Identification of an Upstream Enhancer in the Mouse Lamininα1 Gene Defining Its High Level of Expression in Parietal Endoderm Cells

Laminin-1 is the major component of the embryonic basement membrane and consists of (cid:1) 1, (cid:2) 1, and (cid:3) 1 chains. The expression of laminin-1 is induced in mouse F9 embryonal carcinoma cells upon differentiation into parietal endoderm through transcriptional up-regula-tion of the genes encoding these subunits. Here, we identified a 435-bp enhancer in the 5 (cid:1) -flanking region of the mouse laminin (cid:1) 1 ( LAMA1 ) gene that activated its transcription in a differentiation-dependent manner. This enhancer was also active in PYS-2 parietal yolk sac-derived cells but not in NIH/3T3 fibroblasts, indicating that it was a parietal endoderm-specific enhancer. This enhancer was also active in Engelbreth-Holm-Swarm (EHS) tumor-derived cells characterized by excessive production of laminin-1 and other basement membrane components, suggesting that EHS tumors have a transcriptional control mechanism similar to that of parietal endoderm cells. Electrophoretic mobility shift analyses revealed four protein binding sites (PBS1-PBS4) in the 435-bp region. However, these DNA-binding proteins were detected not only in parietal endoderm cells ( i.e. differentiated F9 cells, PYS-2

Laminins are the major basement membrane glycoproteins regulating tissue morphogenesis through their effects on the proliferation, migration, and differentiation of various types of cells (1)(2)(3). Laminins consist of three subunit chains, ␣, ␤ and ␥, which are assembled and disulfide-bonded in a cross-shaped structure with three short arms and one long rodlike arm. To date, five ␣ chains (␣1-5), three ␤ chains (␤1-3), and three ␥ chains (␥1-3) have been identified, and these assemble into at least 12 distinct laminin isoforms (4 -7). Among these isoforms, laminin-1 is the major component of the early embryonic basement membrane and has been shown to be required for normal development (8,9).
Mouse F9 embryonal carcinoma cells, a cell culture model of early mammalian embryogenesis, can be induced to differentiate into primitive endoderm-and parietal endoderm-like cells upon treatment with retinoic acid and dibutyryl cAMP with concomitant transcriptional activation of the genes encoding the laminin ␣1, ␤1, and ␥1 chains (10,11). The coordinate expression of these subunit genes during F9 cell differentiation suggests that a common mechanism is operating in their transcriptional regulation. Several studies on the transcriptional regulation of laminin subunit expression during the differentiation of F9 cells have been reported previously (12). In the laminin ␤1 (LAMB1) gene promoter, a retinoic acid-responsive element has been identified previously (13)(14)(15)(16), whereas differentiation-dependent elements in the first intron have been identified in the laminin ␥1 (LAMC1) gene (17). However, the molecular mechanisms mediating the coordinate activation of these genes are poorly understood, and the function of the laminin ␣1 (LAMA1) gene promoter has not been studied in any species.
The ␤1 and ␥1 chains are common components of several laminin isoforms (laminin-1, -2, -6, -8, and -10) and have a wide distribution in basement membranes. In contrast, the laminin ␣1 chain has a restricted tissue distribution and is predominantly expressed in the epithelial basement membrane during embryonic development (5, 18 -21). Moreover, the laminin ␣1 chain expression is thought to be the limiting factor in the secretion of laminin-1, because the ␤1 and ␥1 chains, which are preassembled into a disulfide-linked ␤1-␥1 dimer, cannot be secreted without the trimeric assembly with the ␣1 chain (22). These findings prompted us to investigate the mechanism restricting the laminin ␣1 chain expression in a differentiationdependent and cell type-specific manner.
In this study, we isolated the 5Ј-flanking region of the mouse LAMA1 gene. Using reporter gene assays and deletion analyses, we identified an enhancer in the promoter sequence responsible for laminin ␣1 expression during F9 cell differentiation. This enhancer was also active in the PYS-2 mouse teratocarcinoma cell line that exhibits parietal endoderm phenotypes (23) but not in NIH/3T3 fibroblasts, suggesting that this enhancer functions in a parietal endoderm-specific manner. Interestingly, this enhancer was also active in Engelbreth-Holm-Swarm (EHS) 1 tumor-derived cells, which are characterized by an excessive secretion of laminin-1 and other basement membrane components (24,25). We further demonstrated that the synergy of three cis-elements was required for the enhancer activity. DNA-binding proteins interacting with two of these cis-elements were identified as Sp1/Sp3 and YY1, zinc finger transcription factors widely expressed in many tissues, suggesting that posttranslational modifications of these factors and/or cooperative interactions with other proteins are important for parietal endoderm-specific enhancer activity.

