The Far-upstream Enhancer of the Carbamoyl-phosphate Synthetase I Gene Is Responsible for the Tissue Specificity and Hormone Inducibility of Its Expression*

The role of the proximal promoter and the far-up- stream enhancer in the hepatocyte-specific and hormonal regulation of the carbamoyl-phosphate synthetase I (CPS) gene was investigated in transient transfection assays using primary rat hepatocytes, hepatoma cells, and fibroblasts. These experiments revealed that the activity of the promoter is comparable in all cells tested and is, therefore, not responsible for tissue-specific expression. The 5 (cid:42) -untranslated region of the mRNA is a major, non-tissue specific stimulator of expression in FTO-2B hepatoma cells, acting at the post-transcrip-tional level. A 469-base pair DNA fragment, 6 kilobase pairs upstream of the transcription start-site in the CPS gene, confers strong hormone-dependent tissue specific expression, both in combination with the CPS promoter and a minimized viral thymidine kinase promoter. Se- quences similar to a cyclic AMP-responsive element and a glucocorticosteroid-responsive element were found in the isolated enhancer. Substitutional mutations in these sites strongly affected hormone-induced expression. Analysis of the interaction between the enhancer and parts of the CPS promoter revealed that, in addition to the TATA box, the GAG box, a motif similar to the GC box near the TATA motif, is instrumental in conferring the enhancer activity. Carbamoyl-phosphate synthetase (CPS) 1 is the first enzyme of the ornithine cycle. CPS expression can be detected from the 15th embryonic day onward in the liver of the rat. Expression is initially found in a few hepatocytes only, but toward

The role of the proximal promoter and the far-upstream enhancer in the hepatocyte-specific and hormonal regulation of the carbamoyl-phosphate synthetase I (CPS) gene was investigated in transient transfection assays using primary rat hepatocytes, hepatoma cells, and fibroblasts. These experiments revealed that the activity of the promoter is comparable in all cells tested and is, therefore, not responsible for tissue-specific expression. The 5-untranslated region of the mRNA is a major, non-tissue specific stimulator of expression in FTO-2B hepatoma cells, acting at the post-transcriptional level. A 469-base pair DNA fragment, 6 kilobase pairs upstream of the transcription start-site in the CPS gene, confers strong hormone-dependent tissue specific expression, both in combination with the CPS promoter and a minimized viral thymidine kinase promoter. Sequences similar to a cyclic AMP-responsive element and a glucocorticosteroid-responsive element were found in the isolated enhancer. Substitutional mutations in these sites strongly affected hormone-induced expression. Analysis of the interaction between the enhancer and parts of the CPS promoter revealed that, in addition to the TATA box, the GAG box, a motif similar to the GC box near the TATA motif, is instrumental in conferring the enhancer activity.
Carbamoyl-phosphate synthetase (CPS) 1 is the first enzyme of the ornithine cycle. CPS expression can be detected from the 15th embryonic day onward in the liver of the rat. Expression is initially found in a few hepatocytes only, but toward the end of the fetal period all hepatocytes have been recruited to express CPS (1-3). After birth, the expression gradually becomes confined to the hepatocytes surrounding the portal veins (3,4). The only other cells producing appreciable levels of CPS mRNA and protein are the enterocytes of the small intestine (3,5). After birth, CPS enzyme and mRNA levels change in parallel under all experimental conditions (5,8,10,11), suggesting that hormonal regulation, tissue specificity, and zonal restriction of expression of CPS is regulated at the level of transcription. Accordingly, it was shown that glucocorticosteroids and cyclic AMP enhance transcription of the CPS gene in adult rat hepatocytes (6 -9, 12).
The CPS gene is a single-copy, 110 gene that contains 38 exons and is surrounded by matrix attachment regions (13)(14)(15)(16). The mRNA is 5546 nucleotides in length (excluding the poly(A) tract) and consists of a 140-nt 5Ј-untranslated region, an open reading frame of 4500 nt, and a 3Ј-UTR of 906 nt. Functional analysis of the 5Ј-part of the gene showed that the minimal, fully active promoter is located within the 161 nt upstream of the transcription initiation site (15). DNase I footprint analysis of this region revealed three protected sites (sites I-III (17,18)), but the actual identity of the factors occupying these sites is not known (19). In between the TATA motif at position Ϫ21 and protected site I, a so called "GAG" element was identified (19) that resembles the element recognized by the TFIIIA-like Cys 2 / His 2 zinc finger class of transcription factors, including Sp1, but is not a target of the Sp1 protein itself.
