Characterization of the Aldolase B Intronic Enhancer*

The aldolase B gene is transcribed at a high level in the liver, kidney, and small intestine. This high level of gene expression results from cooperation between a weak but liver-specific promoter and an intronic activator. A deletional study of this activator present in the first intron allowed us to ascribe the maximal enhancer function to a 400-base pair (bp) fragment (+1916 to + 2329). This enhancer is highly liver-specific and enhances the activity of heterologous minimal promoters in a position and distance-independent fashion in transiently transfected Hep G2 hepatoma cells. The aldolase B enhancer is composed of two domains, a 200-bp module (Ba) inactive by itself but which synergizes with another 200-bp module (Bb) that alone retains 25% of the total enhancer activity. The Bb sequence is 76% homologous between human and rat genes and contains several binding sites for liver-enriched nuclear factors. By electrophoretic mobility shift assays, we demonstrated that elements 5 and 7 bind hepatic nuclear factor 1 (HNF1), whereas element 2 binds hepatic nuclear factor 4 (HNF4). A functional analysis of the enhancer whose elements have been mutated demonstrated that mutation of any of the HNF1 sites totally suppressed enhancer activity, whereas mutation of the HNF4-binding site reduced it by 80%.

The aldolase B gene is transcribed at a high level in the liver, kidney, and small intestine. This high level of gene expression results from cooperation between a weak but liver-specific promoter and an intronic activator. A deletional study of this activator present in the first intron allowed us to ascribe the maximal enhancer function to a 400-base pair (bp) fragment (؉1916 to ؉ 2329). This enhancer is highly liver-specific and enhances the activity of heterologous minimal promoters in a position and distance-independent fashion in transiently transfected Hep G2 hepatoma cells. The aldolase B enhancer is composed of two domains, a 200-bp module (Ba) inactive by itself but which synergizes with another 200-bp module (Bb) that alone retains 25% of the total enhancer activity. The Bb sequence is 76% homologous between human and rat genes and contains several binding sites for liver-enriched nuclear factors. By electrophoretic mobility shift assays, we demonstrated that elements 5 and 7 bind hepatic nuclear factor 1 (HNF1), whereas element 2 binds hepatic nuclear factor 4 (HNF4). A functional analysis of the enhancer whose elements have been mutated demonstrated that mutation of any of the HNF1 sites totally suppressed enhancer activity, whereas mutation of the HNF4-binding site reduced it by 80%.
Aldolase B, one of the three known aldolase isoenzymes, is the only expressed isoform in highly differentiated hepatocytes (1) and is also found in kidney and small adult intestine where it is associated with aldolases A or C (2). Aldolase B catalyzes the reversible cleavage of fructose 1-phosphate into dihydroxyacetone phosphate and glyceraldehyde; therefore, it is involved in both glycolytic and gluconeogenic pathways. In human, hereditary fructose intolerance is a potentially fatal autosomal recessive disease resulting from aldolase B deficiency. In addition aldolase B gene transcription is regulated by hormones and diet; it is partially repressed by glucagon and cyclic AMP and stimulated 4-fold by glucose and insulin (3).
The aldolase B gene proximal promoter was shown to be liver cell-specific as judged from transient transfection experiments in the hepatoma cell line Hep G2 and in hepatocytes in primary culture (4). However the promoter activity of the Ϫ192-bp 1 5Ј-flanking fragment was always very low in these cells. Recently we explained this result by a dominant restriction of the transcriptional activity due to binding of the hepatic nuclear factor 3 (HNF3) to the PAB element of the promoter (5). This PAB element binds in a mutually exclusive fashion either hepatic nuclear factor 1 (HNF1), which stimulates promoter activity, or HNF3 that, on the contrary, restrained the aldolase B promoter activity (6). In transgenic mice transgenes directed by the Ϫ232-bp proximal promoter fragment were totally silent. The addition of 1.8 kb of sequences located in the first intron of the aldolase B gene (ϩ685 to ϩ2514) led to a 50-fold stimulation of the promoter activity ex vivo in Hep G2 cells and allowed for a correct, tissue-specific expression in transgenic mice (7).
