Identification of a tissue-specific regulatory element within the murine CD14 gene.

We previously isolated and sequenced the 5'-flanking region of the mouse CD14 (mCD14) gene (Matsuura, K., Setoguchi, M., Nasu, N., Higuchi, Y., Yoshida, S., Akizuki, S., and Yamamoto, S. (1989) Nucleic Acids Res. 17, 2132). To define the regulatory elements that control expression of the mCD14 gene, we analyzed the structure of the 5' end of the gene, including a region further upstream of that determined previously. Sequentially 5'-deleted, chimeric, and point mutated clones were tested for the ability to stimulate chloramphenicol acetyltransferase. An 8-base pair sequence, TGATTCAC, at position -255, which resembled the consensus sequence of the 12-O-tetradecanoylphorbol-13-acetate-responsive element (TRE), enhanced the expression of the chloramphenicol acetyltransferase gene in macrophage (aHINS-B3) and non-macrophage (glioblastoma G203 and myeloma NS1) cells. The enhancing ability of the TRE-like sequence (TLS), however, was markedly reduced in G203 cells but not in aHINS-B3 cells when the TLS was followed by the sequence immediately downstream. The TLS and sequence immediately downstream were capable of binding nuclear proteins which were unique to aHINS-B3 cells and macrophages, suggesting that these unique protein regulate the specific expression of the mCD14 gene. Binding of AP-1 to the TLS was also found in aHINS-B3 and G203 cells. Although it is uncertain whether AP-1 is involved in expression of the mCD14 gene, the effect of AP-1 in non-macrophage cells was inhibited by a nuclear protein which binds to the sequence immediately downstream of the TLS.

ments that control expression of the mCD14 gene, we analyzed the structure of the 5' end of the gene, including a region further upstream of that determined previously. Sequentially 5'-deleted, chimeric, and point mutated clones were tested for the ability to stimulate chloramphenicol acetyltransferase. An 8base pair sequence, TGATTCAC, at position -255, which resembled the consensus sequence of the 12-0tetradecanoylphorbol-13-acetate-responsive element (TRE), enhanced the expression of the chloramphenicol acetyltransferase gene in macrophage (aHINS-B3) and non-macrophage (glioblastoma G203 and myeloma NS1) cells. The enhancing ability of the TRE-like sequence (TLS), however, was markedly reduced in G203 cells but not in aHINS-B3 cells when the TLS was followed by the sequence immediately downstream. The TLS and sequence immediately downstream were capable of binding nuclear proteins which were unique to aHINS-B3 cells and macrophages, suggesting that these unique protein regulate the specific expression of the mCD14 gene. Binding of AP-1 to the TLS was also found in aHINS-B3 and G203 cells. Although it is uncertain whether AP-1 is involved in expression of the mCD14 gene, the effect of AP-1 in non-macrophage cells was inhibited by a nuclear protein which binds to the sequence immediately downstream of the TLS.
CD14 is a myelomonocytic differentiation antigen (1-7) and is the cell surface receptor for the lipopolysaccharidelipopolysaccharide-binding protein complex (8). Interaction of CD14 with the complex activates myelomonocytic cells, leading to tumor necrosis factor production (9).
Cell type-specific regulation of gene transcription appears t o be mediated by the binding of regulatory proteins (transacting factors) to specific cis-regulatory regions of genes that are usually, but not always, located in the 5"flanking region of the gene. Cis-regulatory regions usually consist of short DNA binding sites and a combination of binding sites for * This work was supported in part by Grant 02454169 from the Ministry of Education, Science, and Culture. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" 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 D10912.
to the GenBankTM/EMBL Data Bank with accession number($ ubiquitous and cell-specific factors. Expression of the CD14 gene is highly tissue-specific. It is, therefore, of interest to identify the specific cis-regulatory regions and trans-acting factors that are responsible for the expression of this gene and to assess of their degree of tissue specificity. We already cloned the mCD14l gene and 5'-flanking region (10,11). To begin to characterize the cis-and trans-acting elements involved in the transcriptional control of the mCD14 gene, we analyzed the upstream region of the mCD14 gene. We here describe potential cis-acting regulatory elements in the 5'-flanking region, show that an 8-base pair upstream sequence confers cell type-specific expression onto a reporter gene in an mCD14-positive macrophage cell line, and suggest that unique positive regulatory elements are required to achieve cell-specific expression of the mCD14 gene.

