Independent Promoters Regulate the Expression of Two Amino Terminally Distinct Forms of Latent Transforming Growth Factor-β Binding Protein-1 (LTBP-1) in a Cell Type-specific Manner*

Latent transforming growth factor-β (TGF-β)-binding proteins (LTBPs) are components of the extracellular matrix and large latent TGF-β complexes are secreted by various cells. Human LTBP-1 is known to exist in different forms. LTBP-1L (long) has an amino-terminal extension, which is not found in the smaller LTBP-1S isoform. To study the formation and transcriptional regulation of LTBP-1S and LTBP-1L isoforms, we determined the nucleotide sequences of their 5′-flanking regions. The upstream regions of both isoforms are devoid of TATA boxes but contain other putative binding sites for several transcription factors. Genomic sequencing revealed that LTBP-1L transcript is alternatively spliced to an internal splice acceptor inside exon 1 ofLTBP-1S and thus defined the genomic organization of the isoforms. Reporter gene analysis of upstream regions indicated the presence of independent, functional promoters, which regulate the transcription of the isoforms by cell-specific manner. Deletion analyses of the promoter regions revealed specific elements modulating their basal and cell type-specific expression. In SV-40 virus-transformed WI-38 lung fibroblasts a regulatory element repressed the transcription of LTBP-1S by a cell-specific manner. In amniotic epithelial cells, transcription of the LTBP-1Sreporter gene construct was down-regulated by a distal upstream element. mRNA levels of the isoforms of LTBP-1 were stimulated in response to TGF-β1 in WI-38 cells. However, since TGF-β1 failed to stimulate the transcription of LTBP-1reporter gene constructs, TGF-β1 may mediate the induction of the isoforms by post-transcriptional mechanisms. Chromosomal localization of the LTBP-1 gene was refined to 2p22–24.

Transforming growth factor-␤s (TGF-␤s) 1 are multifunctional growth factors which belong to the growing TGF-␤ su-perfamily of structurally related proteins (1). Members of the TGF-␤ superfamily have diverse effects on cellular functions. They regulate cell growth and differentiation and have important roles in developmental processes. TGF-␤s are, furthermore, essential in maintaining the balance between extracellular matrix production and proteolysis (2)(3)(4).
In the majority of normal, non-transformed cells, TGF-␤1 is secreted as a large latent complex (5)(6)(7)(8)(9)(10). During secretion TGF-␤1 is cleaved from its COOH-terminal latency associated propeptide, which remains noncovalently bound to the mature growth factor conferring latency to the complex. In the large latent complex TGF-␤1 is attached by disulfide bonding of its latency associated propeptide-part to a high molecular weight latent transforming growth factor ␤-binding protein, LTBP. LTBPs are members of LTBP/fibrillin family, which includes fibrillins -1 and -2 (11,12) and . Fibrillins, which are components of elastic tissue 10-nm microfibrils, resemble LTBPs in protein domain structure, both consisting of multiple copies of conserved 8-cysteine repeats and epidermal growth factor like (EGF-like) repeats. In the large latent complex the association between the latent TGF-␤1 and LTBP-1 is mediated by the third 8-cysteine repeat of LTBP-1 (18,19). Immunofluorescence and electron microscopic studies suggest also that LTBPs have roles as structural extracellular matrix proteins, since at least LTBP-1 and LTBP-2 have been observed to localize to 10-nm extracellular matrix microfibrils, similar to those containing fibrillins (20,21).
Recent studies have revealed several variable forms of LTBPs generated by alternative splicing (22)(23)(24)(25). At least LTBP-1 and -4 have been detected to possess amino-terminally distinct forms (26,27,17). Alternatively spliced exons have also been detected at the 5Ј-end of Fibrillin-1 (28). LTBP-1L (long) is a splice variant, which contains an amino-terminal extension of 346 amino acids not found in the smaller LTBP-1S (small) isoform (27). The mRNA expression patterns of the isoforms in different human tissues suggest that they are regulated in a tissue-specific manner. LTBP-1L is mainly expressed in heart, placenta, kidney, and prostate, whereas LTBP-1S has a wider expression pattern and appears also in lung, skeletal muscle, testis, and ovary (27). The molecular mechanisms that lead to their diverse expression patterns are not known.
LTBP-1 is considered to play an important role in the assembly and efficient secretion of the small latent TGF-␤1 complex (29,30). After secretion LTBP-1 is known to target TGF-␤1 to the extracellular matrix (9), which probably serves as a storage place, from which the growth factor can be activated effectively. The longer, amino-terminal extended isoform of LTBP-1 is known to associate more efficiently than LTBP-1S with the extracellular matrix (27). Interestingly, studies of transformed cells have revealed a decrease in the production and secretion of LTBP-1, or even its total absence in these cells. Certain osteosarcoma cells secrete TGF-␤1 in a small latent complex lacking LTBP-1 (8). Immunohistochemical analysis of prostatic carcinoma indicated that in malignant tissue TGF-␤1 is produced without association with LTBP-1 (31). In human gastrointestinal carcinomas no expression of LTBP-1 was detected and TGF-␤1 remained in the cytosol of cancer cells suggesting defects in the secretory pathway of the growth factor (32).
The present study was carried out to determine how the two diverse transcripts of LTBP-1 are formed and to examine the putative differences in their transcriptional regulation in normal and transformed cells. Characterization and functional analyses of the upstream regions of LTBP-1 isoforms revealed that their transcription is regulated independently from distinct promoter regions. Determination of the genomic structure of LTBP-1L splice site further refined the mechanism of how the independently regulated transcripts of LTBP-1S and -1L are formed. Functional analyses of their regulatory regions revealed elements affecting their constitutive and cell-type specific transcription, including differential regulation in transformed cells compared with normal cells. We, furthermore, determined the transcriptional responses of these genes to TGF-␤1 and certain other growth factors, and mapped the chromosomal localization of the LTBP-1 gene by fluorescence in situ hybridization to region 2p22-24.

