Involvement of a unique carbohydrate-responsive factor in the glucose regulation of rat liver fatty-acid synthase gene transcription.

Refeeding carbohydrate to fasted rats induces the transcription of genes encoding enzymes of fatty acid biosynthesis, e.g. fatty-acid synthase (FAS). Part of this transcriptional induction is mediated by insulin. An insulin response element has been described for the fatty-acid synthase gene region of -600 to +65, but the 2-3-fold increase in fatty-acid synthase promoter activity attributable to this region is small compared with the 20-30-fold induction in fatty-acid synthase gene transcription observed in fasted rats refed carbohydrate. We have previously reported that the fatty-acid synthase gene region between -7382 and -6970 was essential for achieving high in vivo rates of gene transcription. The studies of the current report demonstrate that the region of -7382 to -6970 of the fatty-acid synthase gene contains a carbohydrate response element (CHO-RE(FAS)) with a palindrome sequence (CATGTGn(5)GGCGTG) that is nearly identical to the CHO-RE of the l-type pyruvate kinase and S(14) genes. The glucose responsiveness imparted by CHO-RE(FAS) was independent of insulin. Moreover, CHO-RE(FAS) conferred glucose responsiveness to a heterologous promoter (i.e. l-type pyruvate kinase). Electrophoretic mobility shift assays demonstrated that CHO-RE(FAS) readily bound a unique hepatic ChoRF and that CHO-RE(FAS) competed with the CHO-RE of the l-type pyruvate kinase and S(14) genes for ChoRF binding. In vivo footprinting revealed that fasting reduced and refeeding increased ChoRF binding to CHO-RE(FAS). Thus, carbohydrate responsiveness of rat liver fatty-acid synthase appears to require both insulin and glucose signaling pathways. More importantly, a unique hepatic ChoRF has now been shown to recognize glucose responsive sequences that are common to three different genes: fatty-acid synthase, l-type pyruvate kinase, and S(14).

The consumption of a high carbohydrate diet increases the hepatic production of malonyl-CoA and its subsequent utiliza-tion for de novo fatty acid biosynthesis (1). In addition to being the substrate for fatty acid biosynthesis, malonyl-CoA governs the rate of fatty acid oxidation by virtue of its ability to function as a negative metabolite effector of carnitine palmitoyltransferase (2). The concentration of malonyl-CoA is determined by its rate of synthesis by acetly-CoA carboxylase and its rate of utilization by fatty-acid synthase. Carbohydrate ingestion causes a coordinate induction in the expression of both hepatic enzymes (3)(4)(5)(6)(7)(8)(9)(10). In the case of fatty-acid synthase the amount of protein is largely determined by mRNA abundance for fattyacid synthase, and this in turn is determined by the rate of fatty-acid synthase gene transcription (5)(6)(7)(8)(9)(10). The carbohydrate induction of hepatic fatty-acid synthase gene transcription requires both insulin and glucocorticoids (5, 8, 10 -12). On the other hand, fatty-acid synthase expression is suppressed by polyunsaturated fatty acids and sterols (9,(12)(13)(14) and by the administration of glucagon and growth hormone (15,16). The transcriptional response of the hepatic fatty-acid synthase gene to carbohydrate reportedly involves a glucose response region located within the first intron (17) as well as insulin and cAMP response sequences located between Ϫ444/Ϫ278, Ϫ278/Ϫ131, and Ϫ120/Ϫ50 (15, 18 -26). Functional studies have revealed that the sequence between Ϫ70 and Ϫ57 imparts insulin responsiveness to the fatty-acid synthase promoter (19), while the region of Ϫ99 to Ϫ92 may contain the cAMP target (15). Transcription factors that interact with the insulin response region include USF-1, 1 USF-2, SREBP-1, NF-Y, and Sp1 (19 -26). Moreover, the hepatic content of SREBP-1 and the DNA binding activity of Sp1 are increased by insulin and glucose ingestion and decreased by fasting (20,26). In addition, the binding of SREBP-1 to its DNA recognition sequence within the insulin response region of fatty-acid synthase may enhance the binding of Sp1 to its response element at Ϫ80, which may in turn enhance the interaction of NF-Y with its recognition sequence at Ϫ90 (14,24).