EXPERIMENTAL PROCEDURES
Materials-Restriction enzymes were purchased from Takara, Toyobo, Bethesda Research Laboratories, and Wako Pure Chemicals. Agarose, ultrapure DNA grade, and DNA ligation kit were from Takara. A 5' deletion kit was obtained from Nippongene. Radioactive nucleotides [ o ( -~~P ]~C T P (3,000 Ci/mM) and [Y-~*P]ATP (6,000 Ci/ mM) were obtained from Du Pont-New England Nuclear. Chemicals used for DNA sequencing were obtained from Toyobo. X-ray film (XAR-351) was obtained from Kodak.
Cell Lines, Cells, and Cell Cultures-Murine macrophage cell lines aHINS-B3 (12-14) and 5774, murine glioblastoma line G203, kindly supplied by Dr. J. Kuratsu (Department of Neurosurgery, Kumamoto University), and the murine myeloma cell line PS/NSl/l-Ag-l (NS1) were grown in Dulbecco's modified Eagle's medium containing 10% fetal calf serum. Murine peritoneal macrophages were obtained from the peritoneal cavity of ICR mice after injection with peptone (lo%, 1 ml) 4 days previously. Mouse liver, kidney, spleen, heart, and skeletal muscle were obtained from ICR mice.
Molecular Cloning and Sequencing-The 5"upstream sequence up to -4 kilobases was cloned from a bacteriophage EMBL3 murine liver library (15) using "P-labeled cDNA containing the mCD14 insert (16, 17) and partially sequenced as described previously (12,18). SI Nuclease Mapping-S1 nuclease mapping was carried out according to the method of Berk and Sharp (19). First, total RNA from aHINS-B3 cells was hybridized with an end-labeled NheI-NcoI fragment (see Fig. 1) in 40 mM Pipes (pH 6.4), 0.4 M NaCI, 1 mM EDTA, and 80% formamide for 16 h at 50" C. Following hybridization, the reaction was diluted 10-fold with S1 nuclease (60 units), and the reaction mixture was incubated for 35 min a t 37" C. The reaction mixture was terminated by the addition of termination buffer (2.5 M ammonium acetate and 50 mM EDTA), and the DNA-RNA hybrids were precipitated with ethanol, dissolved in the loading buffer, heated to 90" C, and resolved on a 6% acrylamide, 7 M urea sequencing gel.
Constructions of mCD14-CAT Plasmids"118pHlO clone containing the 739-base pair nucleotides corresponding to bases -678 to 61 of the 5"upstream and exon 1 sequence of the mCD14 gene was digested with KpnI-SalI. The resulting insert was 5' deleted using a kit (Takara) according to the manufacturer's instructions. Briefly, the insert DNA was digested with exonuclease 111. Aliquots (2.5 pl) were removed at 1-min intervals and mixed with 50 p1 of 2 X mung bean nuclease buffer. After heating for 5 min a t 65 "C, the samples were incubated with mung bean nuclease for 60 min a t 37 "C. The 5'deleted samples were then digested with the Klenow fragment followed by T4 DNA ligase treatment. T o determine the extent of deletion into the 5' end of the promoter, each mutant was sequenced as described previously (18). Selected mutants was digested with Sac1 and Hind111 and inserted into the Sad-Hind111 sites of the multicloning site containing pSVmCAT vector which was produced from the pSVOCAT vector. Plasmid DNA was prepared by alkaline lysis folplasmid DNA. lowed by two centrifugation steps through CsCl to isolate supercoiled DNA Transfections and CAT Assay-Plasmid DNA was transfected into the mouse macrophage line aHINS-B3 and the mouse glioblastoma line G203 by calcium phosphate coprecipitation as described previously. Cells were seeded a t 2 X lo6 cells/plate. For transient transfections, a mixture of the test CAT hybrid gene (20 pg equivalent) and the transfection control plasmid (20 pg) was precipitated in Hepes-buffered saline (pH 7.1) and then added to the plates. After 4 h, the cells were given a 2-min glycerol shock followed by a wash with ice-cold phosphate-buffered saline. Fresh medium was then