Upstream tissue inhibitor of metalloproteinases-1 (TIMP-1) element-1, a novel and essential regulatory DNA motif in the human TIMP-1 gene promoter, directly interacts with a 30-kDa nuclear protein.

Elevated expression of the tissue inhibitor of metalloproteinases-1 (TIMP-1) protein and mRNA has been reported in human diseases including cancers and tissue fibrosis. Regulation of TIMP-1 gene expression is mainly mediated at the level of gene transcription and involves the activation of several well known transcription factors including those belonging to the AP-1, STAT, and Pea3/Ets families. In the current study, we have used DNase-1 footprinting to identify a new regulatory element (5'-TGTGGTTTCCG-3') present in the human TIMP-1 gene promoter. Mutagenesis and transfection studies in culture-activated rat hepatic stellate cells and the human Jurkat T cell line demonstrated that the new element named upstream TIMP-1 element-1 (UTE-1) is essential for transcriptional activity of the human TIMP-1 promoter. Electrophoretic mobility shift assay studies revealed that UTE-1 can form protein-DNA complexes of distinct mobilities with nuclear extracts from a variety of mammalian cell types and showed that induction of a high mobility UTE-1 complex is associated with culture activation of freshly isolated rat hepatic stellate cells. A combination of UV-cross-linking and Southwestern blotting techniques demonstrated that UTE-1 directly interacts with a 30-kDa nuclear protein that appears to be present in all cell types tested. We conclude that UTE-1 is a novel regulatory element that in combination with its cellular binding proteins may be an important component of the mechanisms controlling TIMP-1 expression in normal and pathological states.

The complex interactions between cells and components of their surrounding extracellular matrix (ECM) 1 are critical for many cellular properties including division, migration, differentiation, and death (1). Remodeling of the ECM occurs during embryonic development and during certain normal physiological processes such as wound healing (2,3). However, restructuring of the ECM has also been implicated in various human pathologies including impaired wound healing and tissue fibrosis (4), rheumatoid arthritis (5), restenosis following balloon angioplasty (6), atherosclerosis (7), and tumor development, invasion and metastasis (1,8). Under normal physiological conditions, a fine balance of ECM-degrading matrix metalloproteinases and their naturally occurring inhibitors, the tissue inhibitors of metalloproteinases (TIMPs), controls the rate of ECM turnover (5, 9 -11). Since the TIMP family of protease inhibitors (TIMP-1, TIMP-2, TIMP-3, and TIMP-4) are key regulators of matrix metalloproteinase activity and therefore ECM degradation (11), there is great interest in understanding the cellular mechanisms that control their expression. The prototypic member of the TIMP family, TIMP-1, is of particular interest, since, in addition to its classical role as a broad specific inhibitor of matrix metalloproteinases (12), TIMP-1 has also been reported to possess growth factor-like (13,14) and anti-apoptotic properties (15)(16)(17)(18). Guedez et al. (17) recently showed that TIMP-1 inhibits apoptosis and promotes survival of Burkitts' lymphoma cell lines. These observations support work implicating TIMP-1 in cancer, tumor invasion, metastasis, and angiogenesis (19 -23).
TIMP-1 gene transcription and mRNA expression can be stimulated by a wide variety of agents including serum, growth factors, phorbol esters, cytokines, and viruses (24 -29). Activation of hepatic stellate cells (HSCs), a key event in the pathophysiology of liver fibrosis (4), is also accompanied by induction of TIMP-1 promoter activity and mRNA expression (30,31). Nuclear run-on transcription assays employed in some of these studies demonstrate that induction occurs primarily at the level of gene transcription, although TIMP-1 expression in U937 cells appears to be regulated at least in part by changes in mRNA stability (32).
