Characterization of cis-acting elements of the gene for macrophage-stimulating protein from the human. The involvement of positive and negative regulatory elements.

To analyze the promotor region of the human macrophage-stimulating protein (MSP) gene, the 5′-flanking region of this gene was cloned. The major initiation site was determined at T located 49 base pairs upstream of the translation initiation site by primer extention with mRNA from HepG2 and Hep3B cells. There was no TATA sequence in this region. Transient transfection assay with 5′-deletion constructs showed that the transcription of this gene was regulated by positive and negative regulatory elements (PRE and NRE). The PRE (−34 to +2) was essential for the maximal transcription of this gene, and the NRE (−141 to −34) appeared to be responsible for the tissue-specific expression of the gene. The PRE contained the CCAAT sequence and a mutation from CCAAT to CTGAT resulted in a significant loss of the transcriptional activity. Electrophoretic mobility shift assay suggested that two different proteins bound to the PRE (MSP-PRE-binding protein-1 (MSP-PREB1) and 2). MSP-PREB1 and 2 were detected in various cell types, and the CCAAT sequence was involved in these bindings. These findings indicate that MSP-PREB1 and 2 are positive regulators. Further characterization also revealed that MSP-PREB2 was identical to CCAAT binding transcription factor, also known as NF-Y.

Macrophage-stimulating protein (MSP) 1 was originally purified as a human serum protein that made mouse resident peritoneal macrophages capable of responding to the chemoattractant C5a (1). MSP is a heterodimeric protein that consists of a 53-kDa ␣-chain and a 25-kDa ␤-chain. The cloning of MSP cDNA revealed that MSP was synthesized as a biologically inactive single chain pro-MSP (2). It can be converted to the biologically active disulfide-linked heterodimer by serine proteases (3,4). MSP has high amino acid sequence similarity to hepatocyte growth factor/scatter factor (HGF/SF) (2,5). The genomic sequence for MSP was independently cloned by Han et al. (6). Recently, the proteins Ron (7) and STK (8) were identified as the receptors for human and mouse MSP, respectively (9 -12). MSP receptor has high amino acid sequence similarities to the c-met HGF/SF receptor and a member of receptor tyrosine kinase family. The concentration of pro-MSP in human plasma is optimal for biological activity, which is expressed when pro-MSP is cleaved to the MSP heterodimer. Therefore, it is likely that the expression of biological activity is controlled by the activation of pro-MSP and also by the expression of MSP receptors on the surface of target cells. Consistent with this view is the fact that the plasma concentration of pro-MSP is stable and does not increase like acute phase proteins of which concentrations increase in response to noxious stimuli (12).
We previously reported that the MSP mRNA was strongly expressed in the liver and weakly in the kidney and pancreas (2). The cloning of the MSP cDNA from human hepatoma cell line, HepG2, indicated that MSP was produced by hepatocytes. MSP mRNA was also detected at a lower level in the lung, adrenal, placenta and diaphragm during development in maternal rats in addition to the liver (13). Recently, transient expression of MSP was detected in the neural tube during development in the chicken (14). These findings suggest that the expression of MSP mRNA is controlled by factors expressed predominantly by hepatocytes or at the particular stages of the development. Recent studies also indicated that the MSP expression was up-regulated during liver regeneration after partial hepatectomy and hepatitis (15) and down-regulated during fluminant hepatic failure (16).
The transcriptional mechanisms of the proteins produced specifically or predominantly by hepatocytes have been investigated (reviewed in Ref. 17). The transcription of these genes for albumin (18), transthyretin (19), and ␣ 1 -antitrypsin (19,20) are regulated by positive regulators such as CCAAT/enhancerbinding protein-␣ (C/EBP␣), or hepatocyte nuclear factor (HNF) -1, HNF-3, or HNF-4. In contrast, the involvement of negative regulation is reported for apolipoproteins genes that are specifically expressed in the liver (21). The purpose of this study was to investigate the mechanisms that regulate the transcription of the human MSP gene. We characterized the cis-elements of the human MSP gene and found that the transcription of the human MSP gene was controlled by positive and negative regulatory elements.

