Functional Analysis of the Mouse a-Fetoprotein Enhancers and Their Subfragments in Primary Mouse Hepatocyte Cultures*

We have compared the activities of mouse a-fetopro- tein (AFP) enhancers I, 11, and I11 with their minimal enhancer fragments (Mers) I, 11, and I11 and with the entire 7-kilobase pair enhancer domain by transient expression assay in primary fetal mouse liver cells. The level of expression directed by the AFP promoter [p(-lOOS)AFPcat] alone is stimulated at least 10-fold by the entire AFP enhancer domain (-1009 to -6983). Enhancer I can drive the level of chloramphenicol acetyltransferase activity equivalent to that of the entire enhancer domain, whereas the increase in activity by enhancers I1 and I11 is significantly lower (1.5-fold). MersI, 11, and I11 all mediate a greater increase in activity than their corresponding enhancer regions. The increase with MerI is 16-fold. Using DNase I we identified 3 protein-binding regions in MerI; site Ia binds liver and brain nuclear proteins; site Ib binds liver, kidney, and brain nuclear proteins as well as purified C/EBP; site IC binds liver and kidney nuclear proteins. Site-specific mutation of Ia, Ib, or IC showed a 10-25% reduction in chloramphenicol acetyltransferase expression; deletion of the C/EBP-binding site in Ib showed a 45% reduction in activity and mutation of all 3 sites (Ia, Ib, Cell Culture, DNA Transfection, and Measurement of CAT and 8- Galactosidase Activity-These procedures are described in a previous report by Zhang et al. (14).

AFP gene beginning at the site of transcriptional initiation and extending to -7.6 kilobase pairs upstream of the gene. The promoter region has been localized up to -1009 bp, although the proximal promoter which extends to -202 bp contains all the information needed to mediate maximal tissue-specific expression in transient expression assays (10)(11)(12).
The three enhancer domains, i.e. enhancers I, 11, and 111, have been localized within the region from -1.0 to -7.6 kilobase pairs. Their boundaries were initially established by transient expression assay of DNA fragments produced by BamHI restriction digestion (2,9). Furthermore, this assay was used to identify minimal enhancer regions (Mers), which range in size from -200 to 300 bp, and which are localized within each of the enhancer fragments (3). Hammer et al. (4) demonstrated that at least one of the three enhancer regions is required for tissue specificity as indicated by their function in transgenic mice. However, the postnatal repression of the AFP gene has been shown to be a function of the promoter, and/or portions of the structural gene (1,6,8). Thus, both tissue specificity and developmental repression are associated with AFP promoter function, whereas tissue specificity alone is associated with enhancer function. Although all three enhancers exhibit equal levels of activity in HepG2 cells, similar studies with transgenic mice have shown that enhancer I is more active in the livers of transgenic animals, than enhancer I1 and that enhancer I11 has very little activity (4).
The AFP promoter region contains protein-binding sites where trans-acting factors interact with cis-acting DNA sequences to mediate regulation of its tissue-specific expression (10,(12)(13)(14)(15). Evidence for the existence of these multiple protein-binding sites in the proximal promoter of the AFP gene has been obtained by DNase I protection analyses (12)(13)(14)(15) and by transient expression analyses using deletion mutations and site-specific mutations of each of the binding sites (2,10,14). Through these studies it has been shown that the proximal promoter alone, up to -202 bp, is sufficient to mediate tissue-specific transcription of a reporter gene in hepatoma cells (10,11) and in primary cultures of fetal liver cells (14). By using site-specific mutation in each of the binding sites it was shown by transient expression assay that promoter-mediated regulation of the AFP gene requires the combinatorial action of these multiple cis-and trans-acting elements in the proximal promoter.
Recently Zhang et al. (14) observed that by linking enhancer I to proximal promoter fragments (-202 bp) with C/EBPbinding site mutations, the reduced activity caused by these mutations is rescued, i.e. CAT gene expression is returned to normal levels. We hypothesized that enhancer I may provide site(s) for binding of trans-acting factor(s) which can rescue the C/EBP mutations in the promoter, possibly a C/EBPbinding site. On the basis of these observations we felt that it Functional Analysis of Mouse AFP Enhancers 10677 was important to elucidate the DNA-protein interactions of the AFP enhancers and their role in enhancer function. To do this we have sequenced the entire enhancer region (from -1009 to -6983 bp), assayed the enhancers I, 11, and I11 as well as MerI, 11, and I11 activities in primary fetal liver culture, and performed DNase I protection analyses to identify the binding sites in MerI using normal fetal liver and adult liver nuclear extracts. We also report on experiments to define the role of the trans-acting factors and cis-acting binding sites of MerI in augmenting AFP gene expression, using site-specific and deletion mutations.
