The upstream regulatory region of the carbamoyl-phosphate synthetase I gene controls its tissue-specific, developmental, and hormonal regulation in vivo.

The carbamoyl-phosphate synthetase I gene is expressed in the periportal region of the liver, where it is activated by glucocorticosteroids and glucagon (via cyclic AMP), and in the crypts of the intestinal mucosa. The enhancer of the gene is located 6.3 kilobase pairs upstream of the transcription start site and has been shown to direct the hormone-dependent hepatocyte-specific expression in vitro. To analyze the function of the upstream region in vivo, three groups of transgenic mice were generated. In the first group the promoter drives expression of the reporter gene, whereas the promoter and upstream region including the far upstream enhancer drive expression of the reporter gene in the second group. In the third group the far upstream enhancer was directly coupled to a minimized promoter fragment. Reporter-gene expression was virtually undetectable in the first group. In the second group spatial, temporal, and hormonal regulation of expression of the reporter gene and the endogenous carbamoyl-phosphate synthetase gene were identical. The third group showed liver-specific periportal reporter gene expression, but failed to activate expression in the intestine. These results show that the upstream region of the carbamoyl-phosphate synthetase gene controls four characteristics of its expression: tissue specificity, spatial pattern of expression within the liver and intestine, hormone sensitivity, and developmental regulation. Within the upstream region, the far upstream enhancer at −6.3 kilobase pairs is the determinant of the characteristic hepatocyte-specific periportal expression pattern of carbamoyl-phosphate synthetase.

The ornithine cycle converts ammonia, originating mainly from amino acid metabolism, into urea for excretion. This conversion is vital, because ammonia is toxic to vertebrates. All five ornithine cycle enzymes are active predominantly in the periportal hepatocytes (1,2,68). Parts of the cycle, however, are also found in the intestinal mucosa and the kidney, thereby creating an interorgan biosynthetic pathway for arginine biosynthesis (reviewed in Refs. 1, 3, and 4). The first and, under most conditions, major flux-determining step of the ornithine cycle, the conversion of ammonia, bicarbonate, and ATP into carbamoyl phosphate, is catalyzed by carbamoyl-phosphate synthetase I (CPS; EC 6.3.4.16) 1 (1,5). The expression of CPS mRNA can be detected in the liver of the rat from embryonic day 15 (ED15) onwards. Expression is initially found only in a few hepatocytes, but toward the end of the fetal period all hepatocytes have been recruited to express CPS (6 -8). Subsequently, the expression gradually becomes confined to the hepatocytes surrounding the portal veins (8 -10). CPS enzyme and mRNA levels in liver change in parallel under experimental conditions that change circulating glucocorticosteroid and hepatic cyclic AMP levels (3,(11)(12)(13)(14), suggesting that the zonal restriction and hormonal modulation of CPS expression are regulated at the level of transcription. Accordingly, it was shown that glucocorticosteroids and cyclic AMP enhance transcription of the CPS gene in prenatal and adult hepatocytes (3, 14 -17). The only other cells producing CPS mRNA and protein are the enterocytes of the small intestine that express CPS from ED13 onward (8,11). The regulation of the expression of CPS in the liver and the intestine differs in the sense that the developmental changes in both organs do not occur in parallel and that hormonal modulation is not seen in the intestine (3,11).
The CPS gene is a single-copy, 110-kilobase pair (kbp) gene that contains 38 exons and is surrounded by matrix-attachment regions (18 -21). The mRNA is 5546 nucleotides (nt) 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-untranslated region of 906 nt (18). 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 (20). DNase I footprint analysis of this region revealed three protected sites (22,23), but the actual identity of the factors occupying these sites is not known (24). Between the TATA motif at position Ϫ21 and protected site I, a so-called "GAG" element was identified (24), which resembles the element recognized by the TFIIIA-like class of transcription factors, which includes Sp1, but is not a target of Sp1 itself. The promoter alone does not confer tissue specificity in transient transfection experiments (25).
