Identification and Characterization of a Transcriptional Regulator for the lck Proximal Promoter*

The lck gene encodes a protein-tyrosine kinase that plays a key role in signaling mediated through T cell receptor (TCR) and pre-TCR complexes. Transcription of the lck gene is regulated by two independent promoter elements: the proximal and distal promoters. Previous studies employing transgenic mice demonstrated that the sequence between −584 and −240 from the transcription start site in the mouse lck proximal promoter is required for its tissue-specific expression in the thymus. In this study, we demonstrate that a Krüppel-like zinc finger protein, mtβ (BFCOL1, BERF-1, ZBP-89, ZNF148), previously cloned as a protein that binds to the CD3δ gene enhancer, binds to the −365 to −328 region of thelck proximal promoter. mtβ is ubiquitously expressed in various cell lines and mouse tissues. Overexpressed mtβ is more active in T-lineage cells than B-lineage cells for transactivating an artificial promoter consisting of the mtβ binding site and a TATA box. Activity of the lck proximal promoter was significantly impaired by mutating the mtβ binding site or by reducing mtβ protein expression level by using antisense mRNA. Our results indicate that mtβ activity is regulated in a tissue-specific manner and that mtβ is a critical transactivator for the lck proximal promoter.

The lck gene encodes a lymphocyte-specific protein-tyrosine kinase, p56 lck , a member of the src kinase family (1). It has been demonstrated by co-immunoprecipitation that p56 lck associates with the cytoplasmic domains of CD4 and CD8 coreceptors (2) and with the acidic region of the IL-2 1 receptor ␤-chain in T cells (3). By a series of biochemical analysis, it has been shown that p56 lck plays a key role in signal transduction mediated through the T cell receptor (TCR) complex in mature T cells (4,5). It also contributes to signaling through the pre-TCR complex, thereby playing an essential role in thymocyte development. lck-deficient mice and transgenic mice overexpressing a dominant negative form of p56 lckck exhibit severe impairment in the expansion of CD4/CD8-double negative immature thymocytes (6,7). A simple doubling of wild type p56 lck expression levels in immature thymocytes in transgenic mice was sufficient to block maturation of thymocytes (8). These findings suggest that the transcriptional control of the lck gene must be tightly regulated to express adequate amounts of p56 lck at the right developmental stage during thymopoiesis.
The lck gene is transcribed from two structurally unrelated promoters (9 -13). The lck proximal promoter is positioned immediately adjacent to the first coding exon, and is active in the thymus, but is essentially silent in peripheral T cells. The distal promoter is located far 5Ј-upstream from the proximal promoter and is active during all developmental stages of Tlineage cells. Since the proximal promoter becomes active only at an early developmental stage of T-lymphopoiesis (14,15), and since the level of p56 lck greatly influences thymocyte maturation (6 -8), the transcriptional regulators of this promoter play a critical role in the developmental program for T-lineage cells.
The 5Ј-flanking sequence of the lck proximal promoter that is critical for the thymocyte-specific and developmental stagespecific expression has been defined by transgenic mouse models (16). Transgenic animals bearing truncations in the mouse lck proximal promoter revealed that as little as 584 bases of the 5Ј-flanking sequence can confer appropriate developmentally regulated expression of heterologous reporter genes. The 5Ј sequence critical for the promoter activity contains several binding sites for nuclear proteins. Among those nuclear proteins, "B-factor," which binds to the G-rich stretch within the Ϫ365 to Ϫ328 region was reported as a candidate for the critical transcriptional regulator. B-factor is only found in cells expressing the lck transcript derived from the proximal promoter, namely thymocytes and thymoma cell lines such as LSTRA and EL4 (16).
In this study, we characterized the B-factor and identified an 86-kDa Krü ppel-type zinc finger protein, which had been cloned previously as a binding protein to the CD3␦ gene enhancer, as a component of the B-factor. The NH 2 -terminal half of the protein is 90% identical to ht␤, a 49-kDa protein that binds to the human TCR V␤8.1 gene promoter and the TCR ␣ gene silencer (17), indicating that mt␤ is the murine homologue of ht␤, and the reported amino acid sequence of ht␤ is a part of its full-length protein. mt␤ is ubiquitously expressed in various cell lines and tissues. We re-evaluated distribution of the Bfactor and found it is also expressed in various cell lines and tissues. However, the transcriptional activity of mt␤ measured by reporter constructs carrying the B-factor binding site is observed only in T-lineage cells. The transcription from the lck proximal promoter is greatly impaired by introducing mutations in the B-factor binding site or by expression of mt␤ antisense mRNAs. Our results demonstrate that mt␤ is one of the critical transactivators driving the lck proximal promoter and that its activity is regulated in a tissue-specific manner.