Structural features of the human, murine, and rat TIMP-1 gene promoters have been described previously (31). All three promoters lack a classical TATA box and contain an evolutionary conserved 22-bp serum response element located 75 bp upstream of the major transcription start site (24,26,29,31,33). The serum response element, which is critical for responsiveness of the promoter to growth factors, cytokines, and viruses, is composed of binding sites for AP-1 (Fos/Jun), signal transducer and activator of transcription, and Pea3 (Ets) transcription factors. Mutagenesis studies performed in several primary and transformed cell lines have established that the AP-1 site within the serum response element is critical for basal and inducible transcription (26,27,29,33,34). We have recently reported that this AP-1 site is critical for the function of a minimal active 162-bp human TIMP-1 promoter in human fibroblasts and culture-activated rat HSCs (31,33). The molecular events underlying HSC activation and induction of TIMP-1 can be studied in a cell culture model in which freshly isolated primary HSCs cultured for several days on tissue culture plastic undergo a similar phenotypic transformation to that observed in vivo (4,30,31). Studies using this culture model showed that induction of TIMP-1 promoter activity occurred at culture time points (5 days or more) associated with morphological and biochemical activation of rat HSCs and with induced expression of JunD, Fra2, and FosB, implicating these AP-1 proteins in the control of TIMP-1 expression in activated HSCs (31).
Whereas the role of the AP-1 containing serum response element site in the regulation of TIMP-1 gene transcription is well characterized, much of the 162-bp minimal active human TIMP-1 promoter remains functionally undefined. We have therefore used DNase I footprinting (35) to identify DNA-protein interactions occurring on the promoter in cells expressing high levels of TIMP-1. We report the identification of previously undescribed DNA-protein interactions at a novel regulatory sequence named upstream TIMP-1 element 1 (UTE-1). The site appears to be essential for transcriptional activity in a variety of cell types including activated primary human and rat HSCs as well as transformed human cell lines. We describe the characterization of nuclear proteins capable of interacting with the UTE-1 sequence in a specific manner and the identification of a 30-kDa species that directly binds to UTE-1 DNA.

EXPERIMENTAL PROCEDURES
HSC Isolation and Culture-HSCs were isolated from the livers of normal Harlan Sprague Dawley rats (400 Ϯ 50 g) by sequential perfusion with pronase and collagenase as described previously (36). HSCs were separated from the cell suspension over an 11.5% Optiprep (Nycomed Pharma AS, Oslo) gradient followed by elutriation. HSCs were seeded onto plastic and cultured in Dulbecco's modified Eagle's medium (Life Technologies, Inc.) supplemented with 16% fetal calf serum (Life Technologies) and maintained at 37°C in an atmosphere of 5% CO 2 .
Human HSCs were isolated from the livers of adult male patients following partial hepatectomy. Sequential perfusion with pronase and collagenase was carried out as described for isolation of rat HSCs (36). HSCs were separated from the cell suspension over an 11.5% Optiprep gradient and cultured onto plastic in Dulbecco's modified Eagle's medium supplemented with 16% fetal calf serum and maintained at 37°C in an atmosphere of 5% CO 2 .
Nuclear Extract Preparation-Nuclear extracts were prepared from cells using a modified version of the protocol described by Dignam et al. (37). In brief, cells (10 7 ) were harvested into 5 ml of ice-cold phosphatebuffered saline by centrifugation at 2500 rpm. The pellets were resuspended into 100 l of Dignam buffer A containing 0.2% Nonidet P-40 and 0.5 mM 4-(2-aminoethyl)benzenesulfonyl fluoride. The lysate was centrifuged at 8000 rpm for 10 s to pellet the Nonidet P-40-insoluble material, and the supernatant was removed. The pellet was resuspended into 50 l of Dignam buffer C, incubated on ice for 10 min with occasional vortexing to disrupt the nuclear membranes. Extracts were centrifuged for 1 min at 8000 rpm, the supernatants were removed, and the pellets were discarded. The protein content of the nuclear extracts was determined using the Bio-Rad DC assay kit (Bio-Rad).