Two MspI restriction sites upstream of the transcription start site were found to be differentially methylated (15). The site at Ϫ6.3 kbp becomes demethylated shortly after birth in liver and intestine. The other site, at Ϫ4.0 kbp relative to the transcription start site is fully methylated in liver and partly demethylated in postnatal small intestine. A 4-kbp fragment containing the Ϫ6.3-kbp MspI site responds to cyclic AMP and dexamethasone in transient expression assays, giving a 10-fold rise in reporter gene expression in FTO-2B hepatoma cells (15).
The regulatory regions of the CPS gene were analyzed with respect to their role in tissue-specific expression and hormone sensitivity. A 469-bp far-upstream enhancer fragment was found to be responsible for hormone-dependent tissue-specific expression of the CPS gene. It was also investigated whether the proximal promoter and the 5Ј-UTR of the mRNA contributed to tissue-specific expression. Exploration of the interaction of the enhancer fragment with parts of the CPS promoter revealed an important role for the GAG element in the transduction of the hormonally induced activation signal from the enhancer.

EXPERIMENTAL PROCEDURES
Cell Culture-FTO-2B rat hepatoma cells (20), Rat-1 fibroblasts (21,22), Chinese hamster ovary cells (CHO-K1), and Caco-2 cells (23) were cultured in Dulbecco's modified Eagle's medium/F-12 (Life Technologies, Inc.), supplemented with 10% fetal calf serum (FCS; Life Technologies, Inc.) and MH1C1 hepatoma cells were cultured in Ham's F-10 medium (Life Technologies, Inc.), supplemented with 17.5% FCS. The FCS was selected for its inability to induce CPS expression in cultured hepatocytes. All cells were cultured in a 5% CO 2 /air atmosphere at * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EMBL Data Bank with accession number(s) X90476.
DNA Transfection-Exponentially growing cells were transfected by electroporation (24). The capacity of CPS gene sequences to direct expression of a reporter gene was analyzed by cloning into the vector pLT1 or pT81luc. The pLT1 vector, based on pBluescript SKϩ (Stratagene), contains the firefly luciferase gene (25) as a reporter in conjunction with SV40 small t-antigen intron and polyadenylation signal optimized for expression (26). The vector pT81luc (27) contains the firefly luciferase reporter gene driven by the first 81 base pairs of the viral thymidine kinase promoter (28) and is referred to as "minimized TK promoter." To transfect equimolar amounts, 3 g/kbp of CsCl-purified supercoiled plasmid was used. As a control the vector pLT1 without inserted promoter sequences was used. This vector produces 70 ϫ 10 3 relative light units/mg of protein, which is 1.4 Ϯ 0.4% (S.E., n ϭ 6) of the activity of pLT1 in which the minimal, fully active CPS promoter (Ϫ161 to ϩ138 nt with respect to the transcription start site) is inserted (background: 0.1 ϫ 10 3 relative light units/mg of protein). Differences in transfection efficiency were corrected for by co-transfecting 3 g of vector pRSVcat (RSV LTR (29)). After each transfection the cell suspension was divided into equal parts, one being grown in culture medium and the other(s) in culture medium supplemented with 100 nM dexamethasone or with 100 nM dexamethasone, 1 mM dibutyryl cyclic AMP (Bt 2 cAMP; Boehringer) and 0.1 mM 3-isobutyl-1-methyl-xanthine (IBMX, Sigma) (7). 44 hours after transfection cells were harvested, lysed in 100 mM KH 2 PO 4 /K 2 HPO 4 , pH 7.6, 0.1% Triton X-100, and tested for CAT activity (30), luciferase activity (31), and protein concentration (bicinchoninic acid reagents, Pierce).
Determination of Transfection Efficiency-FTO-2B hepatoma cells and Rat-1 fibroblasts were transiently transfected as described (24) with 31 g of luciferase construct and 25 g of pRSV-n-LacZ, encoding ␤-galactosidase carrying a nuclear translocation signal. 44 hours after transfection, luciferase activity and protein concentration were determined in one well. Cells in the other well were stained for galactosidase activity (32). The specific luciferase activity driven by the promoter of interest (well 1), divided by the transfection efficiency (well 2), is a measure for the activity per transfected cell (24).