Scanning CPS gene sequences between 12 kbp upstream and 4 kbp downstream of the transcription-initiation site in in vitro transfection experiments revealed the existence of only a single enhancer region (20,25,26). This 469-base pair (bp) region harbors a cyclic AMP-response element (CRE) and a glucocorticosteroid-response element (GRE), and confers strong hormone-dependent, tissue-specific expression (25). Analysis of the interaction between the enhancer and the CPS promoter revealed that, in addition to the TATA-box, the GAG-box is instrumental in conferring the enhancer activity (25,26).
To study the regulatory potential of the upstream region including the far upstream enhancer in vivo, two groups of transgenic mice were generated: CPSK mice in which a 1.8-kbp upstream region including the proximal promoter drives reporter gene expression and CPSL mice in which a 12-kbp 5Ј-upstream region, including both the promoter and the far upstream enhancer, drives reporter gene expression. Four characteristics were used to assess the regulatory role of these sequences in CPS expression in vivo: organ specificity, topography of expression within the liver and the small intestine, responsiveness to hormonal treatment, and developmental changes in reporter mRNA levels. To test the hypothesis, that the far upstream enhancer is the major determinant in the regulation of CPS gene expression, CPSE mice were generated, in which the luciferase reporter gene is driven by the 469-bp enhancer in conjunction to the minimized CPS promoter.

EXPERIMENTAL PROCEDURES
Animal Care-Animals were housed with a 12-h light and 12-h dark cycle, and permitted ad libitum access to water and standard pellettype diet. This study was performed in accordance with the Dutch guidelines for the use of experimental animals.
Generation and Screening of Transgenic Mice-Three constructs were used for the generation of transgenic mice (see Fig. 1). The constructs are based on a pBluescript-derived vector containing the promoterless chloramphenicol acetyltransferase (CAT) reporter gene (CPSK and CPSL) or the promoterless modified luciferase gene (CPSE; 1.8-kbp HindIII/EcoRI lucϩ fragment of pSPlucϩ; Promega, Madison, WI) in conjunction with the SV40 small t-antigen intron and polyadenylation signal optimized for expression (20,27). The CAT-reporter gene constructs are flanked by NotI restriction sites, and the lucϩ reporter gene cassette is flanked by KpnI (5Ј) and SstI (3Ј) to allow removal of vector sequences. The CPSK construct contains genomic rat CPS I gene sequences from Ϫ1.8 kbp (BamHI) to 138 bp (MnlI) upstream of the CAT gene. The CPSL construct contains genomic rat CPS I gene sequences from Ϫ12 kbp (BamHI) to 138 bp (MnlI) upstream of the CAT gene. The CPSE construct contains the 469-bp genomic rat CPS I enhancer fragment (25) in conjunction to genomic rat CPS I gene sequences from Ϫ161 bp (PstI) to 138 bp (MnlI) upstream of the lucϩ gene (see Fig. 1). The isolated NotI fragments (CPSK and CPSL) or KpnI/SstI fragment (CPSE) were injected into the pronuclei of zygotes of C57Bl6/N mice (CPSK and CPSL) or FVB mice (CPSE), and the injected embryos were implanted in pseudopregnant foster mothers (28). Transgenic offspring were identified by polymerase chain reaction (PCR) on tail-tip lysates (29), using primers specific for SV40 sequences (SV40i, CAG GCA TAG AGT GTC TGC; SV40pA, CTG GGG ATC CAG ACA TGA). Copy numbers of constructs integrated in the genome were determined by Southern blot analysis of 10 g of tail-tip DNA digested with PstI. Calibration standards consisted of 10 -500 pg of the 1.4-kbp PstI fragment of the SV40 sequences of the expression vector hidden in 10 g of tail-tip DNA of a nontransgenic mouse.
Treatment of Animals-To induce CPS expression in the liver and to reverse its portocentral gradient of expression (13), 48-h starved or diabetic CPSL animals were injected intraperitoneally with dexamethasone phosphate (3 mg/kg body weight) 16 h before sacrifice.
