Regulation of Rat Hepatic L-Pyruvate Kinase Promoter Composition and Activity by Glucose, n-3 Polyunsaturated Fatty Acids, and Peroxisome Proliferator-activated Receptor-α Agonist*

Carbohydrate regulatory element-binding protein (ChREBP), MAX-like factor X (MLX), and hepatic nuclear factor-4α (HNF-4α) are key transcription factors involved in the glucose-mediated induction of hepatic L-type pyruvate kinase (L-PK) gene transcription. n-3 polyunsaturated fatty acids (PUFA) and WY14643 (peroxisome proliferator-activated receptor α (PPARα) agonist) interfere with glucose-stimulated L-PK gene transcription in vivo and in rat primary hepatocytes. Feeding rats a diet containing n-3 PUFA or WY14643 suppressed hepatic mRNAL-PK but did not suppress hepatic ChREBP or HNF-4α nuclear abundance. Hepatic MLX nuclear abundance, however, was suppressed by n-3 PUFA but not WY14643. In rat primary hepatocytes, glucose-stimulated accumulation of mRNALPK and L-PK promoter activity correlated with increased ChREBP nuclear abundance. This treatment also increased L-PK promoter occupancy by RNA polymerase II (RNA pol II), acetylated histone H3 (Ac-H3), and acetylated histone H4 (Ac-H4) but did not significantly impact L-PK promoter occupancy by ChREBP or HNF-4α. Inhibition of L-PK promoter activity by n-3 PUFA correlated with suppressed RNA pol II, Ac-H3, and Ac-H4 occupancy on the L-PK promoter. Although n-3 PUFA transiently suppressed ChREBP and MLX nuclear abundance, this treatment did not impact ChREBP-LPK promoter interaction. HNF4α-LPK promoter interaction was transiently suppressed by n-3 PUFA. Inhibition of L-PK promoter activity by WY14643 correlated with a transient decline in ChREBP nuclear abundance and decreased Ac-H4 interaction with the L-PK promoter. WY14643, however, had no impact on MLX nuclear abundance or HNF4α-LPK promoter interaction. Although overexpressed ChREBP or HNF-4α did not relieve n-3 PUFA suppression of L-PK gene expression, overexpressed MLX fully abrogated n-3 PUFA suppression of L-PK promoter activity and mRNAL-PK abundance. Overexpressed ChREBP, but not MLX, relieved the WY14643 inhibition of L-PK. In conclusion, n-3 PUFA and WY14643/PPARα target different transcription factors to control L-PK gene transcription. MLX, the heterodimer partner for ChREBP, has emerged as a novel target for n-3 PUFA regulation.

other glucose-regulated genes are not known to require HNF-4␣. Thus, the role of HNF-4␣ in glucose-regulated gene expression appears unique to L-PK. HNF-4␣, however, has been implicated in n-3 PUFA and WY14643 control of L-PK gene transcription because the L-PK DR1 element is required for both n-3 PUFA and WY14643 suppression of LPK gene transcription (5,6). Unlike WY14643, n-3 PUFA suppression of L-PK gene transcription does not require functional PPAR␣. Thus, n-3 PUFA and WY14643 utilize separate mechanisms to control L-PK gene expression.
Despite the progress in defining the transcription factors that control L-PK gene transcription, there remains a poor understanding of how these factors interact on the L-PK promoter in cells. Moreover, the molecular basis of n-3 PUFA or WY14643 antagonism of glucose-mediated induction of L-PK gene transcription remains unknown. To address these issues, we examined the glucose, n-3 PUFA, and WY14643 regulation of hepatic nuclear abundance of ChREBP, MLX, and HNF-4␣ in vivo and in cultured rat primary hepatocytes. The chromatin immunoprecipitation (ChIP) assay was used for a detailed analysis of L-PK promoter composition in response to glucose, n-3 PUFA, and WY14643 challenge of rat primary hepatocytes. These studies coupled with gene expression analyses have clarified the effects of glucose, n-3 PUFA, and WY14643 on L-PK promoter activity and composition. More importantly, these studies have identified a new n-3 PUFA-regulated transcription factor that is involved in the control of L-PK gene transcription.

MATERIALS AND METHODS
Animals and Primary Hepatocytes-All procedures for the use and care of animals for laboratory research have been approved by the All University Committee for Animal Use and Care at Michigan State University. Male Sprague-Dawley rats (Charles River Breeding Laboratories, Kalamazoo, MI) were maintained on Harlan-Teklad laboratory chow (catalog number 8640) and water ad libitum. Protocols for mealfeeding rats high carbohydrate diets supplemented with olive oil (10% w/w), fish oil (10% w/w), or olive oil (10% w/w) ϩ WY14643 (0.1% w/w) were described previously (28).
Rat primary hepatocytes were prepared from male Sprague-Dawley rats fed Teklad chow (ad libitum), cultured on BioCoat (collagen type I) plates (BD Biosciences), and treated with insulin (Invitrogen), fatty acids (Nu-Chek Prep, Elysian, MN), lactate, glucose, or PPAR␣ agonist WY14643 (Chemsyn Labs, KS) as described previously (28). For the ChIP assay and RNA and protein extraction, cells were plated onto 10-cm type I collagencoated plates (BD Biosciences) at 10 7 cells/plate in Williams E (Invitrogen), 10 mM lactate, 10 nM dexamethasone (Sigma), 1 M insulin (Invitrogen), and 10% fetal bovine serum (Invitrogen). For transfection studies, cells were plated in the same media onto 6-well type 1 collagen-coated plates at 1.5 ϫ 10 6 cells/well. The ratio of culture media to cell number was maintained constant for the different plating conditions. For treatments, hepatocytes were incubated in medium (Williams E ϩ 25 mM glucose, 10 nM dexamethasone, 1 M insulin, 50 M bovine serum albumin, and no serum) in the absence and presence of 250 M fatty acids (18:1,n-9 or 20:5,n-3) or 100 M WY14643. Bovine serum albumin used in these studies is endotoxinand fatty acid-free (Serological Proteins, Inc., Kankakee, IL).
Reporter Plasmids-The reporter plasmid TK-MH100X4-LUC contains four copies of the Gal4-binding element upstream of the thymidine kinase promoter driving luciferase expression (28). phRG-Luc expressing Renilla luciferase was obtained from Promega (Madison, WI); phRG-Luc serves as an internal control for transfection efficiency. The pM-HNF4␣1-LBD expression vector was constructed by PCR amplification of the E region, i.e. the ligand binding domain (LBD), of HNF-4␣1 (forward primer AATTCATCAGCACGCGGAGGTCAAC-TAC; reverse primer GGGGATCCCGGCAGACCCCAAGCAG-CATCTC) using pBS-HNF-4␣1 (obtained from F. Sladek, University of California, Riverside) as template. The HNF-4␣1 LBD was ligated inframe with the Gal4 DNA binding domain of the pM vector (Clontech). Transcription of the Gal4-HNF4␣ fusion protein is driven by the SV40 early promoter. The sequence of the construct was verified by DNA sequencing. L-PK-Luc was constructed by excising the Ϫ196to ϩ12-bp region from LPK-CAT (5) with BamHI and XhoI and subcloning into TOPOII (Invitrogen). After amplification, the TOPO-PK plasmid was digested with XhoI and SacI to release the L-PK promoter; the fragment was ligated into pGL3-basic vector (Promega).