Plasmid Construction-The full-length mt␤ cDNA 2 was subcloned into the EcoRI site of pcDNA3 (Stratagene), a eukaryotic expression vector driven by the human cytomegalovirus enhancer and promoter, resulting in pcDNA3-mt␤. For the mt␤ antisense plasmid (pcDNA3-ASmt␤), the full-length mt␤ cDNA was subcloned in the opposite direction into the EcoRI site of pcDNA3. Various truncated fragments from the mouse lck proximal promoter were subcloned into the pGL2-Basic plasmid (Promega), which has a firefly luciferase gene without promoter or enhancer. For Ϫ3200/pGL2, the NotI-BamHI (positions Ϫ3200 to ϩ37) fragment of the p1017 plasmid (20) containing the entire lck proximal promoter region was blunt-ended and ligated to the XhoI, HindIII-digested, blunt-ended pGL2-Basic plasmid. For Ϫ433GL2, two SmaI fragments (position Ϫ3200 to Ϫ1675 and Ϫ1675 to Ϫ433) were removed from Ϫ3200/pGL2 and self-ligated. For Ϫ240/ pGL2, two KpnI fragments (position Ϫ3200 to Ϫ584 and Ϫ584 to Ϫ240) were removed from Ϫ3200/pGL2. For Ϫ584/pGL2, the KpnI fragment (position Ϫ584 to Ϫ240) from Ϫ3200/pGL2 was inserted into the KpnI site of Ϫ240/pGL2. To construct reporter plasmids carrying the B-factor binding site and TATA box, Ϫ365/Ϫ328 and Ϫ365/Ϫ328mut oligonucleotides (see section above) were inserted upstream of the TATA box of pLuc-S (gift from Drs. P. Doerfler and M. Busslinger (21)), resulting in pLuc-wild and pLuc-mut, respectively. Point mutations were introduced into the mt␤ binding sites of Ϫ3200/pGL2 and Ϫ433/pGL2 by PCR-based directed mutagenesis using Ϫ365/Ϫ328mut oligonucleotides to generate Ϫ3200-mut/pGL2 and Ϫ433-mut/pGL2, respectively.
RNA Isolation and Northern Blot Analysis-Total RNA was extracted from various cell lines using the acid guanidine isothiocyanatephenol-chloroform method. Fifteen micrograms of total RNA were fractionated through electrophoresis on 1% agarose gel in the presence of 0.66 M formaldehyde, transferred to nylon membranes (GeneScreen, DuPont). Mouse multiple tissue Northern blot was purchased from OriGene Technology (Rockville, MA). Membranes carrying RNA were hybridized with a 2.0-kilobase EcoRI fragment of mt␤ cDNA labeled with [␣-32 P]dCTP by the random priming method. After hybridization, membranes were analyzed using a BAS1000 Bio-Image Analyzer (Fuji Film, Tokyo, Japan). After removing the mt␤ probe, the membranes were re-hybridized with a human ␤-actin probe to normalize the amount of RNA loaded per lane.
Generation of Polyclonal Antibodies-A 1135-base pair EcoRV-ApaI fragment encoding Ser 51 to Gly 428 and a 1510-base pair ScaI-NotI fragment encoding Thr 446 to Gly 769 of mt␤ protein were subcloned into the bacterial expression vectors pGEX-4T-1 and pGEX-4T-2 (Amersham Pharmacia Biotech), respectively. The resulting plasmids were used to transform BL21(DE3)/pLysS (Novagen, Madison, WI). Recombinant glutathione S-transferase-mt␤ fusion proteins were induced with 1 mM isopropyl-␤-D-thiogalactopyranoside and affinity-purified by binding to glutathione-linked Sepharose beads (Amersham Pharmacia Biotech). The fusion proteins were further purified by gel filtration and used to immunize rabbits. Rabbit polyclonal anti-mt␤ antibodies were immunopurified on Sepharose-4B beads covalently coupled with the respective glutathione S-transferase-mt␤ fusion proteins used as immunogens.