Diabetes was induced by intraperitoneal injection of streptozocin (Sigma; 200 mg/kg body weight) in 24-h starved animals. Thereafter the animals were allowed free access to food. 30 h after streptozotocin treatment dexamethasone phosphate (3 mg/kg of body weight) was injected intraperitoneally, and 16 h before sacrifice the diabetic condition was diagnosed by measuring glucosuria (Haemo-Glucotest 1-44R; Boehringer Mannheim). Treated animals showed concentrations higher than 44 mM, whereas untreated animals showed concentrations below the detection limit (1 mM).
CAT Assay-CAT activity in the organs was determined in tissue homogenates as described by Seed and Sheen (30) with modifications (31). The detection limit for CAT activity was defined as three times background (50 cpm), which is equivalent to approximately 0.05 milliunit. Calibration curves for the amount of CAT activity present in the samples consisted of 0 -50 milliunits of CAT (Sigma), supplemented with 100 g of total liver protein from a negative C57Bl6/N mouse. CAT activity is expressed as picomoles of product/min⅐mg protein (37°C).
Luciferase Assay-To determine the luciferase activity, organs were immediately frozen in liquid nitrogen after isolation. Homogenates were prepared at 0°C by use of a Potter-Elvehjem type homogenizer in 100 mM KPO 4 (pH 7.8), 10% glycerol, 0.2% Triton X-100, 1 mM EDTA, and 1 mM dithiotreitol. After centrifugation at 4°C in a microcentrifuge (Beckman) for 5 min, total protein content in the supernatant was measured using the bicinchoninic acid protein assay reagent of Pierce. In Situ Hybridization-Freshly isolated tissues and embryos were fixed and embedded as described (31). Serial sections of 7-m thickness were probed for the presence of CAT, CPS, luciferase, glutamine synthetase (GS) and phosphoenolpyruvate carboxykinase (PEPCK) mRNAs by in situ hybridization with the respective 35 S-labeled cRNAs as described (32) with modifications (33). The CAT probe, a 1065-bp fragment (Ksp632I/HindIII) of pCT1 and encoding the CAT gene, was subcloned into pBluescript. Double-labeled ([ 35  Hatched box, the CPS promoter (Ϫ161 to 138); white box, the CAT or luciferase reporter gene and SV40 sequences added for RNA processing. Black box, the 469-bp far upstream enhancer element. Gray box, the translated part of the first exon of the CPS gene (ϩ139 to 266) that is absent in the constructs. exposure to autoradiographic emulsion (Ilford Nuclear Research Emulsion G-5 (Cheshire, UK)) for 6 days and development for 8 min.
General Methods-Chromosomal DNA isolation, Southern blot analysis, PCR, subcloning, restriction analysis, RNA isolation, and Northern blot analysis were performed as described (36). For quantitation of total RNA, glycogen and/or residual DNA was removed by precipitation in 2 M LiCl (16 h, 4°C). Quantitative Northern blot analysis was performed using a Phosphor Imager (Molecular Dynamics, Sunnyvale, CA).

RESULTS
Transgenic Lines and Organ-specific Expression of the Reporter Gene-To determine the function of the promoter and upstream region in vivo, "CPSK" and "CPSL" mice were generated ( Fig. 1). Ten CPSK founder mice and five CPSL founder mice were available. No CAT activity was detected in the livers (Ͻ0.05 milliunit/mg) of the CPSK founder mice.
One line, CPSL line 1 (CPSL1), yielded high levels of CAT activity in the liver (6384 Ϯ 762 milliunits in homozygous animals), less activity in the small intestine (5.0 Ϯ 0.6 milliunits in homozygous animals), and no activity in spleen, kid-ney, or colon. The copy number of constructs integrated in the diploid genome of homozygotes was 28. Four other CPSL lines (lines [2][3][4][5] showed no activity in the organs tested, except for CPSL line 2, which showed a relatively strong CAT activity of 309 Ϯ 57 milliunits/mg protein in the small intestine. At least 30 copies per diploid genome were estimated to have become integrated in this line. Southern blot analysis revealed that no rearrangements or deletions occurred in the transgenic DNA of line 2 (not shown). In situ hybridization of the liver of line 3 showed a very weak, periportally localized expression of CAT mRNA (not shown).
