Differential Cooperation between Regulatory Sequences Required for Human CD53 Gene Expression*

CD53 is a tetraspanin protein mostly expressed in to the lymphoid-myeloid lineage. We have characterized the humanCD53 gene regulatory region. Within the proximal 2 kilobases, and with opposite transcriptional orientation, is located the promoter-enhancer of a second gene, which does not affectCD53. Twenty-four copies of a CA dinucleotide repeat separate these two gene promoters. The proximal enhanceosome of the human CD53 gene is comprised between residues −266 and +84, and can be subdivided into four major subregions, two of them within exon 1. Mutational analysis identified several cooperating sequences. An Sp1 and an ets-1 site, at positions −115 and +62, respectively, are essential for transcriptional competence in all cell lines. Five other regulatory sequences have a dual role, activator or down-regulator, depending on the cell line. At the end of the non-coding exon 1, +64 to +83, there is a second ets-1 regulatory element, which is required for high level of transcription, in cooperation with the Sp1 site, in K562 and Molt-4, but not in Namalwa cells, where it functions as a repressor. This Sp1 site also cooperates with another ets-1/PU.1 site at −172. Different cell types use different regulatory sequences in the enhanceosome for the expression of the same gene.

Gene regulation in a specific cell type requires the cooperation of several cis-acting DNA regulatory sequences, which are binding sites for proteins that transmit molecular signals to genes (1). These sequences bind regulatory proteins and form complexes known as enhanceosomes (2,3). The CD53 gene codes for an antigen that belongs to the tetraspanin family of membrane proteins. These proteins are very hydrophobic and consist of four transmembrane domains with polar residues, and have a small and a large extra cellular loop; the latter appears to be responsible for the functional specificity of each individual tetraspanin protein (4,5). CD53 was originally reported to be a pan leukocyte antigen, whose expression is mostly restricted to the lymphoid-myeloid lineage (4). Recent data suggests that tetraspanin proteins play an important role as co-stimulatory molecules in several cell types (5). Ligation of the CD53 antigen with monoclonal antibodies modulates several biological processes. Among them are intracellular calcium mobilization in human B-cells and monocytes (6) and rat macrophages (7), induction of homotypic adhesion (8,9), and in rat macrophages also induces the expression of the inducible form of nitric-oxide synthase (7). No ligand is known for any tetraspanin antigen. Tetraspanin antigens form a protein complex with several integrins on the cell membrane, which might require a coordinated expression of their genes (10 -13).
Loss of CD53 antigen surface level has been reported in a family with CD53 deficiency, which is characterized by the occurrence of recurrent and heterogeneous infectious diseases caused by viruses, bacteria, and fungi (14), a syndrome similar to the clinical manifestations of defects in leukocyte adhesion properties (15). Also, down-regulation of several tetraspanin antigen gene expression, such as CD9, CD82/KAI1, and CD63, has been correlated with poor prognosis in several types of tumors, including breast, melanoma, lung, prostate, pancreas, and esophageal carcinoma (16 -26). Reintroduction of these antigens, such as CD9 or CD82 acted as a brake reducing cell motility (27). These effects might be a consequence of the tetraspan-integrin interaction and their corresponding effects on cellular motility and adhesion (13). In all these situations, the correlation was performed for a unique member of the tetraspanin family. However, an individual cell expresses simultaneously from 5 to 10 members of this protein family (10,11,28), where they form a complex in the membrane that is different depending on cell type (10,11,13,29,30). This occurs both in normal cells (10), lymphomas (31), and carcinomas. 1 The reduction in levels of an individual protein represents a change in the composition of such complexes (32). Therefore, the role of low-level expression of any given antigen should be considered in the context of the multiple antigen expression.
To understand the coordinated expression of genes coding for tetraspanin antigens, which are located on different chromosomes, it is necessary to characterize their gene regulatory regions. Very few promoters of this growing gene family have been characterized. Only the CD9 promoter (33,34) and the CD63 promoter (35,36) have been studied in some detail, and the genomic location of the CD53 transcription initiation site has also been identified (37). In this report we have cloned and characterized the promoter region of the human CD53 gene. By mutational analysis several cis-acting DNA regulatory sequences were identified, which cooperate and contribute to CD53 gene expression in different cell types. Several of these new regulatory elements are located within the non-coding * This work was supported in part by Ministerio de Ciencia y Tecnología Grant SAF2000/0169 and Junta de Castilla y León Grant CSI-1/ 01. 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  exon 1. Some of these regulatory sequences function as positive or negative elements depending on the structure of the enhanceosome in each cell type. The basic CD53 regulatory region has an Sp1 sequence that cooperates with several ets-1 sequences. Also the promoter-enhancer of a very proximal second gene does not interfere with the CD53 promoter activity.