EXPERIMENTAL PROCEDURES
Cell Lines-Hela, HOS (osteosarcoma), SK-RC29 (renal cell carcinoma), and A172 cells (glioblastoma), were cultured in Dulbecco's modified Eagle's medium (Life Technologies, Inc.) with 10% heat-inactivated fetal calf serum (FCS; Hyclone, Logan, UT). HepG2 hepatoma cells were cultured in RPMI 1640 medium (Life Technologies, Inc.) with 10% FCS, and Hep3B hepatoma cells were cultured in minimum essential * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) U43376.
RNA Analysis-Total RNA was extracted by TRIzol Reagent (Life Technologies, Inc.). Ten micrograms of total RNA were electrophoresed and blotted onto a Hybond-N ϩ nylon membrane (Amersham Corp.). The blots were hybridized with 32 P-labeled MSP 9.13B cDNA probe (2) or ␤-actin cDNA probe. After hybridization, blots were washed, then exposed to XAR-5 x-ray films (Eastman Kodak) with an intensifying screen at Ϫ80°C.
Cloning of the 5Ј-Flanking Region of the Human MSP Gene-A human fibroblast genomic library was purchased from Stratagene (La Jolla, CA). One million clones were screened with the 32 P-labeled MSP 9.13B cDNA probe (2). Sixteen positive clones after the second screening were amplified, purified, and analyzed for restriction enzyme mapping with HindIII (Life Technologies, Inc.). One of the clones, clone 8b, contained the 5.3-kb MSP 5Ј-flanking region. This 5.3-kb fragment was subcloned into the HindIII site of Bluescript II, SK(Ϫ) (Stratagene) (termed pBlue5.3k) and used for DNA sequencing and preparation of chloramphenicol acetyltransferase (CAT) constructs.
Construction of CAT Plasmids-The pCAT basic (Promega, Madison, WI) was used to construct MSP-CAT hybrids. To obtain pMD223, the DNA fragment covering the region from Ϫ223 to ϩ59 of the 5Ј-flanking sequence of the MSP gene was amplified by polymerase chain reaction (PCR) with oligonucleotide primers that corresponded to nucleotides from Ϫ223 to Ϫ204 of the sense strand (5Ј-GGCTGCAGAGGGGTT-TCACCCC3-Ј) and from ϩ40 to ϩ59 of the antisense strand (5Ј-GGTCTAGACCTTCTGGCTGGAGGCTGCA-3Ј). A XbaI linker was incorporated in the 5Ј-end of the antisense primer. The pBlue5.3k was used as the PCR template. The 280-bp PCR product digested with PstI and XbaI was cloned into the PstI-XbaI site of the pCAT basic vector. To obtain the pMD5.3k, the DNA fragment covering from Ϫ5.3k to Ϫ223 of the 5Ј-flanking sequence was cut out from pBlue5.3k with HindIII and PstI (partial digestion was required for PstI digestion). The fragment was gel purified and cloned into the HindIII-PstI site of the pMD223. The pMD1.5k and pMD1.0k were constructed from the pMD5.3k by removing fragments with appropriate restriction enzymes.
To obtain the pMD141, pMD34, pMD34M1 pMD34M2, pMD34M3, pMD34M4, pMD34M5, pMD34M6, and pM2, DNA fragments covering from Ϫ141 to ϩ59, Ϫ34 to ϩ59, or ϩ2 to ϩ59 were amplified by PCR with sense oligonucleotide primers that corresponded to the region from with an antisense probe from ϩ40 to ϩ59 (5Ј-GGTCTAGACCTTCTG-GCTGGAGGCTGCA-3Ј). PstI or XbaI linkers were incorporated in the 5Ј-ends of these primers, and the mutated bases are shown by lower case letters. The pMD223 was used as the PCR template. The PCR products were subcloned into the PstI-XbaI site of the pCAT basic vector after a digestion with PstI and XbaI.