Preparation of Cell Lysates and Western Blotting-Cells were harvested and boiled in SDS-PAGE sample buffer (50 mM Tris, pH 6.8, 100 mM dithiothreitol, 2% SDS, 0.1% bromphenol blue, 10% glycerol) for 5 min. After centrifugation at 15,000 ϫ g for 15 min, resulting clear cell lysates were subjected to SDS-8% PAGE and then electrophoretically transferred to a polyvinylidene difluoride membrane (Immobilon, Millipore, Bedford, MA) in transfer buffer (25 mM Tris, 200 mM glycine, 10% methanol). After blocking with 5% bovine serum albumin in TBS overnight at 4°C, membranes were incubated with appropriately diluted primary antibodies. Membranes were washed with TBS containing 0.1% Tween 20 (TBS-T) and further incubated with horseradish peroxidase-conjugated antibodies against rabbit IgG (Cappel Organon Technica, Durham, NC). After washing, bound antibodies on membranes were detected using an ECL detection system (Amersham Pharmacia Biotech, Buckinghamshire, United Kingdom) and x-ray film (Fuji Film). Splenic B cells were purified by negative sorting using anti-CD43 monoclonal antibody and a magnetic cell sorting system (Miltenyi Biotec, Bergisch Gladbach, Germany). Splenic T cells were purified by negative sorting using anti-B220 and anti-Mac1 monoclonal antibodies and magnetic cell sorting.
Transfections and Luciferase Assays-Cells (3 ϫ 10 6 ) were suspended 2 K. Georgopoulos, unpublished data. in 0.2 ml of Opti-MEM (Life Technologies, Inc.) and transferred to a 4-mm gap cuvette and mixed with 10 g of reporter firefly luciferase plasmids, 1 g of renilla luciferase plasmid pRL-TK (Promega). In some experiments, 10 g of pcDNA3-mt␤ or a control pcDNA3 was added in addition to the reporter plasmids. Cells were transfected at 960 microfarads and 250 V using a Gene Pulser electroporation apparatus (Bio-Rad). Cells were harvested 12 h after transfection, and luciferase activity in cell lysates was measured using the Dual luciferase assay system (Promega) according to the manufacturer's recommendation. The firefly luciferase activity was normalized by the renilla luciferase activity to normalize for the transfection efficiency of each sample. In experiments using pcDNA3-ASmt␤, EL4 cells were transfected with 15 g of pcDNA3-ASmt␤ or control pcDNA3, together with 3 g of reporter firefly luciferase plasmids and 1 g of pRL-TK. Cells were harvested 30 h after transfection, and luciferase activity was measured. To analyze the amounts of mt␤ protein and the endogenous lck transcripts, cells were transfected with 10 g of pcDNA3-ASmt␤ or pcDNA3, together with 1 g of pEGFP-N1 (CLONTECH). Cells expressing GFP were sorted using a FACS Vantage cell sorter (Becton-Dickinson) at 30 or 36 h after transfection. Sorted cells were lysed and subjected to immunoblot analysis using anti-mt␤ or anti-tubulin antibodies or were subjected to RT-PCR analysis.

Identification of a Nuclear Protein
Binding to the lck Proximal Promoter-A nuclear factor termed "B-factor" binds to a G-rich stretch located at Ϫ365 to Ϫ328 from the transcriptional initiation site of the lck proximal promoter, and its expression correlates well with the activity of the promoter (16). Because lck expression driven by the proximal promoter occurs early during lymphopoiesis (14,15), and the level of p56 lck greatly influences thymocyte maturation (6 -8), we hypothesized that the B-factor would be one of the transcription factors playing critical roles in early lymphopoiesis. Ikaros regulates the early lymphopoiesis or the commitment for lymphocytes as demonstrated in the Ikaros-deficient mice (22). Ikaros has been shown to bind to a G-rich sequence of the CD3␦ gene enhancer (23), although the high affinity binding sites for Ikaros are not G-rich (24). We examined whether the B-factor contains Ikaros or Ikaros-related proteins by EMSAs. Nuclear extract of a thymoma cell line, LSTRA, contains the B-factor binding to the Ϫ365 to Ϫ328 G-rich sequence (Ϫ365/Ϫ328) of the lck proximal promoter (Fig. 1A) as shown previously (16). The binding was specific, since it was competed by unlabeled Ϫ365/Ϫ328 oligonucleotides, but not by the Ϫ365/Ϫ328 oligonucleotides carrying mutations in the G-rich sequence (Ϫ365/Ϫ328mut). As shown in Fig. 1B, the binding of B-factor to the Ϫ365/Ϫ328 probe was competed by ␦A sequences, a functional element of the CD3␦ gene enhancer. The ␦A consists of a CRE (cyclic AMP response element)-like region and a G-rich site similar to the B-factor binding site (25). The binding of B-factor to ␦A was mediated by the G-rich site, because mutation at the G-rich site (␦A-G) but not at the CRE-like site (␦A-CRE) abrogated the competition with the Ϫ365/Ϫ328 probe. This binding characteristic of B-factor to ␦A sequence was similar to that of Ikaros (23). However, the high affinity Ikaros binding oligonucleotides (IkarosBS) (24) did not show any competition with Ϫ365/Ϫ328. During the cloning of Ikaros, a cDNA clone encoding a zinc finger protein (mt␤, see below) was simultaneously cloned by its ability to bind to the CD3␦ enhancer. 2 We examined whether antibodies against Ikaros or mt␤ could react with B-factor. Anti-Ikaros antibodies did not affect B-factor complex formation (Fig. 1C), confirming that Ikaros is not a component of B-factor. Interestingly, anti-mt␤ antibodies efficiently supershifted the B-factor complex (Fig. 1D). A similar result was obtained using nuclear extracts of EL4, a lymphoma cell line in which the lck proximal promoter is active (Fig. 1E). These results demonstrate that mt␤ is a component of the B-factor that binds to the Ϫ365 to Ϫ328 region of the mouse lck proximal promoter.