added and the incubation continued for 40-48 h. Cells were washed three times with phosphate-buffered saline, incubated with 40 mM Tris-HC1 (pH 7.4), 150 mM NaCl, 1 mM EDTA for 5 min on ice, harvested by scraping and centrifugation, and resuspended in 180 ~1 of 0.25 M Tris-HC1 (pH 8.0). Extracts were prepared by sonication or by freeze-thaw and centrifuged. The myeloma cell line NS1 was transfected by the DEAE-dextran procedure essentially as described by Grosschedl and Baltimore (20). Briefly, a total of 4 X lo6 cells was washed, resuspended, and incubated in 2 ml of 25 mM Tris, 137 mM NaC1,5 mM KC1,0.3 mM Na2HP04-7H20, 1 mM MgC12,l mM CaCh (pH 7.4) containing 0.5 mg/ml DEAE-dextran (Pharmacia LKB Biotechnology Inc.) and 10 pg of plasmid for 30 min a t 37 "C. After incubation, the cells were added with 5 ml of culture medium and incubated for 10 min a t 37 "C, then collected by centrifugation and suspended in 10 ml of medium. After 48 h in culture, cells were washed with phosphate-buffered saline, freeze-thawed five times in 0.25 M Tris-HC1 (pH 7.5), and heated to 65 "C for 5 min to inactivate deacetylases. Supernatants were collected and assayed for protein according to the method of Bradford (21). CAT was assayed as described previously (22). The extract was incubated with ['4C]chloramphenicol and 0.2 mg of acetyl-CoA for 1 h a t 37 "C, and the products were separated by thin-layer chromatography.
Extract Preparation-Nuclear extracts from a variety of cells were prepared as described by Schreiber et ai. (23) with slight modification.

DNA Sequence Analysis of the 5' End of the mCD14 Gene-
We reported previously that the mCD14 gene (10) and the 5'upstream sequence extended 623 nucleotides upstream of the initiation codon (position l), which was defined by S1 nuclease mapping and primer extension analysis (Fig. l , A and  B ) . To analyze the cis-regulatory structure, a further 723 nucleotides upstream of the 5' end sequence were sequenced (Fig. 2 4 ) .
A long repeat that had 90% homology with the nucleotide sequence at position -152 to 1 containing several TATA-and CAAT-like and IRF-1 binding sequences, was found at position -1285 to -1134. Alu-like Bl-DZ and B1-C sequences were found at positions -1154 to -1001 and -799 to -704, respectively (33,341.  (Fig. 2) was hybridized to total RNA from aHINS-B3. After digestion of single-stranded RNA with S1 nuclease, the resultant protected fragments were analyzed by electrophoresis. Dideoxynucleotide sequencing reactions were electrophoresed in parallel as markers. Lane I , marker (end-labeled pUC19 Hepa I1 digest); lane 2, products of S1 nuclease digestion in the presence of total RNA from aHINS-B3 cells. The arrow indicates the site corresponding the transcriptional site. B, total RNA from aHINS-B3 cells was annealed to a 20-base oligonucleotide specific for a 5' region of mCD14 mRNA. The oligomers were extended and analyzed as described under "Experimental Procedures." Dideoxysequencing reactions were electrophoresed in parallel as markers. Lane 1, aHINS-B3 RNA with reverse transcriptase; the arrow indicates the transcriptional site (nucleotide 1, see Fig. 2). digestion of the HindIII-Hind111 fragment of the mCD14 gene (118pH10, 739 bases containing a promoter and an exon 1 region) with exonuclease I11 followed by mung bean nuclease digestion. This resulted in promoter fragments of 681,541,373,364,331,318,306,260, and 109 base pairs. The HindIII-HindIII and deletion fragments were introduced into a CAT vector ( Fig. 3; see Fig. 5). The contructs were transfected into aHINS-B3, which express mCD14 and the glioblastoma cell line G203 and NS1. G203 and NS1 cells were used as an mCD14-negative background that does not substantially activate mCD14-specific promoters.