Northern blot analysis of liver and small intestine of CPSL1 showed that both CPS and CAT mRNA are present in the liver, and at about 10% of the hepatic level in the intestine ( Fig. 2 and see Fig. 5). This means that the large difference in CAT activity between liver and intestine (more than 1000-fold) is only partly due to the difference in steady-state mRNA levels and also results from differences in post-transcriptional regulation. The major CAT mRNA products formed had the expected size of 1.6 kbp. Neither hetero-nor homozygous mice showed any signs of aberrant phenotype, indicating that no DNA integration in essential genomic regions, including the CPS gene itself, occurred. No deletions or rearrangements of the integrated DNA were observed by Southern blot analysis.
Localization of Reporter mRNA in Liver and Small Intestine-To gain further insight into the role of the 5Ј-regulatory region of the CPS gene with respect to its characteristic intrahepatic pattern of expression, the localization of the CAT mRNA in homozygous mice of the CPSL1 line was visualized by in situ hybridization. As shown in Fig. 3, A-C, the distribution of CAT mRNA was identical to the expression pattern of CPS: It was only found in the periportal region of the liver and absent in the GS-positive hepatocytes surrounding the central veins. In situ hybridizations of the small intestine of CPSL1 showed that the CAT and CPS mRNA were both expressed mainly in the crypts of the intestinal epithelium (Fig. 4). A similar expression pattern of CAT and CPS mRNA in the intestine was observed in CPSL line 2 (not shown). PEPCK mRNA, as a control, was only accumulated in the upper half of the villus.
Hormone Responsiveness of Reporter mRNA Expression-Administration of glucocorticosteroids (dexamethasone) to starved rats not only induced CPS mRNA, but also reversed its lobular portocentral gradient in expression (13). Under this condition, the layer of GS-positive cells around the central veins did express CPS. To investigate the role of the 5Ј-upstream region in hormone responsiveness, starved mice were treated with dexamethasone. In situ hybridization showed that both CPS and CAT mRNA concentrations were highest adjacent to the GS-positive hepatocytes surrounding the central veins (Fig. 3, D-F), that is, the mRNA gradient of both CAT and CPS was reversed compared to the control condition (Fig. 3A). In contrast to the expression pattern in rat liver, neither CPS nor CAT mRNA were present in the GS-positive hepatocytes.
To gain a more quantitative insight into the hormone response, mRNA concentrations in the liver of homozygous dexamethasone-treated diabetic mice of the CPSL1 line were compared to those in control mice by Northern blot analysis. CPS gene expression in rats was shown to be stimulated under these conditions in the liver, but not in the intestine (11). Fig. 5 shows that both CPS and CAT mRNA levels were increased 3-4-fold in liver when compared to controls, whereas the levels in the intestine were decreased. This analysis shows that the hormone responsiveness of the CAT reporter gene is identical to that of the CPS gene in both liver and intestine.
Expression of the Reporter Gene during Development-In rat, CPS mRNA can be detected in liver from ED15 onward, and in intestine from ED13 onward. To investigate whether the 5Јupstream region of the CPS gene determines the developmental onset of CPS expression, homozygous ED14 embryos of CPSL1 were investigated. In situ hybridizations clearly showed that both CPS and CAT mRNA were expressed more abundantly in the small intestine than in liver at this stage of development (Fig. 6). In liver, the CPS probe caused a relatively weak signal when compared to the CAT probe. GS, as a control, was negative in intestine but positive in liver (37,38). In situ hybridization of whole body slices of neonates (0 neonatal days; ND0) further showed that the distribution of the expression of CAT mRNA was restricted to the liver and the enterocytes of the small intestine, that is, similar to the pattern of CPS mRNA (Fig. 7). The CAT signal in the liver had, however, become weak when compared to the signal of CPS.