EXPERIMENTAL PROCEDURES
Cell Culture-F9 and NIH/3T3 cells were obtained from Health Science Research Resources Bank (Osaka, Japan). PYS-2 cells were kindly provided by Dr. Atsuhiko Oohira (Institute for Developmental Research, Aichi Human Service Center, Aichi, Japan). These cells were cultured in high glucose Dulbecco's modified Eagle's medium (Sigma) supplemented with 10% fetal calf serum in an atmosphere of 95% air, 5% CO 2 , and 100% humidity. The differentiation of F9 cells was induced by adding 0.1 M all-trans-retinoic acid (Sigma) and 1 mM dibutyryl cAMP (Sigma) into the medium. EHS tumor-derived cells were prepared cultured as described previously (26) with minor modifications. 2 Isolation of the Mouse LAMA1 Genomic Clones-A mouse RPCI-23 bacterial artificial chromosome (BAC) library was screened using a BAC/PAC library screening kit (GenoTechs, Tsukuba, Japan). The oligonucleotide primers used for screening were: 5Ј-GAGTGTGCTCTTC-CCAGCTC-3Ј and 5Ј-CCCCTGGAGGACAGACCT-3Ј. Genomic DNA fragments containing exon 1 of the mouse LAMA1 gene were digested with restriction enzymes, subcloned into pBluescript II (Stratagene, La Jolla, CA), and sequenced. All DNA sequencing was carried out using an ABI Prism dye terminator cycle sequencing kit and a model 3100 DNA sequencer (PE Applied Biosystems, Foster City, CA).
Luciferase Reporter Plasmid Construction and Site-directed Mutagenesis-A 9.4-kb XbaI fragment containing 2.5 kb of the 5Ј-flanking sequences of the mouse LAMA1 genomic DNA was subcloned into the XbaI site of pBluescript II. An XbaI and blunted EcoNI fragment then was inserted into the pGL3-Basic vector (Promega, Madison WI) to generate a Ϫ2527/Ϫ30 (relative to the initiation codon) plasmid. To generate the longest promoter construct, Ϫ6198/Ϫ30, a blunted NsiI and SpeI fragment was inserted into the Ϫ2527/Ϫ30 plasmid (see Fig.  1). All of the 5Ј-deletion constructs were generated in these plasmids by using the endogenous restriction sites and the appropriate restriction sites in the polylinker.
Site-directed mutagenesis of four potential protein binding sites was carried out in the Ϫ3516/Ϫ3082 plasmid using the GeneEditor TM in vitro site-directed mutagenesis system (Promega) and/or the Gene-Tailor TM site-directed mutagenesis system (Invitrogen). The sequences of the mutagenic primers are available upon request. All of the mutants were verified by sequencing.
Transfection and Reporter Gene Assays-Cells in 24-well plates at 50 -70% confluency were transfected using the Effectene transfection reagent (Qiagen) with 200 ng of reporter plasmid and 20 ng of the Renilla luciferase expression vector phRL-null (Promega) as an internal control. 48 h later, the cells were harvested in Passive lysis buffer (Promega), and the lysates were assayed for luciferase activity using the dual-luciferase reporter assay system (Promega). Firefly luciferase activities of various mouse LAMA1 promoter constructs were normalized to that of the Renilla luciferase and expressed based on the activity of the pGL3-Basic or pGL3-Promoter plasmid as 1. The data are expressed as the mean values Ϯ S.E. of at least three experiments (duplicate samples). The p values were obtained using Student's t test.
Electrophoretic Mobility Shift Assays (EMSA)-Nuclear extracts of various cells were prepared as described previously (27). Singlestranded oligonucleotides were annealed at a concentration of 10 M in annealing buffer (1 mM Tris-HCl (pH 7.5), 1 mM MgCl 2 , and 5 mM NaCl) at 95°C for 5 min and then cooled to room temperature. Doublestranded DNA was end-labeled with [␣-32 P]dCTP and the DNA polymerase Klenow fragment (Invitrogen). Labeled DNA was separated from free dCTP by filtration through a ProbeQuant TM G-50 Micro Column (Amersham Biosciences).