EXPERIMENTAL PROCEDURES
Growth Factors and Other Reagents-Thermostable AmpliTaq polymerase was from Perkin-Elmer (Branschburg, NJ) and SuperScript II RNase H-Reverse Transcriptase from Life Technologies (Gaithersburg, MD). All the DNA modifying enzymes were purchased from New England Biolabs (Beverly, MA). Reagents for dideoxy sequencing and radioactive labels were from Amersham Pharmacia Biotech (Uppsala, Sweden). Human EGF was obtained from Sigma. Human hepatocyte growth factor, human vascular endothelial growth factor (VEGF), and human basic fibroblast growth factor were purchased from R&D Systems (Minneapolis, MN). Recombinant human small latent TGF-␤1 was a gift from Dr. P. Puolakkainen (Oncogen, Seattle, WA).
Cloning of the Genomic 5Ј-Flanking Regions of LTBP-1S and LTBP-1L-Genome Walker Kit (CLONTECH, Palo Alto, CA) was used to identify the upstream region of LTBP-1S. Two nested gene-specific primers corresponding to nucleotides from ϩ182 to ϩ214 and ϩ156 to ϩ185 of human LTBP-1S cDNA (GenBank accession number M34057) were used together with kit adapter primers in amplifying the gene upstream region by nested polymerase chain reactions. Nested PCR reactions were performed for 35 cycles using 4-min extension times, otherwise following the manufacturer's instructions. Amplification products from differentially restricted genomic DNA pools were cloned into pGEM-T cloning vector (Promega Corp., Madison, WI) and used as templates in DNA sequencing. Sequencing was performed in both directions by an ALFexpress automated sequencer (Amersham Pharmacia Biotech). An additional upstream region of LTBP-1S was obtained by using adapter primers of the kit and another set of nested genespecific oligonucleotides designed based on the sequence obtained from the first amplification. The primer sequences were: 5Ј-TGCATTC-CACGTATCAAGTGGTCGTGTAACCTCTACTTG-3Ј and 5Ј-CTCTAG-TGACGGGCAGCTCATTACCCACAAAAGCA-3Ј. Nested PCR reactions, cloning, and sequencing were performed as before.
Genomic DNA for human LTBP-1L was cloned from FIX II human placenta genomic library (Stratagene, La Jolla, CA). The library was screened using a polymerase chain reaction-derived probe corresponding to nucleotides from ϩ1 to ϩ307 of human LTBP-1L cDNA (GenBank™ accession number L48925). The probe was labeled with [␣-32 P]dCTP using the random priming kit (Amersham Pharmacia Biotech) and hybridized to filters in 5 ϫ SSPE, 50% formamide, 5 ϫ Denhardt's, 0.25% sodium dodecyl sulfate at 42°C overnight. DNA from positive clones were extracted and cloned into pBluescript II KS(ϩ) cloning vector (Stratagene) as a NotI restriction fragment. A restriction map of the desired clone was built up by digesting the clone with single cutters of the multiple cloning site of pBluescript II KS(ϩ) and detecting specific fragments by Southern blotting. A shorter 8.4-kb SacI restriction fragment including the 5Ј-flanking region of LTBP-1L was further subcloned into pBluescript II KS(ϩ) and sequenced as described for LTBP-1S.
Primer Extension-Primer extending from ϩ1 to ϩ24 nucleotides relative to the translation start site of LTBP-1S was end-labeled with [␥-32 P]ATP using T4 polynucleotide kinase. Total RNA from WI-38 human embryonic lung fibroblasts (CCL-75, American Type Culture Collection, Rockville, MD) was extracted using a commercial RNeasy Kit (Qiagen, Hilden, Germany). 20 g of total RNA and 200 fmol of the end-labeled oligonucleotide were mixed and denatured at 70°C for 10 min. The oligonucleotide was then extended with SuperScript II RNase H-reverse transcriptase at 48°C for 50 min. The extension products were separated by an 8% denaturing urea polyacrylamide gel. Their sizes were determined by comparison with sequencing ladders obtained from dideoxy sequencing of LTBP-1S using the same primer as in the primer extension reaction.
Ribonuclease Protection Assay-The 5Ј-end of LTBP-1L transcript was determined by ribonuclease protection analysis following the instructions of a commercial RPA III kit (Ambion Inc., Austin, TX). AscI-AgeI and PvuII-DraI fragments of the upstream region of LTBP-1L spanning nucleotides from Ϫ513 to Ϫ21 and from Ϫ398 to Ϫ163, respectively, relative to the translation initiation site of LTBP-1L, were cloned into pGEM-7Zf vector (Promega), which contains the recognition site of T7 polymerase. The plasmids were further linearized and antisense RNA probes were synthesized with T7 RNA polymerase in the presence of [␣-32 P]UTP. Messenger RNA from WI-38 cells was extracted using Oligotext Direct mRNA Midi Kit (Qiagen), and 0.6 g of mRNA was hybridized with the appropriate antisense RNA probe. Following hybridization overnight, unpaired, single-stranded probes were digested with RNase A/RNase T1 Mix and the samples were subsequently precipitated. Protected fragments were analyzed in a denaturing 6% polyacrylamide gel and their sizes were determined by comparing with M13 phage sequencing ladder.