The results of the various constructs and deletions are presented as the percentage of chloramphenicol conversion compared with that in plasmid pSV2CAT which contains the CAT gene transcribed from the simian virus 40 early promoter. CAT activities were corrected for variations in transfection efficiencies, by cotransfecting the pCHllO containing the @-galactosidase gene and normalized with respect to those obtained with pSV2CAT. The region including only 48 base pairs immediately upstream of the transcription start site was capable of significantly directing CAT synthesis in aHINS-B3 and G203 cells. The relative amount of reporter gene expression changed appreciably by progressively including additional 5' sequences up to -257 in aHINS-B3 but not in G203 cells. Further elongation of the 5"upstream sequence diminished the reporter gene expression. These results indicate that the 5"flanking region from positions -257 to -48, particularly -257 to -198, contains a cis-regulatory sequence specific for aHINS-B3 cells. To examine whether the upstream long repetitive sequence (ULR) had promoter activity similar to that of the downstream repeat, a CAT clone containing the XbaI-EcoRI fragment corresponding to positions -1308 to -1045 was constructed. The ULR, however, showed no CAT activity.
As stated above, the TLS was found at nucleotide position -255 to -249, which in the 5' region of 257CAT, demonstrated the highest CAT efficiency. We therefore predicted that TLS would play a role on the expression of the mCD14 gene. An oligonucleotide trimer with the sequence GATCCGTGATTCACGTGATTCACGTGATTCACT designated TLS-3 was synthesized and tested for CAT activity. TLS, when introduced upstream of the CAT gene in 48CAT, showed highly efficient CAT activity in aHINS-B3 cells (Fig.  4). The efficiency was also quite high in G203 (Fig. 4) and NS1 cells (data not shown). Similar results were obtained with the TLS-3R-CAT clone containing TLS-3 upstream of the CAT gene in 48CAT, in an inverted orientation. Specifically enhanced effect in aHINS-B3 cells with 257CAT suggested that a negative cis-regulatory element existed downstream of the TLS sequence. An oligonucleotide trimer with a sequence containing that immediately downstream of the TLS, designated TLSCT-3, was synthesized and examined for enhancing activity. CAT efficiency of TLSCT-3, when introduced upstream and downstream of the CAT gene in 48CAT, was high in aHINS-B3 cells (Figs. 4 and 5) but not in G203 (Fig. 4) and NS1 cells (data not shown). TLSCT-3 also had significant CAT activity in aHINS-B3 cells when inserted downstream of the CAT gene in 48CAT in inverted orientation.
Point mutation of the first nucleotide sequence T to G of the TLS in 257CAT abolished enhancing activity. A similar finding has been reported for TRE (35). TLS is similar to the enhancer core sequence TGATTCAG in the glutathione transferase P gene, in which the core palindromic sequence is presented three nucleotides apart upstream (35,36). Mutation of the eighth nucleotide, C to G, also greatly eliminated the activity in aHINS-B3 cells, suggesting that TLS is different from the enhancer sequence of the glutathione transferase P gene (Fig. 4). The mutation gave no effect on the activity in G203 cells although the level was low.