Despite the fact that the proximal promoter region of the fatty-acid synthase gene contains numerous cis-acting elements that interact with transcription factors that appear to be regulated by insulin and glucose, the 2-3-fold enhancement of fatty-acid synthase promoter activity attributable to the insulin and glucose response elements is relatively weak in comparison with the 20 -30-fold induction in hepatic fatty-acid synthase gene transcription observed when fasted rats are refed carbohydrate or when diabetic rats are administered insulin (5)(6)10). The inability of the proximal promoter regions to yield in vivo rates of fatty-acid synthase gene transcription indicated that the fatty-acid synthase gene contained unidentified enhancer sequences that were essential to the nutritional regulation of hepatic fatty-acid synthase gene transcription. In an earlier report we determined that the distal region of the fatty-acid synthase gene between Ϫ7382 and Ϫ6970 contained elements that imparted to the fatty-acid synthase promoter a rate of transcriptional activity that was comparable with that observed in fasted-refed rats (27). Sequence analysis revealed that this region contained a 17-base pair segment that was homologous to the ChoRE of the L-type pyruvate kinase and S 14 genes (28). This observation led us to hypothesize that the 17-base pair segment within the distal enhancer region of the fatty-acid synthase gene is a functional ChoRE and that this candidate ChoRE may bind the recently identified ChoRF (25,29).

EXPERIMENTAL PROCEDURES
Primary Hepatocyte Culture and Transfection-Primary hepatocytes were isolated from male Harlan Sprague-Dawley rats (180 -220 g) using the collagenase perfusion method (30). After a 4 -6 h attachment period, hepatocytes were transfected using Lipofectin reagent (Life Technologies, Inc.) in Waymouth's medium containing 5.5 mM glucose, 0.1 M dexamethasone, and 1 M insulin for 12-14 h. After transfection cells were incubated for 48 h in Waymouth's medium containing either 5.5 or 27.5 mM glucose, with or without 0.1 M dexamethasone and 1 M insulin. Cells were harvested, and the chloramphenicol acetyltransferase or luciferase reporter activity was quantified.
Plasmid Constructs-The parent plasmid Ϫ250FAS.PGL2 was described previously (14). The fatty-acid synthase gene 5Ј-deletion series were inserted in a unique BamHI site of the Ϫ250FAS.PGL2 to generate Ϫ7382FAS.LUC, Ϫ7242FAS.LUC, and Ϫ7150FAS.LUC (27). The 2ϫFASChoRE.PK construct was created by inserting into the HindIII site of PK(Ϫ96)LUC (25) two copies of an oligonucleotide containing the fatty-acid synthase sequence from Ϫ7218 to Ϫ7191 flanked on each side by a HindIII site. To prepare the Ϫ7382/Ϫ6970FAS.PK construct, the fragment Ϫ7382/Ϫ6970 was excised from a previously described construct (27) with HindIII and inserted into PK(Ϫ96)LUC.
In Vivo DNase I Footprinting Analysis-Nuclei from rat liver were isolated by the method of Jump et al. (31) and stored at Ϫ80°C until use. The digestion of nuclei with DNase I was performed with slight modifications of the method described previously (27). DNA (200 g) was incubated with 0.25 unit of DNase I at 30°C for 10 min in a total volume of 400 l. The digestion with proteinase K was performed at 37°C overnight without RNase A digestion. The in vivo footprinting method was based on that of Mueller and Wold (32). The sequences of primer sets are: 5Ј-CCTCACTATCTGCACCTCCC-3Ј, 5Ј-TGGCATCTC-TGGGCTGCTGTACC-3Ј, and 5Ј-GGCATCTCTGGGCTGCTGTACCTT-AGC-3Ј for the reverse reaction. The linker sequence was identical to that used by Mueller and Wold (32). After extracting DNA, the first primer extension was performed in the following condition: 5 g of cleaved genomic DNA, 5 pmol of primer, 40 mM NaCl, 10 mM Tris-Cl (pH 8.9), 5 mM Mg 2 SO 4 , 0.01% gelatin, 0.2 mM dNTPs, and 0.5 unit of Vent DNA polymerase (New England Biolab) in a total volume of 30 l. in the total volume of 23 l. The temperature program used was 30 cycles of 95°C for 20 s followed by 75°C for 2 min. The third genespecific primer (1.7 pmol/sample) was end-labeled using T4 polynucleotide kinase (United State Biochemical) and [␥-32 P]ATP (PerkinElmer Life Sciences, 6,000 Ci/mmol). The polymerase chain reaction-amplified DNA was labeled in the following reaction mixture: 5 l of the polymerase chain reaction reaction mix, 40 nM concentration of the endlabeled primer, 100 nM concentration each of fresh dNTPs, 0.25 unit of Vent and its buffer (New England Biolab) in final volume of 25 l. The reaction was repeated twice at 77°C for 10 min after denaturing at 95°C for 1 min. The labeled DNA was separated by denaturing polyacrylamide gel electrophoresis using a gel with 6% polyacrylamide, 42% urea, 0.8 mm thick and 40 cm long. The DNA fragments were visualized by autoradiography after drying the gel.
