Hepatocyte-specific interplay of transcription factors at the far-upstream enhancer of the carbamoylphosphate synthetase gene upon glucocorticoid induction Maarten Hoogenkamp 1 , Ingrid C. Gaemers 1 , Onard J. L. M. Schoneveld 1 , Atze T. Das 1 , Thierry Grange 2 and Wouter H. Lamers 1 1 AMC Liver Center, Academic Medical Center, University of Amsterdam, the Netherlands 2 Institut Jacques Monod du CNRS, Universites Paris 6-7, Paris, France Many of the metabolic functions of the liver are divi- ded in a complementary fashion among the periportal and pericentral hepatocytes. The expression of enzymes involved in amino acid degradation and gluconeogene- sis is largely confined to the periportal hepatocytes and regulated by intracellular cAMP levels, in combination Keywords carbamoylphosphate synthetase-I; FoxA; glucocorticoid receptor; in vivo footprinting; liver Correspondence W. H. Lamers, AMC Liver Center, Academic Medical Center, University of Amsterdam, Meibergdreef 69-71, 1105 BK, Amsterdam, the Netherlands Fax: +31 205669190 Tel: +31 205665405 E-mail: w.h.lamers@amc.uva.nl (Received 26 July 2006, revised 12 October 2006, accepted 27 October 2006) doi:10.1111/j.1742-4658.2006.05561.x Carbamoylphosphate synthetase-I is the flux-determining enzyme of the ornithine cycle, and neutralizes toxic ammonia by converting it to urea. An 80 bp glucocorticoid response unit located 6.3 kb upstream of the trans- cription start site mediates hormone responsiveness and liver-specific expression of carbamoylphosphate synthetase-I. The glucocorticoid response unit consists of response elements for the glucocorticoid receptor, forkhead box A, CCAAT ⁄ enhancer-binding protein, and an unidentified protein. With only four transcription factor response elements, the car- bamoylphosphate synthetase-I glucocorticoid response unit is a relatively simple unit. The relationship between carbamoylphosphate synthetase-I expression and in vivo occupancy of the response elements was examined by comparing a carbamoylphosphate synthetase-I-expressing hepatoma cell line with a carbamoylphosphate synthetase-I-negative fibroblast cell line. DNaseI hypersensitivity assays revealed an open chromatin configuration of the carbamoylphosphate synthetase-I enhancer in hepatoma cells only. In vivo footprinting assays showed that the accessory transcription factors of the glucocorticoid response unit bound to their response elements in car- bamoylphosphate synthetase-I-positive cells, irrespective of whether car- bamoylphosphate synthetase-I expression was induced with hormones. In contrast, the binding of glucocorticoid receptor to the carbamoylphosphate synthetase-I glucocorticoid response unit was dependent on treatment of the cells with glucocorticoids. Only forkhead box A was exclusively present in hepatoma cells, and therefore appears to be an important determinant of the observed tissue specificity of carbamoylphosphate synthetase-I expression. As the glucocorticoid receptor is the only DNA-binding protein specifically recruited to the glucocorticoid response unit upon stimulation by glucocorticoids, it is likely to be directly responsible for the transcrip- tional activation mediated by the glucocorticoid response unit. Abbreviations C ⁄ EBP, CCAAT ⁄ enhancer-binding protein; CPS, carbamoylphosphate synthetase-I; CRU, cAMP response unit; FoxA, forkhead box A; GR, glucocorticoid receptor; GRE, glucocorticoid receptor response element; GRU, glucocorticoid response unit; LM, ligation-mediated; PEPCK, phosphoenolpyruvate carboxykinase; PFK-2, 6-phosphofructo-2-kinase; PKA, cAMP-dependent protein kinase; TAT, tyrosine aminotransferase. FEBS Journal 274 (2007) 37–45 ª 2006 The Authors Journal compilation ª 2006 FEBS 37 with glucocorticoids. One of these enzymes is car- bamoylphosphate synthetase-I (CPS; EC 6.3.4.16), which mediates the rate-determining step of the orni- thine cycle that converts ammonia into urea [1]. Hepatocyte-specific expression of CPS is regulated by a distal enhancer, located 6.3 kb upstream of the transcription start site, in combination with the pro- moter region [2]. The distal enhancer is composed of two functional units, i.e. an upstream cAMP-response unit (CRU), covering 150–200 bp, and 100 bp further downstream, a glucocorticoid response unit (GRU) of approximately 80 bp. Transient transfection experi- ments have shown that the functions of these two units are well separated. The CPS CRU is the sole mediator of cAMP-dependent transcriptional activity (O. J. L. M. Schoneveld, M. Hoogenkamp, J. M. P. Stallen, I. C. Gaemers & W. H. Lamers, unpublished results). On the other hand, constructs containing the CPS GRU and the elements at the promoter show approximately 70% of the maximal induction of reporter gene activity after addition of dexamethasone alone. The combina- tion of dexamethasone and