Cell Transfection and CAT Assay-Cells were transiently transfected with 10 g of the each CAT construct and 5 g of pSV-␤ galactosidase plasmid DNA per 100 mm tissue culture plate by a calcium phosphate precipitation technique (22). After 12-h incubation with each DNAcalcium phosphate precipitate, the cells were rinsed with phosphatebuffered saline and cultured in fresh medium for 48 h. The cells were harvested, resuspended in 100 ml of 250 mM Tris-HCl buffer, pH 8.0, and lysed by five cycles of freezing and thawing. The lysates were spun by 12,000 ϫ g for 10 min at 4°C in an Eppendorf microcentrifuge, and the supernatants were transferred to new tubes. To normalize the transfection efficiency for each individual sample, ␤-galactosidase ac-tivity was assayed by using 10 l of each supernatant. CAT assay was performed by using the CAT Enzyme Assay System (Promega).
Preparation of Nuclear Extracts and EMSA-Nuclear extracts were prepared by the previously reported method (23). The protein concentration was determined by a protein assay kit (Bio-Rad), and aliquots of the nuclear extracts were frozen at Ϫ80°C. EMSA was carried out by using 5% polyacrylamide gels in 0.5 ϫ Tris borate-EDTA buffer (90 mM Tris borate, 2 mM EDTA) (23). The probes for this study are shown in Table I. Each binding reaction was performed at 20°C for 30 min in 10 l of 1 ϫ binding buffer (12 mM Hepes, pH 8.0, 60 mM KCl, 2.5 mM EDTA, 14% glycerol, 2 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride) containing 10 fmol of each probe end-labeled with ␥-[ 32 P]ATP and T4 polynucleotide kinase, 1 g of poly(dI⅐dC), and 5 g of each nuclear extract. For the competition assay, 10-, 30-, or 50-fold excess amount of the appropriate unlabeled probe was added to the binding reaction mixture. For antibody inhibition assay, nuclear extracts were preincubated with an antiserum against NF-YB (kindly provided by Drs. D. Mathis, C. Benoist, and R. Mantovani) (24) or preimmune serum at 20°C for 20 min. Gel electrophoresis was carried out at 200 V for 2 h after a prerun at 200 V for 30 min.

Expression of MSP mRNA in Tumor Cell Lines of Different
Tissue Origins-Tissue-specific expression of MSP mRNA was previously reported (2,6,13). Constitutive expression of MSP mRNA was detected in the liver and human MSP cDNA was consequently cloned from a human hepatoma cell line, HepG2 (2). In the present study, the expression of MSP mRNA was investigated by Northern blot analysis in six human tumor cell lines of different tissue origins. As shown in Fig. 1 Cloning and Characterization of the 5Ј-Flanking Region of the MSP Gene-After screening approximately one million individual clones from a human fibroblast genomic library, 40 positive clones were obtained. Sixteen clones were amplified and purified for further characterization. Restriction enzyme mapping with HindIII revealed that clones 8b and 14a contained the MSP gene (2,6,25). The other clones were from MSP pseudogenes located on chromosome 1 2 (2, 25). upstream from the ATG translation initiation codon appeared to be the major transcription initiation site of the human MSP gene (Fig. 3).