For substitution mutagenesis, the XbaI-EcoRI minimum enhancer region I (MerI) fragment from pUIC19 was inserted into M13mp19, mpMerI. Three substitution mutants mpMIA, mpMIB, and mpMIC, in which the mutation sites correspond to the sites of nuclear protein protection regions Ia, Ib, and IC on the AFP MerI were prepared by oligonucleotide-directed mutagenesis of the mpMerI with synthetic oligonucleotides that paired with 12 nucleotides of the wild-type sequence on either side of 6-8-nucleotide target sequence. The base sequence for each substitution (Fig. 3) consisted of a XhoI recognition sequence. The mutations were confirmed by XhoI restriction endonuclease site analysis and dideoxy sequencing. All the substitution mutant fragments were isolated by digestion with BamHI and KpnI and used to replace the wild-type BamHI and KpnI fragment in the pMerI(-1009)AFPcat.
In pMID(-1009)AFPcat sequences from -2317 to -2275 bp of the MerI enhancer fragment were deleted. This construct was generated by digestion with the slow form of Ba131 exonuclease (International Biotechnologies, Inc.) from the XhoI site in pMIB(-1009)AFPcat.
Plasmid Construction and End Labeling of DNA Fragments for DNase I Footprinting-The DNA fragment containing AFP MerI (AseI -2574 to -2190 bp NdeI), was excised from the AFP enhancer domain with restriction endonucleases, blunt ended with the Klenow fragment of DNA polymerase I, and inserted into the SmaI site of pUC19. This construct was named pUIC19. The orientation of each insert in pUC19 is as shown in Fig. 2. These plasmids were constructed and purified using standard protocols.
To label DNA fragments used for DNase I footprinting, the above plasmids were either 3' end-labeled at the EcoRI site or at the XbaI site with [a-32P]dATP, [w3'P]dTTP, dGTP, and dCTP for either coding or noncoding strand using the Klenow fragment of DNA polymerase I. The labeled plasmids were further digested by a second restriction endonuclease, either XbaI or EcoRI. The labeled DNA fragments were separated from the plasmid vectors by polyacrylamide gel electrophoresis (5%), eluted from gel slices in 0.3 M sodium acetate, pH 7.2, and purified by passing through Elutip-d columns purchased from Schleicher & Schuell.
Preparation of Nuclear Extracts from Different Tissues of Adult and Fetal Mouse and DNase I Footprinting-Procedures for the preparation of nuclear extracts and DNase I footprinting are the same as described by Zhang et al. (13,14). Analysis of C/EBPa mRNA Leuel.-Total RNA was isolated from fetal liver, adult liver, and spleen, and fetal liver primary cultures by lysis in guanidine hydrochloride followed by centrifugation through a CsCl cushion as has been described (16). The RNA was resolved by electrophoresis on formaldehyde-agarose gels, followed by transfer to nitrocellulose (17)(18). The nitrocellulose filters were then baked for 1-2 h at 80 "C under vacuum. Filters were probed for C/EBPaspecific mRNA using 32P-labeled probe prepared with a nick translation kit (Bethesda Research Laboratories, Gaithersburg, MD). The 32P-labeled probe was prepared by nick translation of the rat C/EBP cDNA (L39) clone described by Landschultz et al. (19). Prehybridization was at 42 "C in 50% formamide, 5 X SSC (1 X S s c = 0.15 M sodium chloride, 0.015 M sodium citrate), 50 mM phosphate, pH 6.5, 200 pg/ml sheared and denatured salmon sperm DNA, 0.1% SDS, and 4 X Denhardt's solution (17). Hybridizations were performed in the same buffer for 24 h at 42 "C. Filters were washed 4 times at room temperature in 2 X SSC, 0.1% SDS for 15 min and 3 times at 52 "C in 0.1 X SSC, 0.1% SDS for 45 min. Hybridizations were detected by autoradiography using Kodak X-Omat AR x-ray film, and were quantitated by densitometer scanning.
