Fig. 3). Mutation of the putative CRE within 9A to generate 9A CREmt-2 resulted in a reduction of basal and PKA-treated CAT activity but no loss of PKA inducibility (3.0 Ϯ 0.5-fold). The promoter activity of 9A CREmt-2, which contains an intact CCAAT box, was markedly different from that of the 10A (Ϫ60 bp) deletion construct. The 10A construct lacks both the CRE and CCAAT box, has little basal promoter activity with respect to 9A, and also shows no response to PKA. A double mutation of both the CCAAT box and the CRE within 9A (9A CCAATmt-1/CREmt-2, see Fig. 3) resulted in a reduction of both basal and PKA-treated activity equivalent to that measured in 10A (Ϫ60 bp) and caused complete loss of PKA inducibility (1.0 Ϯ 0.1fold). Thus, the CCAAT box and CRE both play a role in cAMP (PKA) responsiveness, and they are functionally redundant in this regard. In contrast, the Y box, CCAAT box and CRE contribute independently to basal promoter activity.
Effect of Orientation or Spacing of the CCAAT Box with Respect to the CRE on Basal Promoter Activity and PKA Inducibility-The CCAAT box region (Ϫ90 through Ϫ80) was inverted to determine whether the orientation of this element contributed to the basal or PKA response. The inversion of the CCAAT box enhanced basal activity compared with 9A (190 Ϯ 10%, p Ͻ 0.05) but did not affect PKA inducibility (Fig. 4). The spacing between the CCAAT box and CRE is important for the PKA response. Insertion of 5 bp between the CCAAT box and the CRE resulted in a significant reduction in both basal activity (50 Ϯ 10%, p Ͻ 0.05) and PKA inducibility (1.9 Ϯ 0.3-fold, p Ͻ 0.05) with respect to 9A (Fig. 4). A 10-bp insertion had no effect on the PKA inducibility, but this did enhance the basal activity of the promoter (170 Ϯ 10%, p Ͻ 0.05). These observations suggest that the orientation or spacing of the CCAAT box with respect to the CRE affects basal activity, whereas spacing alone is critical for PKA induction.
Protein Binding to the HKII CRE-Using a probe that contains the rat HKII CRE (see Table I) and L6 cell nuclear extracts, three specific DNA-protein complexes (1, 2, and 3) were identified with EMSA (see Fig. 5A, lane a). All three FIG. 4. Effect of orientation or spacing of the CCAAT box with respect to the CRE on basal activity and PKA inducibility of the rat HKII gene promoter. Each HKIICAT plasmid construct (10 g) was transiently transfected into L6 myotubes with 5 g of pRSVCat␣ (PKA-treated) or pRSVNeo (basal), and the CAT activity of each plate of cells was measured as described under "Materials and Methods." The data are expressed as described in the legend to Fig. 1. The basal and PKA-treated CAT activity of each construct is shown as mean ϩ S.E. (%), determined in at least five independent experiments using two replicate plates for each construct. The PKA inducibility (PKA-treated/ basal) of each construct is shown as the -fold induction Ϯ S.E. The construct 9A-inv has an inverted CCAAT box. Constructs 9Aϩ5 and 9Aϩ10 have 5-and 10-bp insertions, respectively, between the CCAAT box and the CRE. complexes were nearly eliminated when the HKII CRE (100fold molar excess) was used to compete against itself for protein binding (see Fig. 5A, lane b) or when the somatostatin CRE was employed (Fig. 5A, lane h). These complexes were not observed when an L6 cell cytoplasmic extract was used (data not shown). Other weaker complexes were occasionally observed, but these probably do not reflect the binding of specific complexes to the rat HKII CRE based on the following criteria: 1) the complexes were observed with both nuclear and cytoplasmic extracts; 2) the formation of these complexes was not specifically competed for by rat HKII CRE; and 3) formation of the complexes was reduced by the inclusion of single-stranded DNA in the incubation. The three CRE-specific complexes were not observed when a probe containing the same mutation (9A CREmt-2, Table I) used to determine the functional role of the rat HKII CRE was employed in EMSA (data not shown); nor did this mutant CRE compete with the wild type CRE probe for protein binding (Fig. 5A, lane c). Thus, a CRE mutation that results in the loss of promoter activity also results in a loss of specific protein binding to this DNA element. The HKII CCAAT and Y boxes, the MHC class II E␣Y box, and the C/EBP sequence did not compete for the formation of these three complexes (Fig. 5A,  lanes d-g, i, and j).