The purpose of this work was to delineate the minimal intronic fragment responsible for the enhancer activity and to characterize DNA elements and cognate trans-acting factors involved in both activity and tissue specificity of this enhancer. Among various DNA elements detected in a 400-bp fragment endowed with a liver cell-specific enhancer activity, two HNF1binding sites were shown to be indispensable for this activity. In addition, a conserved HNF4-binding site also behaved as a positive cis-acting element of the enhancer.

MATERIALS AND METHODS
Plasmid Constructions-For generation of the internal deletion, we started with the Ϫ232 A100B/CAT construct previously described (7). The Ϫ232A100B1200/CAT and Ϫ232A100B600/CAT plasmids were obtained by excision of a fragment between sites StuI (located within the B fragment) and KpnI or BamHI (located within the plasmid linkers). The Ϫ232A100Ba/CAT, Ϫ232A100Bb/CAT, Ϫ232A100Bc/CAT, and Ϫ232A100Baϩb/CAT plasmids were obtained by cloning in both orientations, into the SmaI site of the Ϫ232A100/CAT plasmid (7), the fragment of interest generated by polymerase chain reaction. The aϩb polymerase chain reaction fragment was also subcloned in the AflIII site (located upstream from the promoter) or in the ClaI site (located downstream of the CAT gene) of the previously described pECAT vector (4). Then the various promoter fragments Ϫ232 A100 (7), Ϫ194 to ϩ 14 (4) of the aldolase B gene, or Ϫ183 to ϩ 11 of the pyruvate kinase gene (8), or Ϫ105 to ϩ51 of the herpes simplex thymidine kinase (9) gene were excised and subcloned in one or both of these two plasmids.
Plasmids with block mutations in elements 5 and 7 or deletions in elements 2 and 4 were constructed by inserting the mutated fragments, obtained by a two-step polymerase chain reaction procedure (5,10), in the SmaI site of the Ϫ232A100 CAT plasmid. Sequence details on the block mutations are given in Fig. 4.
All constructs were checked by DNA sequencing. The primer sequences used are available upon request.
Cell Culture and Transient Transfection-Hep G2 cells were grown in Dulbecco's modified medium in the presence of 10% (v/v) fetal calf serum, 1 M L-triiodothyronin, 1 M dexamethasone, 10 nM insulin, at 37°C in 5% (v/v) CO 2. Mouse 3T6 cells were grown under the same conditions without hormones.
Transfection were carried out by the calcium phosphate method (11), in experimental conditions previously described (5). In each experiment 7.5 g of the CAT plasmids and 2 g of the luciferase plasmid were cotransfected. The pRSV luciferase standardization plasmid was used to monitor variations in transfection efficacy. Chloramphenicol acetyl-* This work was supported in part by grants from La Ligue Nationale Contre le Cancer, l'Association de Recherche sur le Cancer et le Ministère de l'Education Nationale, and la Technologie et de la Recherche. 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.
Generation and Analysis of Transgenic Mice-The DNA constructs were digested with restriction enzymes ClaI (cutting in 3Ј in the vector) and HindIII (cutting in 5Ј in the plasmid linker). The fragments of interest were isolated by electrophoresis, electroeluted, and purified by using elutip-d columns (Schleicher & Schuell), and then microinjected into fertilized mouse eggs according to Gordon and Ruddle (24). The progeny was analyzed for the presence of the transgene by Southern blot.