EXPERIMENTAL PROCEDURES
Genomic Cloning-To isolate the CD53 gene promoter we used a human genomic library from a 3-year-old caucasian male made by partial EcoRI digestion and cloned in phage FIX (Stratagene, San Diego, CA). The library was screened with a cDNA clone made by primer extension of the 5Ј-untranslated region of human CD53 message kindly provided by V. Horejsi (Czech Academy of Sciences, Prague) (37). The positive genomic clones were mapped by partial digestion of the inserts with several restriction enzymes, followed by hybridization to end-labeled T3 or T7 primers, present in the cloning vector, as probes. Conditions for primer hybridization and filter washes have been previously reported (38). Location of the non-translated first exon was done by Southern blot hybridization under conditions previously reported (39).
Deletions by Exonuclease III and Mung Bean Nuclease and DNA Sequencing-The genomic clone was digested with selected restriction enzymes and subcloned in plasmid pBluescript SKII(Ϫ) The subclones were linearized and nested deletions were made using exonuclease III and mung bean nuclease with a commercial kit (Stratagene). The nucleotide sequence was determined by the dideoxynucleotide termination method according to the T7 DNA polymerase Sequenase kit (Amersham Pharmacia Biotech). The products were analyzed in 7% polyacrylamide gels with urea. The dried gel was exposed to Fuji-RX film (Fuji, Japan). Alternatively they were sequenced with a Dye Terminator Cycle Sequencing kit (PerkinElmer Life Sciences) using Thermus aquaticus FS polymerase and universal M13-forward and reverse primers. Sequences were resolved in an ABI PRISM 377 automatic DNA sequencer and the results were processed with the ABI Analysis software (version 2.1). Nucleotide sequence analysis was performed using the PCGene and OMIGA packages from Oxford Molecular (Oxford, United Kingdom). The nucleotide sequence with the two promoters, CD53 and the new transcriptional unit has GenBank TM accession number AJ243474. To detect specific DNA motifs specific for regulatory elements, such as enhancers and transcription factors, the sequence was analyzed with the Transfac 4.0 program (Gesellschaft fü r Biotechnolosgische Forschung, Braunschweig, Germany) (40).
RNA Preparation and Primer Extension Analysis-Total RNA was extracted following the guanidinium thiocyanate-chloroform method (41), as previously described (42). To extend the novel gene an antisense oligonucleotide (GENA2) was used, 5Ј-GAGAGCTCGTGAGACAGAAC-TAG-3Ј (positions Ϫ2175 to Ϫ2152 in the sequence). The extension was performed with Moloney murine leukemia virus-reverse transcriptase according to the manufactures instructions (Life Technologies, Gaithersburg, MD). The extended products were analyzed in a 6% polyacrylamide, 7 M urea sequencing gel. The radioactivity in the gel was detected by autoradiography with Kodak XAR-5 film, or directly with a FUJIBAS1000 PhosphorImager (Fuji, Fuji, Japan).
Luciferase Reporter Gene Constructions-To characterize the regulatory region, DNA fragments were prepared by polymerase chain reaction amplification. Selected regions of the human CD53 promoter sequence were amplified by polymerase chain reaction using specific oligonucleotides to defined regions upstream and downstream of the transcription start site, which are indicated in the experiments, designed based on the CD53 genomic sequence, and ϳ30 nucleotides long. Pairs of primers were used for the polymerase chain reaction for amplification of selected genomic regions. These primers contained, at their ends, the suitable restriction sites, MluI and SmaI, for subcloning into the appropriate luciferase reporter vector of the pGL2 family (Promega, Madison, WI). All the constructs were cloned in the pGL2-Basic, that lacks both SV40 enhancer and promoter sequences, and in some cases in the pGL2-Promoter vector that lacks the SV40 enhancer sequence, but retains the SV40 promoter. As positive control, we used the pGL2-control plasmid containing both SV40 promoter and enhancer sequences. The empty pGL2-Basic that lacks promoter or enhancer sequences was also used as negative control. To introduce point mutations at specific nucleotide locations we used the Quick Mutagenesis kit from Stratagene. To generate the mutant, two complementary primers containing the desired mutation in its center were designed and used to copy a target sequence with the Pfu polymerase, in such a way that the whole plasmid was copied. The input DNA was digested with DpnI that only cuts the input plasmid of bacterial origin because of its methylation, but does not cut the DNA made by the Pfu polymerase. The remaining DNA was used to transfect DH5␣F Escherichia coli strain and isolate new plasmid constructs with the desired mutation. All mutations were confirmed by nucleotide sequence. For internal control of transfection and normalization we used plasmid pCMV-gal (Invitrogen, San Diego, CA) with the ␤-galactosidase gene. Also we used the dual luciferase Renilla system for normalization (Promega, Madison, WI). The generated light was detected with an OPTOCOM-1 luminometer (MGM Instruments, Inc., Hamden, CT).