Cell Culture, DNA Transfection, and Measurement of CAT and 8-Galactosidase Activity-These procedures are described in a previous report by Zhang et al. (14). Although the three enhancer fragments and their minimum enhancer regions showed very strong activity in the HepG2 hepatoma cell line, analysis of livers of transgenic mice indicated that the three enhancer fragments were not functionally equivalent in vivo (4). In the transgenic liver, enhancer I has higher activity compared to enhancer 11, and enhancer I11 has little activity. Since studies of AFP enhancer activity in tissue culture have been limited to hepatoma cells, we conducted experiments to compare the in vivo activities with activities in normal fetal cells in culture. To do this, we determined the activity of the entire AFP enhancer domain and each subfragment by transient expression analysis in primary fetal mouse liver cells in culture (Fig. 1). The results indicate that enhancer I stimulates the promoter (-1009 bp) 9.8-fold; that enhancers I1 and I11 stimulate the promoter 1.5-and 1.6-fold, respectively, and that the whole enhancer from -7 to -1 kb stimulates promoter activity 10.7-fold. The data also show that MersI, 11, and I11 stimulate the promoter 15.6-, 2.7-, and 3.4-fold, respectively. These data indicate that the enhancer I fragment can augment the activity of the CAT gene to a level equivalent to the activity of the entire 7-kb AFP enhancer region, and that the minimum enhancer fragment, MerI, can augment activity to a level that is 1.5-fold greater than that of the enhancer I fragment. Transient expression assay in the mouse hepatoma cell line (BWTG3) gave results similar to those obtained with primary fetal liver cells in culture ( Fig. 1). In contrast, results from experiments in which HepG2 cells were used as host cells indicate no significant differences in the activities of the three enhancer and Mer fragments ( Fig. 1 and Refs. 2 and 3). Furthermore, the 7-kb AFP enhancer region did not show any activity in either primary fetal kidney cell cultures or in NIH3T3 fibroblast cells (18). Since enhancer I and MerI showed the strongest enhancer activity in primary fetal liver cells and in transgenic mice, we conducted experiments to identify the protein-binding sites of MerI and the role of these DNA-protein complexes in MerI enhancer activity.

Functional Analysis of Mouse AFP Enhancers
Localization of Fetal and Adult Nuclear Protein-binding Sites in AFP Minimum Enhancer I (MerI)-The ability of cis-acting enhancer or promoter regions to regulate gene expression involves the interaction of tram-acting factors with their specific DNA-binding sites. Using site-specific mutations we have shown that hepatocyte nuclear factor-1 (HNF-l), C/EBP, and nuclear factor-1 are essential for maximal activity of the AFP promoter (14). To study the DNA and protein components that contribute to AFP enhancer activity, fetal and adult liver nuclear protein extracts were used to perform DNase I footprinting with the MerI DNA fragment. These analyses revealed that there are three protein-binding regions in MerI, which we named regions Ia, Ib, and IC (Fig. 2). The DNA sequence of each site identified by DNase I protection is indicated by the double underlined sequences in Fig. 3. The footprinting with adult and fetal mouse liver nuclear extracts showed similar protection patterns indicating there are no apparent developmental differences in protein binding patterns detected within this enhancer region. Sequence analysis of the DNase I-protected regions of MerI (Ia, Ib, and IC) revealed homologies with consensus sequences of well characterized tram-acting factor-binding sites as well as binding sites not previously described. The DNA sequence (-2238)GTCATGTGGCA(-2228), in region Ia (Fig. 3) is homologous with the E3 site of the immunoglobulin H(p) heavy-chain enhancer and immunoglobulin K light chain promoter (20)(21)(22)(23). Furthermore, the sequence TGGCA, also in region Ia is one-half of the nuclear factor-1 consensus sequence. Both of these sequences are potential binding sites for ubiquitous proteins reported to be important for enhancer and promoter activity. The sequence CACACAAA, immediately upstream of region Ia, is one of several that is also repeated in MersII and -111.2 However, our protection analyses show that these sequences are not protein-binding sites in any of the Mers.