RESULTS
Delineation of the Activating Sequence in the Intronic B Element of the Aldolase B Gene-In our previous studies we identified a 1.8-kb B region (ϩ685 to ϩ2514) localized in the first intron of the aldolase B gene that was absolutely required for transgene expression in the liver of transgenic mice (7). In transient transfection experiments in hepatoma Hep G2 cells, this B fragment stimulated about 50-fold the basal activity of a 232-bp proximal aldolase B promoter (4). To determine the cis-active sequences in this region, a series of deleted mutants were constructed, and their activity was tested in transient transfection experiments in Hep G2 cells (Fig. 1). Taking advantage of a unique StuI restriction site in the B fragment, we first analyzed the effects of the upstream 1200-bp (ϩ685 to ϩ1915) and of the downstream 600-bp (ϩ1916 to ϩ2514) parts of the B fragment. The B1200 subfragment did not change expression of the reference construct devoid of the B element (Ϫ232A100CAT construct) whereas, in contrast, the B600 downstream subfragment led to a 120-fold stimulation of the basal activity. This subfragment seemed to be more efficacious than the complete B fragment, perhaps due to its closer position with respect to the minimal promoter. Further subdivision of these 600 bp in three short DNA fragments of 200 bp each, designated fragments Ba, Bb, and Bc, showed that fragments Ba and Bc alone were totally inactive, whereas the Bb fragment (ϩ2118 to ϩ2329) retained 25% of the activation observed with the B600 subfragment. Finally association of the fragments B (aϩb) restored the full activation reached with the B600 fragment ( Fig. 1). These results indicated that the 400-bp region, spanning from ϩ1916 to ϩ2329 bp, is able to recapitulate the enhancer activity of the intronic B element of the aldolase B gene. This 400-bp enhancer can be divided into downstream 200 bp, conferring by themselves part of the enhancer activity, and upstream 200 bp by themselves inactive but cooperating with the downstream part to confer a full enhancer activity.
To verify in vivo the relevance of results obtained ex vivo, we generated transgenic mice harboring the constructs studied above ( Table I). The transgene bearing the upstream B1200 subfragment was totally inactive in all 7 lines obtained. In contrast, transgenes including either the downstream B600 subfragment or parts B(aϩb) of this subfragment were detectably active in 8 out of 10 lines studied and were specifically expressed in the liver and kidney but not in the spleen and brain. As previously reported (7), transgene expression is highly dependent on a position effect, thus explaining the various levels of transgene activity and their total inactivity in two lines (once with each construct). In any case, these results confirm in vivo the delineation of the enhancer region of the aldolase B gene established from ex vivo experiments.
The B(aϩb) Fragments Constitute a Transcriptional Liver Cell-specific Enhancer-The next question was whether the B(aϩb) fragment had all canonical properties of an enhancer and whether it was liver cell-specific. To answer this question we placed the aϩb fragment in both orientations in its normal intronic position or in a distal position, 1.2 kb upstream from the promoter or 1 kb downstream of the CAT gene. The activity  (11), we used 7.5 g of CAT plasmid and 2 g of standardization pRSVluciferase plasmid per 6-cm diameter dish. The CAT activity was standardized by the luciferase activity to take into account the variations in the transfection efficiency. The results are shown as fold enhancement with respect to the activity of the enhancerless Ϫ232A100 CAT plasmid. The data are presented as the means Ϯ S.E. of at least three separate experiments.
of all these constructs was tested by transient transfections in Hep G2 cells, and the results are reported in Table II. The fold activation observed was totally independent of the forward or backward orientation of the (aϩb) fragment and almost independent of its position with respect to the cap site, either upstream from the promoter or downstream of the CAT gene or in its natural intronic position, in agreement with canonical enhancer properties (25). However, when the enhancer strength was tested on a construct consisting of the aldolase B promoter spanning from Ϫ194 to ϩ 14 bp, and lacking the first 100 bp and the last 120 bp of the intronic sequence (i.e. both splice sites), the level of activation was reduced. We do not know if this reduction results from a specific cooperation between the enhancer and intronic sequences located in the extreme 5Ј and 3Ј parts of the first intron or from a general increase in the level of transgene expression linked to the presence of a functional intron, already documented in mice but not in transient transfection experiments (26).