Construction of Reporter Plasmids-The genome walking amplification product, which contained the 5Ј-flanking region of LTBP-1S, was transferred from pGEM-T cloning vector to pGL3 basic luciferase reporter gene plasmid (Promega) by first subcloning it to pBluescript II KS(ϩ). The fragment was cloned using EcoRI and SacI restriction endonucleases, which recognize the vector polylinkers. The 3Ј-region of the fragment, which contained the coding sequences of LTBP-1S, was subsequently replaced with a polymerase chain reaction-derived product corresponding to nucleotides from Ϫ1339 to Ϫ1 relative to the translation start site of LTBP-1S. Finally the fragment covering only the upstream region of LTBP-1S was moved from pBluescript II KS(ϩ) to pGL3 basic reporter vector by digesting the vector polylinkers with XhoI and SacI restriction endonucleases. The construct was named "L-1S PROM." Four 5Ј-deletion constructs of L-1S PROM were generated by double digestion with EcoRI, which recognizes a sequence in the vector polylinker, and with the indicated restriction endonucleases followed by blunt end religation. Starting from position Ϫ1 relative to the translation initiation codon of LTBP-1S "L-1S⌬MscI" construct contained 1375 bp, "L-1S⌬BsmBI" 977 bp, "L-1S⌬PvuII" 520 bp, and "L-1S⌬BlpI" 163 bp of the upstream region of LTBP-1S (Fig. 5). An internal deletion to the L-1S PROM construct was generated by digesting it with BsmBI and PvuII endonucleases and religating blunted ends. The construct was named "L-1S⌬BsmBI-PvuII." "L-1S/L PacI-SacI" construct was built by ligating a PacI-SacI restriction fragment from the upstream region of LTBP-1L to the recognition site of EcoRI in the upstream region of the "L-1S PROM" constuct ( Fig. 5A). "L-1S PROM REV" reporter plasmid containing the upstream region of LTBP-1S in reverse orientation was constructed by inverting the insert using NotI and BglII restriction endonucleases.
The LTBP-1L 5Ј-flanking region was cloned into pGL3 basic by first moving the entire 8.4-kb SacI restriction fragment from pBluescript II KS(ϩ) to pGL3 basic reporter plasmid. In order to include the outmost 3Ј-end of the LTBP-1L upstream region, a large 3Ј-region, which contained the coding sequences of LTBP-1L was replaced with an AgeI-Bsp102I fragment of the upstream region of LTBP-1L. The fragment corresponds to nucleotides from Ϫ25 to Ϫ5 relative to its translation start site (Fig. 1B). Several 5Ј-deletion constructs of this 2265-bp construct named L-1L PROM were further generated by digestions with the following enzymes and blunt end religation. "L-1L⌬PacI" construct, which is 1465 bp in size, was made by using SacI, which recognizes a sequence in the vector polylinker, and PacI restriction endonucleases. "L-1L⌬KpnI" (851 bp) was constructed by using KpnI, which has recognition sites in both the vector and the gene upstream region. "L-1L⌬AatI" (457 bp) and "L-1L⌬PvuII" (161 bp) constructs were gener-ated by double digesting with KpnI together with AatI or PvuII restriction endonucleases (Fig. 6). In addition, a reporter plasmid containing the 2265-bp upstream region of LTBP-1L in reverse orientation was constructed using KpnI and XhoI restriction endonucleases. The ends of all the constructs were finally sequenced using universal primers of the pGL3 basic vector.
Cells to be transfected were cultured in 35-mm diameter plates. Upon reaching 50 -80% confluence WI-38, WI-38/VA13, and HA cells were co-transfected with the total 3 g of each construct plasmid and pRL-TK control plasmid using FuGENE 6 transfection reagent (Roche Molecular Biochemicals, Mannheim, Germany) according to the manufacturer's instructions. HT-1080 cells were rinsed twice with serum and antibiotic-free Dulbecco's modified Eagle's medium and transfected similarly using LipofectAMINE transfection reagent (Life Technologies, Inc.). The pRL-TK plasmid contains the Renilla luciferase gene of seapansy under control of the constitutively expressed tymidine kinase promoter, and is used in dual luciferase assays to normalize the transfection efficiency. After 24 h the following concentrations of growth factors were added to WI-38 fibroblasts under serum-free conditions: TGF-␤1 (5 ng/ml), hepatocyte growth factor (20 ng/ml), basic fibroblast growth factor (20 ng/ml), EGF (30 ng/ml), or VEGF (50 ng/ml). Subsequently, 48 h after transfection, the cells were washed twice with phosphate-buffered saline (0.14 M NaCl, 10 mM sodium phosphate buffer, pH 7.4) and lysed with 500 l of Passive Lysis Buffer (Dual Luciferase Kit, Promega). Finally, 20 l of the lysates were subjected for activity measurements by using a Dual Luciferase Kit (Promega) and Digene DCR-1 luminometer (MGM Instruments, Inc., Hamden, CT).
TGF-␤1 Induction and Northern Hybridization Analysis-Human amniotic epithelial cells, WI-38 lung fibroblasts, SV-40 WI-38/VA13 virus-transformed fibroblasts, and HT-1080 osteosarcoma cells were cultured in 100-cm 2 flasks to 70% confluence. For growth factor induction WI-38 cells were washed several times by serum-free medium. Subsequently TGF-␤1 (5 ng/ml) was added to cells and incubated for 24 h under serum-free conditions. Total RNA was extracted from all the cell lines by the commercial RNeasy kit (Qiagen) following the manufacturer's instructions. For Northern hybridization analysis 10 g of total RNA was electrophoresed on a 0.8% formaldehyde-agarose gel and transferred to Hybond-N nylon filter (Amersham Pharmacia Biotech) in 20 ϫ SSC. A probe for hybridizations corresponding to nucleotides from ϩ1756 to ϩ2593 of human LTBP-1S cDNA was radioactively labeled with [␣-32 P]dCTP. The hybridizations were performed using Express Hyb hybridization solution (CLONTECH) according to the manufacturer's instructions and the filters were washed under high stringency conditions followed by autoradiography. Similar hybridizations were performed with radioactively labeled glyceraldehyde-3-phosphate dehydrogenase cDNA probe.