Nuclear extracts (5 g) and, when indicated, unlabeled oligonucleotide competitors were preincubated in 23 l of the gel mobility shift assay buffer (10 mM HEPES-KOH (pH 7.9), 50 mM KCl, 0.6 mM EDTA, 5 mM MgCl 2 , 10% glycerol, 5 mM dithiothreitol, 0.7 mM phenylmethylsulfonyl fluoride, 2 g/l pepstatin A, 2 g/l leupeptin, and 87 ng/l poly(dI-dC) (Amersham Biosciences)) for 10 min on ice. An oligonucleotide probe (1 ϫ 10 5 cpm) was added to the mixture and incubated for an additional 30 min at room temperature. For antibody supershift analyses, 1 l of antibody was added and the incubation was continued for an additional hour. The antibodies used for the supershift analyses were raised against Sp1 (PEP 2, Santa Cruz Biotechnology, Santa Cruz, CA), Sp2 (K-20, Santa Cruz Biotechnology), Sp3 (D-20, Santa Cruz Biotechnology), and YY1 (H-414, Santa Cruz Biotechnology). DNAprotein complexes were separated from the free probe by 5% nondenaturing polyacrylamide gel electrophoresis. After electrophoresis, the gel was blotted onto Whatman 3MM paper, dried, and analyzed using a BAS2000 Image Analyzer (Fuji film, Tokyo, Japan).

RESULTS
Promoter Activity of the 5Ј-Flanking Sequence of the Mouse LAMA1 Gene-A LAMA1 genomic clone was isolated from a mouse BAC genomic library, and a 6.2-kb DNA fragment containing the 5Ј-flanking region of the mouse LAMA1 gene was subcloned and fully sequenced. This sequence is available through the GenBank TM data base (GenBank TM accession number AB097426). Previously, Sasaki et al. (28) estimated the 5Ј-untranslated region of the mouse laminin ␣1 transcript to be 128-bp long. According to this finding, neither a TATA box nor a CCAAT box was found proximal to the putative transcription start site. Mouse and human (GenBank TM accession number AC021879) LAMA1 genes display a high degree of sequence conservation in the proximal promoter regions (Ϫ200 ϳϩ1), suggesting that the LAMA1 proximal promoter regions contain binding sites for the transcription factors necessary for basal expression in rodents and humans.
To identify the cis-regulatory elements controlling the mouse LAMA1 gene transcription, a series of reporter plasmids driven by the 5Ј-flanking region of the LAMA1 gene of different lengths were constructed and transfected into mouse F9 cells before and after induction of differentiation by retinoic acid and dibutyryl cAMP (Figs. 1A-C). When compared with the Ϫ103/ Ϫ30 plasmid, the Ϫ178/Ϫ30 plasmid showed significantly higher activities in both undifferentiated and differentiated F9 cells (designated F9-stem and F9-PE cells, respectively), indicating that the basal promoter activity is localized within the Ϫ103 to Ϫ178 region (FspI to SfoI). Six other reporter plasmids with longer 5Ј sequences (i.e. Ϫ237/Ϫ30, Ϫ676/Ϫ30, Ϫ1036/ Ϫ30, Ϫ2046/Ϫ30, Ϫ2527/Ϫ30, and Ϫ2888/Ϫ30) showed transcriptional activity similar to Ϫ178/Ϫ30 in both F9-stem and F9-PE cells. Intriguingly, the transcriptional activity in F9-PE cells was dramatically elevated when the 5Ј-flanking region was extended to Ϫ3516, although such potentiation in transcriptional activity was not observed in F9-stem cells. These results indicate that the 630-bp region encompassing Ϫ3516 to Ϫ2888 contains an enhancer that is only effective in F9 cells in the differentiated state. Because differentiated F9 cells exhibit a parietal endoderm-like phenotype, we examined the transcriptional activity of these deletion constructs in the PYS-2 parietal yolk sac-derived cells as well as EHS tumor-derived cells that secrete a large amount of laminin-1 (Fig. 1, D and E).