D-E. Zhang and J. Papaconstantinou, unpublished data.  The sequence of MerI and the binding sites for liver ( L ) , kidney ( K ) , and brain ( B ) nuclear proteins, and the C/ EBP-binding site. The lines below the sequence denote the sequences of the coding and noncoding strands protected by liver nuclear proteins; the lines above the sequence denote the sequences of the coding strand protected by kidney and brain nuclear proteins and by purified C/EBP. The mutations introduced into regions Ia, Ib, and IC are shown immediately below the sequence they replace. ence of Box 2, which is 14 nucleotides downstream of Box 1 and which shows homology to Box 1 (Fig. 3). The DNase I protection assay clearly shows that Box 2 is not a proteinbinding site. The sequence TATTGA'/TTT is the consensus sequence for hepatocyte nuclear factor-3 (HNF-3) (24). This sequence is found in region IC of MerI (-24OO)TATTG(A) CTTC(-2392) although it is missing an A at -2395 (Fig. 3).

G T~T A G A~M U X C C U X T T G T T T A W X P . T A T T +~T
The DNase I protection assay indicates that only the 5' end of this sequence is protected.
Localization of Tissue-specific Protein-binding Sites in Merl-Experiments were done to determine if any of the protected regions within MerI are the binding sites for tissuespecific proteins. Nuclear extracts from adult mouse liver, kidney, and brain were used to perform DNase I protection assays. As shown in Fig. 4, the 3' end of region Ib is protected by nuclear proteins from all three tissue types, indicating that the proteins binding to these regions may be ubiquitous factors. There is one region which is only protected by nuclear proteins from the liver which is depicted as region Ib (Fig. 3). Analysis of the liver-specific binding site does not reveal sequences that might bind with HNF-1, which is a well characterized liver-specific trans-acting factor (25)(26)(27). On the other hand, C/EBP, which has been shown to bind to three regions of the AFP proximal promoter (13,14), is present in relatively high abundance in the liver and has been reported to interact with several different DNA motifs. Therefore, we felt that this might be an excellent candidate for binding to the liver-specific enhancer-binding sites and we used purified C/EBP to perform DNase I footprinting with AFP MerI as shown in Fig. 5. The results clearly indicate that C/EBP binds to sequences from -2286 to -2316 of MerI which correlate with the liver-specific binding site shown in Fig. 4.
Functional Analysis of the Trans-acting Factors That Bind t o Merl-Our experiments indicate that MerI is an important functional region within enhancer I of the AFP gene and that there are three different regions within MerI protected by at least four protein factors, one of which is C/EBP (Fig. 3). In these studies we chose to use primary liver cell cultures from 18-20-day fetal livers as host cells to examine the regulatory role of trans-acting factors that interact with the AFP enhancer I. However, the DNA-protein interaction of fetal mouse liver proteins with enhancer DNA fragments were done with proteins extracted from 18-20-day fetal liver nuclei. Since the cells from which nuclear proteins were extracted were not cultured, there is the possibility that the levels (or activity) of trans-acting factors may be altered in cultured cells, thus affecting the levels of transient expression. Xanthopoulos et al. (28) reported, for example, that there is a gradual decrease of C/EBP mRNA levels in primary adult rat hepatocyte cultures. Since we were not able to extract nuclear proteins from the cultured mouse hepatocytes, to address this question we analyzed C/EBP mRNA levels in these cultures to determine whether levels of this trans-acting factor are altered. Total cellular RNA was isolated from adult rat liver, adult mouse liver, and spleen, from primary fetal liver cells in culture, and from mouse hepatoma cells (BWTG3) in culture. These RNAs were hybridized to rat ["2P]C/EBP cDNA. As shown in Fig. 6 the level of C/EBP mRNA, in primary cultures, is the same throughout the 96-h culture period (and higher than the levels in BWTG3 cells). DNase I footprinting analyses using nuclear protein from BWTG3 cells exhibit strong protection of C/EBP-binding sites in both AFP and albumin promoters? These data support the use of fetal liver cells in culture for studies of the role of nuclear protein in the function of the AFP enhancer. To study the importance of factors binding to this region in enhancing AFP promoter activity, oligonucleotide-directed site-specific mutagenesis was used to generate mutations a t each of the three binding sites. The wild-type protection pattern is shown in panel W of Fig. 7 and the mutated sequences are shown in Fig. 3. DNase I footprinting (as shown in panels A and C of Fig. 7) shows that these mutations abolish binding of the corresponding nuclear protein(s) to their binding sites. The 6. An analysis of the C/EBPa mRNA levels in fetal mouse liver primary cultures. Fetal mouse liver cells were plated immediately after dispersal ( t = 0) and mRNA levels were determined at 24-h intervals for 4 days ( t = 24-96). Agarose formaldehyde gels were run using 15 pg of total RNA and Northern hybridization analyses was performed as described under "Materials and Methods." C/EBP mRNA levels were also determined for adult mouse liver and spleen and for adult rat liver. The hybridizat,ion signals were quantitated by densitometric scanning.
