Transcriptional Suppression of the Transferrin Gene by Hypolipidemic Peroxisome Proliferators*

Activation of gene expression by hypolipidemic per- oxisome proliferators ( e.g. native and substituted long chain fatty acids, aryloxyalkanoic fibrate drugs) is ac- companied by transcriptional suppression of liver transferrin gene in treated animals or human hepatoma cell line. Transcriptional suppression of liver transferrin by hypolipidemic peroxisome proliferators re- sults from ( a ) displacement of hepatic nuclear factor (HNF)-4 from the transferrin promoter by nonproduc- tive binding of the peroxisome proliferator-activated re-ceptor-retinoic acid X receptor heterodimer to the ( (cid:50) 76/ (cid:50) 52) PRI promoter element of the human transferrin gene and ( b ) suppression of liver HNF-4 gene expression by hypolipidemic peroxisome proliferators with a concomitant decrease in its availability for binding to the transferrin PRI promoter element. HNF-4 gene suppression and its displacement from the transferrin promoter result in eliminating HNF-4-enhanced transcription of transferrin. Liver transferrin suppression by hypolipi- demic peroxisome proliferators may result in reduced iron availability as well as modulation of transferrin- induced differentiation processes. Transcriptional suppression of HNF-4-enhanced liver genes ( e.g. apolipopro- tein C-III, transferrin) may complement the pleiotropic biological effect exerted by hypolipidemic peroxisome proliferators. Transferrin and CAT activities. Results ex-*

Activation of gene expression by hypolipidemic peroxisome proliferators (e.g. native and substituted long chain fatty acids, aryloxyalkanoic fibrate drugs) is accompanied by transcriptional suppression of liver transferrin gene in treated animals or human hepatoma cell line. Transcriptional suppression of liver transferrin by hypolipidemic peroxisome proliferators results from (a) displacement of hepatic nuclear factor (HNF)-4 from the transferrin promoter by nonproductive binding of the peroxisome proliferator-activated receptor-retinoic acid X receptor heterodimer to the (؊76/ ؊52) PRI promoter element of the human transferrin gene and (b) suppression of liver HNF-4 gene expression by hypolipidemic peroxisome proliferators with a concomitant decrease in its availability for binding to the transferrin PRI promoter element. HNF-4 gene suppression and its displacement from the transferrin promoter result in eliminating HNF-4-enhanced transcription of transferrin. Liver transferrin suppression by hypolipidemic peroxisome proliferators may result in reduced iron availability as well as modulation of transferrininduced differentiation processes. Transcriptional suppression of HNF-4-enhanced liver genes (e.g. apolipoprotein C-III, transferrin) may complement the pleiotropic biological effect exerted by hypolipidemic peroxisome proliferators.
Transferrin (Tf) 1 is highly expressed in the adult mammalian liver and is secreted by hepatocytes into the serum where it functions as an iron transport protein and growth factor for a variety of cells (reviewed in Refs. 1 and 2). Liver Tf expression was reported to be activated by steroid hormones (3) and iron deficiency (3,4). Tf is synthesized to a lower extent by Sertoli cells in the adult testis as well as by adult brain oligodentrocytes, astrocytes, and epithelial cells of the choroid plexus, where it is involved in the maturation of germinal cells and in central nervous system proliferation and differentiation processes.
In experiments to be reported elsewhere, 2 serum iron, iron binding capacity and plasma Tf were found to be 50% reduced in rats treated by xenobiotic amphipathic carboxylates (e.g. aryloxyalkanoic acids (bezafibrate), substituted long chain dicarboxylic acids (Medica 16)) known collectively as hypolipidemic drugs/peroxisome proliferators (HD/PPs) (reviewed in Ref. 5). HD/PPs have previously been reported to activate the expression of a variety of discrete genes (e.g. peroxisomal ␤-oxidation genes (6), -oxidation P450IV genes (7), liver genes coding for thyroid hormone-dependent activities (8), and others) as a result of transcriptional activation mediated by binding of peroxisome proliferators-activated receptors (PPARs) to sequence-specific PPAR-activated response elements (PPREs) in the respective promoters (9 -13). 3 PPAR binding to PPREs requires the retinoic acid X receptor (RXR) for forming the high affinity PPAR⅐RXR heterodimer (14). The putative binding of HD/PPs to PPARs and the role of HD/PPs in initiating the binding of PPAR/RXR to PPREs still remains to be investigated (9).