A dramatic increase in the transcriptional activity was also detected with the Ϫ3516/Ϫ30 construct, but not with the Ϫ2888/Ϫ30 construct, in both PYS-2 and EHS tumor-derived cells, whereas a basal promoter activity was also detectable within the Ϫ103 to Ϫ178 region. These results suggest that the 630-bp region from Ϫ3516 to Ϫ2888 harbors an enhancer activity closely associated with parietal endoderm cells. Although the exact origin of the EHS tumor has not been determined, overproduction of extracellular matrix proteins similar to those of Reichert's membrane (29) as well as gene expression profiles determined by microarray analysis 3 indicates that EHS tumor cells are also parietal endoderm-like cells, lending support for the parietal endoderm-specific enhancer activity within the Ϫ2888/Ϫ3516 630-bp region.
Characterization of a Cell Type-specific Enhancer-To further localize the region critical for the enhancer activity, an 800-bp AflII fragment from Ϫ3684 to Ϫ2892 and its 5Ј-and 3Ј-deletion constructs were tested directly for their enhancer activity using the heterologous SV40 promoter. The 800-bp AflII fragment showed high enhancer activity (i.e. a 100-fold increase relative to the basal promoter activity) in EHS tumor-derived cells (Fig. 2) as well as in F9-PE and PYS-2 cells (data not shown). Deletion from the 3Ј-end to Ϫ3516 (AflII-SacI fragment) and from the 5Ј-end to Ϫ3082 (BglII-AflII fragment) abolished the enhancer activity. In contrast, a 435-bp SacI-BglII fragment covering nucleotides Ϫ3516 to Ϫ3082 retained the full enhancer activity, although further deletion constructs (Ϫ3516/Ϫ3214 and Ϫ3214/Ϫ3082) did not. These results indicate that both regions (SacI-XmnI and XmnI-BglII) contain the regulatory element required for the high expression of LAMA1 in EHS tumor-derived cells, making it likely that the 435-bp region from nucleotides Ϫ3516 to Ϫ3082 is sufficient for the enhancer activity in EHS tumor-derived cells. Similar results were also obtained with F9-PE and PYS-2 cells (data not shown).
To verify the activity of the 435-bp region as a cell typespecific enhancer, this fragment was cloned 5Ј to the SV40 promoter in both the forward and reverse orientation or as two copies in tandem and their enhancer activities were examined in F9-stem cells, F9-PE cells, PYS-2 cells, EHS tumor-derived cells, and NIH/3T3 fibroblasts (Fig. 3). The 435-bp fragment conferred high luciferase activity independent of its orientation in F9-PE, PYS-2, and EHS tumor-derived cells but not in F9-stem and NIH/3T3 fibroblasts. The tandem repeat of the 3 S. Futaki and Y. Hayashi, unpublished observations. 435-bp fragment was more potent than a single copy in the enhancer activity. Given that the enhancer activity was only detected in cells with parietal endoderm phenotypes, we concluded that the 435-bp SacI-BglII (Ϫ3516 to Ϫ3082) fragment acts as a parietal endoderm-specific enhancer.
Characterization of Nuclear Protein Binding by EMSA-To determine the regions in the 435-bp enhancer that interact with DNA-binding proteins, we prepared a series of overlapping double-stranded oligonucleotides (data not shown) altogether covering the whole segment and used them as probes for EMSA analyses. Among 24 sets of oligonucleotides, four oligonucleotides designated protein binding sites (PBS) 1-4 formed DNA-protein complexes with nuclear extracts from EHS tumor-derived cells (Fig. 4). All of the four DNA-protein complexes were detected not only with nuclear extracts from F9-PE and PYS-2 cells but also with those from F9-stem and NIH/3T3 cells, implying that the binding proteins are not unique to parietal endoderm cells.
To narrow down the enhancer activity within these four DNA segments, a series of mutant double-stranded oligonucleotides with 6-bp substitutions were used as competitors for the complex formation of a 32 P-labeled probe and nuclear proteins (Fig.  5). For PBS1, an excess amount of unlabeled oligonucleotides mut1-2 and mut1-3 competed with the protein binding, whereas mut1-1 failed to compete. These results indicate that a substituted sequence in mut1-1 (ATTAAG) is critical for the DNA-protein complex formation. Similarly, the nucleotide sequences substituted in mut2-3 (TAGGTG), mut3-1 (CCATCC), and mut4Ϫ2 (ATAATG) were identified to be critical for protein binding in PBS2, PBS3, and PBS4, respectively.