Gel Shift Analysis of Proteids) That Specifically Bind to TLS and Its Downstream Sequence-To examine the specific binding ability of nuclear protein to TLS and its downstream sequence, band shift assay was performed. Three oligonucleotides GATCCTGTGATTCACTCT, and GATCCTTTTCC-TGTACT corresponding to positions -257 to -245, and -246 to -234, designated A and B, respectively, and TRE were used in competition experiments. Three distinct bands I, 11, and I11 were identified with probe A when incubated with nuclear extract from aHINS-B3 cells (Fig. 6, A and B). Poly(dI.dC) was unable to inhibit the complex formation even at high concentrations (Fig. 6A). G203 nuclear extract formed a single band similar to band I, both of which were competed with TRE nucleotides (Fig. 6, B and D). Incubation of probe A with nuclear extracts from peritoneal macrophages,

IRF-1 -30 TATA -20 -63 A C A G A G G A A G G G A C A G G G T G C C C C C A G G A r r A C A T A A A
-1 5774, and spleen gave rise to one major band similar to band nuclear protein slightly formed bands comparable to bands I I1 (Fig. 7 A , C, and D and E ) , but those from non-macrophage and 11. Band I1 was competed with both oligomers A and B, cell lines and tissues including the liver, kidney, heart, and suggesting that band I1 bound to the TLS and sequence skeletal muscle did not (Fig. 7, C and D), although NS1 immediately downstream. The competition efficiency of oli- pSV2CAT contains SV40 promoter only; B78CAT contains residues -678 to 61; 480CAT contains residues -480 to 61; 312CAT contains residues -312 to 61; 257CAT contains residues -257 to 61; 199CAT contains residues -199 to 61; 48CAT contains residues -48 to 61; TLSCT-3-48CAT contains TLSCT trimer sequence immediately upstream of the CAT gene in 48CAT; 4RCAT-TLSCT-3 contains TLSCT trimer sequence immediately downstream of the CAT gene in 48CAT.

IFN IRF-1 Hindl I I 18 C C T G C A A G C T C G C T G T~G~~C T G T A A A C G A A A C A A A C T T
gonucleotide B appeared to be slightly higher than that of A. In addition, the formation of band I1 was significantly competed with T R E when a high concentration of T R E was included in the binding reactions. This may simply be attributable to that band I1 protein has affinity with TRE. Alternatively, it is possible to assume that band I1 protein has affinity with AP-1 since band I was effectively competed with B as well as T R E (Fig. 6 B ) . Band 111-forming protein, although the amount seemed less abundant, was also found in nuclear extracts from aHINS-B3 and macrophages but not in those from non-macrophage cells, suggesting that the protein also plays a role in CD14 expression.
A band corresponding to band I1 was formed with probe B when aHINS-B3, macrophage, and spleen nuclear extracts were used ( Fig. 6C and Fig. 7, B, C, and E ) , whereas probe B formed another band (band IV) when C203, NS1, liver, and kidney cell nuclear extracts were used ( Fig. 6D and Fig. 7 , C and E ) . The band was discriminated from band I1 since it was exclusively formed with probe B. Gel shift assays with probe B using aHINS-B3 and macrophage nuclear extracts also showed that band I was not formed, suggesting that AP-1 released from band 11-forming protein could not bind to probe B. These results are summarized in Table I.

DISCUSSION
Upstream of the mCD14 gene contains a variety of potential cis-regulatory elements including TATA-and CAAT-like sequences, IRF-1 binding sequences, a motif found in upstream sequences of interferon-inducible genes, VDRE-like sequence, as well as the TLS and TRE sequences. In addition, the repeated nucleotide sequence with the sequence a t position -1 to -153 containing several TATA-, CAAT-like, and IRF-1 binding sequences was found a t position -1285 to -1134.