Fluorescence in Situ Hybridization-A polymerase chain reactionderived fragment corresponding to bases from ϩ91 to ϩ1389 of human LTBP-1S cDNA was used as a probe in selecting human genomic PAC clone (Genome Systems Inc., St. Louis, MO). The obtained PAC clones were verified by polymerase chain reaction using oligonucleotides specific for LTBP-1S. The probes for hybridization were prepared by nick translation using the PAC clones as templates and biotin-l4-dATP as a label. Human lymphocyte metaphase spreads were treated with pepsin (0.2 mg/ml in 0.01 M HCl) at 37°C for 10 min. Chromosomes were denatured in 70% formamide, 2 ϫ SSC, pH 7.0, at 64°C for 2 min. Hybridization signals from preparations were detected by indirect immunofluorescence using avidin-tetramethylrhodamine isothiocyanate. The slides were counterstained with 4,6Ј-diamidino-2-phenylindole. Fluorescent signals were analyzed by an Olympus (Tokyo, Japan) fluorescence microscope equipped with an ISIS digital image analysis system (MetaSystems GmbH, Altlussheim, Germany). The chromosome identity was confirmed by chromosome-specific painting following the instructions of the manufacturer (Cambio, Cambridge, United Kingdom).

Cloning of the 5Ј-Flanking Regions of LTBP-1S and LTBP-
1L-The genomic upstream region of LTBP-1S was amplified by nested polymerase chain reaction using the Genome Walker Kit and oligonucleotides designed based on LTBP-1S cDNA. The amplification products obtained from differentially restricted genomic DNA pools of the kit were cloned into pGEM-T cloning vector for sequencing. Sequencing of five overlapping clones of the 5Ј-flanking region of LTBP-1S resulted in a sequence of 1.75 kb upstream from its translation initiation site (Figs. 1, A and B). An additional 0.5-kb sequence further toward the 5Ј-flanking region was amplified by another nested PCR reaction using oligonucleotides complementary to part of the sequence obtained from the first PCR reaction. The genomic clones contained also the nucleotide sequence surrounding a previously reported splice site of the NH 2 -terminal extended form, LTBP-1L (27). Sequencing of the alternative splice site from the genomic clones revealed that the LTBP-1L transcript is generated by using an intraexonic splice acceptor site in the first exon of LTBP-1S. Therefore, the genomic region upstream from the alternative splice site functions as part of the first exon for LTBP-1S and as an intron for LTBP-1L (see Fig. 9).
Screening of FIX II human placenta genomic library with a PCR-derived probe corresponding to bases from ϩ1 to ϩ307 of LTBP-1L cDNA resulted in several positive phage clones. A clone named CL 6.9, which contained the upstream region of LTBP-1L, was further characterized by Southern blotting (Fig.  2A). An 8.4-kb SacI restriction fragment of CL 6.9 was subcloned into pBluescript KS II(ϩ) vector and used as a template in automated sequencing. The subclone was found to contain 2.3 kb of LTBP-1L upstream region (Fig. 2, A and B).
Identification of the Transcription Initiation Sites of LTBP-1S and LTBP-1L-Primer extension studies were performed in order to determine the exact transcription initiation codon of LTBP-1S. We localized the transcription initiation site by using an end-labeled oligonucleotide, which extended from ϩ1 to ϩ24 nucleotides relative to the translation initiation site. Two clear extension products, 64 and 102 bp in size, were detected when total RNA from WI-38 human embryonic lung fibroblasts was used as template (Fig. 3A). According to these extension products the transcription initiation sites of LTBP-1S are situated at Ϫ40 and Ϫ88 bp relative to its translation initiation codon.
Ribonuclease protection assay was used to determine the nucleotide sequence of the 5Ј-end of the LTBP-1L transcript since the attempts using primer extension analysis were not successful. Two antisense riboprobes spanning from nucleotide Ϫ513 to Ϫ22 and Ϫ398 to Ϫ163 relative to the translation initiation codon of LTBP-1L were hybridized with mRNA extracted from WI-38 cells. The unprotected RNA was subsequently degraded and the products were separated on a sequencing gel. One major 33-bp protected fragment of the probe extending from Ϫ513 to Ϫ22 was detected (Fig. 3B). The other riboprobe was completely digested by RNases (data not shown). The protected fragment corresponds to a transcription initiation site of LTBP-1L situating at Ϫ54 nucleotides relative to its translation initiation site.
Sequence Analysis of the 5Ј-Flanking Regions of LTBP-1S and LTBP-1L-Sequence analyses of the 5Ј-flanking regions of LTBP-1L and LTBP-1S revealed no core promoter motifs such as TATA or CAAT boxes in the proper vicinity of the identified transcription initiation sites. For LTBP-1S this observation was supported by the presence of more than one transcription initiation site in the gene upstream region. However, multiple putative recognition sites for sequence-specific DNA-binding transcription factors were detected in the upstream regions of both isoforms by computer aided analysis (34,35). These include the binding sites for Sp1, AP-1, AP-2, NF-1, Oct-1, CREB, NF-B, GATA-1, and CAAT box-binding protein (Figs. 1B and 2B). In the upstream region of LTBP-1S we identified a poten-tial nuclear factor-binding site, usually designated as TGF-␤ inhibitory element (TIE) or TGF-␤ responsive control element (TCE) (Fig. 1B). The TIE was first identified in the upstream regions of the TGF-␤1 inhibited genes (36). In transglutaminase promoter this element mediates both negative and posi-  Fig. 1A. B, nucleotide sequence of the upstream region of LTBP-1L. The symbols are the same as for LTBP-1S in Fig. 1B. In addition, boxes are used to mark also the recognition sites of the restriction endonucleases used in generating probes for ribonuclease protection studies.
tive effects of TGF-␤1 depending on cell type (37). Also another TGF-␤ responsive element, a recently discovered Smad-binding element (SBE) (38) was identified in the upstream region of LTBP-1S (Fig. 1B). SBE regulates the expression of JunB in response to some TGF-␤ superfamily members. The upstream region of LTBP-1L appeared to have a very high GC content, which is typical of abundantly expressed housekeeping genes.