Contribution of Individual Elements to Enhancer Activity-
We next examined the contribution of these putative enhancer elements to the overall enhancer activity of the 435-bp fragment by introducing mutations in the 6-bp core sequences of the PBS1, PBS2, PBS3, and PBS4 segments (Fig. 6). Mutation at PBS1 had no significant effects on the 435-bp enhancer activity, although mutation in PBS2, PBS3, and PBS4 reduced the enhancer activity by 72%, 93%, and 48%, respectively. Double mutations in these three elements resulted in further reduction of the enhancer activity to 2-5% of the control, and mutations of all three sites completely abolished the enhancer activity. Similar results were observed in PYS-2 cells, but not in NIH/3T3 cells (data not shown). Mutation in the PBS3 element alone eliminated more than 90% of the enhancer activity, suggesting that PBS3 is the most critical for the enhancer activity. In contrast, mutation in the PBS4 element had only a modest effect. These data are consistent with the results that the XmnI-BglII fragment (Ϫ3214/Ϫ3082) lacking PBS1 through PBS3 showed little, if any, enhancer activity, whereas the SacI-XmnI fragment (Ϫ3516/Ϫ3214) lacking only PBS4 had significant enhancer activity (Fig. 2). Together, these results indicate that synergy of three protein binding sites (PBS2, PBS3, and PBS4) accounts for the bulk of the activity of the 435-bp enhancer.
Computer analyses using the TFSEARCH program (30) suggested that PBS2 and PBS4 contained putative binding motifs for Sp1-like (GTGTGG) and YY1 (TAATGG) transcription factors, respectively. To test whether these factors were responsible for the observed protein binding to PBS2 and PBS4, we performed competition and supershift assays. Competitor oli- gonucleotides with an authentic Sp1 (GGGGCGGGGC) or YY1 (GCGGCCATCT) binding motif abolished protein binding to PBS2 and PBS4, respectively, although those with mutations in the Sp1 or YY1 motif failed to compete (Fig. 7, A and B). Furthermore, antibodies to Sp1 and Sp3 produced supershifted complexes, whereas antibodies to YY1 inhibited PBS4-protein complex formation. Therefore, it seems probable that Sp1/Sp3 and YY1 bind to the PBS2 or PBS4 sequences, respectively. DISCUSSION Parietal endoderm derives from the primitive endoderm, which in turn derives from the inner cell mass of the blastocyst at 4.0 -4.5 days post coitum in the mouse (31). Parietal endoderm cells are the major fetal components of the yolk sac, synthesizing large amounts of laminin and collagen IV, which are incorporated into Reichert's membrane (32). Reichert's membrane plays a critical role in the maternofetal exchange of nutrients (33) and is important for the postgastrulation development of the murine embryo. Because parietal endoderm cells continually secrete large amounts of Reichert's membrane components during development, they may be regarded as an active in vivo protein biosynthetic system. However, the regulatory mechanisms of genes encoding Reichert's membrane components remain poorly understood. Elucidation of such mechanisms could have a significant impact on developing a system for the large scale biosynthesis of basement membrane components in vitro.
In this study, we have cloned the promoter region of the mouse LAMA1 gene and identified the distal enhancer (Ϫ3516 to Ϫ3082) responsible for the expression of the laminin ␣1 chain during the parietal endoderm differentiation of F9 cells. The enhancer was also active in PYS-2 and EHS tumor-derived cells but not in NIH/3T3 cells, suggesting that the enhancer activity is parietal endoderm-specific. Consistent with the definition of an enhancer, the 435-bp sequence enhanced luciferase gene expression in either the forward or reverse orientation from the heterologous SV40 promoter. By EMSA analysis, four protein binding sites (PBS1-PBS4) were identified in the 435-bp enhancer. Although the proteins binding to these elements were detected not only in parietal endoderm cells but also in undifferentiated F9 and NIH/3T3 cells, three of these elements (PBS2, PBS3, and PBS4) appear to be essential for the parietal endoderm-specific enhancer activity. The proteins binding to PBS2 were identified as Sp1/Sp3, and the proteins binding to PBS4 were identified as YY1.