To compare the developmental changes in CPS and CAT gene expression in the liver and intestine more quantitatively, Northern blot analysis was performed using total RNA of livers and intestines of homozygous transgenics ranging in age from 4 days before birth (ED15) to 2 months after birth. The developmental changes in hepatic CPS mRNA concentrations were found to be similar to those of the CAT reporter gene (Fig. 8A). For both mRNAs, upsurges in concentration were found perinatally and just before weaning. The difference in mRNA levels at ND0, also seen in the in situ hybridization (Fig. 7), and in the preweaning period, probably reflects the difference in half-life of both messengers. CPS mRNA is relatively stable with a half-life of 6 -12 h (39), whereas CAT mRNA is thought to be less stable (40). In contrast to the liver, the relative CPS mRNA levels in intestine decreased linearly with time, without any fluctuations (Fig. 8B). The CAT mRNA levels showed a similar pattern, except that its concentration was low compared to that of CPS mRNA in the perinatal period. This small discrepancy was also observed in liver.
The Far Upstream Enhancer in Vivo-Within the 12-kbp upstream region, present in the CPSL construct, a single region of 469 bp showed hepatocyte-specific and hormone-dependent enhancer activity in transient transfections (20,25,26). To test the activity of this 469-bp enhancer fragment in vivo, construct CPSE (Fig. 1) was used to generate transgenics. Eight lines were tested for organ-specific luciferase reportergene activity (Table I). Five lines showed liver-specific luciferase activity, and three lines showed no activity in the organs tested. No relationship was found between the luciferase expression levels and the copy numbers of inserts integrated in the genome (Table I). Lines 4 and 5 showed a strong activity in the liver, while a weak activity (Ͻ1%) was found in other organs (Fig. 9). Northern blot analysis revealed presence of luciferase reporter mRNA of the expected size (2.5 kbp). In situ hybridization of serial sections of the liver of CPSE line 5 showed that the luciferase and CPS mRNA were both localized in the periportal region, and not in the pericentral region where GS is expressed (Fig. 10). Luciferase mRNA in liver sections of line 4 was also localized in the periportal region (not shown), although the signal was close to the background level. Therefore the 469-bp enhancer, in combination with the minimized promoter, has the capacity to activate the reporter gene in the periportally localized hepatocytes.
Intracellular Localization of the Reporter mRNA-Although the hepatic expression pattern of CPS and CAT mRNA in CPSL1 and the CPS and luciferase mRNA in CPSE lines 4 and 5 were identical, their intracellular localization differed. CAT and luciferase mRNA was seen mainly inside the nuclei, whereas CPS was found, as expected, in the cytosol. Other transgenics, in which the CAT/SV40 reporter gene is driven by the GS promoter/enhancer (31) or in which a rat GS/SV40 hybrid gene is driven by the PEPCK promoter/enhancer 2 showed a similar subcellular distribution. This observation might therefore result from the use of the SV40 small t intron in the reporter gene transcript. Reverse transcriptase-PCR analysis of total RNA obtained from transiently transfected hepatoma cells expressing the same reporter gene construct showed that the major product is the unspliced transcript and, therefore, probably retained in the nucleus (Fig. 11). A poor splicing efficiency, therefore, appears to explain why the major part of the CAT mRNA in CPSL1 and CPSE line 4 and 5 mice was seen in the nucleus. DISCUSSION The Promoter Region of the CPS Gene Is Not Sufficient to Drive Gene Expression in Vivo-Transient transfections in hepatoma cells and fibroblasts already showed the promoter to be active, but not able to confer tissue specificity (25). Reporter gene expression in transgenics harboring the promoter region (Ϫ1.8 kbp to 138 bp; CPSK) was hardly detectable and did not correspond to the organ distribution of CPS expression. Similar negative results were obtained with transgenics in which the proximal promoter of the glutamine synthetase gene drove reporter gene expression (31). Addition of upstream sequences, including the far upstream enhancer, resulted in high levels of reporter-gene expression. This shows that the proximal promoter alone is not sufficient to drive gene expression in vivo. Possibly, higher order structures in the chromosomal DNA, like nucleosomes, result in promoters that depend on enhancers for their activation (41,42).  Fig. 3A). GS expression is expressed in liver, but absent in intestine. li, liver; i, intestine; k, kidney; c, colon. Magnification, ϫ 50. developmental appearance, were met, showing that the major determinants of the regulation of CPS gene expression reside within 12 kbp of the upstream region of the gene. Both the hormone responsiveness and the developmental changes in CPS gene expression differ between liver and intestine (Figs. 5 and 8). In liver, CPS mRNA levels respond to the developmental and adaptive changes in circulating glucocorticosteroids and intracellular cyclic AMP concentrations, whereas little response is seen in the intestine. The linear postnatal decrease in intestinal CPS (and CAT) mRNA during development (Fig. 8B) correlates with the concomitant retreat of these mRNAs from the villus to the crypts. 3 Since total small-intestine homogenates were used in this assay, fewer CPS mRNA expressing enterocytes will be present in assays of older animals.