Anti-CREB antibody disrupted formation of complexes 1 and 2 (Fig. 5B, lane ␣CREB), and anti-ATF-1 antibody disrupted formation of complexes 2 and 3 (Fig. 5B, lane ␣ATF-1). Nonimmune serum had no effect on DNA-protein complex formation (Fig. 5B, control). These results suggest that the HKII CRE binds CREB and ATF-1 and that these two factors ac-count for the three specific complexes observed. We provisionally identify complex 1 as a CREB homodimer, complex 2 as a CREB⅐ATF-1 heterodimer, and complex 3 as an ATF-1 homodimer.
Protein Binding to the HKII CCAAT and Y Boxes-The HKII CCAAT box, when inverted, is nearly identical to the HKII Y box (see Fig. 2B), which suggests that a common factor may bind to both elements. HKII CCAAT and Y box probes were used with L6 cell nuclear extracts to identify specific DNAprotein complexes by EMSA. A single specific complex was observed with both probes (Fig. 6, compare panels A and B, lane  a). This complex was not observed when cytoplasmic extracts prepared from L6 cells were used in the EMSA (data not shown). Other weaker complexes were occasionally observed, but these complexes do not appear to be specific for the HKII CCAAT and Y box elements (see above for criteria used to determine nonspecific binding). Protein binding was nearly eliminated when a 100-molar excess of the HKII Y box was used to compete with itself or with the HKII CCAAT box probe (Fig. 6, A and B, lane f). Similarly, the HKII CCAAT box effectively competed with itself for protein binding but was a less effective competitor with the HKII Y box probe (Fig. 6, A  and B, lane d). CCAAT or Y box probes that contain mutations shown to reduce HKII promoter activity (see above and Table  I), formed no specific DNA-protein complexes (data not shown). The mutant HKII CCAAT and Y boxes were also unable to compete for DNA binding with either of the wild type probes (Fig. 6, A and B, lanes e and g); thus, the loss of promoter activity observed with these mutations correlates with loss of specific protein binding. Also, the ability of the Y box to compete for protein binding to the CCAAT box suggests that both elements bind the same protein.
The promoter for the murine MHC class II E␣ gene contains a well characterized Y box that binds the transcription factor NF-Y (47,48). However, a Y box is also an inverted CCAAT box; thus, one of the CCAAT box binding factors, such as C/EBP, CTF/NF1, or YB-1, may bind to the HKII Y box and/or CCAAT box. To determine which factor(s) binds to the rat HKII elements, we first showed that an E␣ Y box probe binds a factor from L6 cell nuclear extract and forms a complex similar to that observed for the HKII elements (Fig. 6C, lane a). A 100-fold molar excess of the E␣ Y box competed for factor binding to the HKII CCAAT and Y box probes and to itself (Fig. 6, A-C, lane j). A 100-fold molar excess of the HKII CCAAT or Y box also competes for factor binding to the E␣ Y box probe (see Fig. 6C,  lanes d and f, respectively), and the mutant HKII CCAAT and Y boxes failed to compete (Fig. 6C, lanes e and g). As shown in Fig. 6D, the HKII elements competed with the E␣ Y box for protein binding in a concentration-dependent fashion. The HKII and E␣ Y boxes compete about equally well, but the HKII CCAAT box is about 50% less effective. The single nucleotide difference between the Y box consensus sequence and an inverted HKII CCAAT box (Fig. 2B), may explain why the HKII CCAAT box is a weaker competitor for this common factor (49,50). A 100-fold excess of the C/EBP element, the HKII CRE, or the somatostatin CRE did not compete for specific protein binding to the HKII CCAAT box, HKII Y box, or the E␣Y box probes (Fig. 6, A-C, lanes b, h, and i).