DNase I Footprinting-The TIMP-1 minimal promoter (positions Ϫ102 to ϩ60) was 5Ј-end-labeled with [␣-32 P]dATP (3000 Ci/mmol, Amersham Pharmacia Biotech) using avian myeloblastosis virus reverse transcriptase (10 units; Promega). Nuclear extract (10 g) from HSCs was incubated with 1 g of poly(dI-dC) nonspecific DNA competitor (Sigma) for 10 min, and the radiolabeled TIMP-1 minimal promoter (100 ng) was then added and incubated for a further 20 min. DNase I (0.05 units; type 4, bovine pancreas; Sigma) was added to the reaction mixture and incubated for 1 min. All reactions were carried out at 4°C. DNase I footprinting reactions were resolved on a 8% denaturing poly-acrylamide gel alongside a purine marker ladder.
Electrophoretic Mobility Shift Assay (EMSA)-A 20-base pair oligonucleotide (UTE-1) was constructed to the region of the minimal promoter protected from DNase I digestion in the DNase I footprinting experiments. The UTE-1 sense strand oligonucleotide (5Ј-AGGCCTGT-GGTTTCCGCACC-3Ј) was 5Ј-end-labeled with [␥-32 P]ATP (3000 Ci mmol Ϫ1 ; Amersham Pharmacia Biotech) using T4 polynucleotide kinase (Amersham Pharmacia Biotech). The labeled UTE-1 sense strand was then annealed to the UTE-1 antisense strand (5Ј-GGTGCG-GAAACCACAGGCCT-3Ј). EMSA reactions were performed as described previously (31,38). The standard EMSA reaction consisted of an initial incubation of the nuclear extract (5 g) with 1 g of poly(dI-dC) in a total volume of 18 l for 10 min. The annealed radiolabeled oligonucleotide probe (2 l; 0.1 ng l Ϫ1 ) was added to the reaction and incubated for a further 20 min. All reactions were carried out at 4°C. For competition assays excess unlabeled oligonucleotide (2-20 ng; Table  I) was added to the reaction mixture with the nuclear extract and poly(dI-dC) for the initial incubation, prior to the addition the probe. EMSA reaction mixtures were resolved by electrophoresis on an 8% nondenaturing polyacrylamide gel (37:5:1).
Chloramphenicol Acetyltransferase (CAT) Reporter Gene Assay-A CAT reporter plasmid (pBLCAT3) containing the TIMP-1 162-bp minimal promoter (31) cloned into the HindIII and PstI site, upstream of the CAT gene, was used in all functional studies. The UTE-1 mutant sequence was generated by polymerase chain reaction-mediated mutagenesis of the wild type minimal promoter using the following oligonucleotide primers: mutant forward primer (5Ј-CCTGGAGGCCCAGT-AGCTCCACACCCGCTG-3Ј) and reverse primer (5Ј-CAAGCTGCAGC-CCAGCTCCGGTCCCTGCTG-3Ј). Primary human and rat HSCs (8-day activated) seeded at 1 ϫ 10 6 /well on a six-well plate and Jurkat T cells (5 ϫ 10 6 ) were transfected using 1 g of purified plasmid DNA (Qiagen Maxiprep, Qiagen) with the transfection reagent Effectene (Qiagen) according to the manufacturer's instructions and left in contact with the cells for 48 h. CAT assays were performed as described by Gorman (39). In brief, cell extracts were prepared by repeated freeze-thaw cycles and normalized for protein content using the Bio-Rad DC protein assay, and CAT activities were determined using [ 14 C]chloramphenicol (Amersham Pharmacia Biotech) and acetyl-CoA (Sigma). Acetylated products were separated by thin layer chromatography and quantitated by phosphor imaging. CAT activities were normalized to the amount of DNA taken up by cultures, determined using a modification of Hirt's assay (40).
Ultraviolet Cross-linking-The UTE-1 oligonucleotide was radiolabeled as described previously for EMSA. Nuclear extracts (10 g) were incubated with 2 g of poly(dI-dC) in a total volume of 8 l for 10 min; the oligonucleotide probe (2 l; 0.1 ng l Ϫ1 ) was then added to the reaction and incubated for a further 20 min. DNA-protein complexes were cross-linked by UV radiation for 1 h at 1200 KJ cm Ϫ2 ( ϭ 305 nm) and resolved by electrophoresis on 12.5% SDS-polyacrylamide gel (20 mA, 1 h). All reactions were carried out at 4°C.