Isolation and Transfection of Primary Rat Hepatocytes-Adult Wistar rats were obtained from the Broekman Institute, B.V., Someren, The Netherlands. Hepatocytes were freshly prepared from 17-day-old fetuses as described (33), except that DNase was omitted to prevent degradation of plasmid DNA during transfection. Cells were transfected as described, and plated in rat tail collagen-coated dishes containing Dulbecco's modified Eagle's medium/F-12 and 10% FCS. Nonadhering cells were removed after 5 h. The cultures were incubated in Dulbecco's modified Eagle's medium/F-12 and 2.5% FCS in the absence or presence of added hormones, and harvested after 20 h.
ExoIII Deletions-ExoIII-deletion analysis of the 4-kbp fragment that contains the upstream enhancer (15) (Fig. 5) was performed as described in Promega Protocols and Applications Guide. Constructs B, C, D, and E are derived from construct A, in which the 4-kbp enhancer fragment is cloned in positive direction upstream of the 299-bp CPS promoter-containing fragment (Ϫ161 to ϩ138). Constructs F and G are derived from a plasmid in which the 4-kbp enhancer fragment was cloned in the opposite direction upstream of the CPS promoter. Constructs H, I, and J are subclones derived from constructs D and F, and are cloned in positive direction upstream of the 299-bp CPS promotercontaining fragment.
Site-directed Mutagenesis and Construction of Expression Plasmids-pRc/RSV-CREB (expression vector expressing CREB) and pRc/RSV-CBP (expression vector encoding full-length mouse CBP) were gifts from R. H. Goodman, 6RGR (expression vector encoding glucocorticoid receptor) was a gift from K. R. Yamamoto, and MT-CEV␣ (expression vector encoding C␣) was a gift from G. S. McKnight. Sitedirected mutagenesis was performed using the "megaprimer" method (34) with modifications (35).
All CPS promoter constructs are derived from the 299-bp PstI-HindIII promoter-containing fragment (position Ϫ161 to ϩ138, Fig. 1, HindIII site derived from polylinker of pBluescript SKϩ) described in Ref. 15. To delete sites III and II from the CPS promoter, a 238-bp AflIII-HindIII fragment (position Ϫ100 to ϩ138, Fig. 1) was cloned into pLT1. For deletion of sites III, II, and I, primer CAGCCCCTCCTC-CCTCTAGAATGTCCAGAGATG (complementary to sense-strand) was used to create a XbaI site (underlined) at position Ϫ74 (Fig. 1). The resulting 212-bp XbaI-HindIII fragment was cloned into pLT1. To delete sites III, II, I and the GAG box, a 176-bp AccI-HindIII fragment (position Ϫ38 to ϩ138, Fig. 1) was cloned into pLT1.
Deletions in the 5Ј-UTR of the mRNA were made by creating HindIII sites in the 5Ј-UTR at 50, 89, and 115 nt downstream of the transcription start site of construct 2 (see Fig. 1, construct 2) with oligonucleotides GGGAAGGAAAGCTTTGTGGAGAC, CATGAAAGCTTGTTGTC-CAATTTGC and GTGACTAAGCTTAAATCACAAATATCTC, respectively (all complementary to the sense strand; HindIII sites are underlined). The T7 primer of pBluescript was used as complementary primer. All PCR products were sequenced. The putative CRE and GRE element in the 469-bp enhancer fragment (see Fig. 5, construct J) was mutated after subcloning into pBluescript SKϩ. Primer TTACTTTA-GAATCATATTGAGGACTTATTA was used to disrupt the putative CRE and CATCAGAGAAGTTTGATCTGCTCAGCACAT to disrupt the putative GRE. Underlined nucleotides were substituted (see Fig. 6). The T3 and T7 primers were used to fill in both sides flanking the mutagenesis primer. The PCR products were sequenced in both directions.