The activity of the B(aϩb) enhancer fragment was also tested on heterologous promoters, either the liver-specific Ϫ183-bp proximal promoter of the L-type pyruvate kinase gene (8) or the ubiquitous 105-bp promoter of the thymidine kinase (tk) gene (9). The B(aϩb) fragment enhanced the activity of these promoters by 18-and 15-fold, respectively (Table II). It is noteworthy that stimulation of the L-type pyruvate kinase promoter by the aldolase B enhancer was approximately similar to that by the SV 40 enhancer, previously reported (8).
To determine whether the enhancer activity of the aϩb fragment was by itself specific to liver cells, the constructs containing either the aldolase B or the L-type pyruvate kinase or the tk promoter, with or without the B(aϩb) enhancer, were transiently transfected in mouse 3T6 cells that do not express the aldolase B gene. The enhancer was unable to turn on the liver-specific aldolase B or L-type pyruvate kinase promoters as well to activate the ubiquitous tk promoter. These results indicate that the aldolase B(aϩb) enhancer was clearly cell-specific.
Analysis of Protein-DNA Interactions-A computer analysis of the aϩb enhancer sequence using the recently published Matinspector program (27) was performed. Only the analysis of the Bb (ϩ2118 to ϩ2329) enhancer fragment gave relevant information indicating potential binding sites for liver-enriched nuclear factors such as HNF1, HNF3, HNF4, and CAAT/ enhancer binding protein (C/EBP) and for the ubiquitous AP1 complex. Since this Bb short fragment alone also retained part of the enhancer function and is 76% conserved between human and rat aldolase B genes, we focused our attention on these 200 bp. To confirm that elements of the Bb fragment actually interact with DNA-binding proteins, we first used in vivo DNase I footprinting experiments. The in vivo footprint revealed protein occupancy all over the fragment (not shown), such that it was rather difficult to deduce from this pattern well delineated windows. However, we used this experiment together with the identification of potential binding sites to design seven oligonucleotides that were used for gel shift assay experiments. Fig.  2 summarizes features of sequence analysis and in vivo footprinting experiments and shows the elements whose binding activity was then individually analyzed by gel shift assays. We found that elements 1 and 3 bind factors present in both liver and brain nuclear extracts; in contrast, elements 2, 4, 5, and 7 have a different binding activity in liver and brain (Fig. 3). To determine whether elements 2, 4, 5, and 7 bind previously identified liver-specific transcription factors, we used oligonucleotides of known binding specificity as competitor in the gel retardation experiments. Binding to element 2 was only competed for by the HNF4 oligonucleotide (Fig. 3, a), whereas binding to elements 5 and 7 were highly competed for by the HNF1 oligonucleotide and by each other (Fig. 3, c and d). In addition, binding activity of element 2 was specifically supershifted by anti-HNF4 antibodies (Fig. 3, a), whereas binding activities of elements 5 and 7 were both supershifted by anti-HNF1 antibodies (Fig. 3, c and d).   B(a ϩ b) intronic enhancer fragment in transiently transfected cells Constructs with different promoters, with or without the B(a ϩ b) fragment in different positions, were transfected into either Hep G2 hepatoma cells or 3T6 fibroblasts; the CAT activity was measured and normalized by the luciferase activity generated by the cotransfected pRSV luciferase plasmid. The Ϫ232A100 CAT construct contains 232 bp of the aldolase B promoter plus 100 bp of the intronic element A (see Fig. 1). The Ϫ190 CAT construct contains 190 bp of the aldolase B promoter without splice sites. The Ϫ183 PK CAT construct is driven by the L-type pyruvate kinase promoter. The Ϫ105 tk CAT construct is driven by the HSV thymidine kinase promoter. The position of the B(a ϩ b) enhancer fragment present in the first intron of the aldolase B gene is indicated with respect to the promoter and to the CAT gene. In an intronic position in Ϫ232A100 CAT constructs, this fragment was studied in both orientations (a ϩ b) and (b ϩ a). The results are expressed as fold enhancement with respect to the enhancerless constructs. When more than four independent experiments were performed, the results are given as means Ϯ S.E.; otherwise, the means are given, and in parentheses are the number of experiments.