Functional Analysis of the 5Ј-Flanking Regions of LTBP-1S and LTBP-1L-LTBP-1S and -1L 5Ј-flanking regions were in-serted to pGL3 basic vector, upstream of the luciferase reporter gene. The constructs were transfected to different normal and transformed cell lines to test whether the upstream regions of LTBP-1S and LTBP-1L contain functional promoters and whether they are capable of directing gene expression in a cell-specific manner. Human amniotic epithelial cells were used on the basis of the lower expression level of LTBP-1S in human placenta compared with several other tissues (27). Normal human WI-38 lung fibroblasts are a widely used strain, which has an SV-40 virus-transformed subline WI-38/VA13. The strains were studied since WI-38/VA13 cells are known to lack the LTBP-1 containing fibers observed in normal WI-38 fibroblasts (21). HT-1080 fibrosarcoma cells were another transformed cell line used in the study.
The upstream regions of LTBP-1S and LTBP-1L exhibited strong activation of the reporter gene in comparison to promotorless pGL3 basic vector (Figs. 5A and 6A). In addition, reversion of the upstream regions resulted in loss of the promoter activities (Figs. 5 and 6) proving that they contain functional, orientation dependent promoters. Comparison of the promoter activities in different cell lines (Fig. 4A) revealed that in WI-38 lung fibroblasts the upstream region of LTBP-1S activated the expression of the luciferase gene about 10 times more effectively than that of LTBP-1L. In human amniotic epithelial cells the expression of the reporter gene driven by LTBP-1S promoter was down-regulated since it was only about 1% of its expression level in WI-38 fibroblasts (Fig. 4A). The upstream region of LTBP-1S was able to activate the reporter gene even less than the upstream region of LTBP-1L in these cells (Fig.  4A). Similar differences in the expression levels of LTBP-1 isoforms between the cell lines were detected in Northern blot analysis (Fig. 4B). These results prove that the upstream regions of LTBP-1S and LTBP-1L are capable of directing cellspecific expression of the reporter gene.
In WI-38/VA13 cells, which are SV-40 virus-transformed counterparts of normal WI-38 fibroblasts, the reporter gene activities of both isoforms were diminished clearly when compared with normal WI-38 fibroblasts (Fig. 4A). In normal WI-38 fibroblasts the promoter of LTBP-1S stimulated the expression of luciferase approximately 100 times more efficiently and the promoter of LTBP-1L 6 times more efficiently than in transformed WI-38/VA13 fibroblasts. A clear decrease in the expression levels was also detected in HT-1080 fibrosarcoma cells. Northern blot analysis supported these observations (Fig. 4B). These data suggest that malignant transformation has negative impact on the transcription of both isoforms of LTBP-1.

Identification of Regulatory Elements in the Promoter Regions of LTBP-1S and LTBP-1L
-In order to characterize the specific regions of the promoters, which are responsible of the basal and cell-specific transcription of LTBP-1 isoforms, a series of deletion constructs were made and transiently transfected to selected cell lines. The constructs contained serial 5Ј-deletions of LTBP-1S and LTBP-1L upstream regions fused to the luciferase reporter gene (Figs. 5 and 6).
When fragments from the 5Ј-end of the longest 1751-bp construct of LTBP-1S were systematically deleted the transcription efficiency of the luciferase gene decreased gradually in WI-38 fibroblasts (Fig. 5A). This finding indicates that in WI-38 cells the effective basal expression of LTBP-1S seems to be regulated by additive cis-acting elements, which are evenly distributed throughout the gene upstream region. In human amniotic epithelial cells, which express LTBP-1S at a remarkably lower level, a deletion of an upstream region from nucleotide Ϫ1751 to Ϫ1375 had no remarkable effect on the promoter activity of LTBP-1S upstream region (Fig. 5B). Since in WI-38 fibroblasts the same deletion decreased the activity (Fig.   FIG. 3.

Identification of transcription initiation sites for LTBP-1S and LTBP-1L.
A, total RNA from human embryonic lung fibroblasts was extracted and used as template in primer extension analysis. An end labeled oligonucleotide, covering bases from ϩ1 to ϩ24 relative to the translation initiation site of LTBP-1S, was extended by reverse transcription. The extension products were resolved on 8% urea-polyacrylamide sequencing gel, and the sizes of the extended products were determined by comparison with sequencing ladders obtained from dideoxy sequencing of LTBP-1S using the same primer ( lanes  1-4). Extension products of 64 and 102 bp in size were identified (lane 5). The products correspond to transcription initiation sites located at Ϫ40 and Ϫ88 nucleotides relative to the translation initiation codon of LTPB-1S. The transcription initiation sites are marked by boxes to the shown nucleotide sequence of the upstream region of LTBP-1S. B, antisense RNA probes complementary to nucleotides from Ϫ513 to Ϫ22 and from Ϫ398 to Ϫ163 relative to the translation initiation codon of LTBP-1L were hybridized to mRNA extracted from WI-38 fibroblasts. The riboprobes were subsequently digested with ribonucleases, and the protected fragments were separated on 6% denaturing polyacrylamide sequencing gel. Their sizes were determined by comparison to M13 phage sequencing ladder (bases shown on the left). One major protected fragment, 33 nucleotides in size, was identified corresponding to the transcription initiation site at Ϫ54 nucleotides relative to the translation initiation site of LTBP-1L. 5A), the region could contain a binding site for a positive regulator, which is missing from amniotic epithelial cells. Another possibility is that the region contains two regulatory sites, a positive element and an adjacent binding site of a negative regulator specific only for epithelial cells. Furthermore, deletion of the upstream region of LTBP-1S extending from nucleotide Ϫ1375 to Ϫ977 stimulated the transcription of the reporter gene in epithelial cells (Fig. 5B). In WI-38 fibro-blasts the same deletion decreased its transcription (Fig. 5A) suggesting that the region could contain a binding site for another cell-specific negative regulator, which could cause the observed down-regulation of LTBP-1S in amniotic (HA) epithelial cells. Additional 5Ј-deletions repressed the promoter activity as in WI-38 cells.