The E␣ Y box binds NF-Y, a multisubunit DNA-binding protein (51). The similar DNA-protein complex formation observed between the E␣ Y box and the HKII Y and CCAAT boxes suggested that the HKII elements may also bind NF-Y. A monoclonal antibody raised against the A subunit and a polyclonal antibody raised against the B subunit were used to determine whether NF-Y forms the complex observed with the  Table I. Three specific bands (numbered 1-3) representative of protein binding to each probe are indicated by arrows. Panel B, effect of anti-CREB and anti-ATF-1 antibodies on protein complex formation with the rat HKII CRE. The rat HKII CRE 28wt (HKII CRE) probe was incubated with an L6 cell nuclear extract with or without the anti-CREB or ATF-1 antibodies, as described under "Materials and Methods." Specific bands representative of the proteins that bind to the probes are indicated by the numbered arrows (1)(2)(3). Each panel represents typical results obtained from three independent experiments. HKII CCAAT and Y boxes and L6 nuclear extracts. Anti-NF-YA antibody resulted in similar disruptions of complex formation with the HKII CCAAT or Y box probes and the E␣ Y box probe (Fig. 7, ␣NF-YA7 lanes). Anti-NF-YB antibody completely disrupted complex formation by the HKII CCAAT or Y box probes and the E␣ Y box probe (Fig. 7, ␣NF-YB lanes). Nonimmune serum had no effect on the complex formation by any of the probes (Fig. 7, control lanes). These results suggest that the HKII CCAAT box and Y box bind the transcription factor NF-Y and that this factor may be important for both the basal and cAMP-inducible expression of the HKII gene. DISCUSSION The promoter for the rat HKII gene was isolated and used to construct HKII-CAT fusion genes as a means of identifying cis-acting DNA elements important for HKII gene expression. The HKII-CAT fusion gene constructs were transiently expressed following transfection into differentiated L6 cells. Ichihara et al. (52) used a reporter gene similar to the 4A construct employed in this paper to direct the expression of a CAT fusion gene in the SR-3Y1-2 transformed fibroblast cell line. Mathupala et al. (53) recently reported that approximately 4.3 kb of the rat HKII promoter fused to a luciferase reporter gene directed basal expression in AS-30D hepatoma cells and in cultured hepatocytes. The basal promoter activity directed by this construct was comparable in magnitude in both cell types; however, glucose, insulin, or glucagon increased fusion gene expression only in the hepatoma cells (53).
We chose to study HKII gene expression in L6 cells, a cell line representative of skeletal muscle, which is a primary site of expression of the gene (4,5). We have previously used these cells to study the regulation of HKII gene transcription by insulin and cAMP (6,8) and now extend these earlier studies by identifying three cis-acting DNA elements in the HKII promoter, a Y box, a CCAAT box and a CRE, that are required for basal gene expression. Two of these elements, the CCAAT box and CRE, are required for the induction of the gene by cAMP and have redundant, nonadditive roles in mediating this response.
The Y box element, originally defined as an inverted CCAAT motif (CTGATTGGYY) in the major histocompatibility complex (MHC) class II genes, plays an important role in tissue-specific gene expression (for review, see Refs. 47 and 48). The binding protein NF-Y, also called CBF or CP-1 (42,54), stimulates gene transcription when bound to the Y box (55)(56)(57). NF-Y is highly conserved among species and is composed of several subunits, referred to as NF-YA (CBF-B), NF-YB (CBF-A), and CBF-C; all these subunits bind to DNA as a complex (51, 58 -60). Although the MHC class II genes are expressed in a cell-specific pattern, NF-Y is ubiquitously expressed (51, 61, 62), and it also binds to Y box motifs in non-class II genes including those that encode thymidine kinase (63) Table I. The specific complex representing protein binding to each probe is indicated by arrows. D, comparison of the affinity of the HKII CCAAT and Y boxes to the E␣ Y box for protein complex formation. The murine MHC class II E␣ Y box 28wt (E␣Y) probe was incubated with L6 cell nuclear extract and the rat HKII CCAAT 28wt (HKII CCAAT), rat HKII Y box 28wt (HKII Y), or E␣Y competitors as described under "Materials and Methods." The amount of each competitor was titrated from a 3-100-fold molar excess. The intensity of the bands was measured by densitometry. Protein binding, expressed as the percentage bound with competitors relative to that with no competitors, is shown as mean Ϯ S.E. as determined in three independent experiments.