RNA Isolation and Quantification-5 ϫ 10 6 FTO-2B hepatoma cells were transiently transfected with 30 g of construct UTR 50, UTR 115, or UTR 138 as described and divided into equal parts. After 48 h the luciferase activity and protein content was determined in the cells of one well, while the cells of the other well were lysed in guanidinium isothiocyanate and loaded onto CsCl cushions for total RNA isolation (36). In addition, the RNA samples were treated with 3 units of RNasefree DNase I (RQ1, Promega) for 1 h at 37°C in the presence of 56 units of RNasin (Promega), extracted with phenol and chloroform, and precipitated two times in 2 M LiCl. The concentration of the RNA was determined at 260 nm and the integrity by formaldehyde-agarose gels. The concentration of luciferase-SV40 mRNA hybrids was determined by reverse transcriptase-PCR (avian myeloblastosis virus-reverse transcriptase, Promega), using a mimic DNA construct as internal standard (37). The mimic construct contains identical SV40 sequences, but, in addition, a 129-bp unrelated fragment. The primers (SV40ϩ, CTGTG-GTGTGACATAATTGG; and SV40Ϫ, TACTAAACACAGCATGACTCA) recognize sites flanking the intron, leading to amplification of a 263-bp fragment from the mRNA, a 328-bp fragment from the unspliced RNA or from the plasmid DNA, and a 465-bp fragment from the mimic construct. In each reaction 1 g of total RNA was used. The products of the RT-PCR reaction were separated on 3% Metaphor-agarose (FMC).

RESULTS
Three regions in the CPS gene appeared to be involved in the regulation of expression, viz. the 161-bp proximal promoter, the 138-bp 5Ј-UTR of the mRNA, and the 4-kbp far-upstream enhancer fragment (15). These regions were systematically investigated for their contribution to tissue specificity and response to physiologically relevant hormones.
The Proximal Promoter of the CPS Gene-Two approaches were used to investigate whether the 299-bp CPS promotercontaining fragment (from Ϫ161 to ϩ138 relative to the transcription start site (15)) confers tissue-specific expression. As a first approach, 5Ј sequences of the CPS promoter were deleted, and the resulting derivatives were tested in FTO-2B hepatoma cells and Rat-1 fibroblasts (Fig. 1). Elements upstream of the GAG box did not substantially alter reporter gene expression in either FTO-2B or Rat-1 cells. Deletion of the GAG box, however, decreased activity to 30 -40%. In both cell lines the CPS promoter appeared to be sensitive to added hormones, possibly due to enhanced expression of general transcription factors. The minimal fully active CPS promoter was 50 times more active in FTO-2B cells than the minimized TK promoter.
Since the activity of the CPS promoter was similar in hepatoma and fibroblast cells (Fig. 1), this sequence probably does not confer tissue specificity. To establish this conclusion more firmly, luciferase activity per transfected cell was determined. This approach (Fig. 2) clearly demonstrated that the activity of the CPS promoter was quantitatively comparable in FTO-2B and Rat-1 cells and, hence, does not confer tissue specificity. The minimized TK promoter, containing only the proximal GC box and a TATA box, was less active in the hepatoma cell lines than in the fibroblasts. The GC box, target of the Sp1 protein and the main determinant of the strength of this promoter (28,39) may therefore be hardly functional in hepatoma cell lines, in accordance with the relatively low concentrations of Sp1 in hepatocytes (40).
The 5Ј-Untranslated Region-A promoter fragment lacking the 5Ј-untranslated region of the mRNA, showed a dramatically lower expression in transfection studies (15). Several pro- Constructs harboring parts of the 5Ј-UTR were transiently transfected in FTO-2B hepatoma cells. The bars indicate the normalized specific luciferase activity ("Experimental Procedures") relative to the activity obtained from the construct containing the full-length 5Ј-UTR and 100 bp of the CPS promoter (UTR 138 ϭ construct 2, Fig. 1), which is set to 1. Other constructs harbor the same upstream promoter sequences, but lack 5Ј-UTR regions ϩ115 to ϩ138 (U 115), ϩ89 to ϩ138 (U 89), ϩ50 to ϩ138 (U 50), and ϩ7 to ϩ138 (U 7) relative to the transcription start site. The small ORF (position 97-108) is present in constructs U 115 and U 138, and absent in the other constructs. Black bars indicate the activity of FTO-2B cells incubated in the absence of added hormones, hatched bars in the presence of dexamethasone, Bt 2 cAMP, and IBMX ("Experimental Procedures"). Error bars indicate the S.E. of at least three independent transfections. B, activity of the 5Ј-UTR of the CPS mRNA in the context of the minimized TK promoter (Ϫ81 to ϩ52). The 5Ј-UTR (ϩ3 to ϩ138, black box) was cloned into the BglII site (position ϩ52) of the TK promoter (gray) in both directions (arrows). The activity of the TK promoter was set to 1. Black bars indicate the activity in the absence of added hormones, hatched bars in the presence of dexamethasone, Bt 2 cAMP, and IBMX. Error bars indicate the S.E. of at least three independent transfections.  