In transformed WI-38/VA13 fibroblasts the longest L-1S PROM (1751 bp) reporter gene construct stimulated the luciferase expression most effectively, and the transcriptional activity of the shorter 1375-bp L-1S⌬MscI construct was decreased as in normal WI-38 cells (Fig. 5C). However, in transformed fibroblasts removal of the region extending from nucleotide Ϫ1375 to Ϫ520 had no remarkable effect on the promoter activity (Fig. 5C), whereas in normal WI-38 cells the same deletion repressed it clearly (Fig. 5A). To refine that the slight difference observed between the activities of 977-bp L-1S⌬BsmBI and 520-bp L-1S⌬PvuII constructs in transformed cells (Fig. 5C) is not due to a negative regulatory factor we constructed a deletion construct L-1S⌬PvuII-BsmBI. The activity of L-1S⌬PvuII-BsmBI, which was missing the region from Ϫ977 to Ϫ520, was tested in transformed WI-38/VA13 cells and found to exhibit as strong promoter activity as undeleted L-1S PROM in the same cells (Fig. 5C). This finding supported the idea that transformed WI-38/VA13 fibroblasts are devoid of positive regulators, which recognize the region extending from Ϫ1375 to Ϫ520 bases of the upstream region of LTBP-1S and stimulate its transcription only in normal WI-38 fibroblasts. The lack of positive regulators in transformed cells would further cause negative impact on the overall transcription level of LTBP-1S in transformed cells. However, the same effect could be caused in transformed fibroblasts by negative regulatory factors, which would repress the activity of continually active positive elements situating at the upstream region from Ϫ1375 to Ϫ520 bases of LTBP-1S.
Deletions of the 5Ј-end of the 2265-bp LTBP-1L upstream region increased the reporter gene activity in WI-38 cells (Fig.  6A). When the region between nucleotides Ϫ2265 and Ϫ1465 was removed, the luciferase activity was about 2-fold compared with that of 2265-bp L-1L PROM construct (Fig. 6A). This indicates that the region most likely contains an element which is able to down-regulate the expression of LTBP-1L in WI-38 fibroblasts. To further confirm this hypothesis, the putative negative regulatory region between nucleotides Ϫ2265 and Ϫ1465 was cloned to the upstream region of L-1S PROM construct and tested for its ability to affect the transcription driven from LTBP-1S promoter. When compared with the transcriptional activity of L-1S PROM in WI-38 fibroblasts the construct L-1S/L-1L PacI-SacI, which contains the putative negative regulatory element, exhibited about 70% of the activity of L-1S PROM supporting the earlier observation (Fig. 5A). When in both normal and transformed WI-38 fibroblasts the region between nucleotides Ϫ457 and Ϫ161 of LTBP-1L was deleted, the expression of the reporter gene was clearly reduced (Figs. 6, A  and B). This indicates that the region extending from Ϫ1 to nucleotide Ϫ457 contains a minimal promoter, necessary for efficient basal gene transcription. Otherwise in transformed WI-38/VA13 fibroblasts no such unambiguous, negatively regulating element, which could contribute to observed 6-fold higher transcription levels of LTBP-1L in non-transformed cells, could be identified (Fig. 6B). fected with L-1S PROM and L-1L PROM luciferase constructs and treated with the indicated growth factors (Fig. 7, A and B). TGF-␤1 had no remarkable effect on the transcription rate of LTBP-1L luciferase construct but, unexpectedly, it decreased the expression of LTBP-1S luciferase construct by about 60% (Fig. 7A). The presence of the TGF-␤ inhibitory element/TGF-␤ responsive control element typical of several TGF-␤ inhibited genes (36,37) failed to explain the down-regulation of LTBP-1S since L-1S⌬BsmBI reporter gene construct, which is lacking the putative TGF-␤ inhibitory element/TGF-␤ responsive control element-binding site, was also inhibited by TGF-␤1 to some extent (data not shown). Human basic fibroblast growth factor was detected to increase the transcription driven by the LTBP-1S promoter while the other growth factors tested, including human hepatocyte growth factor, human EGF, and human vascular endothelial growth factor (VEGF), had no remarkable effects on the promoter activities of either of the isoforms of LTBP-1. To further examine the impact of TGF-␤1 on the expression of LTBP-1 isoforms, we performed Northern hybridization analysis. Two mRNA species were detected in WI-38 cells, corresponding to mRNAs of LTBP-1L and LTBP-1S (Fig. 7C). In Northern hybridization analysis, unlike in reporter gene assays, the expression levels of both isoforms of LTBP-1 were induced in response to 24 h treatment with TGF-␤1 (Fig. 7C). On the basis of these observations it seems likely that TGF-␤1 mediates its inducible effects on the expression of LTBP-1 isoforms by post-transcriptional mechanisms, rather than affecting the transcription rates of the genes. Whether the post-transcriptional mechanisms include mRNA stabilization or reduction of mRNA degradation remains to be determined.

Effects of TGF-␤1 on Promoter Activities and mRNA Levels of LTBP-1S and LTBP-1L-We
Chromosomal Localization of LTBP-1 Gene-Chromosomal localization of the human LTBP-1 gene was analyzed by using LTBP-1-specific human PAC clone as a probe. The clone was obtained by screening human genomic DNA PAC library with a PCR product specific for LTBP-1S cDNA. The obtained human genomic LTBP-1 PAC clone was labeled with biotin and hybridized with human leukocyte metaphase preparations. The hybridization studies were performed with three different PAC clones specific for LTBP-1. Fluorescent signals were detected on chromosome 2, at the region 2p22-24 (Fig. 8). The localization of the signals to chromosome 2 was finally verified by painting with a chromosome 2 specific probe (not shown).