mt␤ Is a Krü ppel-type Zinc Finger Protein-The deduced amino acid sequence of mt␤ contains an amino-terminal acidic domain, four tandem C2H2 Krü ppel-type zinc finger motifs (26), and two basic domains, located upstream and downstream of the zinc finger cluster (data not shown). A data base search by BLAST identified a human homologue, ht␤, a 49-kDa pro-FIG. 1. Characterization of the nuclear protein, "B-factor," that binds to the sequence from ؊365 to ؊328 of the lck proximal promoter. A, the B-factor (arrow) in nuclear extracts from LSTRA. EMSA was performed using a fragment containing the sequence from Ϫ365 to Ϫ328 of the lck proximal promoter as a probe. The Ϫ365 to Ϫ328 (Ϫ365/Ϫ328) oligonucleotides or the oligonucleotides carrying mutations in the G-rich region (Ϫ365/Ϫ328mut) were used as competitors (10-and 50-fold molar excess over the labeled probe) to determine the binding specificity of the B-factor. B, the B-factor binds to ␦A, the core enhancer sequences of the CD3␦ gene enhancer. Unlabeled oligonucleotides with various G-rich sequences were used as competitors. ␦A-CRE, ␦A carrying the mutation in the CRE binding site; ␦A-G, ␦A carrying the mutation in the G-rich sequence; and Ikaros BS, a high affinity Ikaros binding sequence. C, Ikaros is not a component of the B-factor. Anti-Ikaros antiserum did not react with the B-factor. D, zinc finger protein; mt␤ that binds to ␦A sequence is a component of the B-factor. The B-factor complex was supershifted by anti-mt␤ antibodies. E, the B-factor present in nuclear extracts from EL4.
tein that binds to the V␤8.1 promoter and the V␣ silencer of the T cell receptor genes (17). The cDNA sequence encoding the NH 2 -terminal half of the mt␤ is 90% identical to ht␤, and their deduced amino acid sequences are 95% identical. The 3Ј-half of the mt␤ coding region has 91% identity with the 3Ј-untranslated region of the reported ht␤ cDNA. These indicate that mt␤ is the murine homologue of ht␤ and that the reported ht␤ cDNA sequence has a one-base deletion that causes a frameshift and a premature stop codon. During this study, several cDNAs that have identical sequences with mt␤ have been reported: BFCOL1 that binds to the proximal promoters of the type I collagen genes (27) and BERF-1, a 89-kDa protein that binds to a muscle-specific enhancer of the ␤-enolase gene (28). In addition, the rat and human homologue of the protein, ZBP-89, has been shown to bind to promoter regions of the gastrin gene (29) and the ornithine decarboxylase promoter (30). It has subsequently been reported that the same zinc finger protein also binds to the p21WAF1 gene (31), the matrix metalloproteinase-3 gene (32), the pT␣ gene (33), and the vimentin gene (34).