In Vitro DNase I Footprinting Analysis-Hepatic nuclear protein extracts for use in in vitro footprinting and electromobility shift assays were prepared from male Harlan Sprague-Dawley rats that had been fasted for 48 h or fasted for 48 h and refed a high carbohydrate fat-free diet for 6 h (33,25). The Ϫ7282/Ϫ6970 fragment of the fatty-acid synthase gene was released as a NcoI/HpaII fragment. The antisense strand was labeled by filling in the NcoI site using Klenow fragment in the presence of [␣-32 P]dCTP. Labeled DNA (ϳ50,000 cpm) was incubated on ice for 10 min with 1 g of poly(dI-dC) and various amounts of nuclear protein extracts in a final volume of 50 l (25 mM Tris-HCl (pH 8), 50 mM KCl, 6.25 mM MgCl 2 , 0.5 mM EDTA, 10% glycerol, 0.5 mM dithiothreitol). Subsequent to the addition of 50 l of 5 mM CaCl 2 and 10 mM MgCl 2 , the binding reaction was treated at room temperature (22°C) with 1 unit of RQ1 RNase-free DNase I (Promega). The reaction was stopped after 1 min by addition of 90 l of 83 mM EDTA-EGTA and 5 l of proteinase K (10 g/l). After a 30-min incubation at 42°C the reaction was ethanol-precipitated and the precipitate resuspended in 80% formamide. The DNA fragments were separated by electrophoresis using a 6% polyacrylamide, 7 M urea sequencing gel.
Electrophoretic Mobility Shift Assay-The following single-stranded oligonucleotides containing the candidate fatty-acid synthase ChoRE were synthesized for use in gel shift assay: FAS-ChoRE (Ϫ7221/Ϫ7190), 5Ј-CTTCCTGCATGTGCCACAGGCGTGTCACCCTC-3Ј and 3Ј-GAAG-GACGTACACGGTGTCCGCACAGTGGGAG-5Ј. Complementary strands were annealed by combining equal amounts of each oligonucleotide, heating to 95°C for 5 min, and cooling to 25°C. The annealed oligonucleotide was end-labeled with [␥-32 P]ATP using T4 polynucleotide kinase and incubated on ice with nuclear protein extract for 30 min in 10 mM HEPES (pH 7.9), 75 mM KCl, 1 mM EDTA, 5 mM MgCl 2, 5 mM dithiothreitol, and 10% glycerol. A typical reaction contained 50,000 cpm of labeled oligonucleotide (0.5-1 ng). For the experiments shown in Figs. 4 and 5, electrophoretic mobility shift assay was performed as described previously (29) using a fraction of liver nuclear extract enriched for ChoRF binding by heparin-Sepharose and salmon sperm DNA-Sepharose chromatography. After incubation samples were subjected to electrophoresis on a 5% nondenaturing polyacrylamide gel in a Tris-glycine buffer system. DNA-protein complexes were visualized by autoradiography after drying the gel.