DISCUSSION
In the current work we have determined the nucleotide sequences of the 5Ј-flanking regions of human LTBP-1S and LTBP-1L genes and showed that they contain functional promoters, which are able to direct the transcription of the isoforms independently, in an orientation dependent manner. Sequencing of the genomic region surrounding the previously reported alternative splice site of the LTBP-1L transcript (27)  further refined the genomic organization of LTBP-1S and LTBP-1L. On the basis of these data, we propose a splicing model for LTBP-1, which produces two independently regulated isoforms (Fig. 9). According to this model LTBP-1L is transcribed independently from a promoter region situating distal to the promoter of the shorter LTBP-1S isoform. Then, during the post-transcriptional hnRNA processing of LTBP-1L, the first exon of LTBP-1S serves as an intraexonic splice acceptor site for LTBP-1L (Fig. 9). The genomic region containing the first exon of LTBP-1S has thus a dual role since it also functions as an exon for LTBP-1S (see also Ref. 51). The use of alternative promoters, which leads to the production of NH 2terminal variants of a gene is a quite infrequent phenomenon described also for other extracellular matrix proteins, such as the mouse gene for the ␣1 chain of type XVIII collagen (39) and murine laminin ␣3 (40). Alternative splicing using internal splice acceptor sides is another relatively rare mechanism, which is known to be used in generating carboxyl-terminal distinct variants of the human ␣2(VI) collagen gene (41).
The transcription initiation sites of LTBP-1S were mapped to Ϫ40 and Ϫ88 nucleotides upstream from its translation start site (Fig. 3A). The start site at Ϫ88 bp is situated in close vicinity of the reported 5Ј-end of LTBP-1S cDNA (13). The transcription of LTBP-1L was identified to start from Ϫ54 nucleotides upstream from its translation initiation codon (Fig.  3B). No canonical TATA boxes were detected upstream from the transcription initiation sites of either of the isoforms. A recent study of genomic organization of fibrillin-1 describes that the 5Ј-flanking region of porcine fibrillin-1 is also devoid of TATA and CCAAT boxes and appears to be highly GC-rich (42). Similarly, LTBP-1L promoter has high GC content and contains potential binding sites for Sp1 in the vicinity and further upstream from the identified transcription initiation site (Fig.  2B). A putative binding site for CCAAT-binding protein is situated Ϫ385 bp upstream from the transcription initiation site of LTBP-1L. The CCAAT box, which is thought to affect the assembly of the transcription apparatus, is often located about Ϫ80 bp upstream from the transcription start point, but can also function from variable distances.
Reporter gene deletion analysis of the upstream region of LTBP-1S revealed that its constitutive, basal expression is not dramatically affected by any special region, but is rather regulated by several positive upstream elements, which gradually increase the promoter activity in WI-38 fibroblasts (Fig. 5A). The upstream region of LTBP-1S contains potential binding sites for general transcription factors such as CREB, NF-1, Oct-1, NF-B, AP-1, and GATA-1 (Fig. 1B).
In reporter gene analysis of LTBP-1L the shortest 161-bp luciferase construct L-1L PvuII, which contained the identified transcription initiation site at Ϫ54 bp, was able to induce After an additional 24 h the cell lysates were prepared and subjected for luciferase activity measurements. Data (mean Ϯ S.E.) are presented as percentage of the untreated control, which was set at 100%. Results shown represent the averages of at least three independent experiments. C, cultures of WI-38 fibroblasts were treated with TGF-␤1 (5 ng/ml) under serum-free conditions. The cells were harvested after 24 h and total RNA was extracted. Northern hybridization analysis was performed using a probe specific for both LTBP-1S and LTBP-1L (marked by arrows). The lower panel shows the hybridization signals of ubiquitously expressed glyceraldehyde-3-phosphate dehydrogenase (GAPDH) cDNA probe, which was used as a control for RNA loading. transcription by 16-fold when compared with the promoterless construct in WI-38 fibroblasts (Fig. 6A). The promoter region extending further to Ϫ457 nucleotides upstream of LTBP-1L seemed to be sufficient to stimulate the basal transcription in a non-tissue-specific way (Fig. 6, A and B). In addition to several putative binding sites of Sp1 and a binding site for CCAATbinding factor, the region from Ϫ1 to Ϫ457 nucleotides of LTBP-1L contains potential binding sites for Ap-2 transcription factor (43) (Fig. 2B). In WI-38 fibroblasts an upstream element extending from Ϫ2266 to Ϫ1465 of LTBP-1L was observed to contain a negative regulatory region, which downregulated the transcription of not only LTBP-1L but also LTBP-1S (Figs. 5A and 6A). Negative elements, which are able to reduce transcription from several promoters, are likely to hinder the formation of the basal transcription complex. Therefore, the trans-acting factor, which recognizes the negative element in the upstream region of LTBP-1L, may interfere with the formation of the basal transcription apparatus rather than with special activators of the transcription. It cannot, however, be excluded that alternatively spliced LTBP-1S and LTBP-1L could share common transcription activators.
Functional analysis of the upstream regions of LTBP-1S and LTBP-1L demonstrated their differential transcription patterns depending on the cell type (Fig. 4A). Similar differences in their expression patterns were detected by Northern blot analysis (Fig. 4B), suggesting the cell type-specific expression of the isoforms to be regulated by transcriptional means. Therefore the diverse expression patterns of LTBP-1S and -1L observed in different human tissues (7) would also seem to result from utilization of independent, alternative promoters for the isoforms. Reporter gene deletion analysis of LTBP-1S promoter revealed that the gene upstream region from Ϫ1751 to Ϫ1375 nucleotides could contain a regulatory element, which may contribute to its lower expression in amniotic epithelial cells than WI-38 fibroblasts (Fig. 5, A and B). In amniotic epithelial cells the down-regulation of the transcription of LTBP-1S could be caused by a lack of positive regulator(s) present only in WI-38 fibroblasts. Differences in relative concentrations of ubiquitously expressed transcription factors in different cells could mediate the cell-specific expression of LTBP-1S as has been shown for other genes (44,45). The region from nucleotide Ϫ1751 to Ϫ1375 of LTBP-1S contains a putative CREB motif, which is recognized by a ubiquitous CEBP motif present in many cellular promoters. There are also potential binding sites for NF-B and Oct-1 transcription factors. Octamer binding protein-1 (Oct-1) is another ubiquitous factor, which is known to stimulate transcription in non-lymphoid cells (46). It is also possible, however, that the identified regulatory region contains a continuously active positive element and in addition a negative element, which is able to repress the transcription only in amniotic epithelial cells. Another binding site of a negative regulatory factor, which could down-regulate the expression of LTBP-1S in amniotic epithelial cells, was identified in the upstream region extending from Ϫ1375 to Ϫ977 nucleotides (Fig. 5B).