Identification of the MSP Transcriptional Regulatory Region in the 5Ј-Flanking Sequence-To investigate the regions responsible for the MSP gene transcriptional regulation, Hep3B and HOS cells were transiently transfected with CAT constructs containing different lengths of the 5Ј-flanking region of the MSP gene obtained by progressive deletion of the 5Ј-end (Fig. 2B), and then CAT activities in the cell extracts were compared (Fig. 2C). High CAT activity was detected in the extracts of Hep3B cells transfected with pMD5.3k. A deletion of the 5Ј-flanking region from Ϫ5.3k to Ϫ141 had little effect on the CAT activity. An additional deletion to Ϫ34 resulted in a 1.5-fold increase. The removal of an additional 36-bp sequence resulted in a significant loss of the CAT activity. In HOS cells, no significant CAT activity was found in the cells transfected with pMD5.3k or other constructs with progressive deletions up to Ϫ141. However, the deletion to Ϫ34 resulted in a significant increase in CAT activity and the removal of an additional 36-bp sequence resulted in a loss of the CAT activity. These results suggest that the transcription of MSP gene is regulated by two different regulatory elements: a negative regulatory element (NRE) located between Ϫ141 and Ϫ34, and a positive regulatory element (PRE) located between Ϫ34 and ϩ2. The binding of suppressor protein(s) to the NRE of the 5Ј-flanking region of the MSP gene may negatively regulate the expression of MSP mRNA in HOS cells.
The CCAAT Element Is Essential for the Transcription of the MSP Gene-Further experiments were performed to characterize the PRE. Several putative binding sites for trans-activators were found after sequencing the proximal 5Ј-flanking region (Fig. 4A). As described above, the region between bp Ϫ34 and ϩ2 was responsible for positive regulation of the MSP gene transcription in Hep3B cells. In this region, we found a CCAAT motif involved in the binding of a small group of transcription factors such as CCAAT-binding factor (26 -31). To investigate whether the CCAAT sequence was responsible for transcription of the MSP gene, six different mutations were introduced into the CAT construct as shown in Fig. 4B. The pMD34M2, which carried two mutated base pairs in the CCAAT sequence of the pMD34 vector, caused a significant decrease of the CAT activity in the cell extracts of Hep 3B cells transfected with this construct. Other mutations had no effect on the CAT activity.
These results indicate that the CCAAT sequence is the PRE of the MSP gene and is responsible for the positive regulation of human MSP gene transcription.
The positive transcriptional activity of the PRE was further studied in other types of cells transfected with pMD34 or pMD34M2 (Fig. 4C). High CAT activities were detected in the cell extracts of Hep3B, Hela, and HOS cells. Mutation in the CCAAT sequence caused significant decreases of CAT activity. Low but significant amounts of CAT activities were also de- MSP-PREB1 and 2 Bind to the CCAAT Element of the MSP Gene-To examine the binding of positive regulators to the CCAAT element, EMSA was performed. In the assay, an annealed 29-mer oligonucleotide (PRE probe) coding from Ϫ34 to Ϫ6 of the 5Ј-flanking region including the CCAAT sequence was used (Table I). As shown in Fig. 5, four shifted bands (DNA-protein complexes) were found when the nuclear extracts of Hep3B cells and 32 P-labeled PRE probe were incubated (lane 1). Addition of excess unlabeled PRE probe did not inhibit appearance of the complex 4, indicating that the complex 4 was nonspecific (lanes 2 and 3). Addition of a small amount of salmon sperm DNA caused disappearance of the complex 3 (lanes 4 and 5), indicating that the complex 3 was also nonspecific. In contrast, Addition of excess unlabeled PRE but not salmon sperm DNA caused the disappearance of complexes 1 and 2, indicating that these complexes were specific. Thus, the proteins (MSP-PREB1 and 2) in the Hep3B nuclear extracts appeared to bind specifically to the PRE and form the complexes. MSP-PREB1 and 2 did not bind to the 20-mer probe that contained the CCAAT sequence but not the 9-bp of the 3Ј-end of the PRE probe (see Fig. 7B, lanes 4 -6), indicating that 3Ј-sequence following the CCAAT sequence is also important for the protein binding.