Cell Lines, Flow Cytometry, and Transfections-The Molt-4 cell line from a T-cell lymphoma; the K562 cell line from a chronic myelogenous leukemia; and the Namalwa B-cell line, derived from a Burkitt lymphoma, was used for transfection experiments. Cells were grown in RPMI 1640 supplemented with 10% fetal calf serum and antibiotics. The cell lines were transfected by electroporation with 10 g of the corresponding plasmid DNA using a Gene-Pulser apparatus (Bio-Rad). The electroporation conditions were 250 volts and 960 F for Molt-4 cells, 300 volts and 500 F for Namalwa cells, and 280 volts and 960 F for K562 cells.
The phenotype of the cell lines was determined by flow cytometry using a FACScalibur form Becton-Dickinson. The CD53 antigen was detected with monoclonal antibody MEM53, and as secondary antibody we used a fluorescein isothiocyanate-labeled rabbit anti-mouse IgG antibody from Sigma.
Protein Extracts and Western Blots-Cells were grown to a density of 5 ϫ 10 6 cells as indicated. For total protein extracts we used 6 ϫ 10 7 cells that were washed in phosphate-buffered saline and lysed in 600 l of RIPA buffer (1% Triton X-100, 150 mM NaCl, 1% aprotinin, 1% leupeptin, 250 M phenylmethylsulfonyl fluoride, 10 mM Tris-HCl, pH 8.0). The cell suspension was passed through a needle to break the DNA. The extract was centrifuged at 10,000 ϫ g for 10 min at 4°C. The supernatant was used as the total protein extract. The protein concentration was determined using a Bio-Rad protein assay kit.
The proteins were fractionated in a 7.5% polyacrylamide gel loaded with 50 g of total protein. The proteins were transferred to a polyvinylidene difluoride membrane (Millipore, Bedford, CT) in a Bio-Rad Trans-blot cell. The membrane was blocked with 5% skimmed milk in TBS (20 mM Tris-HCl, pH 7.5, 150 mM NaCl), 0.1% Tween buffer.
In Western blots, the first antibody indicated in the figure was used at a 1/1000 dilution, and for detection we used protein A-horseradish peroxidase (Amersham Pharmacia Pharmacia Biotech) at 1/1000 dilution or anti-mouse or anti-goat IgG antibodies coupled to peroxidase. The blots were developed using an ECL chemiluminiscence kit from Amersham.
Antibodies-For the transcription factor Sp1 we used a mouse monoclonal antibody, 1C6, from Santa Cruz (Santa Cruz, CA). For Sp2 and Sp3 we used rabbit polyclonal antibodies, K-20 and D-20, respectively, from Santa Cruz. The ets-1 transcription factor was detected with a rabbit polyclonal antibody, N-276, from Santa Cruz. The PU.1 protein was detected with a rabbit polyclonal antibody, T-21, from Santa Cruz. E4BP4 protein was detected with a goat polyclonal, V-19, and GATA1 transcription factor with goat polyclonal, C-20, both from Santa Cruz. Nuclear factor of activated T cells (NFAT) proteins were detected with a rabbit polyclonal antibody, 06-348, from Upstate Biotechnology (Lake Placid, NY).

Cloning of the Human 5Ј CD53 Gene Regulatory Region-To
characterize the human CD53 gene promoter, located on chromosome region 1p13 (43), we cloned the upstream regulatory region by screening a normal individual genomic library with a partial cDNA probe. The probe, clone pPET/XAG, was derived by primer extension of the poly(A) ϩ RNA, and contained only the 5Ј non-translated sequence (37). With this probe a genomic clone, JHT1, containing 20 kilobase pairs was isolated. The restriction map of the genomic clone is shown in Fig. 1. The location of the region containing the beginning of the CD53 transcriptional unit was determined by hybridization to the same probe used for library screening (Fig. 1). From the phage clone we made several subclones in plasmid pBluescript SK-II containing the XbaI and the EcoRI-HindIII region that comprises the start of the CD53 transcribed sequences. We sequenced a region of 3613 nucleotides (GenBank TM accession number AJ243474) surrounding the CD53 start site (Fig. 2).