Transfection, Fatty Acid Treatments, and the Luciferase Assay-Primary hepatocytes were plated in 6-well collagen-coated plates. Each well received 0.5 g of reporter plasmid (L-PK-LUC or TKMH100X4-LUC) and 0.1 g of internal control plasmid (phRG-LUC), plus Lipofectin (6.6 l/g DNA) or Lipofectamine 2000 (1.5 l/g DNA) (Invitrogen) in serum-free Williams E/lactate media (see above). After an overnight transfection, cells were treated with fatty acids (250 M) for 24 h in serum-free media containing bovine serum albumin (50 M). After treatment, cells were lysed and assayed for luciferase activity using the dual luciferase assay kit (Promega, Madison, WI) with a dual channel Turner Luminometer. Protein concentration of the cell lysate was measured using the Bio-Rad protein reagent with bovine serum albumin as standard. All experiments were run in triplicate and repeated at least one time.
Recombinant Adenovirus-Recombinant adenoviruses expressing ChREBP and MLX␤ were obtained from H. Towle, University of Minnesota, Minneapolis, MN (19). The adenoviruses were constructed using the pAdenoVator-CMV5-IRES-GFP. As such, green fluorescent protein (GFP) is expressed from an internal ribosome entry site. Adenovirus was prepared from Hek293 cell lysates. Virus-containing lysates were titered using the Adeno-X rapid titer kit (Clontech). Confluent primary hepatocytes were infected (5-10 plaque-forming units/cell). After infection, cells were transfected with L-PK-Luc to monitor the effect of overexpressing recombinant proteins on the glucose, 20:5,n-3, or WY14643 control of L-PK-Luc.
RNA Extraction and Northern Blot-RNA was extracted from rat liver or primary hepatocytes and separated on denaturing (formaldehyde) agarose gels, transferred to nitrocellulose, and hybridized with 32 P-labeled L-PK-cDNA (5,29). Hybridization was visualized by Phos-phorImager analysis (Amersham Biosciences). Transcript abundance was also measured by the quantitative reverse transcriptase-PCR (qRT-PCR). Briefly, RNA was extracted from primary hepatocytes (29) and used as template for real time-PCR. Specific primers for each gene (see below) were designed using Primer Express software (Applied Biosystems, Foster City, CA). First strand cDNA was synthesized using the SuperScript II RNase H-reverse transcriptase (Invitrogen). Synthesized cDNA was mixed with 2ϫ SYBR Green PCR Master Mix (Applied Biosystems) and various sets of gene-specific forward and reverse primers and subjected to real time-PCR quantification using the ABI PRISM 7900HT sequence detection system (Applied Biosystems). All reactions were performed in triplicate. The relative amounts of mRNAs were calculated by using the comparative C T method (User Bulletin 2, Applied Biosystems). Cyclophilin was used as a control, and all results were normalized to the abundance of cyclophilin mRNA.
Chromatin Immunoprecipitation Assay-Rat primary hepatocytes were plated at 10 7 per 10-cm plate (type I collagen-coated) and treated with lactate, glucose, fatty acids, or WY14643. At the times indicated in the figures, cells were fixed by adding formaldehyde (1.0% final concentration) directly to the culture media and incubated at room temperature for 10 min (acetyl-histone-H4 and acetyl-histone-H3), 37°C for 15 min (RNA pol II and ChREBP), or 37°C for 30 min (HNF-4␣). Cells were washed once with cold PBS containing protease inhibitors (P8340; Sigma), scraped in PBS, homogenized, and transferred into conical tubes. Cells were pelleted and resuspended in 2 ml of SDS lysis buffer (Upstate, Charlottesville, A) with protease inhibitors (P8340; Sigma). The cell lysate was placed on ice for 10 min. The lysate was sonicated with a Branson Sonifier 250 at power level 6, 60% output, 10 pulses for 10 times. Samples were kept on ice and cooled between pulses. After sonication, cell debris was removed by centrifugation, and the supernatant was retained. Sonicated chromatin was aliquoted and stored at Ϫ80°C. Sonicated chromatin (300 l) was used for each immunoprecipitation (IP) reaction. The IP reaction was performed with Upstate chromatin immunoprecipitation assay kit. The acetyl-histone H3, acetyl-histone H4, and RNA polymerase II antibodies were from Upstate. The HNF-4␣ (H-171) antibody was from Santa Cruz Biotechnology. 300 l of sample was diluted into 2 ml with IP dilution buffer (Upstate); 20 l was withdrawn for input DNA. After preincubation with protein A-agarose beads (Invitrogen) for 30 min and brief centrifugation, the supernatant was incubated with antibody (10 l for Ac-H3, Ac-H4, or pol II, 20 g for HNF4␣, 5 l for ChREBP) at 4°C overnight. Antibody-protein-DNA complex was pulled down with protein A-agarose beads. The complex was washed sequentially with low salt, high salt, lithium washing buffers one time each and two times with 10 mM Tris-EDTA, pH 8.0. Protein-DNA complex was eluted by incubation in elution buffer (0.1 M NaHCO 3 , 1% SDS) at room temperature for 15 min twice. Supernatants were pooled and the DNA-protein cross-links reversed in 200 mM NaCl (final concentration) at 65°C for 4 h. Proteins were digested by proteinase K (Roche Applied Science) at 45°C for 1 h. DNA fragments were purified by phenol/chloroform extraction and ethanol precipitation. Glycogen was used to facilitate DNA precipitation. DNA pellets were resuspended in 50 l of sterile Milli-Q water. DNA samples (5 l) were used as template for the PCR. Primers for PCR amplification of the L-PK promoter are in Table 1: Ϫ288 to ϩ12 bp (GenBank TM accession number X05684) (30); TAT enhancer, Ϫ3730 to Ϫ3431 bp (GenBank TM accession number x16379) (35); phosphoenolpyruvate carboxykinase (PepCk) promoter, Ϫ550 to ϩ67 bp (GenBank TM accession number k03243) (36); fatty-acid synthase ChoRE, Ϫ7221 to Ϫ7125 bp (37); S14 ChoRE, Ϫ1485 to Ϫ1350 bp and the S14 proximal promoter, Ϫ290 to ϩ16 bp (GenBank TM accession number M33553) (18,32); ␤-actin coding region, ϩ2383 to ϩ3091 bp (GenBank TM accession number V01217) (38).
Statistical Analysis-Statistical analysis used Student's t test and analysis of variance plus post hoc Tukey honestly significant difference test.