Southwestern Blotting-The Southwestern blotting technique is a variation of the traditional Western blotting method (35). Following transfer of proteins onto a nitrocellulose membrane DNA binding proteins are detected with a radiolabeled DNA probe (UTE-1 oligonucleotide). Nuclear extracts (30 g) from rat activated HSCs, Jurkat T cells, and Daudi B cells were separated on a 12.5% SDS-polyacrylamide gel and transferred onto a nitrocellulose membrane. The UTE-1 DNA probe (15 ng) was prepared as described for EMSA studies and hybridized to the membrane at room temperature for 30 min.

Identification of the UTE-1 Sequence by DNase I Footprint-
ing-Potential binding sites for transcription factors within the TIMP-1 minimal promoter were identified by DNase I footprinting. Nuclear extract from 1-day and 8-day cultured rat HSCs were used for footprinting in an attempt to identify DNA-protein interactions that are induced during HSC activation. When using nuclear extract from activated 8-day cultured HSCs, a large proportion of the TIMP-1 minimal promoter was protected from the DNase I digestion in comparison with the control track (Ϫ) digested in the absence of nuclear extract (Fig.  1A). The protected regions included the potential Sp1 site, TATA-like box, and a previously undescribed 11-bp region downstream from the AP-1 and Pea-3 sites at nucleotides Ϫ63 to Ϫ53 of the human TIMP-1 gene (Fig. 1B). The 11-bp region was also protected from DNase I digestion by nuclear proteins isolated from 1-day cultured HSCs, although the intensity of the footprint was markedly reduced relative to that observed with 8-day cultured HSCs (Fig. 1A). The protected 11-bp sequence (5Ј-TGTGGTTTCCG-3Ј) was analyzed on sequence data bases (EMBL, GenBank TM , Transfac) in an attempt to identify possible matches with previously characterized transcription factor binding sites. Some sequence homology was found with the AP3 site (5Ј-TGGGACTTTCCA-3Ј) and an AP3-like (AP3-L; 5Ј-TGTGGAAAATCT-3Ј) site that resembles the nuclear factors of activated T cells (NFAT-1; 5Ј-AGGAGGAAAAACT-3Ј) binding site (41). No exact match was identified; therefore, the footprinted region was named upstream TIMP-1 Element 1 (UTE-1).
Investigation of UTE-1-binding Proteins by EMSA-UTE-1protein interactions were investigated using a double-stranded 20-base pair oligonucleotide (5Ј-AGGCCTGTGGTTTCCG-CACC-3Ј), constructed to span the UTE-1 element of the TIMP-1 minimal promoter (Table I). Since TIMP-1 is expressed by a variety of different cell types, it was of interest to establish the expression pattern of UTE-1-binding proteins in various mammalian cell lines as well as primary human and rat HSCs. Fig. 2A shows that a retardation complex was obtained with nuclear extracts from all cell types tested, although there were distinct qualitative and quantitative differences observed between the cells. Culture-activated primary rat HSCs and human Jurkat T cells gave rise to similar high mobility UTE-1 DNA-protein complexes, although the relative abundance was far greater for Jurkat T cells. Culture-activated primary human HSCs assembled a lower mobility complex that was of similar mobility and abundance to UTE-1 complexes observed with Daudi B cell and THP-1 monocyte-macrophage cell lines. Passaging of human HSCs, which gives a higher selectivity for HSCs over other cell types (Kupfer cells in particular) that contaminate the original primary culture, was associated with a change in the mobility and abundance of the UTE-1 binding complex. These changes generated a DNA-protein complex that was more similar to the complex observed with pure (95%) primary rat HSCs. These results suggest that UTE-1-binding proteins probably exist in most mammalian cell types; however, the nature and abundance of the DNA-protein complexes formed at promoters may vary between cell types.