In normal cells TGF-␤1 is secreted as large latent complexes of TGF-␤1⅐latency associated propeptide bound to LTBP-1 (7)(8)(9). These complexes, however, cannot be detected in cancer cells derived from malignant tumors (31,32). In immunohistochemical studies, LTBP-1 was lacking from malignant but not benign prostatic tissue (31) whereas, TGF-␤1 was observed also in prostatic cancer tissue. Similarly, in gastrointestinal carcinomas, the immunoreactivity for LTBP-1 was observed in stromal cells but not in cancer cells (32). In cancer stroma the immunoreactivity for TGF-␤1 was detectable in rough endoplasmic reticulum and perinuclear cisternae of the cells, whereas in cancer cells TGF-␤1 was observed in cytosol suggesting defects in its secretory pathway (32). Our results as well suggest alterations in transcriptional regulation of both isoforms of LTBP-1 as a consequence of cell transformation (Fig. 4, A and B). Since TGF-␤ is growth inhibitory to cells of epithelial and endothelial origin, it is possible that autocrine and paracrine growth inhibitory effects of TGF-␤1 could be diminished in transformed cells by reducing the transcription of LTBP-1 isoforms. Reporter gene deletion analysis of the upstream region of LTBP-1S revealed a regulatory region extending from Ϫ1375 to Ϫ520 bases, which may control the down-regulation of its expression in transformed WI-38/VA13 fibroblasts (Fig. 5C). The transcription may be reduced by two possible regulatory mechanisms. The regulatory region could contain binding sites of positively regulating factors, which are missing from transformed cells in contrast to normal cells. Potential binding sites for AP-1, GATA-1, CEBP, and CREB transcription factors are situated in the region in question (Fig.  1B). Another possibility is that both normal and transformed fibroblasts contain continuously active positive regulatory elements, which are repressed by a negative factor in transformed fibroblasts. Deletion analysis of the promoter of LTBP-1L in transformed WI-38/VA13 fibroblasts did not reveal specific regulatory regions, which could contribute to its lower transcription rate in transformed cells. The transcriptional activities of all the tested reporter gene constructs of LTBP-1L were lower in transformed cells than in normal fibroblasts, suggesting that in transformed fibroblasts its expression is regulated primarily by controlling gene activity.
TGF-␤1 treatment has been shown to induce the expression levels of the mRNAs of both isoforms of LTBP-1 in human foreskin fibroblasts (8). In a recent study TGF-␤1 was shown to increase the expression of LTBP-2 by transcriptional means in human fetal lung fibroblasts (47). Our data from Northern blot and reporter gene analysis in WI-38 fibroblasts revealed that TGF-␤1 induces the expression of mRNAs of LTBP-1 isoforms rather by post-transcriptional mechanisms than transcriptional up-regulation (Fig. 7). Since autoregulation of TGF-␤1 is known to be mediated by transcriptional means using an AP-1-binding site (48,49), the parallel expression of TGF-␤1 and LTBP-1 observed in several tissues (26,29,8) may not involve the same regulatory pathways. The exact nature of the putative post-transcriptional mechanisms needs to be determined FIG. 9. Genomic organization and splicing of LTBP-1 transcripts. The genomic DNA of the isoforms of LTBP-1 is transcribed by RNA polII from two independent promoters situated in the upstream regions of LTBP-1S and LTBP-1L. When the hnRNAs are post-transcriptionally processed by removing the noncoding intron sequences, the coding 5Ј-end containing the predicted signal sequence of LTBP-1S serves as an intron for LTBP-1L, and is subsequently spliced out. LTBP-1L transcript is spliced into the first exon of LTBP-1S, which thus has a dual role in functioning as an exon for LTBP-1S and an intraexonic splice acceptor site for LTBP-1L. Boxes denote the coding sequences of LTBP-1S and LTBP-1L. The number of exons is denoted by n since their number was not determined. The position of the signal sequence of LTBP-1S is indicated relative to the alternative splice site and marked by SS.
by studying the stabilization of cytoplasmic mRNA levels.
The chromosomal localization of the LTBP-1 gene had earlier been determined by human-rodent somatic cell hybrids to region 2p123q2, covering most of chromosome 2 (50). The chromosomal sublocalization of LTBP-1 was in this previous study determined using three hybrid cell lines retaining partially overlapping regions of chromosome 2 (50). We defined the localization of the LTBP-1 gene by fluorescence in situ hybridization also to chromosome 2, but dissimilarly to region 2p22324 at the tips of the short arms of the chromosome (Fig.  8). The FISH studies were performed with three different LTBP-1-specific probes, about 120 kb in size, which all hybridized with both homologues of chromosome 2 in a similar manner.
The existence of at least four members of LTBPs and their diverse mRNA expression patterns in different tissues and developmental stages suggest vast functional properties for LTBPs (reviewed in Ref. 52). The alternatively spliced isoforms and the utilization of multiple promoters further emphasize the complex patterns of gene expression in the LTBP/fibrillin family. The present study is an initial step toward understanding the molecular mechanisms leading to unique expression of LTBP-1S and LTBP-1L isoforms. Identification of the regulatory regions of LTBPs and their functional characterization provides means for more detailed examination of the gene transcription and determination of the diverse roles of the multiple forms of LTBPs.