moter fragments lacking parts of the 5Ј-UTR were therefore transfected into FTO-2B cells. Fig. 3A shows that the entire UTR was needed for high promoter activity. If only 23 nt were deleted (construct U 115), 80% of the activity was lost. In primary hepatocytes the activity of U 115 was 47 Ϯ 5% (S.E., n ϭ 5) of that of construct U 138, showing that the observation does not result from using established cell lines. The ratio between hormone-stimulated expression and basal expression is comparable for all constructs, indicating that the 5Ј-UTR is not involved in the hormonal sensitivity of promoter. The contribution of the 5Ј-UTR to tissue specificity was tested by trans-fection to Rat-1 fibroblasts and MH1C1 hepatoma cells (Table  I). Although promoter activity was most affected by the deletions of the 5Ј-UTR in FTO-2B, the effects were similar in MH1C1 and Rat-1, indicating that it plays no role in tissue specificity. To test whether the 5Ј-UTR is an independent stimulator of expression, it was cloned downstream of the minimized TK promoter. Fig. 3B shows that this fragment (ϩ3 to ϩ138) is not able to enhance expression in combination with this promoter. To discriminate between transcriptional and post-transcriptional effects on expression, the levels of RNA obtained from the constructs containing the full-length UTR and two deletion constructs were determined (Fig. 4). Whereas the relative concentrations of luciferase-SV40 mRNA hybrids were found to be similar, the full-length UTR gave a 4 -5-fold higher expression than both deletion constructs. This result shows that the 5Ј-UTR affects expression at the post-transcriptional, and probably at the translational level.
Delineation of the Element That Confers Tissue Specificity and Hormone Responsiveness in the Far-upstream CPS-enhancer Fragment-A 4-kbp DNA fragment harboring sequences from Ϫ8 to Ϫ4 kbp relative to the transcription start site of the CPS gene was shown to be transcriptionally active in Sites with similarity to binding sites of liver-enriched factors HNF3 (Ca/tAAa/gTCAATA), HNF4 (GGGCCANNNa/ga/gGTCCA), and HNF5 (Ta/gTTTGc/t) are given in italics.

FIG. 4. Luciferase mRNA levels produced by full-length and truncated 5-UTR constructs in FTO-2B cells.
Constructs U 138, U 115, and U 50 (Fig. 3) were transiently expressed in FTO-2B hepatoma cells and assayed by determination of luciferase-SV40 hybrid mRNA concentration ("Experimental Procedures"). Lanes 1, 4, 7, 13, and 19, U 138; lanes 2, 5, 8, 14, and 20, U 115; lanes 3, 6, 9 transient-transfection assays (15). To delineate these cis-acting elements more precisely, the 4-kbp fragment was subjected to an ExoIII-deletion analysis. The fragments from the ExoIII library were tested in combination with the 161-bp CPS promoter (Fig. 5). None of the fragments caused stimulation of basal expression, but the response to both dexamethasone and the combination of dexamethasone and Bt 2 cAMP was found to be delimited by the 5Ј and 3Ј borders of constructs D and F, respectively (Fig. 5). This region alone (construct H in Fig. 5) was as active as either construct D or F. Further deletions confined the enhancer activity to a 469-bp fragment (construct J in Fig. 5). The differentially methylated CCGG (15) site is located within this fragment. A number of sites resembling known liver-enriched factor binding-sites (HNF3, -4, and -5) and hormone-responsive elements (CRE and GRE) were identified (Fig. 6).
Tissue Specificity of the Far-upstream Enhancer-The same enhancer fragments that caused hormonal induction of reporter gene expression in FTO-2B hepatomas, also stimulated hormone-dependent expression in Rat-1 fibroblasts (Fig. 5), but 6-fold less. This cell-specific difference in activation properties of the enhancer became more pronounced when the 469-bp enhancer fragment was tested in combination with the minimized TK promoter (Fig. 7). The expression obtained with this fragment in FTO-2B cells was 340-fold induced, compared to the TK promoter itself, while only a 12-fold induction was seen in Rat-1 cells, i.e. an almost 30-fold difference. These results show that the tissue specificity conferred by the enhancer is independent of the promoter. The strength of the 470-bp enhancer fragment was also tested in two hepatoma cell lines (FTO-2B and MH1C1) and two fibroblast lines (Rat-1 and CHO-K1) and compared to the endogenous CPS expression in these cell lines. A clear relation between the strength of the enhancer (Fig. 8A) and the level of expression of CPS protein in the different cell lines (Fig. 8B) was observed. The difference in the fold stimulation of the luciferase reporter gene and the endogenous CPS gene probably results from the difference in stability of both gene products.