Constructs
Hep G2 3T6 Ϫ183 PK CAT 0 0 Ϫ183 PK CAT ----(a ϩ b) The binding activity of element 4 was more difficult to identify. Competition experiments using oligonucleotides with different affinity for either HNF4 and/or COUP-TF were performed (Fig. 3, b), and none of them totally competed for the binding to element 4. The element 4 binding activity was also insensitive to anti-HNF4 antibodies. In contrast, element 4 as well as element 2 were effective in displacing HNF4 bound to the L3 L-PK site (Fig. 3, b). Therefore element 4 could bind factor(s) of the nuclear receptor superfamily different from HNF4 and could bind HNF4 with a low affinity. These element 4-binding factors are not likely to correspond mainly to COUP-TF (28), which is expressed in the brain as well as in the liver. Moreover element 4, whose sequence was reminiscent of an HNF3 recognition site, failed to bind this nuclear factor in our experiments as judged from competition experiments with an authentic HNF3-binding oligonucleotide. These results establish that the aldolase B enhancer is modular in nature, possessing binding sites for at least two liver-specific transcription factors, HNF1 and HNF4.
Mutational Analysis of the Function of the Liver-specific Protein-binding Sites Present in the Aldolase B Enhancer-The relative contribution of elements 2, 4, 5, and 7 to the enhancer strength in Hep G2 cells was tested by transient transfection of mutant constructs in which each of these element were mutated separately or in combination. The mutations were obtained as described under "Materials and Methods," and the aϩb-mutated fragments were introduced into an intronic position of the Ϫ232 A100 CAT aldolase B vector. We verified by electrophoretic mobility shift assay that mutated element 5 did not bind HNF1 (not shown). Mutation in this element as well as deletion of the other HNF1-binding site, element 7, rendered the enhancer totally inactive. Surprisingly, when both HNF1binding sites were deleted, we observed an activity of the reporter below the basal level for the enhancerless promoter (Fig.  4). Deletion of element 2, bearing sequences binding HNF4, resulted in an 80% decrease in the enhancer activity. Deletion of element 4 had a minor effect upon the enhancer activity as the construct with this mutation retained 50% of the activity of the native enhancer. These results imply that the aldolase B enhancer activity is dependent on binding of the liver-enriched nuclear factors HNF1 and HNF4 to their cognate DNA sequences. A striking feature of the aldolase B enhancer is the absolute requirement of both intact HNF1-binding sites to be functional. DISCUSSION We have previously described the promoter of the aldolase B gene which is liver-specific but needs the presence of an intronic enhancer to be strongly active in transiently transfected hepatoma cells (4). In transgenic mice, any transgenes devoid of this enhancer were totally inactive, whereas transgenes possessing this element were expressed in the liver, kidney, and small intestine (7). Since the active intronic region was previously ascribed to a large 1.8-kb fragment whose cis-acting elements and cognate transcription factors had not been reported, we described here the modular structure of this enhancer, characterized liver-specific nuclear proteins interacting with its cis-active elements, and functionally investigated its properties, its tissue specificity, and the role of elements binding liver-enriched transcription factors, HNF1 and HNF4.
First, deletional studies allowed us to ascribe the active enhancer to a 400-bp fragment spanning from nucleotides ϩ1916 to ϩ2329 with respect to the start site of transcription. This 400-bp active fragment can be subdivided into two parts called Ba and Bb; the former is totally inactive by itself but appears to cooperate with the latter since activity mediated by Bb fragment alone is 25% only of the activity generated by the B(aϩb) 400-bp enhancer fragment. This 400-bp aldolase B enhancer fulfills the requirement of a typical enhancer; it is functional in both orientations and in any position, either upstream or downstream from the promoter as well as in its natural intronic position. Finally, the aldolase B enhancer is able to stimulate heterologous promoters, either liver specific or ubiquitous. However, this stimulation only occurs in liverspecific cells (Hep G2 cells), and not in fibroblasts, indicating that this enhancer is by itself tissue-specific, as is the aldolase B promoter. This tissue specificity has been recently confirmed in transgenic mice harboring a chimeric transgene directed by a promoter silent in the liver (29) plus, in front of it, the 400-bp aldolase B enhancer; this construct was strongly expressed in the liver as well as in the kidney and small intestine tissue structures normally synthesizing aldolase B. 2 Therefore, the 400-bp intronic enhancer of the aldolase B gene appears to play an essential role in both level and tissue specificity of aldolase B gene expression.