Since we have previously reported that transcriptional activity of the human TIMP-1 promoter is induced during HSC activation (31) it was of interest to determine if UTE-1-binding proteins are also induced during the culturing of freshly isolated HSCs. For these experiments, we used rat HSCs previously shown to support transcriptional activity of the human TIMP-1 promoter (31). EMSA reactions with nuclear extracts from different time points of HSC activation on plastic revealed the existence of multiple DNA-protein complexes of dissimilar electrophoretic mobilities (Fig. 2B). Freshly isolated HSCs expressed multiple UTE-1 complexes that upon culturing were replaced by two complexes of lower mobility that persisted for the first 24 h of culture. Subsequent periods of culture were associated with the loss of the low mobility complexes and induction of at least one higher mobility complex that was maximally expressed at day 5 of culture and persisted for at least a further 9 days (data not shown). It is noteworthy that the appearance of the high mobility UTE-1 complex at 3 days and its maximal induction at 5 days are synchronous with the de novo expression and peak induction of TIMP-1 mRNA, respectively (30).
Mapping of Nucleotides Required for Protein Binding to UTE-1 and Comparison with AP3, AP3-L, and NFAT Binding Sites-Sequence specificity of the UTE-1 DNA-protein interaction was tested by competition EMSA reactions. A 50-fold excess of unlabeled wild type (wt) double-stranded oligonucleotide competed out all protein binding to the radiolabeled UTE-1 site using nuclear extracts from activated rat HSCs (Fig. 3A). Hence, binding of nuclear proteins to the UTE-1 sequence is saturable. By contrast, a mutant double-stranded oligonucleotide (M in Fig. 3, B and C)

UTE-1: A Novel Regulatory Sequence in the TIMP-1 Promoter
protein complexes. Similar data were obtained using nuclear extracts from Jurkat T cells and THP-1 monocyte-macrophages (data not shown). Binding sites for the AP3 and AP3-like transcription factors (41) have sequence homology with the UTE-1 site (Table I). To determine if UTE-1 binds these transcription factors, competition EMSA experiments were performed using a 50-fold excess of unlabeled double-stranded oligonucleotides carrying high affinity binding sites for NFAT, AP3, and AP3-L. Fig. 3A shows that the NFAT, AP3, and AP3-L oligonucleotides were unable to compete for UTE-1 DNA-protein interactions. Additional studies confirmed that partial competition of UTE-1 binding was achievable when using a 250 -500-fold excess of the AP3 and AP3-L oligonucleotides; by contrast, wt UTE-1 oligonucleotides totally competed out all binding at only a 5-fold molar excess (data not shown). These data indicate that the UTE-1 site is unlikely to interact with transcription factors previously shown to bind to AP3, AP3-L, or NFAT sites.
To determine nucleotides essential for UTE-1 binding, double-stranded oligonucleotides with 5-bp mutations (large scan, or ls1 to ls7) scanning across the UTE-1 site were synthesized such that they overlapped one another by 3 bp, apart from ls6 and ls7, which overlapped by just 1 bp (Table I). Lack of competition was observed for ls2 through to ls5, whereas ls1, ls6, and ls7 were able to compete as effectively as the wild type UTE-1 sequence (Fig. 3B). These data indicate that nucleotides within the 6-bp sequence 5Ј-Ϫ63 TGTGGT Ϫ58 -3Ј are essential for UTE-1 complex formation in rat HSCs. Similar data were obtained for UTE-1 DNA Jurkat T cells and THP-1 monocytemacrophages (data not shown), indicating that the 6-bp sequence is a general minimal recognition site for UTE-1 cellular binding proteins. To confirm identification of essential binding nucleotides of UTE-1-binding proteins, a second set of scanning mutant double-stranded oligonucleotides containing 2-bp (with 1-bp overlaps) mutations (point scan, or ps) across the sequence 5Ј-T Ϫ63 GTGGTT Ϫ57 -3Ј were employed. When using nuclear extract from activated rat HSCs as a source of UTE-1-binding protein mutant oligonucleotides ps1 to ps5 displayed reduced ability to compete for UTE-1 protein binding relative to the wild type sequence. By contrast, ps6 was an effective competitor. These data suggest that the sequence 5Ј-Ϫ63 TGTGG Ϫ59 -3Ј is the UTE-1 binding site for proteins expressed in activated rat HSCs. Similar fine mutagenesis mapping of the UTE-1 site using nuclear extracts from Jurkat T cells showed that all mutant oligonucleotides were poor competitors including ps6 (Fig. 3D). Lack of competition for protein binding on the wt UTE-1 site using mutant oligonucleotide ps6 was also observed with nuclear extracts from THP-1 cells (data not shown). Hence, a larger UTE-1 binding site than the sequence mapped for UTE-1 binding for rat HSCs appears to be required for interactions with proteins expressed in T cells and monocyte-macrophages.
