INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Removal of ECM¹ structures is an important component of normal tissue remodeling, and excessive or inappropriate ECM degradation is the basis of a number of pathologies. The major effectors of ECM degradation belong to a family of structurally related enzymes called matrix metalloproteinases (MMPs) (1). MMPs share several properties that reflect the multiple levels at which their activity is regulated. These include secretion in an inactive pro-form requiring activation by proteolytic cleavage, requirement for zinc in the active site, and capacity of the active enzyme to be inhibited by members of the TIMP (tissue inhibitor of metalloproteinases) family (2). Currently, there is much interest in the development of drugs to inhibit MMP activity that take advantage of the shared biochemical properties of these enzymes (3).

Although MMP activity is controlled at multiple levels, resident cells of a tissue typically do not make MMPs unless they are needed for remodeling (4). Therefore, regulation of MMP gene expression offers a second possible drug target for controlling inappropriate MMP activity. Expression of MMPs is controlled primarily at the level of transcription (5). A large number of agents and conditions stimulate MMP transcription (6, 7), but only a few with inhibitory properties are known (4, 5). Currently, identification of MMP transcriptional inhibitors can be made in cultured cell models. However, to begin evaluation of compounds with transcriptional inhibitory properties for efficacy in specific disease processes, animal models must be developed that make use of rapid and simple assays. One approach would be to introduce easily assayed reporter genes under control of a specific MMP promoter into transgenic mice. Development of this type of model would be able to rely on the extensive work characterizing the requirements for MMP promoter activity in cultured cells; however, the promoter requirements are likely to be far more complex in animals. We are aware of no studies characterizing promoter sequences in mice for any MMP gene to date.

Gelatinase B (MMP-9) was one of the first members of the MMP family to be characterized (8). This MMP is reactive against native type IV and V collagens, elastin, fibronectin, and denatured collagens of all types (9, 10). At first thought to be specific for polymorphonuclear leukocytes (11) and macrophages (12), it is now known to be expressed by the resident cells of a variety of different tissues (13-16). Expression of the gelatinase B (gelB) gene in tissues is developmentally regulated. Major sites of expression include trophoblast cells surrounding the implanting embryo (17) as well as the brain and skeletal elements of the embryo proper (18-20). In adults, gelB is expressed in brain (21) as well as in bone osteoclasts (18, 22). Expression in adults is also inducible at a number of sites by stress stimuli. For example, gelB is not expressed in the epidermis or dermis of adult skin; however, expression is induced in the basal cell layers of the epidermis (23, 24) or corneal epithelium (25) migrating to resurface a wound. Much like other MMPs, inappropriate or excessive gelB expression is associated with a number of pathological conditions. Sites of pathological expression include the epithelium of ulcerating corneas (26, 27), the brain of Alzheimer's disease patients (21), the tumor cells of metastatic cancers (28), the vessel walls of aortic aneurysms (29), and smooth muscle cells and fibroblasts surrounding inflamed vessels in patients with giant cell arthitides (30).

In cell cultures, gelB expression is stimulated by growth factors and inflammatory cytokines such as transforming growth factor-, interleukin-1, and tumor necrosis factor- and tumor-promoting phorbol esters such as TPA (4). These stimulators also coordinately induce other MMP genes such as collagenase and stromelysin; however, the signaling pathways for gelB appear to be different (31).² The DNA flanking the 5'-end of the gelB gene is highly conserved across the human, mouse, and rabbit species from the transcription start site (+1), upstream through base 600. In common with many other MMPs, a canonical TPA response element that binds transcription factor AP-1 is found close to the transcription start site (33); a second TPA response element is located near the distal end of the conserved region (34-37). At least one of these TPA response elements must be intact for promoter activity in cultured cells (37). Other sequence motifs that confer response to transcription factors NF-B, Sp1, and AP-2 (35-38) play a part in the unique expression characteristics of the gelB gene.

In this report, we advance MMP promoter characterization to the next level by investigating the regulatory sequences of the gelB promoter that confer expression in transgenic mice. We show that sequences between 522 to +19 of the gelB promoter, as characterized in the transgenic mouse line 3445, are necessary and sufficient for appropriate developmental and wound-inducible expression of a lacZ reporter gene. We further show that the expression and activity of three transcription factors that control the gelB promoter in cultured cells (NF-B, AP-2, and Sp1) are selectively induced in the epithelium migrating to close the wound in vivo. Although promoter activity parallels expression of the endogenous gene in cell cultures, we show that cell cultures cannot model many aspects of promoter regulation in vivo. These studies suggest that the transgenic mouse line 3445 might be a useful model for investigating the unique regulation characteristics of gelB gene expression and for identifying and characterizing new drugs that can control gelB gene transcription.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Cell Culture-- The rabbit synovial cell line HIG-82, a gift from Dr. C. H. Evans (39), was cultured as described (14). Primary corneal epithelial cells were isolated and cultured as described (14, 31). The rabbit corneal "epithelial-like" cell line SIRC (American Type Culture Collection, Rockville, MD) and the mouse keratinocyte cell line PAM-212 (a gift from Dr. Paolo Dotto, Massachusetts General Hospital) were cultured as recommended by the suppliers.