The CD53 gene lacks a TATA box, but has a sequence that plays its role, as well as a capping site (37). The sequence was analyzed with the Transfac 4.0 program to detect sequences that are recognized by transcription factors (40). In the sequence, shown in Fig. 2, we have indicated the position of several cis-acting consensus DNA motifs recognized by transcription factors, such as Sp1, ets-1/PU.1, elk-1, an Ets-related factor (44,45), and GATA1, which are likely to be implicated in the regulation of CD53 gene expression. Some of these regulatory elements are located within the non-coding exon 1 (Fig. 2).
There Is a Second Promoter-Enhancer Proximal to Human CD53-The analysis of the nucleotide sequence upstream of the CD53 promoter also detected a putative TATA box located in the Ϫ2000 region approximately; with opposite orientation with respect to the CD53 gene start site (Fig. 1). The presence of such an unknown gene was confirmed by the sequencing of the human genome, which is located upstream and proximal of the human CD53 promoter in chromosome 1p13.3 (46). Before proceeding to characterize the CD53 promoter we confirmed that there is indeed a start site for an unknown transcriptional unit, called gene A, by performing a primer extension assay using as target RNA from two different cell lines, Molt-4 and K562 (data not shown).
To functionally demonstrate that there is a second enhancerpromoter from another gene proximal to CD53 we subcloned it upstream of a luciferase reporter gene in the pGL2-B (basic) vector. The constructs were of increasing length till the CD53 promoter was included in the antisense orientation. In the region from Ϫ1837 to Ϫ2127 there is a very strong enhancerpromoter that is even stronger, in the three cell lines tested, than the positive control containing the SV40 enhancer-promoter used as positive activation control (Fig. 3). As the length of DNA toward the CD53 promoter was increased, there was a significant drop in the activity of this novel enhancer-promoter, the major drop in luciferase activity occurred when the region located between Ϫ1533 to Ϫ489 was included in the construct. This region has 24 copies of a CA dinucleotide repeat which is located between positions Ϫ1361 to Ϫ1313. The CA repeat is thus located between the two promoters. The inclusion of the CD53 enhancer, present in constructs extending to the ϩ137 and ϩ322 positions, did not prevent the drop in the activity of gene A promoter-enhancer (Fig. 3), suggesting that the presence of the CD53 promoter-enhancer does not affect this novel transcriptional unit. The relative strength of the gene A promoter-enhancer (Ϫ1837 to Ϫ2127) is stronger than the SV40 control (pGL2-C vector) in the three cell lines.
The Proximal CD53 Promoter Region Is the Major Determinant of Its Activity-To characterize the proximal regulatory region of the human CD53 gene we used three different cell lines from the lymphoid-myeloid lineage, such as K562 derived from an erythroleukemia, Molt-4 derived from a T-cell lym-FIG. 2. Nucleotide sequence of the proximal region surrounding the transcription start site of CD53 and containing its proximal enhancer-promoter; the location of consensus sequences corresponding to known transcription factors is indicated. This sequence is within the clone comprising the human CD53 gene promoter from position Ϫ2562 to ϩ1051 (EBI/GenBank TM accession number AJ243474), which includes the promoter-enhancer of an unknown gene (gene A in this report). The two gene promoter-enhancers are separated by 24 copies of a CA dinucleotide repeat. phoma, and Namalwa cells derived from a Burkitt lymphoma, a B-cell tumor. The three cell lines expressed CD53 antigen on their surface as determined by flow cytometry analysis (Fig. 4). The expression levels in the three cell lines were also determined at the RNA level by a Northern blot (not shown). Therefore, these different cells lines can be used to ascertain if the same regulatory elements to determine the transcriptional activity of the CD53 promoter in the three cells lines, or alternatively if the proximal enhanceosome or transcriptional complex (2) is different in each cell line, although located in the same genomic region.
First, to identify the location of the major regulatory elements within the human CD53 gene regulatory region, several constructs were made from Ϫ2157 to the ϩ1 position using vectors pGL2-B (without enhancer-promoter) and pGL2-E (lacks the promoter, but has the SV40 enhancer). With both vectors the results obtained were similar, therefore we continued the analysis using only the constructs made in the pGL2-B vector.
The activation of transcription was studied in three cell lines: K562, Molt-4, and Namalwa. The proximal region between nucleotides Ϫ266 and ϩ1 (Fig. 5) appeared to contain most of the sequences required for expression of the CD53 gene, and in the three cell lines, the values were similar to those of the positive control with the SV40 promoter-enhancer. The effect from the Ϫ266 to Ϫ998 sequence was heterogeneous and no conclusion could be drawn from it. Although in Molt-4 cells, the region between Ϫ509 and Ϫ667 resulted in a 2-fold increase in the activity of the proximal CD53 region (Fig. 5). The inclusion of the gene A enhancer-promoter and the CA repeat, up to position Ϫ2157, did not significantly affect the CD53 promoter in K562 cells, but resulted in a drop in activity in Molt-4 and Namalwa cells, although still with a relative high-level transcriptional activity with respect to the empty vector (Fig. 5). We conclude that the major determinant of the transcriptional activity of the human CD53 gene regulatory region is very proximal to the ϩ1 position.