Since some HD/PPs are extensively used in humans as hypolipidemic drugs (15) and since transcriptional suppression, rather than activation, mediated by the HD/PP-PPAR/RXR-PPRE transduction pathway may complement the pleiotropic biological effect exerted by HD/PPs, we became interested in elucidating the mode of action of HD/PPs as putative suppressors of liver Tf. Liver Tf gene suppression by HD/PPs will be shown here to be mediated by PPAR/RXR and to involve the HNF-4 enhancer element of the Tf gene promoter.

EXPERIMENTAL PROCEDURES
Animals and Cultures-Male albino rats weighing 150 -200 g were fed with laboratory chow diet. 0.25% (w/w) Medica 16 was added to their diet where indicated. Hep G2, CV-1, and COS-7 cells were cultured in Dulbecco's modified Eagle's media supplemented with 10% fetal calf serum with either Me 2 SO as vehicle or 120 M Medica 16 added to the culture medium where indicated.
Run-on Transcription-Hep G2 nuclei were prepared according to Ref. 16. Run-on transcription assays were carried out as described previously (17).
Transfection Assays-Transcriptional activity was measured in cells cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum. Cells were transfected for 6 h with the respective CsCl-purified plasmid DNA added by calcium phosphate precipitation, washed, and further cultured for 42 h in the absence or presence of 120 M of Medica 16 added as 1000ϫ stock in dimethyl sulfoxide. The ␤-galactosidase expression vector pRSGAL (2 g) added to each precipitate served as an internal control for transfection. When transfected with variable amounts of expression vectors, total amount of DNA was kept constant for each expression vector by supplementing with the parent pSG5 vector (18). Cell extracts were prepared by freeze-thawing and assayed for ␤-galactosidase and CAT activities. Results are ex-  1 The abbreviations used are: Tf, transferrin; CAT, chloramphenicol acetyltransferase; HD/PP, hypolipidemic drug(s)/peroxisome proliferator; HNF-4, hepatic nuclear factor-4; PPAR, peroxisome proliferatoractivated receptor; PPRE, PPAR-activated response element; RXR, retinoic acid X receptor; TK, thymidine kinase. pressed as fold induction relative to CAT expression observed in cells transfected with the parent pSG5 vector. Each point represents the mean of duplicate cultures differing by no more than 10%.
Gel Mobility Shift Assays-Gel mobility shift assays were carried out using rat nuclear extracts (19), in vitro synthesized transcription factors, or whole COS extracts overexpressing the respectively transfected expression vectors. hRXR and rHNF-4 cDNAs cloned in pSG5 were linearized by XbaI and transcribed (Stratagene) and translated in rabbit reticulocytes (Promega). COS extracts enriched with PPAR, RXR, or HNF-4 were prepared from COS-7 cells transfected for 5 h by calcium phosphate precipitation with 10 g of pSG5, pSG5-PPAR, pSG5-RXR, pSG5-HNF-4, or selected combinations of the above plasmids. Following transfection, cells were incubated for 48 h, harvested, lysed by three cycles of freezing-thawing in 100 l/plate of lysis buffer (600 mM KCl, 10 mM Tris-HCl, pH 7.5, 1 mM dithiothreitol, 1 mM EDTA, 1 mM phenylmethylsulfonyl fluoride, 1 g/ml pepstatin, and 1 g/ml leupeptin), and centrifuged at 20,000 g for 15 min; the supernatants were then aliquoted and stored at Ϫ70°C. For gel shift assays, programmed or unprogrammed reticulocyte lysates (2 l) or whole COS extracts (4 g) as indicated were incubated for 20 min on ice in 11 mM Hepes (pH 7.9) containing 50 mM KCl, 1 mM dithiothreitol, 2.5 mM MgCl 2 , 10% glycerol, 1 g of poly(dI-dC) in a final volume of 20 l. 0.1 ng of the respective 32 P-labeled oligonucleotide was then added, and incubation was continued for an additional 20 min at room temperature. Protein-DNA complexes were resolved on a 5% non-denaturing polyacrylamide gel in 0.5 ϫ TBE.