When compared with rats fasted overnight, feeding rats a high carbohydrate diet supplemented with olive oil induced mRNA L-PK 6-fold. The nuclear abundance of ChREBP and MLX was induced 4-and 2-fold, respectively (Fig. 2). HNF-4␣ nuclear abundance remained unchanged by fasting or refeeding.
The effect of dietary fat and WY14643 on the nuclear abundance of ChREBP, MLX, and HNF-4␣ and L-PK gene expression was examined in rats meal-fed a high carbohydrate diet supplemented with olive oil, fish oil, or olive oil ϩ WY14643 for 7 days (Fig. 3). Olive oil is enriched in n-9 monounsaturated fatty acids, and fish oil is enriched in C 20 -22 n-3 PUFA. Both fish oil and WY14643-supplemented diets suppressed L-PK mRNA (Fig. 3C). However, changes in dietary fat or WY14643 composition had no effect on hepatic nuclear or cytoplasmic ChREBP 5Ј-GGAGGTTTGGCCAATGACCCCTTTGG S14 Ϫ1485and Ϫ1350-bp ChoRE 5Ј 5Ј-AGCTCTCCCAGCCCTGAC 18, 32 3Ј 5Ј-CCTGGTTGTGTAACTCCCTTTG S14 Ϫ290and ϩ19-bp proximal promoter 5Ј 5Ј-GCCTGCAGTCAAGTGTACTGGGT 18, 32 3Ј 5Ј-GCTTCCTTTCTCAGAGACC ␤-Actin ϩ2383and ϩ3091-bp coding region 5Ј 5Ј-TACTCCTGCTTGCTGATCCAC 38 3Ј 5Ј-GGCTACAGCTTCACCACCAC Nutrient Control of L-PK Promoter Composition or HNF4␣. Moreover, these diets did not affect hepatic mRNA ChREBP or mRNA HNF4␣ abundance. Rats fed the high carbohydrate diet supplemented with fish oil, however, had 50% less nuclear MLX when compared with the other diets. The fish oil diet had no effect on mRNA MLX abundance. Cytoplasmic MLX levels were higher in fish oil-fed animals when compared with olive or olive ϩ WY14643-fed animals (Fig. 3A, lower panel) suggesting fish oil treatment may act at a post-translational level to inhibit MLX translocation to the nucleus.
These results indicate that in vivo changes in MLX, and not ChREBP or HNF-4␣ nuclear abundance, correlate with n-3 PUFA suppression of L-PK gene expression. In contrast, WY14643 has no effect on the nuclear abundance of the three main transcription factors controlling glucose-regulated L-PK gene expression.

Glucose, Fatty Acid, and WY14643 Effects on L-PK Expression and ChREBP, MLX, and HNF-4␣ Nuclear Abundance in Rat Primary Hepatocytes
Rat primary hepatocytes were used to examine the time course for glucose, fatty acid, and WY14643 control of mRNA L-PK (Fig. 4) and ChREBP and MLX nuclear abundance (Fig. 5). Primary rat hepatocytes maintained overnight in Williams E medium containing lactate (10 mM), insulin (1 M), and dexamethasone (10 nM) were switched to Williams E medium containing glucose (25 mM), insulin (1 M), dexamethasone (10 nM), and either fatty acids (oleic acid, 18:1,n-9, or eicosapentaenoic acid, 20:5,n-3) at 250 M or WY14643 at 100 M. Cells were then harvested at various times to examine hepatocyte abundance of mRNA L-PK (Fig. 4, A and B). A second group of cells were transfected with an L-PK-luciferase (L-PK-Luc) reporter plasmid and treated as above to monitor effects on L-PK promoter activity (Fig. 4C).
Glucose induced L-PK-Luc activity 8-fold, and 18:1,n-9 did not significantly affect this response. 20:5,n-3 and WY14643, however, suppressed the glucose induction of L-PK-Luc activity by Ͼ75% (Fig. 4C). These studies confirmed previous studies (6) and provided key time course information on the response of the L-PK gene to glucose, fatty acid, and WY14643 challenge.
The effects of glucose, 20:5,n-3, and WY14643 on the nuclear abundance of ChREBP and MLX were examined (Fig. 5). Both ChREBP and MLX were detected in nuclei of hepatocytes maintained in lactate-containing medium (Fig. 5). Switching cells to glucose-containing medium induced nuclear ChREBP with no change in nuclear MLX. Particularly important is the robust (4-, 7-, and 12-fold) glucose-stimulated induction of hepatic nuclear ChREBP at 1.5, 6, and 24 h. 20:5,n-3 and WY14643 had no effect on glucosestimulated ChREBP nuclear abundance at 1.5 or 24 h but modestly atten-FIGURE 1. Schematic of the L-PK promoter. The diagram illustrates the location of key transcription factors and their binding sites on the L-PK promoter. Factors regulated by glucose, fatty acids, and the PPAR␣ agonist WY14643 are identified. Linker scanning studies established that the ChoRE was required for glucose control, whereas the DR-1 element was required for PUFA and PPAR␣ control of L-PK gene transcription (5, 6, 13). FIGURE 2. Effect of fasting and refeeding on hepatic content of ChREBP, MLX, and HNF-4␣. Rats were meal-fed a high carbohydrate diet supplemented with olive oil for 7 days ("Materials and Methods"). Olive oil-fed rats were fasted overnight (white bars) or meal-fed (black bars) the olive oil diet for 4 h. Fasted rats were euthanized at 8 a.m.; fed rats were euthanized 2 h after completion of the meal, i.e. at 2 p.m. Hepatic RNA and nuclear protein extracts were prepared for Northern and Western blotting. Upper panel, immunoblot of hepatic nuclear abundance of ChREBP, MLX, and HNF-4␣ from three separate rats per treatment. Lower panel, L-PK mRNA was measured by Northern analysis and quantified along with the abundance of nuclear ChREBP, MLX, and HNF-4 measured by immunoblot. Quantified results are expressed as fold change from the fasted level, mean Ϯ S.D., n ϭ 3. *, p Ͻ 0.005, Student's t test.
MLX nuclear abundance was transiently suppressed (70%) at 1.5 h by 20:5,n-3. By 6 and 24 h, MLX nuclear abundance was not different from cells receiving glucose alone. WY14643 had no effect on nuclear MLX abundance. As in vivo (Fig. 3), the nuclear abundance of HNF-4␣ in rat primary hepatocytes remained unaffected by 20:5,n-3 or WY14643 treatment (not shown). These results reveal a transient effect of PUFA and WY14643 on ChREBP nuclear abundance and PUFA on MLX nuclear abundance.