cAMP induces the con- struct maximally, whereas such a construct is not sensitive to cAMP alone [3]. The GRU consists of a response element for the ubiquitously expressed glucocorticoid receptor (GR) and three accessory factors. These accessory factors are the liver-enriched transcription factors forkhead box A (FoxA) and CCAAT ⁄ enhancer-binding protein (C ⁄ EBP), whereas the third factor is an unidentified  75 kDa protein denoted P3 [3,4]. The importance of each individual factor has been investigated extensively in transient transfection experiments [2,4]. Mutation analyses have shown that the presence of each of the four elements is essential for the glucocorticoid response. Although there does not seem to be a general rule for how a GRU is organized, the CPS GRU and the GRUs of gluconeogenic genes that are expressed in hepatocytes all contain binding sites for FoxA and C ⁄ EBP [5]. Mutation analyses of the CPS GRU have shown that each of the four elements is essential for the glucocorticoid response and that changes in spa- cing, order or orientation of the elements all cause a strong reduction in gene inducibility [2,4]. Neverthe- less, it is unknown what rules determine their activity. In particular, it is unknown to what extent DNA accessibility to transcription factors and the sequence in which the transcription factors bind play a role. Furthermore, it remains unclear whether accessory fac- tors have to bind first to the GRU, thereby allowing stable binding of GR, or whether it is binding of GR that allows access for the accessory factors [6,7]. With only four transcription factor response ele- ments located within close proximity of each other, the CPS GRU is a relatively simple unit. Therefore, this GRU is an ideal target with which to establish which factors are constitutively bound and which function as the trigger to initiate liver-specific gene expression. In order to investigate the in vivo occupancy of the GRU response elements, we compared CPS-positive FTO-2B hepatoma cells with CPS-negative Rat-1 fibroblasts. We show that the CPS enhancer is in an open confi- guration in FTO-2B cells, whereas the chromatin is not accessible in Rat-1 cells. We further show by in vivo footprinting assays that the accessory factors bind constitutively to their response elements in the CPS-expressing hepatoma cells, but not in CPS-negat- ive fibroblasts. Similarly, GR solely binds to its response element in CPS-expressing cells, but does so only after activation by its ligand. Results Because the expression of CPS in FTO-2B hepatoma cells has previously been shown to be responsive to the hormonal stimuli relevant in vivo [2], these cells were used as a paradigm for CPS-expressing cells, whereas Rat-1 fibroblasts served as CPS-negative control cells (Fig. 1A) [8]. Western blot analysis showed that the GR is expressed at a comparable level in both cell lines (Fig. 1A). C⁄ EBPa DNA-binding activity was only present in nuclear extract from FTO-2B cells, whereas C ⁄ EBPb DNA-binding activity was present in nuclear extracts of both cell lines (Fig. 1B). FoxA1 and FoxA2 DNA-binding activities were found only in FTO-2B nuclear extracts, whereas FoxA3 DNA-binding activity could not be detected in either cell line. The P3 protein is ubiquitously expressed [3]. Local chromatin accessibility at the CPS enhancer was determined by DNaseI hypersensitivity analysis. Figure 2A shows the position of both SstI restriction sites that were used for digestion, as well as the posi- tion of the probe, directly upstream of the ) 5.3 kb SstI site. Untreated samples showed the expected 7.6 kb SstI fragment and an additional, much longer, band, presumably resulting from incomplete SstI diges- tion (Fig. 2B, lanes 1, 5, 9 and 13). Both in untreated and in dexamethasone ⁄ cAMP-treated FTO-2B cells, fragments of 0.7–1.2 kb could be identified at inter- mediate DNaseI concentrations (Fig. 2B, lanes 7 and 15). Thus, chromatin appears to be accessible at the CPS GRU irrespective of hormonal activation. In con- trast, Rat-1 fibroblasts did not exhibit such a hypersen- sitive area, regardless of hormone treatment. The lack of accessibility of the GRU enhancer in Rat-1 cells Transcription factors at the CPS GRU M. Hoogenkamp et al. 38 FEBS Journal 274 (2007) 37–45 ª 2006 The Authors Journal compilation ª 2006 FEBS therefore corresponds with the lack of CPS expression in these cells, as shown in Fig. 1A. We analyzed the binding of transcription factors to the CPS GRU in Rat-1 fibroblasts and FTO-2B hepatoma cells that were or were not treated with dex- amethasone ⁄ cAMP. Alterations in the accessibility of DNA sequences were visualized by ligation-mediated PCR (LM-PCR). Previously, we subjected the CPS enhancer to in vitro footprinting using an end-labeled DNA fragment. The transcription factors producing the footprints