To verify the validity of the use of hepatoma cells as a model for assessing tissue-specific expression, the enhancer was also tested in freshly isolated hepatocytes. In these cells the CPS promoter was stimulated 4.6 -5.0-fold (n ϭ 2) by dexamethasone, Bt 2 cAMP, and IBMX. When combined with the 469-bp enhancer, the expression was reduced to 27 Ϯ 3% (n ϭ 5) of that of the promoter alone in the absence of added hormones. In the presence of hormones, however, the expression was stimulated 63-79-fold (n ϭ 2) compared to the promoter, the hormonal induction being more than 250-fold.
Expression in Intestinal Cells-The enhancer element at Ϫ6.3 kbp is selectively demethylated in CPS-expressing cells. To test whether sequences surrounding the site at Ϫ4.0 kbp which is partially demethylated in the small intestine only (15), have enhancer properties in intestinal cell lines, they were tested in Caco-2 cells (Fig. 9). Only the Ϫ6.3 kbp site-containing fragment conferred enhanced expression in the presence of dexamethasone, Bt 2 cAMP, and IBMX, both in combination with the CPS promoter and with the minimized viral TK promoter fragment. Combined with the data from other cell lines, we conclude that only the fragment harboring the MspI site at Ϫ6.3 kbp has the capacity to enhance CPS gene expression.
Interaction between the Enhancer and the Promoter-In the presence of added hormones, the enhancer stimulates expression via the proximal promoter. To investigate which elements of the promoter mediate this bridging function, a series of CPS promoter deletions was tested in combination with the 1003-bp enhancer fragment (fragment H in Fig. 5) in the FTO-2B hepatoma cell line. Fig. 10 shows that both the TATA box and the GAG box were needed for an optimal transactivation by enhancer elements. When the GAG box was deleted, transactivation decreased to 50% of that of the fully active promoter. Neither sites I-III nor the 5Ј-UTR are involved in the interaction between the promoter and the enhancer fragment. These effects became even more pronounced when the 469-bp fragment (fragment J in Fig. 5) was used. When tested in combination with the TK promoter (Fig. 10), the hormonal stimulation decreased to 25% of that of the fully active CPS promoter. Obviously, the GC box is not able to functionally replace the GAG box.
When the distance between the enhancer and the minimized TK promoter was decreased, luciferase expression levels increased strongly (Fig. 11). Such distance effects were virtually absent in combination with the CPS promoter (Figs. 5 and 11).  Fig. 5). The TATA box (black rectangle), GAG box, and sites I-III of the CPS promoter (19) are indicated. In gray the minimized TK promoter is indicated, with the TATA box (rectangle) and the GC box. Hormonal inducibility is expressed as the ratio of the activity in the presence of added hormones over the activity in absence of added hormones after correction for the effect of hormones on the promoter-deletion construct.
Whether or not sites upstream of the TATA box of the CPS promoter are functioning as coupling elements between the enhancer and the promoter was tested. A promoter fragment in which all sites upstream of the TATA box were deleted, leaving only the TATA box and 5Ј-UTR (Fig. 1, construct 4) was coupled to either the 1003-bp enhancer fragment (H in Fig. 5) or the 469-bp fragment (J in Fig. 5) and transfected to FTO-2B cells (Fig. 11). The truncated promoter was more sensitive to increasing distance than the CPS promoter, but not as sensitive as the TK promoter. These results underline the role of the GAG box.