Glucose, Fatty Acid, and WY14643 Effects on L-PK Promoter Composition
The ChIP assay was used in conjunction with the time course approach to examine changes in L-PK promoter composition during glucose, 20:5,n-3, and WY14643 control of L-PK gene expression (Fig. 6). In primary hepatocytes maintained in Williams E ϩ insulin ϩ lactate, L-PK promoter activity and mRNA L-PK is low (Fig. 4). Little RNA pol II, Ac-H3, and Ac-H4 is bound to the L-PK promoter (Fig. 6A). Despite the absence of RNA pol II, Ac-H3, and Ac-H4 from the L-PK promoter, both ChREBP and HNF-4␣ are bound to the promoter. Technical limitations precluded an accurate measure of MLX on the L-PK promoter. Because ChREBP binding to the L-PK ChoRE requires MLX in vitro (39), our analysis may indirectly reveal the presence of MLX on the L-PK promoter. As such, ChREBP/MLX heterodimers and HNF-4␣ homodimers are bound to the L-PK promoter in the absence of RNA pol II and acetylated histone H3 and H4.
Glucose Effects-Switching primary hepatocytes from lactate to glucose-containing medium increased the abundance of RNA pol II, Ac-H3, and Ac-H4 on the L-PK promoter at 1.5 h. By 6 h, RNA pol II on Rats were meal fed a high carbohydrate diet supplemented with olive oil, fish oil, or olive oil ϩ WY14643 for 7 days ("Materials and Methods"). Hepatic RNA and cytoplasmic and nuclear protein were isolated from animals euthanized 2 h after completion of the meal. A, immunoblots for nuclear HNF-4␣ (HNF-4), ChREBP, and MLX abundance and cytoplasmic ChREBP and MLX. Results are from three separate animals per treatment. B, quantified levels of nuclear protein abundance of ChREBP, MLX, and HNF-4␣ as measured by immunoblot. C, mRNA abundance of L-PK, ChREBP, MLX, and HNF-4␣ as measured by qRT-PCR. Results are expressed as fold change from olive oil-fed rats. Results are the mean Ϯ S.D. for seven different animals per treatment. *, p Յ 0.01, analysis of variance. Quantified results are expressed as fold change in L-PK mRNA from lactate-treated cells, mean Ϯ S.D., n ϭ 4. C, primary hepatocytes in lactate-containing medium were transfected with L-PK-Luc (firefly) and phRG-Luc (Renilla) (see "Materials and Methods") and treated as above with glucose, fatty acids, or WY14643 for 24 h. Cells were harvested for luciferase assays. Results are expressed as the fold change in relative luciferase activity (RLA) from lactate-treated cells, mean Ϯ S.D., n ϭ 4.
the L-PK promoter increased to a maximum (2-fold) above basal values. By 24 h, however, RNA pol II levels on the L-PK promoter returned to near pretreatment levels (Fig. 6, A and B). Ac-H3 and Ac-H4 on the L-PK promoter increased significantly (6-fold) by 6 h and remained elevated 24 h after initiating glucose treatment (Fig. 6C). The effect of glucose on Ac-H3 and Ac-H4 promoter content was specific; the acetylation status of H3 and H4 on the TAT (Fig. 6) and PepCk (not shown) promoters was not affected by glucose. Neither TAT nor PepCk expression is induced by glucose.
In contrast to RNA pol II, Ac-H3, and Ac-H4, the abundance of ChREBP and HNF-4␣ on the L-PK promoter was not significantly affected by glucose treatment. Thus, glucose-stimulated recruitment of RNA pol II to the L-PK promoter and the increased acetylation of H3 and H4 are not accompanied by demonstrable changes in ChREBP or HNF-4␣ promoter occupancy.
Because glucose has no effect on HNF-4␣ nuclear content, we were not surprised to find no major effect of glucose on HNF-4␣ interaction with the L-PK promoter. In contrast, glucose stimulated a significant accumulation of nuclear ChREBP. As such, we expected to measure an increase ChREBP binding to the L-PK promoter using the ChIP assay. Instead, no major change in ChREBP-L-PK promoter interaction was detected. This outcome questioned the efficacy of the ChREBP-ChIP assay to accurately reflect ChREBP-ChoRE interaction. Several control studies were carried out to assess the efficacy of this assay.
␤-Actin is not regulated by glucose or ChREBP; DNA encoding the ␤-actin structural gene was not detected in ChREBP immunoprecipitates (not shown). Transcription of both S14 and FAS is induced by glucose, and ChoREs have been identified in the promoters of both genes (17,40). ChREBP immunoprecipitates contained both the S14 and FAS ChoRE DNA. Glucose did not consistently or significantly alter ChREBP interaction with the S14 or FAS ChoREs in primary hepatocytes (Fig. 6). ChREBP is not bound to the S14 proximal promoter (Ϫ290 to ϩ 19 bp), and this DNA was not detected in the ChREBP immunoprecipitates (not shown). Taken together, these results indicate that the ChIP assay for ChREBP provides an accurate measure of ChREBP-ChoRE containing complexes. 20:5,n-3 Effects-Treating cells with glucose ϩ 20:5,n-3 inhibits glucose-stimulated L-PK gene transcription (6), mRNA L-PK accumulation, and L-PK-Luc activity (Fig. 4). As expected, 20:5,n-3 attenuated recruitment of RNA pol II to the promoter and the acetylation of histones H3 and H4 (Fig. 6 A-C) at 1.5 and 6 h. This effect correlated with the 50% suppression of L-PK mRNA ( Fig. 4) During this time, 20:5,n-3 treatment did not interfere with ChREBP interaction with the L-PK promoter. The effect of 20:5,n-3 on histone acetylation was specific to the L-PK promoter because histone H3 and H4 acetylation status remained unaffected on the TAT (Fig. 6) and PepCk promoters (not shown).
After 24 h of treatment, however, ChREBP, RNA pol II and acetylated H3 were not detected on the L-PK promoter (Fig. 6, A and D). ChREBP interaction with the ChoRE of the S14 and FAS promoter was not consistently affected by 20:5,n-3 treatment at any time point examined (Fig. 6A).
In contrast to ChREBP, 20:5,n-3 treatment induced a rapid (at 1.5 h) but transient (ϳ60%) decline in HNF-4␣ interaction with the L-PK promoter (Fig. 6, A and E). The effect of 20:5,n-3 on HNF4␣-L-PK promoter interaction is specific, because similar changes were not detected on the TAT (Fig.  6) or PepCk promoters (not shown).
WY14643 Effects-Treatment of cells with glucose plus WY14643 significantly suppressed mRNA L-PK and L-PK-Luc activity (Fig. 4) and completely suppressed acetylation of histone H4 on the L-PK promoter (Fig. 7, A and B). WY14643 had no effect on the interaction of HNF-4␣ with either the L-PK or TAT promoters (Fig. 7, A and C). Although RNA pol II abundance on the L-PK promoter was not examined, previous studies established that WY14643 strongly suppressed L-PK gene transcription (6).
Taken together, these results indicate that glucose, 20:5,n-3, and WY14643 effects on L-PK gene transcription can be explained by regulation of RNA pol II and histone acetylation status on the L-PK promoter. Effects of these treatments on L-PK expression, however, cannot be explained by regulation of major changes in ChREBP or HNF-4␣ interaction with the L-PK promoter.