We then determined the cis-acting elements and cognate transcription factors involved in these properties.
The nuclear factors that are mostly responsible for the tissue-specific transcription in differentiated liver cells include HNF1 (30), HNF3 (17), HNF4 (31), HNF6 (32), C/EBP (33), and DBP (34). The experiments with the aldolase B enhancer were first carried out in the Hep G2 hepatoma cells in which the relative amounts of HNF1, HNF3, and HNF4 proteins are close to those found in the fully differentiated liver but which lack C/EBP and DBP (35). In the enhancer sequence a computer search for the presence of putative binding sites for the abovementioned transcription factors was performed, resulting in the identification in fragment Bb of such consensus sequences for factors HNF1, HNF3, and HNF4. Band shift experiments clearly demonstrated that elements 5 and 7 do indeed bind HNF1 nuclear factor, whereas element 2 binds HNF4. The sequence of element 4 could suggest binding of both HNF3 and HNF4/COUP-TF factors. However, the retarded bands observed in a gel shift assay with an element 4-specific probe, detected in the presence of liver but not of brain nuclear ex-2 B. Romagnolo, personal communication. tracts, were specifically displaced neither by an excess of HNF3-specific oligonucleotide nor by anti-HNF4␣ antibodies. COUP-TF factors are known to be ubiquitous, particularly abundant in the brain (36), so that it is unlikely that the complex with element 4 corresponds to these factors. Since the binding activity of element 4 was partly competed for by some HNF4-binding sites and by synthetic DR1 motif, whereas the element 4 itself was able to displace authentic HNF4-containing complexes, we speculate that this element could mainly bind in liver extracts, as yet, unknown members of the nuclear receptor superfamily. Therefore we explored the role of all of the above-mentioned elements, including element 4, in the strength of the aldolase B enhancer by deleting or mutating each of them. Both HNF1-binding sites (elements 5 and 7) were also deleted. Point mutation or deletion of either HNF1-binding site abolishes completely the enhancer function, whereas dele-tion of both sites leads to restriction of the basal aldolase B promoter transcriptional activity. The requirement of both intact HNF1-binding sites in order to maintain the enhancer function in a hepatoma cell line is a striking feature of this enhancer. A common feature of several liver-specific enhancers is the cooperation between two or more different liver-enriched nuclear factors (17,(37)(38)(39)(40). The stringent dependence of the aldolase B enhancer activity upon HNF1-binding sites seems in discrepancy with the report that the aldolase B gene is active in the liver of HNF1-␣-deficient knock-out mice (41). However, transcriptional activity of the aldolase B gene in these mice could be explained by the presence of a compensatory increased amount of HNF1-␤ in the liver of the HNF1-␣-deficient mice, both of these factors binding to the same target and being transcriptional activators (38,42). It is noteworthy that interchangeability of HNF1-␣ and -␤ for sustaining the function of FIG. 3. Gel shift analysis of elements B2, B4, B5, and B7 of the enhancer fragment Bb. The labeled, double-stranded oligonucleotide probes used for nuclear factor binding as well as the competitor oligonucleotides are described under "Materials and Methods." All competition experiments were performed by adding 50-fold excess of unlabeled double-stranded competitors. The lanes (Ϫ) in A, B, and D indicate the absence of competitor; the nature of the competitors is indicated above the corresponding lanes. The arrows indicate position of specific bands (specifically displaced by cognate competitor) and of nonspecific bands (ns) (either nondisplaced, or nonspecifically displaced). A, gel shift assays with liver or brain nuclear extracts. B, gel shift assays