Identification of a Glucose Response Region within the Distal
Promoter of the Fatty-acid Synthase Gene-Nuclei isolated from the livers of rats that had been fasted for 48 h and refed a high glucose diet for 6 h displayed greater sensitivity to DNase I cleavage than did nuclei from fasted rats. Three DNase hypersensitivity sites were identified in the 5Ј-flanking regions of the fatty-acid synthase gene: (a) Ϫ7382 to Ϫ6970; (b) Ϫ600 to Ϫ400; and (c) Ϫ100 to ϩ50 (27). Although the proximal sites (b and c) correspond to the regions that are involved with the gene's response to insulin (18,19,22), these regions do not appear to impart glucose responsiveness to the fatty-acid synthase promoter (Fig. 1, note Ϫ7150 and Ϫ250). However, transfection-reporter analyses using primary cultures of rat hepatocytes revealed that constructs that included the distal region of Ϫ7382 to Ϫ6970 with the proximal insulin elements yielded 7-10-fold more promoter activity than did constructs containing only the insulin response region of the proximal promoter (27). More importantly, the region between Ϫ7382 and Ϫ6970 imparted glucose responsiveness to the fatty-acid synthase promoter (Fig. 1). Specifically, hepatocytes that had been transfected with a luciferase reporter construct that linked the distal enhancer sequences of Ϫ7382 to Ϫ6970 with the proximal insulin response element of the fatty-acid synthase promoter expressed 2-fold more luciferase when the medium contained 27.5 mM glucose than when it contained only 5.5 mM glucose. More importantly this response was independent of insulin. Deleting the sequences between Ϫ7382 and Ϫ7242 had no effect on the glucose stimulation of fatty-acid synthase promoter activity. However, deleting the region between Ϫ7242 and Ϫ7150 completely eliminated the glucose induction of the fatty-acid synthase promoter.
The Distal Glucose Enhancer Region of the Fatty-acid Synthase Gene Contains a ChoRE-Functional analyses indicated that the sequences between Ϫ7242 and Ϫ7150 of the fatty-acid synthase gene contained a glucose response element (Fig. 1). In vitro footprinting using nuclear protein extracts from fed rats revealed that the sequence between Ϫ7240 and Ϫ7120 contained six sites that were protected from DNase I cleavage (Fig.  2). The protected site between Ϫ7214 and Ϫ7190 (sites IV and V) contains the palindrome 5Ј-CATGTG(n 5 )GGCGTG-3Ј that has a high degree of sequence similarity with the ChoRE of the S 14 and L-type pyruvate kinase genes (Fig. 3) (28). The protected site between Ϫ7150 and Ϫ7125 (site I) contains a CCAAT box, which electrophoretic mobility shift and supershift assays revealed bind the transcription factor, C/EBP␣ (data not shown). In addition, this protected area also contains a TGGA(n 6 )GCCAA sequence that may be a recognition sequence for NF-1. The two footprints between Ϫ7180 and Ϫ7170 (sites II and III) and the footprint between Ϫ7235 and Ϫ7220 (site VI) do not appear to contain sequences that correspond to any known nuclear transcription factors.
Given the high homology between the ChoRE in the L-type pyruvate kinase and S 14 genes, and the sequences located between Ϫ7214 and Ϫ7190 of the fatty-acid synthase gene (Fig.  3), we hypothesized that nuclear protein binding to this site would change with fasting and fasting-refeeding. In vivo footprint analysis of the fatty-acid synthase gene between Ϫ7350 and Ϫ7130 revealed that refeeding fasted rats a high glucose diet resulted in protection of the candidate ChoRE site from DNase I cleavage (Fig. 4, sites 2 and 3). Fasting reinstated DNase I cleavage within the ChoRE (Fig. 4). Fasting also resulted in protein binding at two sites: a, Ϫ7210 to Ϫ7190 and b, Ϫ7170 to Ϫ7160 (Fig. 4). Interestingly, site a appears to involve sequences that are components of the ChoRE, and site b was equivalent to site 4 of the fed. Site 1 in the fed animals remains to be identified, but site 5 corresponds to the C/EBP␣ site (Fig. 4).
The ChoRE of the Fatty-acid Synthase Gene Binds a Carbohydrate Response Factor and Imparts Carbohydrate Responsiveness-The ChoRE of the rat L-type pyruvate kinase and S 14 genes binds a novel hepatic carbohydrate responsive transcription factor (ChoRF) (25,29). Electrophoretic mobility shift assays using a partially enriched ChoRF fraction prepared from the hepatic nuclei of fed rats revealed that the ChoRE of the rat fatty-acid synthase gene binds the ChoRF (Fig. 5, lane 1). Protein binding activity for the fatty-acid synthase ChoRE was equal to or even greater than for the ChoRE of L-type pyruvate kinase or two S 14 ChoRE variants that support a glucose response (Fig. 5, lanes 1-3). In addition to binding the novel ChoRF, the ChoRE sequence of the fatty-acid synthase gene also interacts with USF, a transcription factor possibly involved with the carbohydrate/insulin induction of fatty-acid synthase and L-type pyruvate kinase (19,34). The similarity of the ChoRE of fatty-acid synthase to that of the L-type pyruvate kinase gene was reinforced by the observation that the fattyacid synthase ChoRE effectively competed for the binding of the ChoRF to the ChoRE of the L-type pyruvate kinase gene (Fig. 6, lanes 1, 4, and 5). The competitive action of the fattyacid synthase ChoRE was comparable with that observed for the ChoRE of the S 14 gene (Fig. 6, lanes 4 -7). On the other hand, the sterol response element and the hepatic nuclear factor-4 response element were ineffective as competitors for ChoRF binding (Fig. 6, lanes 8 -11).