The Upstream Region of the CPS Gene Determines the Spatiotemporal Pattern of Expression in Liver and
Enhancer activity in the 12-kbp region has been reported in two independent studies (20,26), both establishing the existence of only a single enhancer in this region. Even though the sequence of the enhancer in both studies is identical, its location is not: we have positioned the enhancer at Ϫ6.3 kbp, whereas Goping et al. (26) mapped it at Ϫ10 kbp. Southern blot analysis of rat genomic DNA digested with HpaII and BamHI showed unequivocally that the uniquely demethylated HpaII site in hepatic genomic DNA that characterizes the 469-bp enhancer (25,43) is positioned at Ϫ6.3 kbp. In this analysis the EcoRI/BamHI fragment from the 4.6-kbp EcoRI fragment containing the first exon was used as probe. The erroneous mapping of the enhancer at Ϫ10 kbp by Goping et al. (26) probably arose as a result of an inversion of the HindIII fragment between Ϫ5 and Ϫ11.4 kbp.
The Far Upstream Enhancer Fails to Activate Gene Expression in the Intestine-Analysis of the CPSE transgenics showed that the 469-bp enhancer element determines liver specificity, but fails to drive intestinal expression. This finding is in line with the known difference in regulation of CPS expression in liver and intestine during development and hormonal treatment (Figs. 5 and 8). As a consequence, separate elements should exist within the 12-kbp upstream region that are responsible for liver-specific and intestine-specific expression. The use of distinct regulatory units to regulate expression in different tissues is also seen in other genes. The promoter and proximal upstream region of the ornithine transcarbamoylase gene is sufficient to drive expression in the small intestine, while addition of the far upstream enhancer is necessary for liver-specific expression (44). The hormone-responsive hepatocyte-specific region of the PEPCK gene is located between Ϫ457 and 69 bp relative to the transcription-initiation site, whereas the adipocyte-specific region is found between Ϫ2086 to Ϫ888 bp (45; and references therein). The putative intestine-specific element in the upstream region of the CPS gene was not iden-   (20,25,26). One possible location might be at 4.0 kbp upstream of the transcription start, where a HpaII site is found that is demethylated in intestine only (20). However, this region showed no enhancer activity in transient transfections to the enterocyte-like Caco-2 cells (25). The exact location of the intestine-specific enhancers therefore remains uncertain. The Mechanism Underlying Liver-specific Periportal Expression of the CPS Gene-Transient transfection assays showed that the enhancer is responsible for the hormone-dependent hepatocyte-specific expression of the CPS gene (25). A functional CRE and a functional unit of GREs were identified in this region, as well as sites which show strong similarity with the consensus hepatocyte nuclear factor-3 (HNF-3) binding sites (46) in close proximity to the GREs, and a site similar to a HNF-4 binding site, which is relatively close to the CRE (25). The 469-bp CPS far upstream enhancer combined with the minimized promoter drives reporter-gene expression selectively in the periportal region of the liver. However, the relative contribution of the hormone-mediated signal transduction pathways and of the liver-enriched transcription factordependent cascade in determining the periportally localized expression remains to be established.