Reporter Gene Assays Demonstrated That the UTE-1 Element Is an Essential Functional Site in the TIMP-1 Minimal Promoter-The functional importance of the UTE-1 element in transcriptional regulation of the TIMP-1 gene was investigated by transfection experiments using a CAT reporter gene driven by the minimal promoter (Ϫ102 to ϩ60) from the human TIMP-1 gene (31). Promoter constructs containing the wild type minimal promoter significantly activated transcription in culture-activated rat HSCs, human HSCs, and Jurkat T cells (Fig. 4A). By contrast, a TIMP-1 promoter-CAT reporter construct containing a mutated UTE-1 site (5Ј-Ϫ63 TGTGGTT Ϫ57 -3Ј to 5Ј-CAGTAGC-3Ј) was essentially inactive, displaying similar activity to that of pBLCAT3. Similar results were also observed when wild type and UTE-1 mutant promoter-CAT constructs

UTE-1: A Novel Regulatory Sequence in the TIMP-1 Promoter
were transfected into the THP-1 macrophage cell line, 2 confirming the general requirement of the UTE-1 sequence for TIMP-1 promoter activity. To further characterize the UTE-1 site, constructs containing mutations (5Ј-TGTTATT-3Ј and 5Ј-TGTGAGT-3Ј) affecting the critical core UTE-1 protein binding nucleotides G Ϫ60 G Ϫ59 were transfected into activated rat HSCs. Although the double point mutant constructs were more active than the completely scrambled mutant UTE-1 construct, both were significantly less active (approximately 70%) than the wild type promoter-CAT reporter (Fig. 4B). EMSA experi-2 J. E. Trim and D. A. Mann, unpublished observation. FIG. 3. A, transcription factor sites with similar sequences to the UTE-1 element do not compete for UTE-1-binding proteins. Nuclear extracts were prepared from 8-day activated HSCs and used at concentrations of 5 g in EMSA reactions containing 0.2 ng of radiolabeled double-stranded oligonucleotide UTE-1 probe in the presence of a 50-fold excess of unlabeled oligonucleotide. Negative (Ϫ; no nuclear extract) and positive (ϩ; 5 g of nuclear extract) control reactions were run alongside competition reactions. Competition EMSA reactions were carried out in the presence of nuclear extract and excess unlabeled wild type UTE-1, mutant UTE-1, NFAT, AP3, or AP3-L. B, large scan mutagenesis screening within the UTE-1 element by competition EMSA. 5-g aliquots of nuclear extract isolated from 8-day activated HSCs was incubated for 15 min either alone or with unlabeled wild type (wt) or mutant double-stranded oligonucleotides (10 ng). Mutant oligonucleotides carried either a complete scrambled mutation (M) or scanning 5-bp (lanes 1-7) overlapping mutations (5Ј to 3Ј) in the UTE-1 sequence (Ϫ68 to Ϫ49). Extracts were then incubated for a further 15 min after the addition of 0.2 ng of radiolabeled wild type UTE-1 oligonucleotides. The loss of UTE-1 protein-DNA complex was due to successful competition by the unlabeled oligonucleotides and indicates sequences that are not essential for UTE-1 binding. Point scan mutagenesis screening within the UTE-1 element by competition EMSA using HSC nuclear extracts (C) or Jurkat T cell nuclear extracts (D) is shown. 5-g aliquots of nuclear extract were used in competition EMSAs with 10 ng of unlabeled oligonucleotides containing 2-bp scanning mutations (overlapping by 1 bp) between nucleotides Ϫ62 and Ϫ57. All gels were representative of at least three experiments using nuclear extracts from three independent preparations. UTE-1: A Novel Regulatory Sequence in the TIMP-1 Promoter ments using radiolabeled mutant UTE-1 sites in direct binding assays confirmed lack of significant protein binding to the three mutant UTE-1 sites (M, M2, and M3) used in the CAT reporter studies (data not shown). We therefore conclude that sequencespecific protein-DNA interactions at the UTE-1 site are likely to be important in the regulation of TIMP-1 promoter activity.