in these assays were identified by com- parison with footprints produced by purified proteins [3]. To compare and validate the in vivo footprinting, we also analyzed in vitro DNaseI-treated DNA by LM-PCR. Linearized plasmid containing the CPS enhancer was incubated with either BSA or rat liver nuclear extract prior to DNaseI treatment. Analysis of the in vitro footprinted samples for both strands of the GRU region revealed three clear footprints (Fig. 3, compare lane 1 with lane 2, and lane 7 with lane 8), in agreement with our earlier observations [3]. Very simi- lar results were observed when the in vivo footprinted FTO-2B samples were compared with the Rat-1 sam- ples (Fig. 3). The footprint due to C ⁄ EBP binding was observed only in the FTO-2B samples at positions 322–343. FoxA binding was apparent in the FTO-2B samples at positions 343–357, and highlighted by a characteristic DNaseI hypersensitivity at position 350 on the upper strand and position 347 on the lower strand [7]. Both the in vitro and in vivo experiments showed that this FoxA-specific hypersensitivity was consistently more prominent on the upper strand than 1+ 3.5- tsS I tsS I 9 .21- r e cna h nebk3.6- A esaNDI PMAc/xed - + b k3. 1 –7.0 - sSt- tsS )b k6.7 ( } PMAc/xed 1-taRR1-taB2- OTFB2-OTF 2 1 4 3 6587 9 10 11 12 13 14 16 15 B Fig. 2. DNaseI hypersensitivity of the CPS upstream region. (A) Schematic representation of the upstream region of the CPS gene. A 130 bp 32 P-labeled probe, located upstream of the SstI site at ) 5.3 kb, was used for hybridization. (B) Southern blot of in vivo DNaseI-digested DNA from Rat-1 fibroblasts and FTO-2B hepatoma cells. Cells were left untreated or were treated with dexametha- sone and cAMP. Under each condition, cells were subjected to increasing amounts of DNaseI. The position of the intact SstI–SstI fragment and the GRU are indicated. CPS 160 kDa AB GR 95 kDa Rat-1 FTO-2B Rat-1 FTO-2B C/EBP probe FoxA probe Rat-1 FTO-2B Rat-1 FTO-2B Rat-1 FTO-2B Rat-1 FTO-2B Rat-1 FTO-2B Rat-1 FTO-2B Rat-1 FTO-2B probe probe PBE/C α Antiserum: PBE/C β 1AxoFenon2AxoF3AxoFenon PαBE/CS S PβBE/CSS 1AxoFSS 2AxoFSS 1AxoF Rat-1 FTO-2B SS 1AxoF Fig. 1. Expression of CPS and its regulating transcription factors in Rat-1 and FTO-2B cells. (A) CPS and GR were detected by western blot- ting. For CPS, 32 lg of total protein from Rat-1 or FTO-2B cells was loaded per lane, whereas for GR, 50 lg of total protein was loaded per lane. Amido black staining of the membrane served as loading control. (B) The presence of C ⁄ EBP and FoxA family members was visualized by antibody-mediated supershifts in electrophoretic mobility shift assays. In each panel, the first lane corresponds to free probe, whereas the other lanes correspond to probe incubated with Rat-1 or FTO-2B nuclear extracts. Where indicated, antibody directed against specific members of the C ⁄ EBP and FoxA families of transcription factors was added. The region of the gel showing the supershifted complex with FoxA1 antibody is additionally shown after a longer exposure. ‘SS’ indicates observed supershifts. M. Hoogenkamp et al. Transcription factors at the CPS GRU FEBS Journal 274 (2007) 37–45 ª 2006 The Authors Journal compilation ª 2006 FEBS 39 on the lower strand. Binding of P3 to positions 360– 377 was best detected on the upper strand as protec- tion around position 363 and hypersensitivity at posi- tion 381. Although the in vitro and in vivo footprints are highly similar to each other, the band patterns are not identical. This difference can be attributed to the differences in DNaseI accessibility of naked DNA (in vitro) and DNA in a chromatin context (in vivo) [9]. Interestingly, treatment of the cells with dexametha- sone ⁄ cAMP prior to DNaseI treatment did not result in any changes in the pattern of bands for either of the two cell types, showing that binding of C ⁄ EBP and FoxA to the CPS GRU in FTO-2B cells is independ- ent of these hormonal stimuli. FTO-2B hepatoma cells exhibit a constitutive cAMP-dependent protein kinase (PKA) activity [10]. In the FTO-2B-derived hepatoma cell line WT-8, PKA activity is again fully dependent on cAMP [10]. DNa- seI footprinting of the CPS upstream enhancer gener- ated a near-identical banding pattern in untreated FTO-2B and WT-8 cells, including the prominent FoxA-specific hypersensitive band (Fig. 4, arrow). As already shown for FTO-2B (Fig. 3), treatment of both cell lines with dexamethasone ⁄ cAMP did not alter the banding pattern. This confirms that the binding of the accessory factors to the CPS CRU does not result from PKA activity, but is associated with the hepatic phenotype. As expected [11], DNaseI footprinting was not suit- able for revealing the interaction between the GR and its response element (GRE), but dimethylsulfate foot- printing was (Fig. 5). Comparison of the FTO-2B samples with the