Finally, in addition to binding a unique ChoRF, the ChoRE of the fatty-acid synthase gene was found to impart carbohydrate responsiveness to a basal L-type pyruvate kinase promoter that is otherwise unresponsive to glucose (Table I). Specifically, when a tandem repeat of the fatty-acid synthase ChoRE was inserted into the L-type pyruvate kinase promoter-reporter construct (Ϫ96/ϩ12) PK.LUC (lacking the PK ChoRE sequence), luciferase expression was increased nearly 5-fold; and when ligated to (Ϫ40/ϩ12) PK.LUC, luciferase activity was increased 30-fold (Table I). The greater glucose induction with the (Ϫ40/ ϩ12) PK.LUC construct was due to the fact that the Ϫ40 has lower promoter activity than does the Ϫ96 L-type pyruvate kinase promoter when hepatocytes are maintained in a low glucose media. Finally, the enhancer activity of the fatty-acid synthase ChoRE was very comparable with that of the S 14 ChoRE (Table I). DISCUSSION The cis-acting elements responsible for the carbohydrate/ insulin induction of fatty-acid synthase gene transcription have been attributed to sequences within the proximal promoter region (i.e. Ϫ644 to Ϫ50) of the gene (18 -22). Of particular interest has been the sequence between Ϫ68 and Ϫ57. This area contains an E-box binding site for USF-1 and -2 and a binding site for SREBP-1c that involves sequences on either side of the E-box (14,23). Disruption of either the USF-1 or USF-2 genes reduced the hepatic abundance of fatty-acid synthase mRNA to a level that was only 10 -15% of that found in wild type mice (35). Moreover, an early report by Sul and co-workers (19) indicated that the DNA binding activity and protein abundance of USF-1 was reduced by fasting and increased by carbohydrate refeeding. However, subsequent studies have clearly demonstrated that the abundance of hepatic USFs is not altered by nutritional manipulations (7). In contrast to USF, the hepatic abundance of SREBP-1c was rapidly induced by refeeding fasted rats a high glucose diet, and by administering insulin to diabetic rats, or by treating rat hepa-tocytes in primary culture with insulin (36 -38). This induction of hepatic SREBP-1c expression was paralleled by an increase in hepatic fatty-acid synthase gene transcription (20). The importance of SREBP-1c to fatty-acid synthase gene transcription was further suggested by the observation that disrupting the binding site for SREBP-1c without affecting the E-box site for USF eliminated the insulin/glucose activation of the fatty-acid synthase promoter (24). Thus, the nuclear abundance of SREBP-1c and the binding of SREBP-1c to its recognition sequences within the area of Ϫ68 to Ϫ57 appear to be important determinants for hepatic fatty-acid synthase promoter activity. However, disruption of the SREBP-1 gene only partially prevented the increase in hepatic fatty-acid synthase mRNA that results from feeding fasted rats a high glucose diet (27). This suggested that nutritional regulation of fatty-acid synthase gene transcription required factors other than SREBP-1c. Using transgenic mice to characterize 5Ј-flanking sequences between Ϫ2100 and ϩ67, Sul and co-workers (19,21) recently discovered that a USF binding site at Ϫ332 and an SREBP-1 binding site at Ϫ150 of the fatty-acid synthase gene were essential for insulin and carbohydrate regulation of the gene. Moreover, the investigators proposed that the region between FIG. 2. In vitro footprint of the fattyacid synthase gene between ؊7240 and ؊7120. The fragments of the fattyacid synthase gene between Ϫ7190 and Ϫ7120 (A) and Ϫ7240 and Ϫ7180 (B) were subjected to in vitro DNase I footprinting as described under "Experimental Procedures." The radiolabeled probes were incubated with DNase I plus 0 (lanes N), 10, and 50 g of liver nuclear proteins extract (lanes 1 and 2). The DNase I protected regions are marked by the brackets at the right and are described in C. Ϫ444 and ϩ67 is sufficient to replicate the in vivo insulin and carbohydrate induction of fatty-acid synthase gene transcription. However, a close examination of their data reveals that the transcription of the native fatty-acid synthase gene is ϳ2-3-fold greater than that observed for the reporter gene. Thus elements critical to the in vivo regulation of fatty-acid synthase gene transcription appear to be located outside of the proximal promoter region.