CAT analysis suggests that expression of the mCD14 gene is regulated by a t least two 5"upstream sequences; one is the proximal promoter element a t position -48 to -1 containing a TATA-like sequence, and the other is the enhancer TLS at position -255 to -249. Since levels of enhancement by the proximal promoter element are significantly high in aHINS-B3 cells, the element would be necessary for expression of the mCD14 gene. T L S plays an important role in cell type-specific expression in aHINS-B3 cells. TLS-3 markedly enhances the level of expression of the CAT gene when inserted at the upstream of 48CAT in both orientations. A significant enhancement, however, was observed in G203. In contrast, TLSCT-3 increased the level of CAT gene transcription when inserted at a different position or in inverse orientation in aHINS-B3 but not in G203 cells. Gel shift analysis showed that unique nuclear proteins in addition to AP-1 bound to T L S in aHINS-B3 cells. The major protein, which bound to TLS and the sequence immediately downstream, formed band I1 that was also found in macrophages, macrophage cell lines, and spleen but not in non-macrophage cells. In C203 cells, T L S was exclusively bound with AP-1. In addition, binding of another nuclear protein responsible for band IV to the sequence immediately downstream of TLS was observed. Nuclear proteins in other non-macrophage cells also contained PanelA, gel shift and competition assays using poly(dl.dC) with aHINS-R.7 nuclear extracts. Pand R, gel shift and competition assays using A and I3 fragments and TRE with aHINS-H:3 nuclear extracts. P o n d C , gel shift and competition assays using A and H fragments with aHINS-€3.7 nuclear extracts. Panel 11, gel shilt and competition assays using A and R fragments and THE with G20.7 nuclear extracts. 1. /I. 111, and I V denote the oligonucleotide-protein complexes by gel shift assays. similar probe B-binding proteins. Band 111-forming protein, sequence immediately downstream, contributes to the specific which appeared to be less abundant in macrophages, may play expression of the mCD14 gene in monocytic lineages. TLS is a role in mCD14 expression in aHINS-B3 cells. Taken collec-capable of binding AP-1 in aHINS-BB and G203 cells. Altively, these results suggest that binding of unique nuclear though it is uncertain whether AP-1 is involved in mCD14 proteins, especially band 11-forming protein to TLS and the gene expression, band IV-forming protein in (220.7 cells binds Gel shift and competition assays using A a n d B f r a g m e n t s a n d TRE with nuclear extracts from macrophages, macrophage and non-macrophage cell lines, and various organs. Pond A. gel shift and competition assays using A and H fragments and TRE with macrophage nuclear extracts. Panel H, gel shift and competition assays using A and R fragments with macrophage nuclear extracts. Panel C , gel shift assays using A and €3 fragments with aHINS-H:1. macrophage, Ji74, and NS1 nuclear extracts. Panel I). gel shift assays using A fragment with nuclear extracts from macrophage, heart, liver, kidney, skeletal muscle, and spleen. Pond Fa', gel shift assays using B fragment with nuclear extracts from macrophage, heart, liver, kidney, skeletal muscle, and spleen. I, II. 111, and I V denote the oligonucleotide-protein complexes by gel shift assays. t o immediately downstream of the TLS and could inhibit the effect of AP-1. TPA, however, only had a marginal enhancing effect up to 60 min on expression of the mCD14 gene.' The contribution of AP-1 on expression of the mCD14 gene is probably slight, if any. Importance of band 11and IV-forming proteins in the regulation of mCDl4 expression is also suggested in other cells (Table I). TLS, TGATTCAC is similar to the enhancer core sequence, TGATTCAG of the glutathione transferase P gene, (34) which is highly expressed in chemically induced rat hepatoma cells, precancerous hyperplastic nodules, and to a much lesser extent in other tissues including the lung, kidney, and testis. A palindromic sequence located 5"upstream and situated three nucleotides apart from the GTP enhancer core sequence is required for expression of the glutathione transferase P gene. A palindromic sequence was found in the adjacent region upstream of TLS in the mCD14 gene. It is, however, unlikely that the palindromic sequence is necessary for enhancing ability of expression of the CD14 gene because CAT expression is reduced in 270CAT containing the palindromic sequence.
The enhancer effect of the GTP enhancer and TRE sequence is reportedly abolished when the 5' end nucleotide T is changed to G. Similarly, a change in the 5' end of TLS in the mCD14 gene caused the reduction of CAT expression. The 3' end of the sequence of TLS, C is different from that of GTP enhancer core, which is G. It appears that C is essential in the CD14 gene expression because mutation of C t o G markedly reduces the expression, indicating that TLS is a unique enhancer sequence although it is cross-reactive with AP-1 in addition to CD14-specific transcription factor(s).
Homology of the nucleotide sequence of the 5"upstream repeat with the proximal promoter sequence reached 90%. The ULR, however, failed to show promoter activity. This could be attributed to disruption of TATA activity by a single nucleotide change C for T within the TATA-like sequence of the ULR, namely TAAACTTA uersus TAAATTTA.