Mechanisms Controlling L-PK Promoter Activity
ChREBP, MLX, and HNF-4␣ are important for glucose-regulated L-PK gene transcription. The ChIP analysis, however, failed to provide clues as to how glucose, 20:5,n-3, or WY14643 control RNA pol II inter-action with the L-PK promoter. More subtle mechanisms are likely involved. This section examines three mechanisms as possible explanations for the nutrient control of L-PK gene expression.
As a control study, we first established that AMPK controls L-PK gene expression in primary rat hepatocytes. Metformin and 5-amino-4-imidazole carboxamide ribotide, two AMPK activators, strongly suppressed mRNA L-PK and L-PK promoter activity in rat primary hepatocytes (supplemental Fig. 1S). We next examined AMPK phosphorylation in vivo and in primary rat hepatocytes. Fasting rats overnight elevated cytosolic phospho-AMPK (pAMPK, active form) ϳ4-fold over levels seen in meal-fed rats (supplemental Fig. 2S). Feeding rats n-3 PUFA-or WY14643-supplemented diets did not significantly induce cytosolic pAMPK in liver (in vivo). Nuclear pAMPK levels were induced ϳ30% by fasting and 2-fold by WY14,643. Feeding a fish oil-supplemented diet, however, had no effect on nuclear pAMPK.
In primary rat hepatocytes, time course studies indicated that glucose, 20:5,n-3, or WY14643 did not significantly induce nuclear or cytosolic pAMPK or total AMPK (supplemental Fig. 2S). The results with primary hepatocytes differ from the in vivo studies in two ways. First, pAMPK was not elevated in lactate-treated cells. This result is likely due to the presence of insulin (1 M) in the culture media. Second, WY14643 did not induce nuclear or cytosolic pAMPK in primary hepatocytes, as seen in vivo. This difference is likely due to experimental design. The effect on liver nuclear pAMPK was assessed after 7 days on the WY14643 diet. Livers from WY14643-fed rats were enlarged, i.e. hepatomegaly. Hepatomegaly is commonly seen in rodents fed robust PPAR␣ activators, like WY14643 (44). Rat primary hepatocytes cultured in the absence or presence of WY14643 do not replicate within the 24-h time frame of the experiment. As such, the effect of WY14643 on nuclear pAMPK abundance in vivo is likely not a rapid effect of WY14643 on hepatic function. Taken together, the transient effects of n-3 PUFA and WY14643 on ChREBP and MLX nuclear abundance or HNF-4␣ interaction with the L-PK promoter cannot be explained by enhanced AMPK phosphorylation.
Effects of n-3 PUFA and WY14643 on HNF-4␣1 Transactivation-Treatment of primary hepatocytes with exogenous fatty acids leads to the rapid uptake of fatty acids, conversion to fatty acyl-CoA, and assimilation of fatty acids into complex lipids (45). HNF-4␣ and Ϫ4␥ bind fatty acids and fatty acyl-CoA in vitro (46 -48). HNF-4␣ is implicated as a target of fatty acid regulation because the DR-1 element, which binds HNF-4␣, is required for the n-3 PUFA and WY14643 suppression of LPK gene transcription (6). Moreover, 20:5,n-3 but not WY14643 transiently antagonized HNF-4␣ interaction with the L-PK promoter (Fig.  6). A transfection approach was used to determine whetherHNF-4␣1 transactivation was affected by either fatty acids or WY14643.
This study compared the effects of fatty acids (18:1,n-9 and 20:5,n-3) and WY14643 on two reporter systems as follows: 1) L-PK-Luc and 2) the Gal4-responsive reporter plasmid (TKMH100x4) activated by the Gal4-HNF4␣1 expression vector (Fig. 8). The L-PK-luciferase reporter FIGURE 6. Effects of glucose and 20:5,n-3 on L-PK promoter composition. The ChIP assay was used to examine the composition of the rat L-PK, TAT, FAS, and S14 promoters. HNF-4 binds the TAT enhancer (TAT-Enhancer) Ϫ3062 to Ϫ3579 bp upstream from the transcription start site. The S14 ChoRE is located between Ϫ1.4 and Ϫ1.35 kb from the transcription start site. The fatty-acid synthase ChoRE (FAS-ChoRE) is located between Ϫ7.2 and Ϫ7.1 kb from the transcription start site (Table 1). Rat primary hepatocytes were treated as described in Fig. 5. At the times indicated, cells were treated with formaldehyde for the ChIP assay. Fragmented chromatin was immunoprecipitated (IP) with antibodies against RNA polymerase II (RNA pol II), acetylated histone H3 (Ac-H3), acetylated histone H4 (Ac-H4), ChREBP or HNF-4␣. DNA was extracted from the immunoprecipitated complex, and 5 l of 50 l of purified DNA was used as template for PCR to amplify the L-PK promoter (Ϫ288 to ϩ12 bp) or the TAT enhancer (Ϫ3730 to Ϫ3431 bp), the S14 ChoRE (Ϫ1485 to Ϫ1350 bp) and the FAS-ChoRE (Ϫ7221 to Ϫ7125 bp). Primers for the S14 proximal promoter (Ϫ290 to ϩ 19 bp), the ␤-actin coding (ϩ2383 to ϩ 3091 bp), and the PepCk promoter (Ϫ550 to ϩ 67 bp) were also used ( Table 1). The PCRs for the S14 promoter, ␤-actin coding region, and PepCk promoter did not yield PCR products from chromatin immunoprecipitated with ChREBP antibody. Input, 1% of the sample used for IP was taken out prior to IP to represent total DNA quantity. A, representative PCR products from the input DNA and DNA immunoprecipitated with RNA pol II, Ac-H3, Ac-H4, ChREBP, or HNF-4␣ antibodies. Results of the ChIP assay were quantified (B-E); mean Ϯ S.D. for three independent studies. Glucose-treated cells (filled circles, solid line); glucose ϩ 20:5-treated cells (filled box, dashed line). *, p Ͻ 0.01 versus glucose-treated cells, Student's t test. JULY 7, 2006 • VOLUME 281 • NUMBER 27 gene contains L-PK promoter elements extending to Ϫ197 bp (see Fig.  1 and Fig. 4). Treatment of transfected hepatocytes with glucose (vehicle) or glucose ϩ 18:1,n-9 stimulated L-PK-Luc promoter activity 5-6fold. Treatment with glucose ϩ 20:5,n-3 or glucose ϩ WY14643 attenuated this response by ϳ80%.

Nutrient Control of L-PK Promoter Composition
Transactivation of HNF-4␣1 used an expression vector containing the HNF-4␣1 ligand binding domain (E region) fused to the Gal4-DNA binding domain to create the pM-HNF4␣1-LBD expression vector. The TKMH100x4-Luc reporter plasmid contains Gal4 regulatory elements that bind the Gal4-HNF4␣1-LBD fusion protein.