Rat-1 samples revealed that C⁄ EBP binding to the GRU increased and decreased sensitiv- ity to dimethylsulfate on the upper strand at positions 333 and 326, respectively, whereas changes in reactivity at positions 331, 336 and 340 were seen on the lower strand. FoxA binding resulted in protection of the gua- nines at positions 347 and 351 on the upper strand, and protection at position 352 on the lower strand. These footprints in the FTO-2B samples were not influenced by treatment with dexamethasone ⁄ cAMP, in accordance with the DNaseI-footprinting experiments. At the position of the GRE, differences between non- treated and hormone-treated Rat-1 fibroblasts were not observed on either DNA strand. Moreover, there was no difference between the Rat-1 samples and the untreated FTO-2B hepatoma cells. After the addition of hormones to FTO-2B cells, however, the reactivity of several guanines towards dimethylsulfate was altered. These guanines map within the GRE region that was footprinted by the GR in vitro [3]. On the dnartsr eppu 3P RG Ax oF PB E /C ’5 ’3 B2-OTF ENASB + - + - 1-taR 413 333 56 3 5 43 383 4 3 5 6 ovivni 413 33 3 563 54 3 383 ortivni 21 dnartsrewol PBE /C 3P RG AxoF ’5 ’3 B2-OTF + - + - 1-taR 113 023 043 873 3 53 693 19 01 1 12 ovivni ENASB 693 04 3 873 11 3 353 023 ortivni 8 7 Fig. 3. DNaseI footprinting of the CPS upstream enhancer in FTO-2B and Rat-1 cells. In vitro footprints were obtained by incubating a linea- rized plasmid containing the CPS GRU with 45 lg of BSA or rat liver nuclear extract, after which DNaseI was added. The resulting DNA frag- ments were used as template for LM-PCR. For in vivo footprints, dexamethasone ⁄ cAMP-treated (+) and untreated (–) Rat-1 fibroblasts and FTO-2B hepatoma cells were permeabilized and incubated with DNaseI. After isolation of the DNA, the samples were subjected to LM-PCR. An arrow indicates the FTO-2B-specific hypersensitive site in the FoxA-binding site. Schematic representations of the GRU with its binding sites are included for clarity. Transcription factors at the CPS GRU M. Hoogenkamp et al. 40 FEBS Journal 274 (2007) 37–45 ª 2006 The Authors Journal compilation ª 2006 FEBS upper strand, the bands representing the guanines at positions 383 and 385 were increased 1.6-fold (± 0.13; N ¼ 4) and 1.7-fold (± 0.18; N ¼ 4), respectively, whereas the band at position 392 was reduced 1.5-fold (± 0.05; N ¼ 4). On the lower strand, GR binding resulted in protection over the region covering posi- tions 386–400. The observed bands corresponding to the guanines at positions 386, 395 and 400 were decreased 1.4-fold (± 0.15; N ¼ 3), 1.5-fold (± 0.13; N ¼ 3), and 1.4-fold (± 0.08; N ¼ 3), respectively. Elevated levels of cAMP alone did not lead to altera- tions at the GRE, whereas dexamethasone alone did (Fig. 5C). Five out of six of these changes of reactivity of guanines map within the consensus GR-binding sites within the GRE (Fig. 5D). All the guanines that are known to be affected upon GR binding to a similar GRE [11] showed altered reactivity. Altogether, these findings indicate that these glucocorticoid- induced footprints were due to GR binding. Discussion Transcriptional regulation results from the cooper- ative binding of transcription factors [12]. One of the key determinants of the activation of genes that are under the control of hormone response units is the order and time at which the transcription factors bind to their response elements. With only four response elements for transcription factors located within a stretch of 80 bp, the CPS GRU is ideal for determin- ing which transcription factor-binding events are pre- requisites and which form the final trigger for GRU activity. DNaseI hypersensitivity analysis showed that the DNA region encompassing the CPS enhancer is in an open chromatin configuration in CPS-expressing hepa- toma cells (Fig. 2). An open chromatin configuration of GRU-containing distal enhancer regions appears to be the rule in well-differentiated hepatoma cells [13– 15], including the distal tyrosine aminotransferase (TAT) GRU at ) 5.5 kb, but this does not apply to the more proximal TAT GRU at ) 2.5 kb, which needs prior exposure to glucocorticoids to acquire an open conformation [6,9,15]. In contrast to its accessibility in hepatoma cells, the CPS GRU is not in an open confi- guration in the Rat-1 cell line, which does not express CPS. These findings are in line with the concept that accessibility of an enhancer region to DNaseI corre- lates with expression from that enhancer [16]. The absence of CPS expression in Rat-1 fibroblasts correlates with the absence of FoxA DNA-binding activities (Fig. 1) and a nonaccessible CPS GRU (Fig. 2). FoxA is one of the relatively few transcription factors that can function, in the absence of ATP- dependent complexes, to open up compacted chroma- tin, thereby allowing