There Are Four Subregions in the Proximal Regulatory Region of CD53-To further characterize the proximal regulatory region of the human CD53 gene we included the immediately upstream sequence, as well as the non-coding exon 1 and intron 1, extending from nucleotides Ϫ266 to ϩ84. This region was subdivided into four subregions, A (Ϫ266 to Ϫ168), B (Ϫ168 to ϩ1), D (ϩ1 to ϩ64), and C (ϩ64 to ϩ84), to detect the location of putative cis-acting regulatory elements (Fig. 6). All clones that included intronic sequences beyond the splice donor at ϩ84, such as the ϩ137, ϩ322, and ϩ1040 constructs, were not active because the luciferase reporter gene was deleted as a result of splicing into the SV40 splice acceptor signal present in the reporter vector. Regions B ϩ D (Ϫ163 to ϩ64) appeared to account for the basic expression of the CD53 promoter, as originally reported (37), and deletion of region D in this construct resulted in a very important drop in activity (Ϫ168 to ϩ1 in Fig. 4). The region comprised from position Ϫ266 to ϩ84 has the highest activity, and was very similar to the activity of the construct from Ϫ168 to ϩ84 positions. The inclusion of region C (ϩ64 to ϩ84) in constructs starting either at Ϫ266 or Ϫ168 also resulted in an increase in transcriptional activity, suggesting the existence of another important sequence at the end of exon 1. The fragment from ϩ1 to ϩ84 (D ϩ C) was not active by itself, probably because of the lack of TATA-like box (Fig. 6). If the CD53 TATA-like box was included (Ϫ25 to ϩ84) there was also no recovery of activity (Fig. 6). But in the construct that extended from Ϫ266 to ϩ64, there was a partial recovery of the activity (Fig. 6). These data indicates that within exon 1 there are at least two important regulatory sequences, one between ϩ1 and ϩ64 (D), and the other between ϩ64 and ϩ84 (C). Because this non-coding exon 1 has no TATA box, we tested whether this region by itself could function as an enhancer by activating a heterologous promoter, such as that of SV40. We cloned this CD53 exon 1 region upstream of the SV40 promoter, in the pGL2-P vector (that has the SV40 promoter, but not its enhancer). Neither orientation of this CD53 exon 1 was able to activate transcription directed by the SV40 promoter in Molt-4 or K562 cells (not shown). Thus, the regulatory element appears to be specific and is not active by itself. To be functional this region, which has two regulatory subregions (D ϩ C), needs to act coordinated with other upstream elements located in the proximal regulatory region of human CD53 gene, and that are not present in a heterologous promoter, such as SV40.
Also in the proximal upstream region the deletion of sequence Ϫ266 to Ϫ168 (region A) resulted in a partial loss of activity, more noticeable if the construct reached only the ϩ64 position (Fig. 6). Thus in the proximal upstream region A there is another regulatory sequence in addition to the previously known to be within B (Ϫ168 to ϩ1).
We can conclude that the proximal sequences and the first exon of CD53 can be subdivided into four subregions, each with at least one regulatory element, that contribute to CD53 gene expression. These data obtained by deletion analysis did not detect any significant difference among the three types of cell lines used (Fig. 6).

Identification by Mutagenesis of Cis-acting Sequences That Control CD53 Gene Expression in Different Cell
Types-Gene expression is controlled by the assembly of the enhanceosome o transcription complex where several cis-acting DNA sequences are bound to different regulatory and structural proteins (2, 3). We reasoned that a mutational analysis of the potential cisacting sequences in the region might detect the differences, if any, in the assembly of the transcription complex in each cell line. Therefore, in order to determine if CD53 gene expression was regulated differentially depending on the cell type we performed a mutational analysis of the proximal region. The DNA sequence was analyzed with the Transfac 4.0 program to detect DNA elements that are candidates to be regulatory sequences (40). Among the elements detected, we selected eight to be studied by site-directed mutagenesis, which are distributed within the four subregions previously identified by deletion analysis, and that did not detect differences among the three cell lines (Fig. 6). The selected DNA target sites for mutation were modified by the substitution of two nucleotides that form the core consensus of the site and are required for binding of transcription factor (Table I). Two of them, A5 and C1, have a core sequence related to the ets-1/PU.1 family of transcription factors, and another (B5) a site for an ets-related factor, elk-1. Mutations in DNA sequences that are positive regulators will be detected by a reduction in luciferase activity, and those that play a negative role will be detected by an increase in luciferase activity. These sequences were mutated to identify the putative regulatory elements located in different subregions of the CD53 promoter, as well as determine their relative contribution to CD53 gene expression in specific cell types.