To locate the sequence essential for the binding, a competition assay was performed with mutated PREs. The mutated PREs (M1, M2, M3, M4, M5, and M6) were cut out from the CAT constructs described above (pMD34M1, pMD34M2, pMD34M3, pMD34M4, pMD34M5, and pMD34M6, respective-  ly). As shown in Fig. 6, most of the band 1 and 2 disappeared by competition with 30-fold excess amounts of M1, M3, M4, M5, or M6, but not with the M2 fragment. These results suggested that both MSP-PREB1 and 2, which formed complexes 1 and 2, bound to the M1, M3, M4, M5, and M6 with high affinity, but not to the M2. Therefore, the CCAAT sequence was involved in the high affinity binding of the proteins. EMSA was also performed with nuclear extracts of HepG2, Hep3B, Hela, HOS, A172, and SK-RC 29 cells (Fig. 7A). Binding of MSP-PREB1 and 2 to the PRE probe was observed with all of the nuclear extracts, indicating the ubiquitous expression of the proteins. The results obtained by CAT assay and EMSA strongly suggest that MSP-PREB1 and/or 2 are the positive regulatory protein(s) for MSP gene transcription.
MSP-PREB2 Is Identical to the CCAAT-binding Factor, NF-Y/CBF-To compare MSP-PREB1 and 2 with known CCAATbinding proteins such as NF-Y/CBF (26), C/EBP␣ (30), and NF-I (31), EMSA was performed with oligonucleotide probes carrying the sequences of the specific binding elements for the each protein (Table I). Among the probes, an excess of the nonlabeled CBF probe which carried the CCAAT sequence of the ␣2(I) collagen gene strongly reduced the binding of PREB2, but only weakly reduced the binding of PREB1 (Fig. 7B, lanes  7-9). EMSA was also performed with nuclear extracts of Hep3B cells and the 32 P-labeled CBF probe. As shown in Fig. 7C, lanes  1-3, a shifted band that competed with the excess amount of unlabeled CBF probe was observed, indicating that the band was a complex of NF-Y/CBF binding to the CBF probe. This band also competed with excess amount of the PRE probe (Fig.  7C, lanes 4 -6). A weak band was also detected just above the NF-Y/CBF-CBF probe complex (Fig. 7C, shown by an open  arrowhead). This band might be due to the weak binding of MSP-PREB1 to the ␣2(I) collagen promoter.
It was previously reported that NF-Y/CBF binding to the ␣2(I) collagen promoter was enhanced by addition of 5 mM EDTA to the incubation mixture (32). In our assay, this enhancing effect was also observed (Fig. 7D, lanes 1-5). The binding of MSP-PREB1 and 2 was enhanced by 5 mM EDTA, and decreased in the presence of CaCl 2 (Fig. 7D, lanes 6 -10). It is not likely that MSP-PREB1 is a modified form of MSP-PREB2 by N-linked glycosylation since tunicamycin-treatment of the cells had no effect on the shifted bands (data not shown).
Finally, antibody inhibition assay was performed with an antiserum against the B subunit of NF-Y/CBF. As shown in Fig. 8, the formation of MSP-PREB2-DNA complex was inhibited by the antibody in a dose-dependent manner, indicating that MSP-PREB2 is identical to NF-Y/CBF. The formation of MSP-PREB1-DNA complex was also partially inhibited by the antiserum, suggesting that MSP-PREB1 may be a NF-Y/CBFrelated protein. The C/EBP␣ (Fig. 7B, lanes 10 -12), NF-I (lanes [13][14][15], and Sp1 (lanes 16 -18) probes did not affect on the binding of MSP-PREB1 or 2.

DISCUSSION
In the present study, we investigated the transcriptional mechanisms of the human MSP gene. The 5Ј-flanking region of the human MSP gene was cloned, and the transcription initiation sites were determined by primer extension. As shown in Fig. 3, multiple initiation sites were detected. However, the major extension product started at T located 49 bp upstream of the ATG translation initiation site. Multiple initiation sites are commonly observed in TATA-less genes (33). In fact, a TATA box was not found in the 5Ј-flanking region of the MSP gene (Fig. 4A). These results suggest that transcription of the human MSP gene is regulated by TATA-less promoters.