One approach for identifying gene regions that may play an important role in gene transcription is to manipulate nutritional status and examine changes in chromatin structure (27). Using this approach we found that refeeding glucose to 48-h fasted rats induced changes in chromatin structure within three regions of the hepatic fatty-acid synthase gene: (a) Ϫ7300 and Ϫ6900, (b) Ϫ600 to Ϫ400, and (c) Ϫ150 to ϩ50 (27). Surprisingly, the region between ϩ283 and ϩ303, which reportedly contains a glucose response element (17), did not display a DNase I cleavage site (27). Consistent with this observation we were also unable to demonstrate that the ϩ283 to ϩ303 region imparted glucose responsiveness to the fatty-acid synthase promoter. 2 The two DNase I hypersensitivity sites between Ϫ600 and ϩ50 corresponded to the regions within the fatty-acid synthase promoter, which contain insulin response sequences (18 -22). Moreover, insulin treatment of hepatocytes transfected with constructs containing this region of the proximal promoter region increased fatty-acid synthase promoter activity 2-3-fold, but the proximal promoter region failed to confer glucose responsiveness to the fatty-acid synthase promoter ( Fig. 1). On the other hand, when the distal region of Ϫ7382 to Ϫ6970 was linked to the proximal sequences of the fatty-acid synthase gene, fatty-acid synthase promoter activity was increased 3-4-fold (27), and the transcriptional activity of the promoter was comparable with the rate of gene transcription observed in rats fed a high glucose diet (5)(6)(7)(8)(9). Interestingly, the 3-4-fold enhancement of fatty-acid synthase promoter activity contributed by the sequences between Ϫ7382 and Ϫ6970 was approximately that amount required for the proximal promoter region of Ϫ444 to ϩ67 to achieve in vivo rates of fatty-acid synthase gene transcription (19,21,27).
The above observations suggested to us that perhaps the proximal promoter region imparted insulin control to the fattyacid synthase promoter, while the distal sequences between Ϫ7382 and Ϫ6970 conferred carbohydrate stimulation to the promoter. Consequently the sequences between Ϫ7382 and Ϫ6970 were examined more closely for their role in the glucose induction of fatty-acid synthase gene transcription. Functional mapping studies indicated that the Ϫ7382 to Ϫ6970 region contained glucose-responsive sequences. More importantly, the glucose enhancement of fatty-acid synthase promoter activity appeared to be independent of insulin (Fig. 1). The ability of glucose to enhance fatty-acid synthase promoter activity by an insulin-independent mechanism is consistent with early observations showing that the feeding of fructose to diabetic rats induced hepatic fatty-acid synthase gene expression (39,40).
The glucose-responsive element within the distal region of the fatty-acid synthase gene appears to reside within the area The "mut 3/5" ChoRE is derived from the rat S 14 ChoRE (34) and the "mut 3-6" ChoRE is derived from the mouse S 14 ChoRE (29). Each oligonucleotide was labeled to approximately the same specific activity and used with partially enriched ChoRF fraction from rat liver (29). Arrows indicate the ChoRF and the USF complexes. of Ϫ7242 to Ϫ7150 (Fig. 1). Sequence analysis revealed that this 100-nucleotide region contains candidate recognition sites for several transcription factors, including C/EBP␣ and NF-1, but the sequence located between Ϫ7240 and Ϫ7190 was of particular interest, because it possesses characteristics that are similar to the ChoRE of the L-type pyruvate kinase and S 14 genes. Like the ChoRE of the L-type pyruvate kinase and S 14 genes, the E-box motif of the ChoRE for fatty-acid synthase exists as palindromic sequences that are separated by a 5-bp spacer (Fig. 3). In the case of the fatty-acid synthase ChoRE, one motif contains a five out of six nucleotide match and the other displays a four out of six nucleotide match with the ChoRE of the other two genes. Mutation analysis of the rat S 14 and L-type pyruvate kinase genes indicates that the first 4 bp (CACG) of the E-box are critical to conferring glucose responsiveness, while mutations in the fifth and sixth positions did not disrupt the glucose response (41). Spacing between the E-boxes is critical to the function of the ChoRE, because shortening the spacer to 4 bp resulted in a loss of the glucose response, and lengthening the spacing sequence to 6 bp blunted the glucose response (28).