Recombinant mt␤ Binds to the lck Proximal Promoter-We then asked whether recombinant mt␤ forms the B-factor complex. Recombinant mt␤ expressed in COS7 cells was detected as a band around 105 kDa in immunoblots ( Fig. 2A). An endogenous simian homologue of mt␤ in COS7 cells was detected at the same position as the recombinant mt␤ when the blot was overexposed (data not shown). mt␤ protein appeared to migrate more slowly in SDS-PAGE than its estimated molecular size, as is consistent with the observation for BFCOL1 by Hasegawa et al. (27). In EMSA, a residual amount of the B-factor complex was detected in COS7 cells that derived from the endogenous simian mt␤-homologue protein. Overexpression of mt␤ resulted in a significant increase of the amount of the B-factor complex (Fig. 2B). The entire complex was supershifted by the addition of anti-mt␤ antibodies. These results strongly indicate that mtB by itself, or in combination with proteins present in COS7 cells, forms the B-factor complex that binds to the Ϫ365 to Ϫ328 region of the lck promoter.

Expression of Mt␤ in Various Cell Lines and
Tissues-In a previous study, the strong correlation between the lck proximal promoter activity and amounts of B-factor has been reported (16). We therefore examined the expression of mt␤ mRNA in various cell lines and tissues. As shown in Fig. 3A, two mRNA species, with estimated sizes of 9.0 and 4.2 kilobase, were detected. Mt␤ mRNA expression was observed in all cell lines tested and was independent of the lck proximal promoter activity. In mice, the mt␤ transcripts were ubiquitously expressed in various tissues. The mRNA was abundant in thymus where the proximal promoter is active; however, significant amounts of mRNA were also detected in all tissues, especially in the heart, kidney, and liver (Fig. 3B). We conclude that there is no correlation between lck proximal promoter activity and the expression levels of mt␤ mRNA.
The expression level of mt␤ protein might be controlled by post-transcriptional mechanisms. To test this possibility, we measured mt␤ protein levels by immunoblots of whole cell extracts isolated from various cell lines and primary mouse lymphoid cells. The mt␤ protein was detected in all tested cell lines, LSTRA (thymoma), BAL17 (mature B), Ba/F3 (proB), MTH (mature T), and EL4 (lymphoma) cell lines (Fig. 4A, left  panel). Significant amounts of the mt␤ protein were also detected not only in thymocytes but also in splenic B cells and T cells (Fig. 4A, right panel). The expression of mt␤ protein was further confirmed by EMSAs. The B-factor was detected in the nuclear extract prepared from thymocytes as reported previously (16). We initially failed to detect either mt␤ protein or B-factor complex in extract from total splenocytes. However, the B-factor was present in the nuclear extract prepared from purified splenic B cells as well as mature T cells (Fig. 4B). High proteinase activity in total splenocytes may have caused deg- . Type I and II lck mRNAs are transcribed from the proximal and distal promoters, respectively. The relative ratio of type I and type II mRNAs in EL4 and BAL17 cells were measured by semiquantitative RT-PCR analysis. Serial dilutions (3-fold) of cDNA prepared from each cell line were subjected to PCR using sets of primers for type I (primers A and C) and type II (primers B and C) lck transcripts. ␤-Actin cDNA was amplified (right lower panel) to calibrate the amounts of cDNA templates in each sample. The proximal lck promoter is mainly active in EL4, while the distal promoter is active in BAL17. B, mt␤ activates transcription from an artificial promoter consisting of the mt␤ binding site of the lck proximal promoter and a TATA-box (pLuc-wild) in EL4 but not in BAL17. Cells were transfected with 10 g of reporter plasmid and 10 g of mt␤ expression (ϩ) or vector plasmid (Ϫ). Twelve hours after transfection, cells were harvested, lysed, and subjected to luciferase activity measurement. The luciferase activities are represented as percent activity of that produced by pGL2 control vector driven by the SV40 promoter. In pLuc-mut, mutations were introduced into the mt␤ binding site. The activity produced by a promoterless plasmid (0) is also shown. The results represent mean Ϯ S.D. of multiple independent transfections. C, the binding of mt␤ is critical for the lck proximal promoter activity. Cells were transfected with 10 g of luciferase reporter constructs carrying the various lengths of the lck proximal promoter region (Ϫ3200, Ϫ584, Ϫ433, Ϫ240, and 0) or the mutated promoter sequences (Ϫ3200-mut and Ϫ433-mut). The mt␤ binding site was destroyed by point mutations in the Ϫ3200-mut and Ϫ433-mut reporter constructs. The results are represented as percent luciferase activity observed with pGL2 control vector driven by SV40 promoter. The results represent mean Ϯ S.D. of multiple independent transfections. In each experiment, the luciferase activities were normalized for transfection efficiency by measuring renilla luciferase activities encoded by a co-transfected pRL-TK plasmid. ND, not determined.