Within region A, three sequences were mutated, but only two appeared to modulate CD53 promoter activity, but with different roles depending on cell type. The mutations were introduced in the construct from Ϫ266 to ϩ64 that lacks element C. The mutation of site A3 (PuF core) resulted in a very high increase in luciferase activity in Molt-4 cells, and more moderately in K562, indicating that this sequence plays a negative regulatory role in these two cell lines (Fig. 7). This A3 mutation had no effect in Namalwa cells (Fig. 7). The mutation of site A5 (ets-1 core) had no effect in K562 and Molt-4, but resulted in a very high increase of activity in Namalwa cells, where it plays a negative regulatory role (Fig. 7). Mutation of the A1 site had no effect in any cell line (not shown).
Region B was previously identified as essential for CD53 expression and it must have a basic component required for gene expression. In this region we mutated three different putative target sequences. There is an SP1 consensus sequence (B1) that was mutated in a construct from Ϫ266 to ϩ 84 region. This mutation resulted in a complete loss of activity despite the presence of all the other DNA-binding sites as wild type in the three cell lines (Fig. 7). The deletion of the region with this Sp1 site, construct from Ϫ25 to ϩ84, retaining element C1 as wild type, also resulted in almost complete loss of activity (Fig. 5). These observations suggest that this SP1 site is a basic component of the transcriptional unit in the three cell lines. Mutation of element B3 reduced by half the expression in K562, but mainly resulted in an important increase in activity in the two lymphoid cell lines, Molt-4 and Namalwa (Fig. 7). Mutation of element B5 increased the activity in Namalwa cells, moderately reduced the activity in Molt-4, and had no effect in K562 cells (Fig. 7).
The region comprised between nucleotides ϩ65 and ϩ84 (region C), at the end of the non-coding exon 1 confers a very strong transcriptional activation. This region contains 18 nucleotides and a core recognition sequence (C1) of ets-1/PU.1 family transcription factors. To determine if this element was implicated in transcriptional activation, we replaced the two GG of the target sequence by two AA (nt 2 ϩ71 and ϩ72) (Table  I) in a construct containing the Ϫ266 to ϩ84 region, this mutation resulted in a drop in activity to levels similar to a deletion construct without that region (Ϫ266 to ϩ64), suggesting that it participates in the activation of transcription, but as shown by the construct from Ϫ168 to ϩ64 (Fig. 6), it needs to cooperate with other sequences to be functional. Element C1 appears to play different roles depending on cell type. C1 is required for expression in K562 and Molt-4 cells, and is a strong negative regulator in Namalwa cells since its mutation resulted in a very high increase in transcriptional activity (Fig. 7).
Within region D, we mutated element D3 (a core consensus for GATA-1 or NFAT). This mutation only resulted in an important activation of transcription in Namalwa cells, with no effect on K562 or Molt-4 cells (Fig. 7). At the end of region D, there is an ets-1 core element at position ϩ62 (element D1) that overlaps the beginning of region C, which is absolutely essential for the transcriptional competence of the CD53 promoter, its mutation resulted in a complete loss of activity in the three cell lines (Fig. 7).
The overall picture resulting from this mutational analysis is that different regulatory sequences account for expression of the CD53 gene in the three cell types studied (summarized in 2 The abbreviation used is: nt, nucleotide. FIG. 6. Characterization of the transcriptional activity in the CD53 proximal region, from ؊266 to ؉84, detects four different cooperating elements. Identification of an enhancer at the end of exon 1, and its cooperation with a proximal regulatory region. At the bottom is a diagram describing the regions into which the proximal CD53 regulatory region can be subdivided. Fig. 8). Two of the elements, B1 (Sp1 core) and D1 (ets-1 core), are essential for transcriptional competence in all cell lines. But in this context it is very important to draw attention to the fact that a single cis-acting DNA sequence can have opposite roles depending on the cell type, as is the case for elements C1, B3, B5, D3, A3, and A5.