CPS belongs to a set of liver-specific genes, including the other urea-cycle genes, many gluconeogenetic genes (e.g. PEPCK) and amino acid-catabolizing enzymes (e.g. tyrosine aminotransferase), which are expressed selectively in the periportal region of the liver and which are activated by cyclic AMP and glucocorticosteroids (reviewed in Refs. 2, 3, and 47). Both glucocorticosteroid levels and the ratio of glucagon over insulin decrease along the portocentral axis of the liver lobule (47), and might therefore be involved in establishing the portocentral gradient in gene expression. However, the pericentral enzyme ornithine aminotransferase (48) responds to glucocorticoids and glucagon in a similar way as periportal amino acid-metabolizing enzymes (47,49). Furthermore, the glucagon receptor is preferentially expressed in the pericentral region (50). Finally, the regulatory DNA regions of the periportally expressed ureacycle enzyme ornithine transcarbamoylase gene (51) that are responsible for liver-specific expression, do not contain hormone-responsive elements (44). Therefore, glucagon (via cyclic AMP) and glucocorticosteroids alone are not sufficient to explain selective periportal expression. In addition to the hormone-responsive elements, the activation of the CPS gene, but also the PEPCK (52,53), the tyrosine aminotransferase (54,55), and the ornithine-transcarbamoylase gene (44), is mediated by liver-enriched transcription factors (for reviews, see Refs. 56 -58). However, transcription factors HNF-3␥ and C/EBP␤ have been reported to accumulate only at slightly higher levels in the periportal region, while HNF-1␣ and Dbinding protein are distributed evenly throughout the liver, and HNF-4 and C/EBP␣ are expressed preferentially in the FIG. 12. Schematic model for the activation of hormone-dependent hepatocyte-specific activation of the CPS gene. A, in the presence of hormones (cyclic AMP and glucocorticosteroid), activated transcription factors CREB (hatched) and the glucocorticosteroid receptor (gray) will contribute, in synergism with liver-enriched factors HNF-3 and HNF-4 (46,55), to an activation complex. B, the complex (hatched) interacts with the promoter region through the putative GAG element-binding protein (black) (25,26) and through the CREB-binding protein (CBP) (65,66) and TFIIB (65). A DNA region, which showed a high affinity for the nuclear matrix (matrix-attachment region, MAR) is located midway between both regulatory regions (20), apparently causes no sterical hindrance, and may even function as a hinge. pericentral region (59,60). The contribution of these factors to the development of portocentral gradients in gene expression will therefore be small at most. Unless new factors are involved, the observed portocentral gradients in gene expression must arise from interactions between the factors discussed. Fig. 12 presents such a still hypothetical model for the hormonedependent hepatocyte-specific periportal expression of the CPS gene that is based on the presently available information. Hormone-activated factors, like phosphorylated CRE-binding protein (61,62) and ligand-bound glucocorticosteroid receptor (63) are available in the periportally localized hepatocytes. Glucocorticosteroid activation leads to recruitment of HNF-3 to its site (46,55,64). The transduction of cyclic AMP-dependent liver-specific activation depends on CRE, possibly in synergism with the HNF-4 site (55). This results in the formation of an activation complex at the far upstream enhancer. The complex will interact with the putative GAG element-binding protein (25,26) and with transcription factor IIB (65) via the CREBbinding protein (65,66), leading to initiation of transcription (Ref. 67 and references therein). A region in between the enhancer and the promoter shows affinity for the nuclear matrix in different assays (20). This matrix-attachment region is not present in the CPSE construct, and is therefore not essential for activation. Its position in the CPS gene, midway between the enhancer and the promoter (20) might even facilitate loop formation by functioning as a hinge.
In summary, this report establishes that the 12-kbp upstream region of the CPS gene is essential in the organ specificity of its expression, the pattern of its expression within the liver, and the hormonal and developmental regulation of its expression. Within this upstream region the 469-bp enhancer determines the hepatocyte-specific expression and periportal localization within the liver.