FIG. 4. Expression of mt␤ protein and the B-factor is not restricted in T lineage cells.
A, Western blot analysis of mt␤ in LSTRA (thymoma), BAL17 (mature B), Ba/F3 (pro-B), MTH (mature T), EL4 (lymphoma), thymocytes, purified splenic T cells (95% was CD3 ϩ ), and purified splenic B cells (95% was B220 ϩ ). Cells (2 ϫ 10 5 cells) were boiled in 1 ϫ SDS-PAGE sample buffer, and the insoluble materials were removed by centrifugation. Resulting cell lysates were separated, transferred to membrane, and were probed with anti-mt␤ antibodies. Molecular size markers are indicated on the left. B, EMSA. The B-factor (arrow) was detected in nuclear extracts prepared from thymocytes, purified splenic B cells (95% was B220 ϩ ), BAL17, Ba/F3, MTH, and purified splenic T cells (95% was CD3 ϩ ). Anti-mt␤ antibodies supershifted the Bfactor complexes in all of tested extracts. radation and prevented detection of mt␤ protein and the B-factor complex. Careful preparation of nuclear extracts revealed the presence of the B-factor, even in BAL17, Ba/F3, and MTH cell lines and mature splenic T cells (Fig. 4B).
Lineage-specific Transactivation by mt␤ and Its Critical Function for the lck Proximal Promoter Activity-To clarify the role of mt␤ in transactivation of the lck proximal promoter, various reporter plasmids were constructed and introduced into EL4 or mature BAL17. EL4 expressed mainly type I transcripts (9) transcribed from the proximal promoter, while BAL17 expressed only type II transcripts (9) from the distal promoter (Fig. 5A). First, we studied transactivation of a reporter construct that contains only the B-factor binding site of the proximal promoter and a TATA box (pLuc-wild) (Fig. 5B). The pLuc-wild construct showed significant promoter activity in parental EL4 cells, and the activity was augmented ϳ3-fold by overexpression of mt␤. The promoter activity was not observed with a reporter (pLuc-mut) carrying mutations at the B-factor binding site on which mt␤ does not bind. Interestingly, pLuc-wild did not show significant promoter activity in BAL17, and the activity was not increased when mt␤ was overexpressed. We next studied the activity of reporter constructs with various deletions and mutations in the lck proximal promoter sequences (Fig. 5C). The fragment from Ϫ3200 to 0 of the promoter was active in EL4. Deletion of the fragment up to Ϫ584 did not affect promoter activity, whereas an additional deletion up to Ϫ433 resulted in a 3-fold increase of the activity. Further deletion of the fragment, including the B-factor binding site up to Ϫ240, did not impair the activity. However, the introduction of mutations into the B-factor binding site that abolishes mt␤ binding resulted in a significant reduction of the promoter activity of the Ϫ3200 and Ϫ433 fragments (compare Ϫ3200 versus Ϫ3200-mut and Ϫ433 versus Ϫ433-mut). All reporter constructs with the lck proximal promoter sequence did not show significant promoter activity in BAL17.
To confirm the role of mt␤ in transactivation of the lck proximal promoter, we reduced the protein expression level of mt␤ in EL4 by expressing mt␤ antisense mRNA. As shown in Fig. 6A, the amount of mt␤ protein was reduced to about 50% of control by transfection of the antisense plasmid. The activity of Ϫ3200 lck promoter fragment was significantly diminished, whereas that of the control SV40 promoter activity was not affected (Fig. 6B). Moreover, expression levels of the endogenous lck transcripts were also reduced (Fig. 6C). These results indicate that there is lineage-specific control for the mt␤ activity and that mt␤ plays a critical role in transactivation of the lck proximal promoter.

DISCUSSION
It has been reported that a close correlation exists between transcriptional activities of the lck proximal promoter and the presence of the B-factor complex binding to the G-rich stretch of the promoter (16). In this work, we characterized and identified the B-factor, a potential transcriptional regulator of the lck proximal promoter. Our results indicated mt␤, which has been previously cloned by its binding to the CD3␦ enhancer in vitro, is a component of the B-factor. Anti-mt␤ antibodies supershifted the B-factor complex, and the recombinant mt␤ expressed in cell lines formed the B-factor complex.