Suppression of Tf Transcription by HD/PP-
The 50% reduction in plasma Tf in rats treated with Medica 16 2 was accompanied by a respective decrease in liver rTf mRNA (Fig. 1A), thus indicating that the HD/PP effect could be accounted for by pretranslational inhibition of Tf expression. A decrease in Tf mRNA was similarly observed in human transformed Hep G2 liver cells incubated in the presence of Medica 16 (Fig. 1B), thus pointing to the direct effect exerted by HD/PPs on cells expressing Tf.
Suppression of liver Tf mRNA by Medica 16 was accompanied by inhibition of Tf transcription rate as verified by run-on transcription assays in Hep G2 cell nuclei (Fig. 2). Incubation in the presence of Medica 16 for 72 h resulted in a pronounced inhibition of Tf transcription rate, indicating that reduction in Tf mRNA exerted by Medica 16 treatment may be ascribed to suppression of transcription of liver Tf gene by Medica 16. Essentially similar decreases in Tf mRNA and Tf transcription rates were observed using bezafibrate (not shown), thus generalizing the observed effect to other members of the HD/PP class of compounds.
Tf Transcriptional Suppression Is Mediated by PPAR/RXR Binding to the Tf Proximal Promoter-Transfection experiments in hepatoma cells using 5Ј-and 3Ј-deleted mutants of the human Tf gene promoter have previously implicated the first 620 nucleotides of the Tf promoter in regulating the expression of the Tf gene in liver cells (22,27,28). These studies, complemented by expression from the mouse Tf gene promoter in transgenic mice (29,30), by in vitro transcription assays using the G-free cassette system (31), as well as by in vitro binding assays using liver nuclear extracts (22,32), have indicated that three distinct functional regions were involved in regulating liver Tf transcription by the (Ϫ620/ϩ1) hTf promoter, namely, the (Ϫ76/Ϫ51)Tf-PRI element, the (Ϫ103/Ϫ83) Tf-PRII element, and an upstream promoter region within the (Ϫ620/ Ϫ125) Tf promoter sequence, which modulates the activity of  (17). Newly synthesized 32 P-RNA was hybridized with the hTf cDNA plasmid linearized with BamHI restriction enzyme. The extent of hybridization was normalized to the signal obtained with poly(A) binding protein cDNA. Representative experiment is shown out of three independent experiments. the two proximal promoter elements. Tf gene suppression by HD/PP was therefore further analyzed in Hep G2 cells transfected with the (Ϫ620/ϩ39) hTf promoter linked to a CAT reporter gene and incubated in the presence and absence of added Medica 16. The putative involvement of PPAR and RXR in the Medica 16 effect was verified by cotransfecting the cells with expression vectors for PPAR and RXR, respectively. As shown in Fig. 3, A and B, CAT expression from the (Ϫ620/ ϩ39)hTf-CAT construct was 25-50% inhibited by incubating the cells in the presence of added Medica 16. CAT expression in the presence of Medica 16 was further inhibited by cotransfecting the cells with expression vectors for both PPAR and RXR, indicating that the PPAR⅐RXR heterodimer was involved in Tf gene suppression by Medica 16. Inhibition of CAT expression by Medica 16 in the absence of cotransfected PPAR and RXR may therefore be accounted for by endogenous PPAR and RXR in Hep G2 cells. It is noteworthy that similarly to other genes affected by the PPAR/RXR transduction pathway, cotransfecting the cells with expression vectors for PPAR and RXR resulted in inhibition of CAT expression from the (Ϫ620/ϩ39)hTf-CAT construct even in the absence of added Medica 16. Ligandindependent transcriptional modulation of Tf (Fig. 3A) and other genes (11-13) 3 by PPAR/RXR might reflect the presence of an endogenous PPAR activator or the constitutive capacity of the PPAR⅐RXR heterodimer to modulate transcription of concerned genes. The role played by Medica 16 in Tf gene suppression by PPAR/RXR was better exemplified using PPAR-G (12) (Fig. 3B), having a lower constitutive (ligand-independent) activity as compared with PPAR due, perhaps, to its lower affinity for the endogenous HD/PP ligand. These results may therefore indicate that transcriptional suppression of Tf by Medica 16 is mediated by PPAR/RXR affecting an element within the (Ϫ620/ϩ39) Tf promoter sequence.