with liver nuclear extracts, competition with various oligonucleotides reproducing known binding sites. C, panels a, b, and c, supershift assays using 1 l of specific anti-HNF1 or anti-HNF4 antibodies; in the control (Ϫ), we used 1 l of nonimmune serum. D, panel b, gel shift assays using L-PK HNF4 oligonucleotide (Ϫ), competition with either itself, or element 2, or element 4. Panel a, gel shift assay characterization of element B2; panel b, gel shift assay characterization of element B4; panel c, gel shift assay characterization of element B5; panel d, gel shift assay characterization of element B7. enhancers possessing HNF1-binding sites is not a general phenomenon since knock-out HNF1-␣-deficient mice are totally defective in expression of the phenylalanine hydroxylase gene whose distal enhancer (HSS-III), located 3500 bp upstream from the promoter, contains two HNF1-binding sites (43). The phenylalanine hydroxylase gene also contains two additional putative HNF1-binding sites around position Ϫ1200 (HSS-II), and its chromatin structure is closed in HNF1-␣-deficient mice (44). Therefore, it could be that the hypothetical role of HNF1-␣ in remodeling chromatin structure of the phenylalanine hydroxylase gene is not ensured by residual HNF1-␤, at least at its concentration in the liver. In the kidney, by contrast, the amount of HNF1-␤ is higher than in the liver, and the phenylalanine hydroxylase gene remains active in HNF1-␣ Ϫ/Ϫ mice. Therefore it could be that relatively high concentrations of either HNF1-␣ or ␤ are required to open chromatin structure of the phenylalanine hydroxylase gene, whereas residual HNF1-␤ in the liver of HNF1-␣ Ϫ/Ϫ mice would be sufficient to sustain activity of the aldolase B enhancer whose open chromatin structure could depend on other factors.
The HNF4-binding site (element 2) appears also to be essential for the aldolase B gene enhancer activity since its deletion results in an 80% reduction of transcriptional activity, which is fully consistent with the report by Stoffel et al. (45) that the aldolase B gene is silent in HNF4-deficient knock-out murine embryos. In contrast, deletion of element 4 binding non-identified factors has a more limited effect (50% of residual activity).
It is noteworthy that the demonstration of HNF1 and HNF4 as indispensable factors for a full enhancer function does not signify that these factors are sufficient to sustain this function. In fact, cotransfection of fibroblasts with Ϫ105 tk B400 construct and both HNF4 and HNF1 expression vectors did not allow for stimulation of the reporter CAT gene. It is now increasingly recognized that transcription activators work with coactivators sometimes needed to bridge them, from a removed binding location, with the proximal promoter complex. This could explain why the binding of HNF1 and HNF4 to the aldolase B enhancer in nonhepatic cells is insufficient to activate the gene, some of the coactivators/bridging factors being more or less tissue-specific.
In conclusion, specificity and activity of the intronic aldolase B gene enhancer strongly relies on binding sites for liverspecific HNF1 and HNF4 factors. Surprisingly, either of the two HNF1-binding sites is crucially required for the enhancer activity. Although the aldolase B gene proximal promoter includes binding sites for HNF1 (element PAB Ϫ126 to Ϫ104 bp (46)) and, as recently demonstrated, for HNF4 (element PE, located from Ϫ99 to Ϫ81 bp), 3 a specific cooperation between these factors bound to the promoter and the enhancer does not seem to be crucial since the enhancer also works very well with the ubiquitous Ϫ105 tk promoter. Probably through its multiple elements for both liver-specific and non-identified ubiquitous factors, the aldolase B enhancer seems to be a novel strong and highly tissue-specific enhancer useful for target gene expression to the liver, proximal tubules of the kidney, and enterocytes of the small intestine.