access for other transcription factors [17]. It is therefore tempting to speculate that these features underlie the absence of binding of tran- scription factors (Fig. 3) and the lack of CPS expres- sion in Rat-1 cells. In vivo footprinting of FTO-2B hepatoma cells, on the other hand, showed constitutive binding of FoxA and C ⁄ EBP to the CPS GRU (Fig. 3), whereas binding of GR was conditional, i.e. dependent on treatment of the cells with dexametha- sone ⁄ cAMP (Fig. 5). Treatment of Rat-1 fibroblasts with dexamethasone ⁄ cAMP did not result in binding of GR, even though GR is abundantly present in these cells (Fig. 1). In line with experiments showing that GR binding to the ) 2.5 kb TAT GRU can only be detected by genomic footprinting when the accessory factors are bound to this GRU [11], these data indicate that prior binding of accessory transcription factors is necessary to stabilize the interaction between GR and the CPS GRU. In vitro experiments with the phosphoenolpyruvate carboxykinase (PEPCK) GRU showed that binding of COUP-TF and especially FoxA increased the affinity of the low-affinity PEPCK GRE for GR and decreased its dissociation rate [18]. dnartsreppu 3P RG AxoF PBE/C ’5 ’ 3 B2-OTF + - + - 8-TW 413 333 56 3 543 3 83 4321 Fig. 4. DNaseI footprinting of the CPS upstream enhancer in FTO- 2B and WT-8 hepatoma cells. For in vivo footprints, FTO-2B and WT-8 hepatoma cells, either untreated (–) or treated with dexa- methasone ⁄ cAMP (+), were permeabilized and incubated with DNaseI. After isolation of the DNA, the samples were subjected to LM-PCR. An arrow indicates the hypersensitive site in the FoxA- binding site. Schematic representations of the GRU with its binding sites are included for clarity. M. Hoogenkamp et al. Transcription factors at the CPS GRU FEBS Journal 274 (2007) 37–45 ª 2006 The Authors Journal compilation ª 2006 FEBS 41 For both the PEPCK and the CPS GRUs, the distance between the binding sites for GR and FoxA is of crit- ical importance for GRU activity, which probably reflects a direct interaction between the two factors [4,19]. Although it remains to be established to what extent these findings in cell lines can be extrapolated to pri- mary hepatocytes, the absence of FoxA from fibroblasts is in agreement with the concept that FoxA binding is an early event in the opening of the chromatin of he- patocytes [17]. FoxA binds in a glucocorticoid-inde- pendent manner on the CPS GRU (Figs 3 and 5), the PEPCK GRU [20] and the distal TAT GRU at ) 5.5 kb [9,15]. Although FoxA binds to the proximal TAT GRU at ) 2.5 kb in unstimulated FTO-2B cells, its binding is significantly enhanced by prior exposure to glucocorticoids and chromatin remodeling [11], whereas in H4IIE hepatoma cells, its binding to this GRU is strictly dependent upon GR activation [7]. The subse- quent binding of the accessory transcription factors, in turn, stabilizes GR binding (see previous paragraph) [11]. The difference in FoxA binding between the FTO- 2B and H4IIE cell lines could be due to the constitutive PKA activity that is present in the FTO-2B cell line. FoxA binding at the ) 2.5 kb TAT GRU in WT-8 cells is, indeed, very low in the absence of glucocorticoids and can be induced by glucocorticoids and PKA activa- tion in an additive manner [11]. However, FoxA bind- ing to the TAT GRU at ) 5.5 kb and the CPS GRU was constitutive in both FTO-2B and WT-8 cells, ren- dering the CPS GRU similar to the distal TAT GRU. Another study in FTO-2B cells, conducted on the con- stitutively open GRU of the 6-phosphofructo-2-kinase (PFK-2) gene, showed that FoxA binding was largely glucocorticoid-dependent, despite the constitutive PKA activity and the glucocorticoid-independent chromatin remodeling [13]. Taken together, these studies show that although these GRUs are all involved in mediating he- patocyte-specific expression and contain binding sites for GR in combination with response elements for the liver-enriched factors FoxA and C ⁄ EBP, they differ in their recruitment of these common transcription fac- tors. Although it is still unclear what determines these differences in the assembly of the transcription factor complex, transient transfection studies suggest that the precise arrangement of the various binding sites within 12 3 0 4 3 273 504 3P 383 401 RG A xoF P BE/C 453 033 ’5 3’ + - + - 1-taRB2-OTF FTO-2B untreated +dex +forskolin +dex & forsk. GRE PMAc/xed 8 76 5 dnarts rewol dnarts reppu + - + - 1 -t a R B2 -OTF 3P RG AxoF PBE/C 33 3 75 3 743 573 10 4 3 83 31 4 5’ ’3 PMAc/xed 3 214 B A C D A -’ 5 G Tt t g A CGA G 3- ct tgT C TT ’ CTCT- ’ 3 G AA C AaacT G ’5- g aacA 832 0 4 0 Fig. 5. Dimethylsulfate footprinting of the CPS upstream enhancer in Rat-1 and FTO-2B cells. Dexamethasone ⁄ cAMP-treated (+) and untreated (–) Rat-1 fibroblasts and FTO-2B hepatoma cells were incubated with 