Effect of Overexpressed ChREBP and MLX on L-PK Gene Expression-n-3 PUFA and WY14643 transiently suppressed ChREBP and MLX nuclear abundance in primary hepatocytes (Figs. 3 and 5). However, in vivo, n-3 PUFA but not WY14643 suppressed MLX nuclear abundance. Glucose, n-3 PUFA, and WY14643 had no effect on HNF-4␣ nuclear abundance in vivo or in rat primary hepatocytes. These studies suggest that transient control of ChREBP or MLX nuclear abundance might be important for the control of L-PK gene expression. To test this possibility, primary hepatocytes were infected with recombinant adenovirus-expressing green fluorescent protein (Ad-GFP) as a control, Ad-MLX, or Ad-ChREBP. After infection, hepatocytes were transfected with L-PK-Luc (Fig. 9A). A second group of cells was infected with recombinant adenovirus but not transfected with L-PK-Luc (Fig. 9, B-E).
In the transfection study, cells were treated with glucose or glucose ϩ 20:5,n-3, or glucose ϩWY14643. In control (Ad-GFP)-infected cells, glucose induced L-PK-Luc activity 5-fold. Infection of cells with Ad-ChREBP induced L-PK-Luc activity 4-fold in hepatocytes cultured in lactate-containing medium. Changing the culture medium to glucose induced L-PK-Luc activity another 4-fold. Changing the culture medium to glucose ϩ 20:5,n-3 suppressed L-PK-Luc activity by Ͼ50% and ϳ90% in control and Ad-ChREBP-infected cells, respectively. Overexpressed ChREBP partially relieved the inhibition of L-PK promoter activity by WY14643 (Fig. 9A). FIGURE 7. Effects of glucose and WY14643 on L-PK promoter composition. The ChIP assay was used to examine the promoter composition of the rat L-PK and TAT promoters following glucose and WY14643 treatment. Rat primary hepatocytes were treated as described in Fig. 4 and harvested for the ChIP assay. Fragmented chromatin was immunoprecipitated (IP) with antibodies against acetylated histone H4 (Ac-H4) and HNF-4␣. DNA was extracted from the immunoprecipitates; 5 l of 50 l of purified DNA was used in a PCR to detect the L-PK promoter (Ϫ288 to ϩ12 bp) and TAT promoter (Ϫ3730 to Ϫ3431 bp). Input, 1% of the samples to be used in IP were taken out prior to IP to represent total DNA quantity. A, representative PCR products form the input and products immunoprecipitated with Ac-H4 or HNF-4␣ antibodies. B and C, PCR products were quantified; mean Ϯ S.D. for three independent studies. Glucose-treated cells (filled circles, solid line); Glucose ϩ WY14643-treated cells (filled box, dashed line). Means Ϯ S.D. for three independent studies. *, p Ͻ 0.01 versus glucose-treated cells, Student's t test. Infection of cells with Ad-MLX induced L-PK-Luc activity 3-fold in hepatocytes maintained in lactate-containing medium. Glucose treatment induced L-PK-Luc activity another 5-fold. Switching the cells to glucose ϩ 20:5,n-3 induced Luc activity 8-fold, whereas switching hepatocytes to glucose ϩ WY14643 did not induce L-PK-Luc activity.
Fatty-acid synthase and S14 are two well established glucose-and fatty acid-regulated genes (4,19,28). The effect of overexpressed ChREBP and MLX on glucose and fatty acid on control of mRNA FAS and mRNA S14 was compared with the effect on mRNA LPK (Fig. 9, C-E). Overexpressed ChREBP induced mRNA LPK , mRNA FAS , and mRNA S14 by 2-and 2-4-fold in hepatocytes cultured in glucose ϩ insulin (not shown). Switching the medium to glucose ϩ 20:5,n-3 suppressed mRNA LPK , mRNA FAS , and mRNA S14 by ϳ50% in cells infected with Ad-GFP or Ad-ChREBP recombinant adenovirus. Overexpressed ChREBP did not relieve the 20:5,n-3 suppression of L-PK, FAS, or S14.
These results indicate that MLX is a target for 20:5,n-3 control of L-PK gene expression. For genes like S14 and FAS that also utilize SREBP-1 as a key transcription factor (2,49), changes in MLX abundance alone cannot relieve 20:5,n-3 suppression.

DISCUSSION
L-PK plays a central role in hepatic carbohydrate and lipid metabolism. Insulin-stimulated glucose metabolism induces L-PK gene expression, L-PK enzyme activity, and the flow of glucose metabolites toward fatty acid synthesis and storage. The inhibitory effect of dietary n-3 PUFA on L-PK gene transcription is part of a larger regulatory network that shifts hepatic metabolism away from lipid synthesis and storage and toward fatty acid oxidation (2). Glucose and n-3 PUFA manage these metabolic pathways by controlling the activity or nuclear abundance of key transcription factors, including ChREBP, MLX, SREBP-1, and PPAR␣. WY14643 and n-3 PUFA activate PPAR␣ by direct binding (50). n-3 PUFA suppresses SREBP-1 nuclear abundance by multiple mechanisms, including enhanced 26 S proteasomal degradation (2,51). Neither PPAR␣/RXR␣ nor SREBP-1 bind the L-PK promoter; neither transcription factor regulates L-PK expression directly (6,49,52). Until FIGURE 9. Effect of overexpressed ChREBP and MLX␤ on L-PK-Luc activity, mRNA L-PK mRNA S14 , and mRNA FAS . A, rat primary hepatocytes were infected with recombinant adenovirus expressing green fluorescent protein (GFP, Ad-GFP, control), ChREBP (Ad-ChREBP), or MLX (Ad-MLX␤) while in Williams E medium containing lactate and insulin. GFP is expressed from an internal ribosome entry site in the Ad-ChREBP and Ad-MLX adenovirus. Expression levels of GFP in primary hepatocytes were verified by fluorescence microscopy. Twenty four hours after infection, cells were transfected withL-PK-Luc and phRG-Luc overnight. The next day cells were maintained in Williams E containing lactate ϩ insulin or switched to Williams E ϩ glucose ϩ insulin without or with 20:5,n-3 or WY14643. Cells were harvested 24 h later for luciferase activity. Results are expressed as fold change in relative luciferase activity (RLA); means Ϯ S.D., n ϭ 3. Results are representative of three separate studies with triplicate samples in each study. B and C, rat primary hepatocytes were infected with recombinant adenovirus as described above. Twenty four hours after infection, cells were switched from the lactate containing medium to the glucose containing medium in the absence or presence of 20:5,n-3 or WY14643. Cells were harvested 24 h later for RNA extraction and measurement of L-PK, FAS, S14, and cyclophilin mRNA abundance by qRT-PCR. The abundance of L-PK, FAS, or S14 mRNA relative to cyclophilin mRNA was calculated by the C t method. B, results are quantified and expressed as fold change in mRNA L-PK -induced virus infection and by glucose, glucose ϩ 20:5, or glucose ϩ WY14643. Results are normalized to the level of mRNA L-PK in the lactate-treated cells infected with Ad-GFP. Results are the means of two separate studies. C-E, the effect of 20:5,n-3 on L-PK, FAS, and S14 mRNA in primary hepatocytes infected with Ad-GFP, Ad-ChREBP, or Ad-MLX was quantified. Cells were treated as above. Results are reported fold change in mRNA; results are normalized to the transcript level in glucose-treated-Ad-GFP-infected cells. Results are the means Ϯ range of duplicate studies. Results are representative of two independent studies. *, p Ͻ 0.05 versus glucose-treated cells, Student's t test. now, the mechanism by which n-3 PUFA and WY14643 controlled L-PK gene transcription remained elusive. This study presents new information to explain how glucose, n-3 PUFA, and WY14643 regulate hepatic L-PK gene transcription through effects on ChREBP and MLX nuclear abundance and L-PK promoter composition.