Mt␤ is an 89-kDa zinc finger protein belonging to the Krü ppel-type subfamily, whose members recognize the GC-rich or GT-rich sequences with their conserved DNA-binding zinc finger domains (26,35). Mt␤ (identical to BFCOL1, BERF-1) and its human and rat homologues (ht␤, ZBP-89, ZFP148) are reported to bind regulatory regions of various genes, such as the V␤8.1 promoter and the V␣ silencer of the T-cell receptor genes (17), the gastrin promoter (29), the type I collagen promoter (27), the ␤-enolase enhancer (28), the ornithine decarboxylase promoter (30), the p21WAF1 promoter (31), the matrix metalloproteinase-3 promoter (32), the pT␣ enhancer (33), and the silencer element of the vimentin gene (34). Our data showing that mt␤ is ubiquitously expressed at both mRNA and protein levels are consistent with previously published reports and the fact that mt␤ functions in the various promoters and enhancers in various tissues. However, it should be noted that mt␤ regulates genes critical for maturation of T cells, such as lck, pT␣ gene, as well as TCR ␣ and ␤ genes. Thus, mt␤ is one of the key transcriptional regulators controlling the T cell development.
In addition to its binding to a broad range of target genes, mt␤ appears to act as both transcriptional activator and repressor. As shown in this study, mt␤ is required for transactivating the proximal lck promoter. It also activates transcrip-FIG. 6. Expression of mt␤ antisense reduces the lck proximal promoter activity. A, reduction of mt␤ protein level by expression of antisense mt␤. EL4 cells were transfected with 10 g of pcDNA3-ASmt␤ or pcDNA3, together with 1 g of pEGFP-N1. Cells expressing GFP were sorted at 36 h after transfection. GFP-positive cells (5 ϫ 10 4 cells) were lysed, and mt␤ protein levels were analyzed by immunoblot. Molecular size markers are indicated on the left. The blots were stripped and subsequently probed with anti-tubulin antibodies to normalize the amount of loaded proteins. The relative expression levels of mt␤ are indicated by mt␤/tubulin ratio, which is set to 1 in cells transfected with vector control plasmid. B, relative luciferase activities of cells expressing mt␤ antisense plasmid. EL4 cells were transfected with 15 g of pcDNA3-ASmt␤ or control pcDNA3, together with 3 g of reporter plasmids (Ϫ3200/pGL2 or pGL2) and 1 g of pRL-TK. Cells were harvested at 30 h after transfection, and luciferase activity in cell lysates was measured. The results represent mean Ϯ S.D. of three independent transfections. C, reduced expression of the endogenous lck transcripts by expression of antisense mt␤. EL4 cells were co-transfected with pcDNA3-ASmt␤ or control pcDNA3, together with pEGFP-N1 and GFP-positive cells were sorted at 30 h after transfection. cDNAs were synthesized, and serial dilutions (2-fold) of cDNA templates were subjected to PCR amplifications for lck, G3PDH, and HGPRT. tion from the V␤8.1 promoter of the TCR gene and counteracts the silencing effect of the TCR ␣ gene silencer (17), and increased promoter activity of the p21WAF1 gene (31) and the matrix metalloproteinase-3 gene (32). Moreover, the binding site of mt␤ appears to be important for the pT␣ enhancer element (33). In contrast, mt␤ represses transcription from the gastrin gene (29), the ␤-enolase gene (28), the ornithine decarboxylase gene (30), and the vimentin gene (34). It is currently unknown how mt␤/BFCOL1/BERF-1 (ht␤/ZBP-89/ZFP148 in humans) manifests opposite activities on transcription of different genes. It has been shown that ZBP-89 competes with Sp1 for binding to the same element in the gastrin promoter (29) and inhibits the activation of the ornithine decarboxylase promoter by Sp1 (30). This might be one of the potential mechanisms by which mt␤/BFCOL1/BERF-1 suppresses transcription from several promoter elements. A fascinating possibility is that mt␤ has different isoforms derived from alternative splicing, and each isoform has distinct transcriptional activities. To test this hypothesis, we performed RT-PCR to amplify various fragments of mt␤ cDNA using several combinations of primers. However, we could not detect any splicing variants of mt␤ cDNA either in thymocytes or splenocytes (data not shown). Another possibility is that interacting proteins exist that determine the DNA binding specificity and the transactivating activities of the mt␤ complex and whose expression is regulated in a tissue-specific manner. It is also possible that mt␤ receives post-transcriptional modifications in a tissue-specific manner. Endogenous as well as overexpressed mt␤ transactivated transcription from an artificial promoter consisting of B-factor binding sites and TATA-box in EL4 but not in BAL17. Moreover, mt␤ generally transactivates genes expressed in T cells such as lck and TCR ␣ and ␤ genes, but represses gastrin, collagen, and ␤-enolase genes expressed in non-T cells. These observations support the idea that there are tissue-specific mechanisms regulating activity of mt␤. Basic Krü ppel-like factor (BKLF), which is widely expressed in various tissues and binds to the CACCC motifs, is also a member of the Krü ppellike zinc finger protein subfamily (36). Although BKLF positively regulates the transcription from a promoter containing a single BKLF binding site, it represses activity of glucocorticoid receptor-mediated activation of a promoter containing three copies of CACCC motifs and glucocorticoid-responsive elements (36,37). The NH 2 -terminal domain of BKLF is responsible for its repressive activity and interacts with a co-repressor protein, murine COOH-terminal-binding protein 2 (mCtBP2) (37). mCtBP2 interacts with BKLF and a number of mammalian transcription factors, such as Evi-1, AREB6, ZEB, and FOG, via the Pro-X-Asp-Leu-Ser (PXDLS) motif on the transcription factors (37). The mt␤ also carries several PXDLS-like motifs (PVDLQ (amino acids 112-116), PKDNS (amino acids 282-286)). mt␤ may associate with mCtBP2 or related molecules and exhibits suppressing activity in cells that fail to support proximal lck promoter activity. Our initial attempts, however, to detect associating molecules using glutathione S-transferase-mt␤ fusion proteins or modification of mt␤ such as phosphorylation have not been successful. mt␤ is critical for the full activation of the lck proximal promoter activity, since the mutation of the mt␤ binding site of the promoter or the reduction of the mt␤ protein level significantly impaired the promoter activity. However, the overexpression of mt␤ in EL4 did not augment the lck proximal activity (data not shown). This may indicate that the coordinated interaction of mt␤ with T cell-specific transcription factors (whose expression level is limiting) is involved in the full activation of the lck proximal promoter in thymocytes. Binding sites for the T cell-specific factors TCF-1, LEF, and TCF-1␣ are present in a region highly homologous between the murine and human proximal promoters (16). TCF-1 expressed ectotopically in BAL17 cells, however, failed to drive the lck proximal promoter activity with endogenous mt␤ (data not shown), suggesting a complex cooperation of multiple transcription factors in transactivating the lck proximal promoter. A Ϫ240 lck promoter fragment lacking a mt␤ binding site is still active in EL4. It should be noted, however, that the Ϫ240 fragment (but not the Ϫ584 promoter fragment carrying a mt␤ binding site) failed to support thymocyte-specific transcription of the lacZ-hGH transgene construct in mice (16). The EL4 cell line might lack a negative regulator expressed in primary cells that suppresses the Ϫ240 promoter activity in the absence of mt␤. Alternatively, EL4 might abundantly express positive transactivators binding to the Ϫ240 fragment whose activity is repressed by proteins bound to the Ϫ584 to Ϫ240 region. The mechanism that accounts for the discrepancy between the Ϫ240 promoter activity in EL4 and that in thymocytes remains unknown.
In addition to the positive regulators directing the lck proximal promoter activity in thymocytes, the silencers suppressing the promoter are likely to play roles in peripheral T cells. It has been reported that A2 complex binding to the sequence from Ϫ477 to Ϫ460 in the murine proximal promoter was detected in extracts from cells negative for the lck proximal promoter activity (16). It has also been reported that the Ϫ474 to Ϫ466 region in the human lck proximal promoter acts as a strong repressor in human tumor cell lines that do not express lck and binds proteins with molecular masses of 35 and 75 kDa (38). Deletion of the Ϫ584 to Ϫ433 region from our luciferase reporter constructs resulted in the enhancement of the promoter activity. These suppressive elements and binding factors are also critical in achieving tissue-specific expression of the lck proximal promoter in concert with the positive regulators, including mt␤.
In summary, we identified a Krü ppel-type zinc finger protein, mt␤, as a transactivator of the lck proximal promoter. mt␤ is ubiquitously expressed and manifests a broad range of activities on various genes. However, mt␤ is presumably a critical transcription factor for the T cell development, since it positively regulates lck and pT␣ genes as well as TCR ␣ and ␤ genes. Overexpressed mt␤ is active only in T-lineage cells, suggesting that there exists tissue-specific regulatory mechanisms to control mt␤ activity. Understanding the function of mt␤ should provide important insight into how T cell development and the thymocyte-specific expression of the lck proximal promoter are controlled and how one DNA-binding protein regulates different promoters positively and negatively.