We next investigated the cis-acting elements regulating the MSP gene transcription by using CAT constructs ligated with progressively deleted 5Ј-flanking sequences, and found positive and negative regulatory elements. The PRE was located in the region from Ϫ34 to ϩ2, and is essential for the maximal transcriptional activity. The NRE was located in the region from Ϫ141 to Ϫ34, and the hepatocyte-specific transcription of the  1 and 4). Competition assays were performed with a 10-or 50-fold excess of the PRE probe (lanes 2 and 3). MSP gene appears to be regulated by this element.
The positive regulatory element contained the CCAAT sequence. Mutation in this sequence resulted in significant decreases of the transcriptional activity in Hep3B, HOS, and Hela cells. EMSA suggested that two different proteins, MSP-PREB1 and 2, bound to the positive regulatory element. The CCAAT sequence is characterized as the binding site for several trans-activators. One of the CCAAT-binding trans-activators, C/EBP␣, is known as a hepatocyte-specific transcriptional factor, and recognizes two cis-regulatory motifs, "TGTGG(A/ T)(A/T)(A/T)G" and "CCAAT" (30). C/EBP␣ is reported to regulate transcription of genes specifically or predominantly expressed in the liver including albumin (18), factor VIII (34), and ␣ 1 -antitrypsin (19,20). However, the present study suggests that C/EBP␣ is not the major factor that regulates the liverspecific expression of the MSP gene because the C/EBP␣ probe did not compete with the binding of either MSP-PREB1 or 2 to the PRE probe in EMSA.
CCAAT-binding transcription factor (CTF)/nuclear factor-I (NF-I) was originally identified as a cellular factor that was essential for the efficient replication of adenovirus DNA (31,35). However, recent studies revealed that CTF/NF-I recognized the sequence "TGG(N)6GCCAA" instead of the CCAAT sequence (36). Since the "TGG(N) 6 " sequence was not conserved on the 5Ј-flanking region of the MSP gene and the CTF/NF-1 probe did not compete the binding of MSP-PREB1 and 2 to the PRE probe, CTF/NF-1 does not appear to regulate the transcription of the MSP gene.
Other CCAAT-binding factors such as NF-Y/CBF (26,32), CP1 (28,29), CCAAT binding protein for the TK gene (CBP/tk) (37), or a 114-kDa CBF for the hsp70 gene (38) were candidates for the trans-activators of the MSP gene. NF-Y and CBF were originally identified as a binding protein to the Y box of the major histocompatibility complex (MHC) class II gene (27) and ␣2(I) collagen gene promoter (26), respectively. This factor consists of 32k-, 40k-, and 40k-subunits (A, B, and C subunits, respectively) (32,39). cDNA cloning of NF-Y and CBF subunits revealed that these factors were identical (39 -42). Recently it was reported that CP1, which was characterized as the binding protein of the ␣-globin promoter (28,29), was also identical. As shown in Figs Table I. C, NF-Y/CBF binding (arrow) to the CBF probe (lanes 1 and 4). This binding was inhibited by a 10-or 50-fold excess of nonlabeled CBF probe (lanes 2 and 3)   PREB1 may be different from NF-Y/CBF. However, there were a few similarities between MSP-PREB1 and 2. The mobilities on the gels were similar. The CCAAT sequence was essential for the binding of the proteins. EDTA enhanced the bindings of the proteins to the PRE probe. The MSP-PREB1 binding was also weakly inhibited by an antiserum against the B subunit of NF-Y/CBF. These facts suggested that MSP-PREB1 might be a NF-Y/CBF-related protein.