0.1% dimethylsulfate. After DNA isolation, the samples were subjected to LM-PCR. (A, B) Upper and lower strands, respectively. Closed diamonds and circles indicate guanines showing, respectively, increased and decreased sensitivity towards dimethylsulfate in hormone-treated FTO-2B cells, whereas open diamonds and circles indicate hormone-independent decreased and increased sensitivity to dimethylsulfate in FTO-2B compared to Rat-1 cells. (C) Upper strand analysis, showing that dexamethasone treatment alone is sufficient to promote GR binding in FTO2B cells. (D) Sequence of the CPS GRE showing the guanine residues that showed altered reactivity to dimethylsulfate following hormonal treatment of FTO2-B cells. The consensus palin- dromic GR-binding site is indicated in capital letters. The guanines in bold letters are those that showed altered reactivity towards dimethyl- sulfate upon GR binding in vitro to a similar GRE [11]. Transcription factors at the CPS GRU M. Hoogenkamp et al. 42 FEBS Journal 274 (2007) 37–45 ª 2006 The Authors Journal compilation ª 2006 FEBS the GRUs and the relative affinities for their cognate factors play a role, presumably in combination with the chromatin structure established at the GRUs prior to glucocorticoid activation [4,19]. Even though the assembly of the transcription fac- tor complex at the CPS GRU seems to be similar to that of the distal TAT GRU at ) 5.5 kb, there appears to be an important difference. The CPS GRU is sufficient to enhance the activity of the basal promoter [4], whereas the distal TAT GRU has to cooperate with the proximal TAT GRU at ) 2.5 kb [15]. This difference may be more apparent than real, however, because we recently showed that the CPS GRU needs to interact with a GRE directly upstream of the core CPS promoter to transactivate this pro- moter [5,21]. For both the proximal TAT GRU and the PFK-2 GRU, it is not clear which transcription factor is directly responsible for transcriptional activation, as GR allows recruitment of FoxA and C ⁄ EBP, which could be the factors interacting with the transcription machinery. Therefore, GR could play only an indirect role by allowing the recruitment of these factors. The distal TAT GRU, where GR is the only DNA-bind- ing protein interacting specifically in the presence of glucocorticoids, does not provide a clear answer to this question, as this GRU does not activate tran- scription on its own. Our present analysis of tran- scription factor recruitment at the CPS GRU therefore provides the first clear evidence that the transcription activation domains of GR play a key role in transcriptional activation mediated by a GRU, as we show that it is the only DNA-binding protein of the triad GR, FoxA and C ⁄ EBP that is specifically recruited upon glucocorticoid stimulation. The acces- sory factors FoxA and C ⁄ EBP presumably allow sta- bilization of GR binding to the GRU, and thereby stabilization of the interaction of the coregulators interacting with the nucleoprotein complex formed at the GRU. This raises the possibility that the acces- sory factors play a similar role in GRUs where they are recruited in a glucocorticoid-dependent manner, and that the differences that are seen in the modali- ties of recruitment do not reflect fundamental differ- ences in their contribution to the function of the GRUs. Experimental procedures Cell culture Rat-1 fibroblasts [22], FTO-2B hepatoma cells [23], and WT-8, an FTO-2B-derived cell line overexpressing the R1a subunit of PKA [10], were grown at 37 °Cin DMEM ⁄ F12 ⁄ 5% CO 2 ⁄ 10% fetal bovine serum. Nuclear extracts Nuclear extracts from rat livers were prepared as previ- ously described [24]. To prepare nuclear extracts from cell lines, cells were detached with trypsin, washed in NaCl ⁄ P i containing 0.25 mm phenylmethanesulfonyl fluoride, and lysed by resuspension in 1.5 mL of 10 mm Hepes (pH 7.6), 10 mm KCl, 1.5 mm MgCl 2 and 0.5 mm dithiothreitol per 2 · 10 7 cells. After 8 min on ice, nuclei were pelleted by centrifugation in an Eppendorf 5417C centrifuge at 20 800 g for 30 s at 4 °C. The pellets were resuspended in 100 lLof20mm Hepes (pH 7.6), 20% glycerol, 420 mm NaCl, 1.5 mm MgCl 2 , 0.2 mm EDTA and 0.5 mm dithio- threitol, and incubated on ice for 20 min. Nuclear debris was spun down at 20 800 g for 2 min at 4 °C (Eppendorf 5417C centrifuge). Western blotting Whole cell extracts were prepared by lysis of cells in 20 mm Tris (pH 7.5), 150 mm NaCl, 1% NP40, 0.5 mm dithiothrei- tol and 0.2 mm phenylmethanesulfonyl fluoride. Insoluble debris was spun down in an Eppendorf centrifuge at 20 800 g for 20 s at 4 °C (Eppendorf 5417C centrifuge). Western blotting was performed as previously described [25]. Electrophoretic mobility shift assays Electrophoretic mobibility shift assays were performed as previously described [4], except that the binding