The promoter element involved in transcriptional suppres- sion of the Tf gene by PPAR/RXR and Medica 16 was further characterized by transfecting Hep G2 cells with 5Ј-deleted constructs of the Tf promoter linked to CAT and analyzing the effect exerted by cotransfected PPAR/RXR and added Medica 16 on expression of the concerned constructs. As shown in Fig.  3C, CAT expression promoted either by the (Ϫ620/ϩ39)hTf or (Ϫ125/ϩ39)hTf promoter sequences, which consist of the Tf-PRI as well as the Tf-PRII elements, was suppressed by cotransfected PPAR/RXR in the absence or presence of Medica 16, thus pointing to a PPAR/RXR responsive element within the (Ϫ125/ϩ39)hTf promoter sequence. Deleting the Tf-PRII element resulted in a 7-fold decrease in CAT expression, reflecting the synergistic role played by Tf-PRI and -II in Tf transcription. However, transcriptional suppression by PPAR/ RXR was still maintained in the (Ϫ82/ϩ39)hTf-CAT construct. On the other hand, deleting both the PRI and PRII elements resulted in loss of suppression by PPAR/RXR, thus indicating that transcriptional suppression of the Tf gene by PPAR/RXR involves the Tf-PRI element.
The role played by the Tf-PRI element in transcriptional suppression of the Tf gene by PPAR/RXR was further verified in Hep G2 cells transfected with (Ϫ620/ϩ39)Tf-CAT constructs mutated in either the PRI ((Ϫ620/ϩ39)hTf(PRI mut)-CAT) or PRII ((Ϫ620/ϩ39)hTf(PRII mut)-CAT) elements (22) and cotransfected with expression vectors for PPAR and RXR. As shown in Fig. 3D, expression from the Tf-PRII mutated construct was 50% inhibited as compared with the respective wild type construct. However, the PRII mutation did not interfere with transcriptional suppression by PPAR/RXR. Expression from the Tf-PRI mutated construct was 90% inhibited as compared with the wild type construct and could not be further suppressed by PPAR/RXR. Hence, suppression of the liver Tf gene by PPAR/RXR appears to be specifically mediated by the Tf-PRI element.
Transfection studies were complemented by studying binding of PPAR and/or RXR to the hTf-PRI sequence using mobility shift analysis. As shown in Fig. 4A, the PPAR⅐RXR heterodimer transcribed and translated in vitro in rabbit reticulocytes, but not the respective individual receptors, indeed binds to the hTf-PRI element. Similarly, PPAR derived from transfected COS cells (Fig. 4B) and complemented by endogenous or transfected RXR specifically binds to the hTf-PRI element and may be supershifted by anti-mPPAR antibody.
HNF-4 Displacement from the hTf-PRI Element by PPAR/ RXR-The hTf gene in the liver system has recently been reported to be transactivated by HNF-4 and to bind HNF-4 to the Tf-PRI element (22). Transcriptional suppression of the Tf gene by PPAR/RXR mediated by the Tf-PRI element (Fig. 3) could therefore be ascribed to PPAR/RXR interference with HNF-4 transcriptional activation of the liver Tf gene.