Estimation of the Molecular Size of UTE-1 DNA-binding Proteins-The ultraviolet cross-linking technique allows visualization of DNA-protein complexes while separating these complexes according to size on a denaturing gel system (35). Proteins of different sizes bound to the UTE-1 oligonucleotide across the time course of stellate cell activation (Fig. 5A). The proteins binding to UTE-1 in quiescent cells were approximately 35 and 47 kDa. In 1-day activated HSCs, we only detected the 47-kDa protein. At 5 and 8 days of activation, there was an absence of both the 35-and 47-kDa proteins, and the appearance of a protein of approximately 30 kDa was observed. In the EMSA studies, it was observed that nuclear extracts from different cell types gave rise to distinct UTE-1 DNA-protein complexes ( Fig. 2A). Southwestern blotting demonstrated that double-stranded UTE-1 oligonucleotides can be used to detect a 30-kDa protein in nuclear extracts from rat HSCs, Jurkat T cells, and THP-1 macrophages (Fig. 5B). Levels of the 30-kDa protein appeared to be substantially higher in Jurkat and THP-1 cells compared with levels detected in rat HSCs. Control UV cross-linking and Southwestern blotting experiments using mutant oligonucleotide probes failed to detect the 30-kDa species; in addition, UTE-1-binding proteins detected by UV-cross linking were competed for by including an 50-fold excess of unlabeled UTE-1 oligonucleotides in the reaction mixture (data not shown). From these studies, we suggest that there may be a common sequence-specific UTE-1 DNAbinding protein present in mammalian cells. DISCUSSION The functional implication of TIMP-1 in a variety of diseases and in particular the up-regulation of TIMP-1 expression in human cancers (17)(18)(19)(20)(21) and cirrhotic liver (30,(42)(43)(44) warrants a detailed study of the factors controlling TIMP-1 expression in mammalian cells. We have previously reported a minimal active promoter for the human TIMP-1 gene that was sufficient to support a high level of transcription in culture-activated rat HSCs but was inert in freshly isolated or quiescent HSCs (31). We have reasoned that the minimal TIMP-1 gene promoter may be used to identify transcription factors with activities that are induced during HSC activation. Transcription factors already identified include the AP-1 family proteins JunD, Fra2, and FosB, which bind to an essential AP-1 binding site at the 5Ј-end of the promoter and that are induced with kinetics similar to those described for induction of TIMP-1 mRNA expression (31). In the present study, we have identified a novel cis-acting regulatory DNA element named UTE-1 that is located downstream of the AP-1 and Pea3 sites and appears to be essential for transcriptional activity of the human TIMP-1 promoter in a variety of cell types including activated rat and human HSCs. We suggest that UTE-1 is a new regulatory DNA element that is an important regulator of transcription of the human TIMP-1 gene.
A combination of DNase I footprinting and competition EMSA experiments delineated the critical nucleotide sequence Nuclear extracts were prepared from sister cultures at different time points of culture activation (0 (quiescent), 1 day, 3 days, 5 days, and 8 days) and used at concentrations of 10 g in UV reactions containing 0.2 ng of radiolabeled double-stranded oligonucleotide UTE-1 probe and then irradiated at 1200 kJ cm Ϫ2 ( ϭ 305 nm) for 1 h. Protein size was calculated using a prestained protein marker ladder. B, Southwestern blotting. 10 g of nuclear extracts prepared from 8-day activated primary rat HSCs, Daudi B cells, Jurkat T cells, and THP-1 macrophages were separated on a 12% SDS-polyacrylamide gel electrophoresis and transferred onto nitrocellulose filters. Filters were then probed for UTE-1-binding proteins using a radiolabeled double-stranded UTE-1 oligonucleotide probe, and the size of the protein recognized was estimated from a prestained protein marker ladder. Gels were representative of three experiments using three independent nuclear extract preparations.