binding of proteins to regions Ia and IC were fully and specifically abolished by the mutation in the corresponding DNA sequence. The 8-nucleotide substitution mutation in the middle of region Ib was observed to partially abolish binding (Fig.  7B) of its protein leaving the region from -2290 to -2276 14 for a detailed promoter analyses). Enhancer activities were analyzed as described in the legend to Fig. 1. still protected. MerI fragments with each mutation were inserted into the plasmid p(-1009)AFPcat, which contains the AFP promoter and transfected into cultures of primary fetal mouse liver cells as mentioned above. The CAT activities from transfection experiments, shown in Fig. 8, indicate that mutation in region Ia (Fig. 8B) resulted in a 10% reduction of MerI activity while the mutations in regions Ib (Fig. 8C) and IC (Fig. 8D) resulted in a 25% reduction of MerI activity, but none of the single site mutations of MerI exhibited a dominant effect on its enhancer activity. Since the mutation of region Ib did not completely abolish binding to that region, a deletion mutation was prepared in which sequences from Functional Analysis of Mouse AFP Enhancers

10681
-2319 to -2276 were removed. The CAT activity of this expression vector was reduced by -40-45% (Fig. 8E). Sitespecific mutation of either Ia (Fig. 8F) or IC (Fig. 8G) in the Ib deleted MerI did not further reduce activity, indicating that approximately 50% of the enhancer activity can be mediated when 2 of the 3 binding sites are mutated. Finally, a triple mutation, i.e. Ia + IC site-specific mutation plus deleted Ib, reduced activity by -75% (Fig. 8H). We interpret these data to indicate that MerI enhancer activity is combinatorial and additive. The persistence of enhancer activity when protein binding at all three sites is abolished may be due to the presence of other important sequences. Alternatively, although mutations abolish protein binding in vitro, as indicated by DNase I protection, binding may not be fully abolished in uivo. The AFP enhancer domain and its subfragments exhibit the same pattern of activity in primary fetal liver cells, in mouse hepatoma cells (BWTG3), and in transgenic mice. In contrast, no significant differences were seen with the same expression vectors in HepG2 cells. Although we do not understand the basis for this difference with the HepG2 cells, preliminary gel shift analyses indicate that there are significant differences in binding activities of HNF-1 and C/EBP from nuclear extracts of HepG2 cells? These data suggest that the altered activity of essential trans-acting factors may be a basis for the ability of all enhancer subfragments to exhibit similar high levels of activity.

DISCUSSION
Sequence analysis of MerI has revealed a variety of potential binding sites for trans-acting factors (3). We have identified those sequences which exhibit protein binding activity, and we have attempted to correlate the functional roles of these protein-DNA complexes through mutation analysis. Box 1, for example, binds purified C/EBP as well as a protein(s) from liver nuclear extract, and both site-specific or deletion mutation of this site reduced enhancer activity by approximately 25 and 40%, respectively. The purified C/EBP used to localize this site is C/EBPa (29). There are other members of the C/EBP family of trans-acting factors, and our preliminary data indicate that both C/EBPa and C/EBPp isoforms bind to this site.5 Region Ia, which is homologous to the pE3 enhancer site of the immunoglobulin K light chain (-2238 to -2228) also exhibits binding activity with liver J. Papaconstantinou and C-C. Hsieh, manuscript in preparation. J. Papaconstantinou, X. Ge, and D-E. Zhang, manuscript in preparation.
nuclear proteins, although we have not identified the protein that binds to this site. Site-specific mutation of region Ia abolishes protein binding activity but does not severely affect MerI enhancer activity (-10% reduction). The protein that binds to the pE3 enhancer site (from immunoglobulin K light chain) is ubiquitous and this sequence has been shown to have enhancer activity in other tissues (20,22). Poliard et al. (12) have demonstrated an interaction between this corresponding region in the rat AFP enhancer and nuclear protein from rat liver and pointed out that TGGCA is also part of the nuclear factor-1 consensus sequence. This suggests that region Ia may be a nonspecific enhancer site and that it works in concert with regions Ib and IC. Godbout et al. (3) showed that enhancer I exhibits tissue specificity since its activity was considerably lower in HeLa cells than in HepG2 and Hep3B. However, the activity of a subfragment of MerI (from -2381 to -2189) which contains regions Ia and Ib, and one of two direct repeats ((-2339)TGAGAGGT(-2332)) in its 5'-flanking region, respectively (Fig. 3), was found to be 10-fold higher in HeLa cells. Their data suggest that a significant level of tissue specificity is lost when region IC is deleted from this element. Our studies have shown that region IC binds proteins from both liver and kidney nuclear extracts. Although these proteins have not been identified, this binding site contains an HNF-3 consensus sequence which may be a binding site for this liver-specific trans-acting factor. These studies suggest that regions Ia and Ib may function as nonspecific enhancers and the region Ic-binding site may confer tissuespecific enhancer activity in enhancer I.