PPAR/RXR interference with HNF-4 binding to the Tf-PRI element was studied by comparing the Tf-PRI binding affinities of HNF-4 and PPAR/RXR using gel shift assays. As shown in Fig. 5, both HNF-4 and PPAR/RXR could bind to the hTf-PRI element sequence with apparent binding affinities of 2.7 Ϯ 0.3 and 3.6 Ϯ 0.6 nM hTf-PRI (mean Ϯ S.D. for three independent experiments), respectively, thus indicating that the two receptors could compete for binding to the concerned element. Their functional interaction with the hTf-PRI element in the context of the heterologous thymidine kinase promoter was analyzed in CV-1 cells transfected with a hTf-PRI-TK-CAT construct consisting of the hTf-PRI element in front of the thymidine kinase promoter (Fig. 6). Cotransfecting these cells with either HNF-4 or PPAR/RXR resulted in 10-and ϫ4-fold activation of CAT expression, indicating that PPAR/RXR binding to the hTf-PRI element in the context of the heterologous TK promoter may result in its transactivation. The functional relationship between PPAR/RXR and HNF-4 within the context of the homologous hTf promoter was studied in CV-1 cells transfected with the (Ϫ125/ϩ39)hTf-CAT construct and cotransfected with expression vectors for HNF-4, PPAR, and RXR. As shown in Fig. 7, HNF-4, but not PPAR/ RXR, activated CAT expression promoted by the homologous (Ϫ125/ϩ39)hTf proximal promoter. Furthermore, cotransfecting the cells with both HNF-4 and PPAR/RXR resulted in eliminating HNF-4 activation of the Tf gene. Hence, PPAR/ RXR binding to the hTf-PRI element in the context of the homologous Tf promoter is nonproductive but interferes with HNF-4 productive binding to this element, resulting in inhibition of HNF-4-enhanced transcriptional transactivation of the liver Tf gene.
HNF-4 Suppression by HD/PP-HD/PPs have recently been reported by us to suppress liver HNF-4 mRNA and protein in HD/PP-treated rats and Hep G2 cells (17). Since HNF-4 is a major liver nuclear transcription factor, which may bind and transactivate Tf gene transcription, its reduced liver content in HD/PP-treated rats could result in its lower availability for binding to the Tf-PRI element with a concomitant reduced expression of the liver Tf gene in treated animals. This mode of action of Medica 16 could complement the direct suppressive mode mediated by HNF-4 displacement from the Tf-PRI element by PPAR/RXR. The availability of HNF-4 for binding and transactivating Tf gene transcription in nuclear extracts derived from Medica 16-treated rats was verified by analyzing the extent of binding of liver nuclear extracts of treated and nontreated rats to the hTf-PRI element. Bound HNF-4 in the respective extracts was identified by the extent of HNF-4 supershifted by added anti-HNF-4 antibodies. As shown in Fig. 8, nuclear extract binding to the Tf-PRI element was significantly reduced in Medica 16-treated rats and specifically accounted for by the lower availability of HNF-4 in nuclear extracts of treated animals. Hence, in addition to HNF-4 displacement from the Tf promoter by PPAR/RXR binding, HNF-4 suppression by Medica 16 may be pivotal to the overall suppressive effect of Medica 16 on Tf gene transcription. DISCUSSION The decrease in plasma Tf observed in rats treated with hypolipidemic peroxisome proliferators was shown here to be accompanied by a decrease in liver Tf mRNA and to result from transcriptional suppression of the liver Tf gene as verified by run-on transcription assays in liver nuclei derived from Hep G2 cells. Since the level of plasma Tf is dominated by liver Tf expression and secretion, transcriptional suppression of liver Tf by HD/PP may account for the reduced plasma Tf levels in treated animals. Furthermore, since Tf transcriptional suppression was similarly observed in human Hep G2 cells incubated in the presence of added HD/PP, Tf suppression could be relevant to dyslipoproteinemic patients treated with HD/PP.
Transcriptional suppression of liver Tf gene by Medica 16 was found here to be related to HNF-4-enhanced transcription of the Tf gene and ascribed to displacement of HNF-4 from the Tf promoter by nonproductive binding of PPAR/RXR to the Tf-PRI (HNF-4 enhancer) element. The PPAR⅐RXR heterodimer behaves in this respect similarly to other previously reported transcription factors, e.g. ARP-1, which may compete with HNF-4 for binding to HNF-4 enhancer elements (33,34). The extent of inhibition of Tf transcription by PPAR/RXR in a specific cell type may therefore be expected to reflect the pre- vailing content of concerned transcription factors and their respective binding affinities for the Tf-PRI element. Since the liver system is highly enriched in HNF-4 as compared with chicken ovalbumin upstream promoter transcription factor or ARP-1 (34), transfection with PPAR/RXR, or activating the endogenous PPAR/RXR by HD/PP or using both intervention modes results in Tf suppression. However, in cell types where the Tf-PRI function is dominated by suppressive transcription factors, HD/PP may exert an apparent transactivation of transcription mediated by displacing the concerned suppressive transcription factors by nonproductive binding of PPAR/RXR. The resultant effect in a specific cell type may be further confounded by competition for RXR between some of the concerned transcription factors in addition to competing for the same promoter element. It should be pointed out, however, that generalizing the mode of action of HD/PP as verified here in cells transiently transfected with Tf-promoted CAT constructs to the endogenous Tf promoter still remains to be complemented by studying the chromatin context of the endogenous gene as well as the role played by additional regulatory sequences of the Tf promoter not present in the transiently transfected promoter constructs used here.