Although the CCAAT motif is essential for the NF-Y/CBF binding to DNA, the upstream and downstream sequences of the CCAAT element also affect on the binding affinity. Recently, Roy and Lee (43) reported that the G-rich sequence, which was conserved downstream of the CCAAT sequence of the ␣2(I) collagen gene and GRP78/Bip gene, was required for the maximal DNA-binding affinity of NF-Y/CBF. This G-rich sequence, however, is not conserved on the Y box of the MHC class II gene, the ␣-globin promoter, or the PRE of the MSP gene. In our study, the affinity of the CBF binding to the ␣2(I) collagen promoter sequence appeared to be higher than that to the PRE of the MSP gene (Fig. 7, B and C). The CCAAT sequence was also essential for the MSP-PREB1 binding to the PRE, but MSP-PREB1 did not bind to the CCAAT element of the ␣2(I) collagen promoter. Therefore, an additional sequence, which is located in either upstream or downstream of the CCAAT sequence of the MSP gene, may be required for the MSP-PREB1 binding. We attempted to locate the additional sequence for MSP-PREB1 binding by both CAT assay and EMSA with mutations, but we failed to find the sequence. To identify the binding motif required for the maximal binding of MSP-PREB1, further analyses will be necessary.
Several questions remained unanswered. CAT assay indicated the most potent enhancer activity was in the CCAAT element, but some enhancer activity was still detectable after the deletion of this sequence. Although EMSA indicated the ubiquitous expression of MSP-PREB1 and 2, CAT assay with mutated constructs showed higher activities in Hep3B, HOS, and Hela cells than in A172 or SK-RC29 cells. Therefore, another positive regulator might be involved in the regulation of the MSP gene transcription. In some TATA-less promoters, the initiator (Inr) element is critical in positioning RNA polymerase II, and sufficient to direct basal levels of transcription (44). We performed CAT assay with additional CAT constructs carrying different deletions. In the pMD141/D6, the sequence between Ϫ5 and ϩ49 was deleted from the pMD141. A significant decrease (1/20) of the CAT activity was observed in the cell extracts of Hep3B cells transfected with pMD141/D6. In contrast, no significant decrease was detected with the pMD141/36 that lacked the sequence between ϩ37 and ϩ49. As shown in Fig. 4A, a sequence similar to the consensus sequence for the Inr ( Ϫ3 YYCAYYYYY ϩ6 ) (45,46) was found between ϩ23 and ϩ31. However, substitutions of base pairs GG (ϩ18, ϩ19) with TT, CA (ϩ25, ϩ26) with AG, or CA (ϩ32, ϩ33) with TT did not affect on the transcriptional activity (data not shown). These results indicated that the region between Ϫ5 and ϩ49 was important for the transcription of this gene but the sequence between ϩ23 and ϩ31 was not responsible for the transcriptional activity. Further investigation of this region will lead us to a better understanding of the positive regulation of the MSP gene.
As described above, the negative regulation of the genes specifically expressed in the liver has been investigated. In the case of apoB gene, the binding of positive regulators, such as C/EBP␣, HNF-3, and HNF-4 to each regulatory element is essential for the maximal transcriptional activity of this gene, and this activity is negatively regulated by repressors that bind to these positive regulatory elements. For example, nuclear proteins, COUP-TFs, are capable of binding to the HNF-4 binding site of the apoB gene, and competitively down-regulate the transcription of this gene.
Our study also suggested the role of the NRE in the hepatocyte-specific expression of the MSP gene. However, the mechanisms for this negative regulation are not known at present. In this region, two putative binding consensus sequences for MyoD (47,48) and AP-2 (49 -51) were found. MyoD is a nuclear protein expressed in skeletal cells and contains a region homologous to the proteins of c-myc family. MyoD regulates the transcription of genes specifically expressed in skeletal muscle by binding to the promoter sequences of the genes. Since MSP expression was detected in diaphragm during development in maternal rats (13), it will be interesting to investigate whether MyoD is involved in the MSP gene transcription in skeletal muscle. AP-2 was detected in several kinds of cells such as Hela cells but not in HepG2 cells (49). Therefore, AP-2 might be involved in the negative regulation of the MSP gene. Further investigation of the NRE is necessary to understand the tissuespecific expression of the MSP gene.