Our analysis has yielded both expected and unexpected results. As expected, fasting and refeeding rats or treating rat primary hepatocytes with glucose induced a robust accumulation of ChREBP in hepatic nuclei and the induction of mRNA L-PK (Figs. 2 and 5). Prior to glucose activation of L-PK gene transcription, the L-PK promoter has no detectable RNA pol II, Ac-H3, or Ac-H4. Surprisingly, however, both ChREBP and HNF-4␣ are bound to the L-PK promoter. Glucose treatment of rat primary hepatocytes stimulates ChREBP accumulation in nuclei (Fig. 5), recruitment of RNA pol II to the L-PK promoter, the acetylation of histones H3 and H4 (Fig. 6), and L-PK promoter activity (Fig. 4). During the time course of L-PK gene activation, ChREBP undergoes modest changes in occupancy on the L-PK promoter. HNF-4␣ interaction with the LPK promoter remains unchanged. These results indicate that factors other than the mere presence of ChREBP on the L-PK promoter are involved in glucose-mediated induction of L-PK gene expression. 20:5,n-3 and WY14643 suppressed L-PK promoter activity and inhibited RNA pol II recruitment to the L-PK promoter and the acetylation of histone H3 and H4 on the LPK promoter (Figs. 6 and 7). 20:5,n-3 and WY14643 transiently suppressed ChREBP and MLX nuclear abundance at 6 and 1.5 h but did not significantly affect ChREBP interaction for 6 h after initiating glucose treatment. Only after 24 h was a decline in ChREBP abundance on the L-PK promoter observed. HNF-4␣ interaction with the L-PK promoter was also transiently suppressed by 20:5,n-3 but not WY14,643. Clearly, the interaction of ChREBP and HNF-4␣ with the L-PK promoter is independent of the presence of RNA pol II, acetylated histones H3 and H4, or the transcriptional status of the L-PK gene.
The dissociation between glucose-stimulated ChREBP nuclear abundance and the lack of significant change in ChREBP occupancy on the L-PK promoter have at least two explanations. First, our measure of ChREBP and HNF-4␣ binding to the L-PK promoter is an artifact. Our control studies, however, validated the ChIP assay for ChREBP and HNF-4␣ (Fig. 6). Second, the ChIP assay when applied to the measure of endogenous proteins on chromatin does not reveal an exchange of one protein for another; it only detects the presence or absence of a protein on chromatin. During the 90-min lag preceding RNA pol II recruitment to the L-PK promoter, glucose stimulated a robust (5-fold) translocation of ChREBP to the nucleus (Fig. 5). This translocation is likely stimulated by glucose metabolism and the generation of xylulose 5-phosphate, a PP2A activator. Dephosphorylated ChREBP moves to the nucleus (22,24). The influx of nascent ChREBP into the nucleus, coupled with its heterodimerization with MLX, likely remodels the L-PK promoter with nascent ChREBP/MLX heterodimers and HNF-4␣ homodimers. Binding of these nascent proteins to the L-PK promoter may induce recruitment of RNA pol II to the promoter and the acetylation of histones H3 and H4. Unfortunately, the differences in ChREBP abundance on the L-PK promoter over this time course did not reach significance.
Gene activation by ligand-activated nuclear receptors involves remodeling of promoter composition (53,54). This process requires a change in the covalent modification status of chromatin-associated proteins. Ubiquitination of transcription factors and their 26 S proteasomal degradation has emerged as an important mechanism controlling promoter composition (53)(54)(55). Inhibitors of the 26 S proteasome significantly affect gene activation. In this regard, preliminary studies with MG132, a 26 S proteasome inhibitor, attenuated glucose activation of L-PK gene transcription (not shown). Whether any proteins on the L-PK promoter are ubiquitinated remains to be determined.
Glucose-stimulated exchange of proteins on the L-PK promoter may not fully explain why L-PK promoter activity is low despite the presence of ChREBP on the promoter. It is equally possible that ChREBP heterodimerization with other basic helix-loop-helix proteins yields a complex that fails to recruit RNA pol II and the associated co-regulators required for gene activation. Although ChREBP binding to ChoRE requires MLX (39), a role for other basic helix-loop-helix proteins forming heterodimer partners with ChREBP has not been excluded.
n-3 PUFA, like 20:5,n-3, inhibits glucose-stimulated L-PK promoter activity and the accumulation of mRNA L-PK by attenuating RNA pol II recruitment to the L-PK promoter and the acetylation of histone H3 and H4 bound to the L-PK promoter. 20:5,n-3 inhibition of RNA pol II recruitment to the L-PK promoter cannot be explained by a major suppression of ChREBP nuclear abundance or its interaction with the L-PK promoter (Figs. 5 and 6). 20:5,n-3 only transiently suppresses ChREBP nuclear abundance (at 6 h) and has no effect on hepatic ChREBP nuclear abundance in vivo (Figs. 3 and 5). Overexpression of ChREBP in rat hepatocytes stimulates L-PK gene expression but does not abrogate 20:5,n-3 suppression of L-PK expression.
In contrast, MLX nuclear abundance is suppressed by n-3 PUFA in vivo (Fig. 3) and is transiently down-regulated at 1.5 h by 20:5,n-3 (Fig.  5). Overexpression of MLX fully abrogates 20:5,n-3 suppression of L-PK gene expression. This finding identifies MLX as a novel target for n-3 PUFA regulation.
Although HNF-4␣ interaction with the L-PK promoter is transiently regulated by 20:5,n-3, we consider this event to be independent of direct effect of fatty acids on HNF-4␣, per se. We base this view on the absence of any effect of fatty acids on HNF-4␣1 transactivation (Fig. 8) or HNF-4␣ interaction with other HNF4-regulated promoters (TAT or PepCk) (Fig. 6) and the fact that other HNF4-regulated genes (TAT and PepCk) are not affected by n-3 PUFA (2, 3). We suspect these transient changes in HNF-4␣ interaction with the L-PK promoter reflect remodeling of L-PK promoter composition during the course of glucose activation.