In addition to suppressing Tf transcription by PPAR/RXR binding to the Tf-PRI element, HD/PP may suppress Tf transcription by suppressing HNF-4 expression. Suppression of HNF-4 transcription by HD/PP has previously been verified by showing that HNF-4 transcription rates and transcript and protein levels were significantly reduced in livers of treated animals (17). These previous results have been confirmed here in the context of the Tf gene by showing that the availability of HNF-4 for binding to the hTf-PRI element was significantly reduced in liver nuclear extracts derived from treated animals. Since HNF-4 expression is positively modulated by HNF-4 itself (35), suppression of HNF-4 gene expression by HD/PP may perhaps result as well from PPAR/RXR binding and displacement of HNF-4 from its putative enhancer in the HNF-4 gene promoter.
The increasing list of promoter elements reported to bind PPAR/RXR and be involved in transcriptional modulation of various genes by PPAR/RXR may call for updating the PPRE consensus sequence. As shown in Table I, PPRE sequences consist of a direct repeat separated by one-nucleotide spacer (DR-1). However, only half of the nucleotides within each repeat sequence are strictly conserved, while others vary considerably. Hence, additional cis and trans parameters other than those dictated by the direct repeat and/or the PPAR⅐RXR heterodimer, respectively, are presumably involved in binding and transactivation mediated by the PPAR/RXR-PPRE transduction pathway. Indeed, in the enoyl-CoA hydratase (37) and the malic enzyme 3 genes, half repeats adjacent to the PPRE direct repeat were shown to modulate PPAR/RXR binding to PPRE or transactivation of PPRE-promoted transcription. In two other cases (e.g. P450IV (12) and malic enzyme 3 ), additional distant upstream promoter sequences were shown to be involved in modulating the PPAR effect. Furthermore, putative PPAR/ RXR interacting proteins similar to those recently reported for the thyroid hormone nuclear receptor (39) may modulate transcriptional activity driven by the PPAR-PPRE basal unit.
Transcriptional suppression of Tf by HD/PP is essentially similar to that recently reported for the liver apolipoprotein C-III gene (17). In both, the suppressive effect appears to be related to HNF-4-enhanced transcription of the concerned gene and to result from displacement of HNF-4 by PPAR/RXR from the HNF-4 element of the concerned gene together with HNF-4 suppression by HD/PP, resulting in its reduced liver availability. Transcriptional suppression by HD/PP therefore complements transcriptional transactivation induced by HD/PP, thus extending the scope of effects of HD/PP as pleiotropic gene modulators. Other HNF-4-activated genes should similarly be considered as candidates for transcriptional suppression by HD/PP.
The functional significance of Tf suppression by xenobiotic HD/PP still remains to be investigated. Tf suppression could result in a decrease in iron availability and, if not compensated by increase in iron saturation or in Tf receptors, could lead to anemia. Slight anemia has indeed been observed in rats under conditions of subchromic treatment with HD/PP. 4 Since HNF-4 has recently been shown to mediate transcriptional activation of the erythropoietin gene by hypoxia (40), Tf suppression by HD/PP could be complemented by erythropoietin suppression similarly mediated by the PPAR/RXR transduction pathway. Moreover, Tf ferric reduction catalyzed by the plasma membrane NADH reductase has recently been reported to initiate intracellular alkalinization and mitogenic growth (reviewed in Ref. 1). Tf suppression by endogenous activators of the PPAR-PPRE transduction pathway could therefore be biologically significant in differentiation processes modulated by Tf availability.