The type of transcription factor that interacts with the ChoRE of the fatty-acid synthase gene and functions as the target for the glucose signal is unknown, but the 5Ј-CACGTG sequence is a consensus binding sequence for basic/helix-loophelix/leucine zipper transcription factors (28). Hasegawa et al. (41) recently reported that a "glucose response element-binding protein" interacted with the CACGTG motifs located in the ChoRE of rat L-type pyruvate kinase, and in the insulin response element (Ϫ71 to Ϫ50) of the fatty-acid synthase gene. However, this unknown protein interacted weakly with the ChoRE of the rat S 14 gene and did not interact at all with comparable motifs in the acetyl-CoA carboxylase and citrate lyase genes. Moreover the insulin response element between Ϫ71 and Ϫ50 is not a glucose response sequence for the fattyacid synthase gene (Fig. 1). In contrast we found that the unique ChoRF of rat liver nuclei (25,29) readily interacted with the ChoRE L-type pyruvate kinase, S 14 , and fatty-acid synthase genes (Fig. 6). While the identity of ChoRF is unclear, it has now been shown to recognize specific glucose-responsive sequences common to three genes: L-type pyruvate kinase, S 14 , and fatty-acid synthase. It also recognizes variant versions of these elements, such as the mut3/5 oligonucleotide (Fig. 6), that retain glucose responsiveness but have greatly reduced binding affinity for USF. On the other hand, mutations of these naturally occurring ChoREs that fail to support a response to glucose do not bind ChoRF. Likewise, ChoRF does not bind to a consensus sterol response element or the adenovirus major late promoter USF binding site. Based on these correlations, ChoRF appears to be a strong candidate for mediating the effects of glucose on lipogenic gene expression. At present the identity of ChoRF is unknown, but it does not appear that ChoRF is identical to SREBP-1c or USF. The migration of the ChoRF complex is significantly slower than the migration of either of these factors. In addition, antibodies to USF or SREBP-1 fail to inhibit binding of ChoRF to the ChoRE (25). We thus propose that ChoRF is a novel hepatic factor, most likely of the basic/ helix-loop-helix family, that is regulated by changes in glucose metabolism.
In summary, we have shown that glucose regulation of fattyacid synthase transcription is mediated at least in part by a distal enhancer located over 7,000 bp upstream from the transcriptional start site. Within this enhancer a specific ChoRE binds the ChoRF and confers glucose regulation. The ChoRF appears to work in conjunction with SREBP-1c, and possibly USF-1 and -2, to impart glucose and insulin responsiveness to the fatty-acid synthase promoter. Hence, regulation of fattyacid synthase expression in response to carbohydrate requires both insulin and glucose signaling pathways working synergistically to provide the overall dramatic effects observed by carbohydrate diet. Recently, a similar situation has been found for the S 14 gene (25). It is not unreasonable to speculate that many other lipogenic enzyme genes regulated by these two effectors may utilize a similar dual regulatory pathway for glucose and insulin stimulation.
FIG. 6. The fatty-acid synthase ChoRE competes for the binding of the ChoRF to the L-type pyruvate kinase ChoRE. Competition experiments were conducted with the L-type pyruvate kinase (L-PK) ChoRE as the radiolabeled probe and cold ChoRE from the fatty-acid synthase (FAS) and the rat S 14 genes; a consensus sterol response element (SRE-1) (43) or the hepatic nuclear factor 4 (HNF4) binding site from the L-type pyruvate kinase promoter (Ϫ146/Ϫ124). The competing oligonucleotides were added as 25ϫ and 50ϫ molar excess to the binding reaction.