reaction contained 20 mm Hepes (pH 7.6), 500 mm KCl, 12% gly- cerol (v ⁄ v), 1 mm EDTA, 1 mm dithiothreitol, 1 mm sper- midine, 0.5 lg of double-stranded poly(dIdC)Ælg )1 nuclear extract, and 0.3 lgÆlL )1 BSA. Double-stranded probes were designed on the basis of the rat CPS GRU sequence (Table 1). Antibodies Rabbit polyclonal antibodies against C ⁄ EBPa (sc-61), C ⁄ EBPb (sc-746) and GR (sc-1003), and goat polyclonal antibodies against FoxA1 (sc-6553), FoxA2 (sc-6554) and FoxA3 (sc-5360), were obtained from Santa Cruz Biotech- nology (Santa Cruz, CA, USA). Rabbit polyclonal anti- body against CPS has been previously described [26]. DNaseI treatment Cells were grown to 70% confluence and supplemented, where indicated, with 100 nm dexamethasone, 1 mm dibuty- ryl-cAMP and 0.1 mm 3-isobutyl-1-methylxanthine (IBMX) M. Hoogenkamp et al. Transcription factors at the CPS GRU FEBS Journal 274 (2007) 37–45 ª 2006 The Authors Journal compilation ª 2006 FEBS 43 2 h before the start of the experiment. DNaseI treatment was performed exactly as previously described [27]. Dimethylsulfate treatment Cells were grown to 70% confluence and exposed, where indicated, to 100 nm dexamethasone, 1 mm dibutyryl- cAMP and 0.1 mm IBMX for 2 h before the start of the experiment. All solutions used for hormone-treated cells contained 100 nm dexamethasone. After exposure for 4 min at room temperature to 0.1% dimethylsulfate in NaCl ⁄ P i , cells were washed three times with NaCl ⁄ P i and lysed in 2.5 mL of 50 mm Tris (pH 8.0), 20 mm EDTA, 1% SDS and 100 lgÆmL )1 proteinase K per 80 cm 2 flask, digested overnight at 55 °C, and further processed as des- cribed [28]. In vitro footprinting In vitro footprinting was performed as previously described [29]. As matrix, linearized plasmid DNA containing the CPS enhancer region was used. Protein binding was per- formed using 45 lg of rat liver nuclear extract or BSA. DNaseI hypersensitivity analysis Thirty micrograms of DNaseI-treated DNA was cut with SstI, separated on a 1% agarose gel, blotted [25], and hybridized to a 130 bp [a- 32 P]ATP-labeled PCR probe. LM-PCR The starting material consisted of 1 lg of genomic DNA for in vivo footprints or 2 ng of plasmid DNA for in vitro footprints. LM-PCR was performed as described [30], except that the linker–ligation mix contained 5% poly- ethyleneglycol-6000. The primers used are described in Table 1. Acknowledgements The research presented in this article was financially supported by ZonMW grant 902-23-250 and by grants to TG of the Association pour la Recherche sur le Cancer and the Ligue contre le Cancer. References 1 Meijer AJ, Lamers WH & Chamuleau RA (1990) Nitro- gen metabolism and ornithine cycle function. Physiol Rev 70, 701–748. 2 Christoffels VM, Van Den Hoff MJ, Moorman AF & Lamers WH (1995) The far-upstream enhancer of the carbamoyl-phosphate synthetase I gene is responsible for the tissue specificity and hormone inducibility of its expression. J Biol Chem 270, 24932–24940. 3 Christoffels VM, Grange T, Kaestner KH, Cole TJ, Darlington GJ, Croniger CM & Lamers WH (1998) Glucocorticoid receptor, C ⁄ EBP, HNF3, and protein kinase A coordinately activate the glucocorticoid response unit of the carbamoylphosphate synthetase I gene. Mol Cell Biol 18, 6305–6315. 4 Schoneveld OJ, Gaemers IC, Das AT, Hoogenkamp M, Renes J, Ruijter JM & Lamers WH (2004) Structural requirements of the glucocorticoid-response unit of the carbamoyl-phosphate synthase gene. Biochem J 382, 463–470. 5 Schoneveld OJ, Gaemers IC & Lamers WH (2004) Mechanisms of glucocorticoid signalling. Biochim Bio- phys Acta 1680, 114–128. 6 Reik A, Schutz G & Stewart AF (1991) Glucocorticoids are required for establishment and maintenance of an alteration in chromatin structure: induction leads to a reversible disruption of nucleosomes over an enhancer. EMBO J 10, 2569–2576. 7 Rigaud G, Roux J, Pictet R & Grange T (1991) In vivo footprinting of rat TAT gene: dynamic interplay between the glucocorticoid receptor and a liver-specific factor. Cell 67, 977–986. 8 Hoogenkamp M, Stallen JM, Lamers WH & Gaemers IC (2006) In vivo footprinting of the carbamoylpho- sphate synthetase I cAMP-response unit indicates important roles for FoxA and PKA in formation of the enhanceosome. Biochimie 88, 1357–1366. 9 Flavin M, Cappabianca L, Kress C, Thomassin H & Grange T (2004) Nature of the accessible chromatin at a glucocorticoid-responsive enhancer. Mol Cell Biol 24, 7891–7901. Table 1. Oligonucleotide sequences used for experiments. EMSA, electrophoretic mobility shift assay. Probes and primers (5¢-to3¢) Probes for EMSA C ⁄ EBP TTTCGAGTCTTGCAAAATCATCA FoxA CATCAGTGTTTGCTCTTGACAAG LM-PCR primers Upper strand Primer 1 TTCTTAAAACTTGACCAAA Primer 2 GGGTACGATGACTAAATGATCGGA Primer 3 (Figs 3 and 5A,B) TCATCAGCAGCCCTTCTTTGCACAAC Primer 3 (Figs 4 and 5C) GACTAAATGATCGGATACGTGCCCATTCT Lower strand Primer 1 CTCAACGTCATTCTAAAGT Primer 2 ACAACATACTTCGAAACTGTGACC Primer 3 TGTCCTGGCACATGACCCGGATCA Transcription factors at the CPS GRU M. Hoogenkamp et al. 44 FEBS Journal 274 (2007) 37–45 ª 2006 The Authors Journal compilation ª 2006 FEBS 10 Jones KW, Shapero MH, Chevrette M & Fournier RE (1991) Subtractive hybridization cloning of a tissue-spe- cific extinguisher: TSE1 encodes a regulatory subunit of protein kinase A. Cell 66, 861–872. 