There are several regions whose sequences suggest they may be potential binding sites, but showed no binding activity. One of these is the sequence within the Box 2 region of MerI (-2271)TTGTTGCAGT(-2262) which is similar to Box 1, which binds C/EBP as well as one of the C/EBP-binding sites of the AFP promoter (13,14). Thus if this region, which is 15 nucleotides downstream of Box 1 is required for enhancer activity, its mechanism of action may not involve a DNAprotein interaction (as detected by our assay). Similarly, the CACAAA motif which is at the 5' end of region Ia-binding site is only partially protected as indicated by our DNase I assay. A similar situation is seen in region IC of MerI which also contains a partially protected HNF-3 consensus sequence. Mutation of this site abolishes protein binding activity and results in a 25% reduction of enhancer activity, and its deletion appears to reduce tissue specificity (3). Whether this is a functional HNF-3 binding site remains to be determined.
Our site-specific and deletion mutation analyses showed that none of the MerI protein-binding sites exhibited dominant tissue-specific enhancer activity and that these regulatory sites work in a combinatorial mechanism. Even though mutations totally abolish protein binding to regions Ia and IC, these mutations only resulted in a 10 and 25% reduction of activity indicating that the other two functional sites can contribute 75-90% of the enhancer activity. Mutation of region Ib (C/EBP-binding site) resulted in a 25% inhibition of CAT expression. However, since proteGion of this site was only partly abolished, this might account for the low level of inhibition seen. Upon deletion of the entire region Ib (mutation Id), enhancer activity was reduced by -45% indicating that site-specific mutation did not fully inactivate enhancer activity of that region.
To further understand the role of the binding sites we constructed MerI fragments with multiple site-specific mutations. Mutation of Ia or IC in the Ib deleted MerI did not significantly affect the level of inhibition above that of the Functional Analysis of deletion of Ib alone. These results suggest that as long as one of the three binding sites of MerI remains intact, this functional site can augment transcription at one-half the control level. Finally, when both Ia and IC are mutated in the Ib deleted MerI, the enhancer activity was reduced by -75-80%. Thus, through the use of both site-specific and deletion mutation we were not able to identify a dominant trans-acting factor essential for enhancer activity. Similar results have been reported with the IgH(p) (30,31) and SV40 (32,33) enhancers.
Our studies suggest that the activity of MerI is achieved through the combinatorial action of the binding of multiple trans-acting factors. The stepwise decline in activity with multiple mutations is indicative of such an additive mechanism of action. Interestingly, persistence of -20-25% of enhancer activity by the triple mutation of MerI indicates that other sequences may play an important role in enhancer activity, or a significant percentage of the remaining activity may be due to promoter driven expression alone.
Using purified C/EBP protein in DNase I protection assays we identified C/EBP protected regions in all three Mers which also correspond to their liver-specific binding sites ( Fig. 5 and data not shown). However, MersII and I11 exhibit relatively low activity, while MerI exhibits a 16-fold enhancement in a homologous system. These data clearly indicate therefore that C/EBP alone is not sufficient to drive the maximal level of transcription of the mouse AFP promoter. Thus the activity of enhancer I and MerI is not totally dependent on C/EBP binding and C/EBP must act in combination with other factors to achieve maximal enhancer activity.
Recent studies have demonstrated that there are several C/ EBP isoforms (28). Although these proteins recognize similar DNA binding motifs, their physiological functions are associated with such processes as terminal differentiation, energy metabolism (34,35), and liver-specific gene activation (36)(37)(38)(39). Thus, although we have identified the C/EBP-binding sites of the AFP enhancer domains using recombinant produced C/EBPa, we are presently doing experiments to identify the C/EBP isoform(s) that bind to this site in vivo.