UTE-1: A Novel Regulatory Sequence in the TIMP-1 Promoter
required for formation of UTE-1 DNA-protein interactions in activated rat HSCs as 5Ј-Ϫ63 TGTGG Ϫ59 -3Ј. Similar competition EMSA experiments with T cells and macrophages suggested that there may be a requirement for a larger binding site with utilization of additional nucleotides at the 3Ј-end of the site. The molecular basis for this apparent difference in binding site requirement between different cell types is not known. Functional importance of UTE-1 for transcriptional regulation of the TIMP-1 gene was demonstrated by the fact that UTE-1 site mutations that prevented protein binding also dramatically reduced promoter activity in the context of transfected TIMP-1 promoter-CAT reporter constructs.
Of major interest was the characterization of nuclear proteins that interact specifically with UTE-1, since these factors are likely to be important regulators of TIMP-1 gene expression in mammalian cells. Although the UTE-1 site has sequence similarities with previously described SV40 AP3 and human immunodeficiency virus-1 AP3/NFAT-like regulatory DNA elements, its binding specificity was distinct from these sequences, and therefore it is unlikely to interact with AP3 or NFAT proteins (41). The lack of strong competition by the AP3-like sequence was surprising given the presence of the sequence 5Ј-TGTGG-3Ј, which is identical to the core UTE-1 binding site. Direct binding studies failed to detect assembly of additional protein-DNA complexes on the AP3-like site that could compete for recruitment of UTE-1-binding proteins. 3 We therefore suggest that the existence of sequence differences in the nucleotides flanking the core UTE-1 and AP3-like sites is the most likely explanation for the AP3-like site acting as a poor competitor for UTE-1 binding.
Southwestern blotting demonstrated that rat HSCs, Jurkat, and THP-1 cells express a 30-kDa protein that directly interacts with the UTE-1 sequence. Moreover, UV cross-linking of UTE-1 DNA and nuclear proteins derived from rat HSCs revealed that a 30-kDa species is induced during the culture activation of these cells. Induction of the 30-kDa species also coincided with the expression of HSC nuclear proteins responsible for assembly of high electrophoretic mobility UTE-1 binding complexes. We propose that the 30-kDa protein detected by Southwestern blotting is responsible for the 30-kDa UV-crosslinked complex and the inducible EMSA complexes observed during HSC activation. EMSA studies suggested that HSCs, T cells, and monocyte-macrophages assemble UTE-1 DNA-protein complexes of different mobilities. It is tempting to speculate that the 30-kDa UTE-1-binding protein identified in each of these three phenotypically distinct cell types either undergoes cell-specific modifications such as protein-protein interactions or alternatively may display cell-specific charge/conformational differences that are lost in the denaturing SDSpolyacrylamide gels.
Function of the TIMP-1 promoter in HSCs requires intact AP-1 and UTE-1 regulatory DNA elements and is only active in fully activated or myofibroblast-like HSCs that express the AP-1 proteins Fra2, FosB, and JunD (31). Coordinated induction of Fra2, FosB, JunD, and the 30-kDa UTE-1-binding protein and the interaction of these proteins with their appropriate binding sites within the minimal active TIMP-1 gene promoter are likely to be key regulatory mechanisms controlling expression of TIMP-1 in activated HSCs. Since the AP-1 and UTE-1 sites are also required for promoter activity in a variety of other cell types, it will be important to direct future studies to understanding the way in which AP-1 and UTE-1 binding proteins cooperate to regulate promoter activity. In addition, cloning and functional characterization of the 30-kDa UTE-1 binding protein we have identified will be a future priority. Such studies will improve our understanding of the transcriptional mechanisms responsible for controlling TIMP-1 expression in normal and pathological states.