Levels of Transcription Factors-The functional differences observed in the three cells lines might be a consequence of differences in the level of transcription factors present in each cell line, or be the result of a differential assembly of the proteins in the proximal enhanceosome. To address this issue we determined the level of several transcription factors in the three cell lines. For this purpose whole cell extracts were prepared and fractionated in denaturing SDS-polyacrylamide gel electrophoresis. The proteins were transferred to a polyvinylidene difluoride membrane, and the filters were analyzed by Western blot with different antibodies specific for transcription factors as indicated in Fig. 9. The level of transcription factors that have a marked differential effect, such as Sp1, ets-1, or PU.1, appear to be similar in the three cell lines. The erythroid transcription factor GATA-1 (47) is restricted to the K562 cell line. The E4BP4 element is expressed at similar levels in the three cell lines (Fig. 9). This element is a known negative regulator of transcription (48), and it exerts this effect in the two lymphoid cell lines, Molt-A and Namalwa (Fig. 7). The blot with antibody against NFAT is not shown because of the very heterogeneous size of this family that contains multiple members. The pattern observed was a smear, that was similar and within the expected size range for this protein family in the three cell lines. The cell lines were also analyzed for other factors of the Sp family, as well as with an antibody against ␤-actin as a control for gel loading.
The similarity in the levels of transcription factors suggests that the functional differences observed in the three cell lines are likely to be a result of a differential protein assembly in the enhanceosome, which is specific and different for each of them. In that way different cells achieve the expression of a common gene.

TABLE I
Mutations introduced in different potential regulatory sequences of the human CD53 proximal promoter region Two nucleotide changes (indicated in bold and underlined) were introduced in the known essential nucleotides of the binding site to make the mutant sequence. Their nucleotide position within the promoter sequence is indicated in parentheses. wt, wild type; mt, mutant.
FIG. 7. Mutagenesis of selected cis-acting DNA regulatory sequences in the human CD53 gene promoter. Effect of the double mutations (Table I)  Diagram illustrating the different functional organization of the regulatory elements in the active CD53 promoter in K562, Molt-4, and Namalwa cell lines. Positive elements are those that result in a loss of activity when they are mutated (squares), negative elements are those that activate transcription when they are mutated (circles), and elements with no effect (triangles). The shaded elements are required for transcriptional competence in the three cell lines.

DISCUSSION
Understanding the regulation of tetraspan antigen genes expression is necessary to know how they are coordinated among themselves and with other proteins with which they interact, such as integrins, to form complexes on the cell membrane. For this purpose we cloned a genomic fragment containing the human CD53 gene promoter (Fig. 1), and identified several regulatory elements within a region surrounding the start site and including the non-coding exon 1 (Fig. 2). Within the genomic clone and ϳ2 kilobases upstream of the CD53 start, by primer extension and luciferase reporter assays it was identified the enhancer-promoter of an unknown gene (Fig. 3), confirmed by the sequence of the human genome (46). This second enhancer-promoter does not interfere with the activity of the CD53 promoter. However, the activity of the novel gene regulatory region drops significantly when the region containing a CA dinucleotide repeat is included in the construct. The 24 CA repeat region can form Z-DNA and perhaps might functionally separate the two transcriptional units (49). The activity of the CD53 promoter remains at a similar level independent of whether the CA repeat and the gene A promoter are present in the construct, except when the ϩ64 to ϩ84 element is included, which is much higher. Close proximity of two gene promoters has been found for other genes. Thus, the murine trkA gene is 2 kilobases from the IRR gene promoter, and it is not affected by its presence (50). Also the interaction between a TATA-less promoter and a gene with conventional TATA, close to each other and with opposite orientation has been studied experimentally (1). In these constructs the TATA promoter containing gene is preferentially active. In the case of the CD53 gene (TATA-less promoter), this promoter is less active than the gene A promoter (with TATA box). TATA-less promoters, as it is the case of the CD53 gene, require the participation of Sp1 elements and members of the Ets family of transcription factors (51).
The transcriptional analysis of the Ϫ266 to ϩ84 region of the CD53 gene which included the non-coding exon 1 in the three cell lines of different lineage allowed the identification of several cooperating regulatory sequences. The same elements contribute to the expression of the CD53 gene, but their effect and combination of elements is specific of the cell type (Fig. 8). An Sp1 element (B1), previously postulated to be important for expression was shown to be essential for CD53 transcription in all cell lines. Both, its deletion (Fig. 6) or mutation (Fig. 7) resulted in a total loss of activity in the three cell lines. This element must be a basic component of the transcription machinery of the human CD53 gene. Sp1 elements are also important for expression of other tetraspan antigens, such as CD63 (52) and CD9 (33).