WY14643 also strongly inhibits glucose-stimulated L-PK gene transcription (6), the acetylation of histone H4 on the L-PK promoter ( Fig.  7), and the accumulation of mRNA L-PK (Fig. 4). Like 20:5,n-3, WY14643 only modestly affects glucose-stimulated ChREBP translocation into the nucleus (Fig. 5). WY14643 had no effect on hepatic ChREBP nuclear abundance in vivo. WY14643 also had no effect on nuclear MLX levels in liver or primary hepatocytes or the interaction of HNF-4␣ with the L-PK promoter (Figs. 3, 5, and 6). WY14643, however, significantly antagonized HNF-4␣1 transactivation (Fig. 8). Overexpressed ChREBP, but not overexpressed MLX, partially reversed the WY14643 suppression of L-PK promoter activity and mRNA L-PK (Fig. 9). HNF-4␣ regulates many hepatic genes, but only a few are inhibited by strong PPAR␣ agonists (WY14643) like L-PK (6), apolipoprotein CIII (56), and Cyp7A (57). A likely explanation for this effect is that PPAR␣ interferes with co-regulator interaction with HNF-4␣ on the L-PK promoter. A test of this hypothesis will first require identifying the co-regulators that interact with the L-PK promoter.
How can n-3 PUFA induce rapid but transient changes in ChREBP and MLX nuclear abundance or HNF-4␣ interaction with the L-PK promoter? A likely scenario is that n-3 PUFA regulates a signal transduction pathway(s). Two groups have reported that fatty acid activation of AMPK may account for fatty acid effects on gene expression (24,43). AMPK regulates ChREBP entry into the nucleus. Once in cells, fatty acids are rapidly assimilated into complex lipids. A key event in this metabolic scheme is the conversion of fatty acids to fatty acyl-CoA, a reaction that requires ATP. Exogenous 20:5,n-3 is rapidly converted to 20:5-CoA in hepatocytes and rapidly assimilated into complex lipids, predominantly as triglycerides (45). We (see the Supplemental Material) and others (58,59) found no evidence for fatty acid effects on AMPK phosphorylation in primary hepatocytes. Fatty acid treatment of primary hepatocytes, however, induces p38 MAPK (60) and ERK phosphorylation (51). We found no effect of p38 MAPK inhibitors on L-PK gene expression (not shown). 22:6,n-3, but not 20:5,n-3, induces ERK phosphorylation (51). We speculate that 20:5,n-3 regulates other signal transduction pathways that impact MLX and ChREBP nuclear abundance and HNF-4␣ interaction with the L-PK promoter. The identity of these other pathways remains to be established.
During the course of our studies, Dentin et al. (58) reported that PUFA interfered with ChREBP translocation to the nucleus in mouse liver and hepatocytes. Our in vivo and primary hepatocytes studies, however, show no effect of n-3 PUFA on ChREBP nuclear abundance in vivo and only a transient effect in primary hepatocytes (Figs. 3 and 5). Several factors can account for the different outcomes reported by Dentin et al. (58) and herein. First, rat primary hepatocytes were purified through Percoll (Amersham Biosciences), whereas Dentin et al. (58) used unfractionated mouse hepatocytes. Rats were meal-fed high carbohydrate diets supplemented with olive or fish oil for 7 days. Animals were euthanized 2 h after completion of the last meal. Dentin et al. (58) maintained mice on a high carbohydrate fat-free diet and switched the animals to a high carbohydrate diet supplemented with fatty acids; animals were euthanized 18 h later. Finally, Dentin et al. (58) did not examine hepatic MLX levels. Animal species or experimental design may account for these differences. Our results indicate that PUFA control of ChREBP is not the sole mechanism for fatty acid regulation of L-PK. MLX has emerged as a target for n-3 PUFA control.
As a result of these studies, we have revised our original interpretation for how n-3 PUFA and WY14643 control L-PK gene transcription. Our earlier studies were based on linker-scanning analysis and indicated that the DR-1 element was required for both WY14643 and n-3 PUFA control of L-PK gene transcription (5, 6) (Fig. 1). This interpretation, however, is confounded by fact that both n-3 PUFA and WY14643 only inhibit glucose-activated L-PK expression and not basal L-PK expression. HNF-4␣ is not a direct target of n-3 PUFA (Fig. 8). n-3 PUFA, however, only transiently affects HNF-4␣ interaction with the L-PK promoter (Fig. 6). In contrast, WY14643 interferes with HNF-4␣1 transactivation (Fig. 8) but has no effect on HNF-4␣ interaction with the L-PK promoter (Fig. 7). A likely explanation for the WY14643 effect is that ligand-activated PPAR␣ interferes with co-activator recruitment to the L-PK promoter. The robust inhibition of histone H4 acetylation on the L-PK promoter supports this concept (Fig. 7). In revision of our original model (Fig. 1), we now show that the PUFA and PPAR␣ cis-regulatory targets should include the ChREBP/MLX and HNF-4␣-binding sites.
The new information reported herein has raised many unanswered questions. First among these is how glucose treatment of hepatocytes induces RNA pol II and histone acetylation on the L-PK promoter. The lack of significant glucose-regulated ChREBP occupancy on the L-PK promoter suggests that something other than ChREBP accumulation in the nucleus is required to induce L-PK gene activation. Finding that the 26 S proteasome inhibitor, MG132, blocks glucose-activated L-PK gene transcription implicates the 26 S proteasome in this regulatory scheme. Analysis of the ubiquitination status of ChREBP, HNF-4␣, and MLX will be important to define how nutrients control L-PK promoter composition and gene transcription. The suppressive effect of n-3 PUFA on MLX nuclear abundance without effects on MLX expression implicates mechanisms that control MLX nuclear import or export. Another heterodimer partner of MLX is MondoA. The subcellular distribution of MondoA:MLX heterodimers is determined by CRM1-dependent nuclear export signals in MondoA that interact with 14-3-3 family members. Leptomycin B, a nuclear export inhibitor, promotes Mon-doA:MLX accumulation in nuclei (61). Leptomycin B, however, does not prevent the n-3 PUFA suppression of MLX abundance. 3 We speculate that n-3 PUFA transiently inhibits MLX import into the nucleus. The mechanism underlying this process requires further analysis.
In summary, glucose stimulates ChREBP translocation into hepatocyte nuclei, recruitment of RNA pol II to the L-PK promoter, and increased histone H3 and H4 acetylation on the L-PK promoter without major changes in promoter-bound ChREBP or HNF-4␣. 20:5,n-3 interferes with this process by transiently suppressing nuclear MLX and ChREBP nuclear abundance. MLX is an obligate heterodimer partner of ChREBP that is required for ChREBP binding to the L-PK ChoRE. Overexpressed MLX eliminates the 20:5,n-3 but not the WY14643 control of L-PK gene expression. Overexpressed ChREBP relieves the WY14643 suppression of L-PK promoter activity but has no effect on n-3 PUFA inhibition of L-PK promoter activity. WY14643 encumbers HNF-4␣ transactivation but sn-3 PUFA does not. Overexpressed HNF-4␣ has no effect on n-3 PUFA or WY14643 suppression of L-PK gene expression. These studies have identified MLX as a key target for n-3 PUFA control of L-PK gene expression. n-3 PUFA and WY14643 clearly use distinct mechanisms to inhibit glucose-activated L-PK gene transcription.