Mutational Analysis of cis-Elements in the Enhancer-Expression of the CPS gene in vivo is regulated at the transcriptional level by both glucocorticosteroids and cyclic AMP (8,12). The activity of the upstream enhancer depends entirely on the presence of these hormones and was therefore investigated for the presence of hormone-responsive elements. The 469-bp enhancer contains sequences that qualify as a potential cAMPresponsive element (CRE, Fig. 6), the binding site for dimers of the CREB family, and a glucocorticosteroid-responsive element (GRE, Fig. 6). When 4 nt of the CRE were substituted (CREmut, Fig. 6), basal expression decreased 60% (Fig. 12A). By binding to the regulatory subunit of protein kinase A, cyclic AMP releases the active catalytic subunit (C␣) (41). Expression vectors for C␣, CREB (51,52), and CBP (53,54) were cotransfected with the wild-type enhancer or the CRE mutant. These additions resulted in a 2-fold increase of the expression signal of the wild-type enhancer, but the CRE mutant no longer conferred this effect (Fig. 12A). To functionally test the existence of a GRE, one nucleotide that has been shown to be essential for receptor binding (42) was substituted in each of the three putative GRE half-sites (Fig. 6). The activity of the mutated enhancer (GREmut) was decreased to 20% of that of the wild-type enhancer when tested in combination with an expression vector encoding the GR in the presence of dexamethasone (Fig. 12B). Interestingly, the fold-induction by dexamethasone of the CRE mutant was not affected. DISCUSSION The aim of this study was the functional characterization of the regulatory regions of the CPS gene with respect to their role in tissue-specific expression and hormone sensitivity. Three regions were functionally analyzed. The CPS promoter, comprising a TATA box, a GAG box, and elements I-III, associated with the binding of presently unknown proteins (19), the 138-bp 5Ј-untranslated region of the mRNA, and the far-upstream enhancer at Ϫ6.3 kbp.
Our data show that, in vitro, the CPS promoter has comparable strength in hepatoma cells, which do express CPS, and in fibroblasts, which do not express CPS, demonstrating that it is not involved in tissue-specific regulation of transcription. Fur- FIG. 11. Effects of distance between promoter and enhancer on the effectiveness of the enhancer. The distance (bp) between the TATA box and the center of the 469-bp enhancer fragment is depicted on the horizontal axis and reporter gene activity on the vertical axis. FTO-2B hepatoma cells were incubated in the presence of the hormones dexamethasone, Bt 2 cAMP, and IBMX. All activities were normalized to the activity of the CPS promoter fragment which was set to 1. Circles (q) represent the activity of the CPS promoter containing elements I-III, the GAG box, and the TATA box (construct 1, Fig. 1). Triangles (å) represent the activity of the CPS promoter containing only the TATA box (construct 4 of Fig. 1) and squares (f) represent the activities of the minimized TK promoter (construct TK in Fig. 1). Promoters were combined with the 469-, 547-, 1003-bp, and the 2-kbp enhancer fragments (Fig. 5, fragments J, I, H, and D, respectively). The solid lines are curves proportional to (bp) q . Theoretically, the exponent q is a measure for the probability of a DNA site to be in the vicinity of the second. Mathematically, for random distribution in three dimensions, q ϭ Ϫ1.5 (the Gaussian limit (55)). q (CPS) ϭ Ϫ0.2, q (truncated CPS) ϭ Ϫ1, q (TK) ϭ Ϫ3.5. thermore, our findings clearly show that sites I, II, and III (Ϫ150 to Ϫ79 nt) are not essential for basal promoter activity or transduction of the hormonally induced activation signal from the enhancer. The GAG box, however, is quantitatively important because its absence reduces both the promoter activity and the effect of the enhancer on the promoter 2-3-fold. These data suggest that the protein(s) binding to the GAG box are involved in transducing the activation signal from the enhancer to the promoter as well as in the functioning of the promoter itself.