Sp1/Sp3 and YY1 have broad tissue distribution and have been implicated in the regulation of several tissue-specific genes as well as housekeeping genes (34 -37). Sp1 binding sites have also been identified in heat shock protein 47 (38)  laminin ␥1 genes (39), both of which are highly expressed in F9 cells differentiated into parietal endoderm cells. However, it remains to be determined whether these sites are involved in their parietal endoderm-specific expression. Supershift analyses with anti-Sp1, anti-Sp2, and anti-Sp3 antibodies revealed that either Sp1 or Sp3 could bind to PBS2. It has been reported that Sp3 can function as a positive regulatory factor or as a repressor of Sp1-mediated transcription depending on its alternatively spliced isoforms (40,41). Further studies are required to determine which isoforms are involved in the 435-bp enhancer activity.
YY1 is also known to act as a transcriptional activator or repressor depending on its promoter context. The transcriptional activity of YY1 appears to be regulated at the posttranslational level, possibly through interaction with other proteins. In fact, a wide variety of transcription factors including Sp1 and nuclear receptor co-activators have been shown to associate with YY1 (34,36,(42)(43)(44). Considering these findings, the parietal endoderm-specific activation of the LAMA1 gene may be controlled by complex transcriptional pathways involving interactions among three ubiquitous factors (Sp1/Sp3, YY1, and an unidentified factor), tissue-specific co-factors, and posttranslational modification such as phosphorylation and acetylation. Recently, it was demonstrated that Akt/protein kinase B activates the transcription of all three chains of laminin-1 as well as type IV collagen (45). It has also been shown that the  DNA binding and transcriptional activities of YY1 and Sp1/Sp3 are regulated by acetylation and deacetylation (44,46,47). It remains to be explored whether Sp1/Sp3, YY1, and an unidentified factor binding to PBS3 are targets of such modifications.
Although parietal endoderm-specific enhancer elements have been identified in the ␣1(IV) and ␣2(IV) collagen genes (48,49), the proteins binding to these elements have not been identified. There is no clear sequence similarity between the enhancer elements in the collagen IV genes and the presently identified 435-bp enhancer. A parietal endoderm-specific enhancer has also been identified in the 5Ј-flanking region of the platelet-derived growth factor ␣ receptor gene (50), the expression of which is also induced in F9 cells during the differentiation into parietal endoderm cells. GATA-4, a member of the GATA transcription factor family, is considered to be responsible for the platelet-derived growth factor ␣ receptor enhancer activity. This is consistent with a recent report that GATA-4 and GATA-6 are key regulators of differentiation of the extraembryonic endoderm (51). The 435-bp enhancer has several GATA-like motifs, but it seems unlikely that these motifs are involved in the DNA-protein complex formation, because the double-stranded oligonucleotides containing the GATA-like motifs did not produce any significantly shifted band in the EMSA analysis and GATA-4 failed to activate the 435-bp enhancer (data not shown). These observations indicate that the parietal endoderm-specific gene expression can be conferred by either GATA-dependent or GATA-independent mechanisms. In search of the parietal endoderm-specific enhancer of the LAMB1 and LAMC1 genes, we cloned ϳ4-kb and ϳ7-kb genomic DNA segments covering the 5Ј-flanking regions of the LAMB1 and LAMC1 genes and examined their enhancer activity in PYS-2 cells. However, none of these DNA segments showed as strong transcriptional activity as the 435-bp enhancer. 4 Further sequences upstream of these region or the introns of the mouse LAMB1 and LAMC1 genes may contain a regulatory element similar to the 435-bp enhancer.
In conclusion, we have identified a parietal endoderm-specific enhancer of the mouse LAMA1 gene, which could explain the increased mRNA levels of laminin-1 during early mouse development. Further characterization of this enhancer, i.e. identification of the nuclear protein(s) binding to PBS3 and/or other factors interacting with Sp1/Sp3 and YY1, will clarify the novel mechanism(s) operating in the regulation of parietal endoderm-specific gene expression. This 435-bp enhancer system may also provide a clue to understanding the molecular basis of the large amount of production of basement membrane components in EHS tumor and parietal endoderm cells.