(67), ␣-collagen (68), and interleukin 4 (57). The present study shows that NF-Y binds two highly homologous sequences, the Y box and the CCAAT box, in the rat HKII promoter. Specific mutations in either the Y box or the CCAAT box that reduce binding of NF-Y have corresponding effects on the transcriptional activity of reporter genes that contain these elements.
The HKII CRE binds both CREB and ATF-1, factors known to affect basal and cAMP responses in a variety of genes (23,24). Mutations of the HKII gene CRE that result in reduced binding of CREB and ATF-1 also result in decreased basal activity of the promoter. These mutations, however, had no effect on the cAMP response. Cyclic AMP induction is only lost when both the HKII CCAAT box and the CRE are mutated. Since mutations that affect only the CCAAT box have no effect on the cAMP response, the HKII CCAAT box-NF-Y complex and the CRE-CREB/ATF-1 complex have a redundant role in the cAMP response. To our knowledge, NF-Y has not been associated with cAMP responsiveness; nor has it been directly implicated as a regulator of genes involved in glucose metabolism. In this regard, it should be noted that an inverted CCAAT box mediates cAMP responsiveness of the human tryptophan hydroxylase gene and fatty acid synthase gene through an interaction with an unknown nuclear factor (25,27). In addition, a CCAAT box mediates the cAMP response of the Gprotein ␣ i2 subunit gene, which lacks a typical CRE (26). It will be interesting to see if NF-Y is involved in these effects.
The apparent role of NF-Y in the cAMP response of the HKII gene was not expected. CREB and ATF-1 bind constitutively to CREs and support basal transcription. The phosphorylation of these proteins by PKA results in an enhanced rate of gene transcription (15,16,24,69). PKA-mediated phosphorylation of NF-Y has not been reported, but the protein does contain several consensus phosphorylation sites (data not shown). Phosphorylation could increase the affinity of binding of NF-Y to the HKII CCAAT box, which has a lower binding affinity for the protein than does the consensus Y box. Alternatively, phosphorylation could enhance the transactivation potential of constitutively bound NF-Y, as happens with CREB.
The orientation of the CCAAT box relative to the CRE did not alter the PKA response, but the introduction of a half turn in the DNA helix between the CCAAT box and the CRE reduces the PKA induction, and insertion of a full turn maintains the response. This suggests that a specific helical orientation of the binding factors NF-Y and CREB/ATF-1 is critical for the PKA response in the rat HKII gene. A similar stereospecific alignment of the Y box and X box is required for MHC class II DRA promoter function (70).
As summarized in Fig. 8, NF-Y bound to the CCAAT box and CREB/ATF-1 bound to the CRE play important roles in both basal and cAMP-induced expression of the rat HKII gene. A third element important for basal HKII gene transcription, the Y box, also binds NF-Y. Finally, a typical TATA box is located near the transcription initiation site. It is not clear which of the combinations of CREB and ATF-1 are involved in the basal and cAMP-mediated responses through the CRE. It is also not clear how NF-Y is involved in these responses. Further experiments should help clarify these points. FIG. 8. Schematic representation of the basal and cAMP response elements and their binding factors in the rat HKII gene promoter. The Y box, CCAAT box, and CRE contribute to the basal activity of the HKII gene promoter, and the CCAAT box and CRE provide a redundant response to cAMP. Homo-and heterodimers of CREB/ATF-1 bind to the CRE. NF-Y binds to the Y and CCAAT boxes, probably as a trimeric complex of A, B, and C subunits (58 -60).