Ets/Pu.1 sequences play a major role in CD53 gene expression. The GGAA is the core recognition sequence for the Ets family of transcription factors, consisting of more than 45 different members, which are very relevant for lymphoid-myeloid gene expression (47,53). One of them, D1, is absolutely essential for gene expression in the three cell lines (Fig. 7). Two Ets elements (A5 and C1) were identified as well as another element (B5) with a consensus for elk-1, a factor that has an Ets domain (45). Some of these Ets target sequences have a dual role, activator or repressor depending on the cell type, suggesting that they modulate the activity of the basic transcription machinery. Thus mutation of the three sites (A5, B5, and C1) resulted in a very high activity of the promoter in Namalwa cells (Fig. 7), where it must have a repressor role, preventing activation. The small region C, ϩ64 to ϩ84, has a canonical Ets core sequence (C1) that requires cooperating with another sequence within the CD53 promoter-enhancer region. This important element functions in cooperation with the Sp1 element (B1) and by itself has no role, and it does not activate transcription from a heterologous SV40 promoter. This C1 element is a down-regulator in Namalwa cells and an activator in K562 and Molt-4 cells. A dual role for Ets/PU.1 proteins has been reported in the regulation of the locus, where it functions as a positive regulator in pre-B cells, and as a negative regulator in mature B cell stages (54).
Previous work postulated that within region B of this report there were three putative binding sites, and Sp1, PuF, and PU.1 sites, that might account for CD53 gene activation (37). In this report we show that they indeed contribute to the basal level of the promoter, but that they need to cooperate with another more distal element in exon 1 (ϩ64 to ϩ84) to achieve a much higher level of expression.
It is interesting to note that integrin genes, proteins that complex with tetraspan antigens on the membrane (32), are also regulated by combination of Sp1 and ets-1 DNA elements (55,56), where PU.1 may play a recruiting role (57). For example, the integrin CD11b requires Sp1 (58) and PU.1 (59) to drive its expression. Cooperation between Sp1 and Ets regulatory sequences have been reported in the regulation of some integrin genes, among these genes are CD11b (58,59), the ␤2 (CD18) (60) and the ␤5 chain (61). Mutation of the Sp1 element dramatically reduces CD18 expression (60), an effect similar to mutation of the Sp1 element (B1) in the human CD53 promoter (Fig. 7). Also some ets-1 DNA recognition sequences have a positive or negative role in these cell lines, such as is the case for PU.1 which functions as a negative element in pre-B cells, and as positive in mature B-cells (54). Also Sp1 and Ets cooperation has been reported in other proteins such as the mannose receptor (62) and the btk kinase (63). The differential FIG. 9. Level of several transcription factors determined by Western blot analysis. Total cellular extracts from the different cell lines, K562, Molt-4, and Namalwa, were used for the analysis. The proteins were fractionated in a 10% polyacrylamide-SDS gel and after transfer to a polyvinylidene difluoride Immobilon-P membrane, the Western blots were analyzed with an antibody against a specific transcription factor. The detection was performed using a chemiluminiscence commercial kit. An antibody against ␤-actin was included as a control for protein loading in the gels. assembly of the regulatory region might be regulated by methylation (64) and the binding of other proteins, such as CTCF (65), which also have a role as insulators of gene transcription (49).
Several regulatory elements contribute to a specific gene expression (66,67). Depending on the type of cell, a given element might have a different role or opposite effects as demonstrated in this report. For the same genomic region to achieve a similar level of expression, a different composition of regulatory elements is needed in each cell line. The different requirements for CD53 gene expression in the three cell lines are illustrated in the diagram of Fig. 8. In general, transcription regulatory protein has been shown to function as positive or negative regulators (68). In this report by identifying simultaneously several regulatory sequences within a promoter and analyzing their role in combination, we have shown that there are three different functional complexes (Fig. 8). Some elements play the same role independently of the cell type. That is the case of element B1, a consensus Sp1 site, which is a basic component of the transcription machinery. Other elements function as positive or negative regulators depending on the cell. Element B5 has no effect in K562 cells, but activates transcription in Molt-4 cells, and is a negative regulator in Namalwa cells. The B3 element is an activator in K562, and a negative regulator in both lymphoid cell lines, this element has a consensus sequence for the E4BP4, which is a known negative regulator in other systems (48). The C1 element, however, is a repressor of gene expression in Namalwa cells. Element D3 is required for high-level transcription in Namalwa cells. This element interacts with members of the ets transcription factors (69). The positive or negative effect of a DNA sequence is a likely consequence of its effect on the protein assembly of the enhanceosome in each specific cell type.
In this report we have shown that the regulation of the human CD53 gene includes exon 1 sequences and is composed of several elements. Essential Sp1 and ets-1 sites are required for transcriptional competence; and several other elements, mainly of the Ets family, have a different role depending on the structure of the enhanceosome in each cell type. We postulate that the combination of Sp1 and Ets transcription factors coordinate the expression of tetraspan genes and the proteins with which they interact on the cell membrane, such as integrins. The CD53 enhanceosome has different protein components, reflected by different DNA sequence roles, which are required to adjust the expression of a single gene to different cell types.