A promoter fragment lacking the 5Ј-UTR of CPS mRNA (ϩ7 to ϩ138) is hardly active in FTO-2B cells (Ref. 15 and Fig. 3). Such a finding can be due to an incorrect mapping of the transcription start site, or to the involvement of downstream sequences in the basal transcription complex (15). These possibilities can now be virtually ruled out because addition of up to 115 nt of the 5Ј-UTR downstream of the transcription start site does not restore expression levels to those obtained with the full-length 5Ј-UTR (Fig. 3A). When combined with the minimized TK promoter (Fig. 3B), the full-length 5Ј-leader of CPS mRNA inhibits, rather than stimulates expression. This result makes the presence of cis-elements in the 5Ј-UTR DNA, which, in combination with a promoter, are able to stimulate transcription, rather unlikely. Quantification of RNA levels ( Fig. 4) shows that the 5Ј-UTR acts at the translational level. The 5Ј-UTR contains a small upstream open reading frame (uORF) of four codons (Met-Arg-Tyr-Leu) starting with the initiation codon at position 97 (position relative to the transcription start site) in a suboptimal context compared to the more downstream initiation codon of the CPS reading frame (50 and 70% similarity to the consensus sequence (43), respectively), followed by three stop codons. uORFs are thought to be able to suppress translation of downstream cistrons (44 -46). Strikingly, other eukaryotic CPS genes of which the 5Ј-UTR sequences are known, the human CPS I gene (47), the shark CPS III gene (48), and the yeast CPA1 gene (46), also contain one or more uORFs. The uORF of the CPA1 gene is known to suppress translation in a regulated manner. Constructs U 115 and U 138 (Fig. 3) both contain the uORF, but differ markedly in luciferase activity. Deletion of the 23 nt between the uORF and the luciferase start codon affects inter-cistronic length, the sequence context at the uORF stop codon, and secondary structure, parameters which were found to be important for translational control (49).
Scanning sequences up to 12 kbp upstream and 4 kbp downstream of the promoter fragment, only one fragment located at 6 kbp from the transcription start site could be found that had the capacity to stimulate reporter gene expression in FTO-2B cells and Rat-1 fibroblasts (15). This enhancer element has now been confined to a 469-bp fragment (Fig. 5). Transgenic mice, harboring the CPS promoter and 12-kbp upstream DNA in combination with the CAT reporter gene, give rise to hepatocyte-specific expression of CAT mRNA, which co-localizes with the endogenous CPS mRNA in the liver. Mice harboring only the proximal promoter show extremely weak CAT activity, which is not tissue-specific. 2 Combination of in vivo and in vitro results suggests that the proximal promoter and the 469-bp far-upstream enhancer are both necessary and sufficient for tissue-specific CPS expression.
Sites homologous to an imperfect CRE, with the structural characteristics of a so called "low affinity site" (50) and a GRE were found to be essential elements in the minimal enhancer fragment (Fig. 12); in the absense of added hormones, the enhancer was inactive, while mutations of the CRE and GRE significantly decreased hormone-dependent enhancer activity. Expression of the construct carrying the CRE mutation was still responsive to glucocorticoids (Fig. 12B), indicating that the CRE and GRE are not functionally linked. On the other hand, when sequences upstream of position 339, including the CRE (Fig. 6) are deleted (construct E of Fig. 5), the enhancer looses all activity. The deleted area contains nearly perfect consensus sequences for HNF3 and, to a lesser extent, for HNF4.
In our transfection analysis the hormonal stimulation by the CPS enhancer was found to be independent of the distance between the CPS promoter and enhancer, whereas this distance was important when combining TK promoter and enhancer (Fig. 11). This promoter-specific effect is probably highly relevant in vivo, because the CPS promoter and the enhancer are approximately 6-kbp apart. One way to explain the differences is to hypothesize that transcriptional activity is related to the probability of the enhancer to contact the promoter. The curves, proportional to (bp) q , represent this probability and exponent q the slope in a double logarithmic plot (Fig. 11). Mathematically, when the promoter and the enhancer are randomly distributed in space, q ϭ Ϫ1.5 (the Gaussian limit (55)). q Ϸ 0 (CPS promoter) means that the probability is ϳ1, i.e. that the promoter and the enhancer are connected independently of distance. q ϭ Ϫ1 (truncated CPS promoter) approaches a random distribution and q ϭ Ϫ3.5 (TK promoter) indicates hindrance of enhancer-promoter interaction. The main difference between the CPS and the truncated CPS promoter is the GAG box (see also Fig. 10), indicating that this motif is instrumental in conferring the hormonal activation of the enhancer to the transcriptional complex. The main difference between the minimized TK promoter and the truncated CPS promoter is the GC box. This GC box might be the actual cause of the distance dependence of the minimized TK promoter in conjunction with the CPS enhancer, possibly by preventing interaction of enhancer and promoter through steric hindrance.
In summary, our data support a model in which the tissuespecific expression of CPS is determined by the far-upstream enhancer. The activity of this enhancer is strictly dependent on the presence of glucocorticoids and cyclic AMP. The activated enhancer complex will stimulate transcription through interaction with factors bound to the GAG box and the TATA region.