11 Espinas ML, Roux J, Pictet R & Grange T (1995) Glucocorticoids and protein kinase A coordinately mod- ulate transcription factor recruitment at a glucocorti- coid-responsive unit. Mol Cell Biol 15, 5346–5354. 12 Carey M (1998) The enhanceosome and transcriptional synergy. Cell 92, 5–8. 13 Zimmermann PL, Pierreux CE, Rigaud G, Rousseau GG & Lemaigre FP (1997) In vivo protein–DNA inter- actions on a glucocorticoid response unit of a liver-spe- cific gene: hormone-induced transcription factor binding to constitutively open chromatin. DNA Cell Biol 16, 713–723. 14 Ip YT, Granner DK & Chalkley R (1989) Hormonal regulation of phosphoenolpyruvate carboxykinase gene expression is mediated through modulation of an already disrupted chromatin structure. Mol Cell Biol 9, 1289–1297. 15 Grange T, Roux J, Rigaud G & Pictet R (1989) Two remote glucocorticoid responsive units interact coopera- tively to promote glucocorticoid induction of rat tyro- sine aminotransferase gene expression. Nucleic Acids Res 17, 8695–8709. 16 Felsenfeld G, Boyes J, Chung J, Clark D & Studitsky V (1996) Chromatin structure and gene expression. Proc Natl Acad Sci USA 93, 9384–9388. 17 Cirillo LA, Lin FR, Cuesta I, Friedman D, Jarnik M & Zaret KS (2002) Opening of compacted chromatin by early developmental transcription factors HNF3 (FoxA) and GATA-4. Mol Cell 9, 279–289. 18 Stafford JM, Wilkinson JC, Beechem JM & Granner DK (2001) Accessory factors facilitate the binding of glucocorticoid receptor to the phosphoenolpyruvate car- boxykinase gene promoter. J Biol Chem 276, 39885– 39891. 19 Sugiyama T, Scott DK, Wang JC & Granner DK (1998) Structural requirements of the glucocorticoid and retinoic acid response units in the phosphoenolpyruvate carboxykinase gene promoter. Mol Endocrinol 12, 1487– 1498. 20 Wang XL, Herzog B, Waltner-Law M, Hall RK, Shiota M & Granner DK (2004) The synergistic effect of dexa- methasone and all-trans-retinoic acid on hepatic phos- phoenolpyruvate carboxykinase gene expression involves the coactivator p300. J Biol Chem 279, 34191– 34200. 21 Schoneveld OJ, Gaemers IC, Hoogenkamp M & Lamers WH (2005) The role of proximal-enhancer elements in the glucocorticoid regulation of carbamoylphosphate synthetase gene transcription from the upstream response unit. Biochimie 87, 1033–1040. 22 Prasad I, Zouzias D & Basilico C (1976) State of the viral DNA in rat cells transformed by polyoma virus. I. Virus rescue and the presence of nonintegrated viral DNA molecules. J Virol 18, 436–444. 23 Killary AM & Fournier RE (1984) A genetic analysis of extinction: trans-dominant loci regulate expression of liver-specific traits in hepatoma hybrid cells. Cell 38, 523–534. 24 Sierra F (1990) A laboratory guide to in vitro transcrip- tion. In Biomethods (Azzi A, Polak JM & Saluz HP, eds), pp. 30–50. Birkhauser-Verlag, Basel. 25 Sambrook J, Fritsch EF & Maniatis T (1989) Molecular Cloning: A Laboratory Manual. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY. 26 Charles R, de Graaf A & Moorman AF (1980) Radio- immunochemical determination of carbamoyl-phosphate synthase (ammonia) content of adult rat liver. Biochim Biophys Acta 629, 36–49. 27 Lefevre P, Lacroix C, Tagoh H, Hoogenkamp M, Melnik S, Ingram R & Bonifer C (2005) Differentiation- dependent alterations in histone methylation and chro- matin architecture at the inducible chicken lysozyme gene. J Biol Chem 280, 27552–27560. 28 Mueller PR, Wold B & Garrity PA (2002) Current Pro- tocols in Molecular Biology. John Wiley & Sons, New York, NY, doi:10.1002/0471142727.mb1503s56. 29 Espinas ML, Roux J, Ghysdael J, Pictet R & Grange T (1994) Participation of Ets transcription factors in the glucocorticoid response of the rat tyrosine aminotrans- ferase gene. Mol Cell Biol 14, 4116–4125. 30 Grange T, Bertrand E, Espinas ML, Fromont-Racine M, Rigaud G, Roux J & Pictet R (1997) In vivo foot- printing of the interaction of proteins with DNA and RNA. Methods 11, 151–163. M. Hoogenkamp et al. Transcription factors at the CPS GRU FEBS Journal 274 (2007) 37–45 ª 2006 The Authors Journal compilation ª 2006 FEBS 45 . Hepatocyte-specific interplay of transcription factors at the far-upstream enhancer of the carbamoylphosphate synthetase gene upon glucocorticoid induction Maarten. the cooper- ative binding of transcription factors [12]. One of the key determinants of the activation of genes that are under the control of hormone response