Construct Development and Transient Transfection Analysis-- Construct Pr43 was created using a two-step cloning strategy. A 541-base pair fragment extending from 522 to +19 of the rabbit gelB gene was PCR-amplified from rabbit genomic clone p36 (37) using the Puc/M13 forward primer (5'-GTA AAA CGA CGG CCA GT-3'; Promega, Madison, WI) and primer BF6 (5'-CGC GGG CTC GAG GGT GAG GGG AGA AGC GCT-3') complementary to bases +19 to +2, with an XhoI site and a GC clamp added to the 5'-end. The resulting PCR product was cut with EcoRI (base 522) and XhoI and cloned into the pNASS expression vector (Promega). This construction is called Pr22. The 4666-base pair fragment from Pr22 containing the 522 to +19 gelB promoter/lacZ fusion gene was next cut with EcoRI/HindIII and cloned into pcDNA3 (Invitrogen, San Diego, CA). The cytomegalovirus promoter from pcDNA3 was then removed by digestion with NruI and HindIII; the remaining plasmid ends were blunt-ended; and the ends were religated to produce Pr43 (see Fig. 1).

Development of Stably Transfected Keratinocyte Cell Lines-- Pr43 plasmid DNA was transfected into SIRC and PAM-212 cells using LipofectAMINE reagent (Life Technologies, Inc.) following the procedure of the manufacturer. Cells were cultured in medium containing the antibiotic drug G418 (Geneticin, Life Technologies, Inc.) at 0.625 mg/ml. Drug-resistant stably transfected cells were identified as discrete surviving colonies after 1 month of selection in culture. To determine whether transfected cells expressed the -gal reporter gene, cells in one of the transfected dishes were fixed in 0.25% glutaraldehyde and stained with X-gal as described (40). About 80% of the drug-resistant colonies stained blue. Single drug-resistant colonies from SIRC and PAM-212 cells were isolated and amplified, and clones expressing -gal were used in cell culture studies. For the experiments, cells were plated in six-well cluster plates at equal density. The following day, the cells were washed in serum-free medium, and each well was replenished with 1 ml of serum-free medium with and without 1 µM TPA (Sigma). The conditioned media were collected after 24 h, and equal volume samples were analyzed by gelatin zymography. gelB was quantitated by densitometry according to our standard methods (41). Equal amounts of protein sample from the control and TPA-treated cell cultures were used for -gal assays using o-nitrophenyl--D-galactopyranoside as the substrate and a spectrophotometric absorbance of 420 nm, according to a standard protocol (40).

Development of Transgenic Mice-- To prepare DNA for generation of transgenic mice, plasmid Pr43 was cut with EcoRI and HindIII, and the larger 4666-base pair band containing the gelB promoter and lacZ gene was gel-purified from agarose using the Glassmax kit (Life Technologies, Inc.). A similar strategy was used for Pr34 and Pr42. Plasmid Pr35 was cut with NarI and HindIII, and the gelB promoter/lacZ fusion gene fragment was purified for microinjections. Transgenic mice were generated in the core facilities of the Yale Cancer Center (New Haven, CT) and Beth Israel Hospital (Boston, MA). Briefly, fertilized F2 hybrid embryos obtained from matings of C57BL6 and SJL mice were injected with DNA fragments (2 µg/ml) and transferred to pseudopregnant Swiss outbred foster mothers. Putative transgenic founder pups born from these foster mothers were analyzed when the pups reached 5 weeks of age. Pups were screened for integration of the transgene into their genomes by Southern blot analysis of DNA from tail biopsies using a -gal DNA fragment as a probe. To determine whether the integrated DNA was transmitted to progeny, founder mice were mated with wild-type CD-1 mice, and tail DNA from the F1 generation progeny was analyzed by PCR. One to two micrograms of tail DNA was subjected to PCR using a forward primer (RM62, 5'-CAC CGG CCC TGA GTC AGC ACT-3') complementary to the rabbit gelB gene promoter sequences between 71 and 91 and a reverse primer (BF1, 5'-CCT CAG TGG ATG TTG CCT TTA CTT-3') specific to the lacZ gene. A positive result was ascertained by the presence of an amplified DNA band of 245 base pairs. The transgene in all cases was found to be incorporated into the germ line since it could be transmitted to offspring according to mendelian genetics, as determined by mating positive heterozygotes with wild-type CD-1 mice (data not shown).

Skin and Corneal Surgeries-- Adult transgenic mice (6-8 weeks old) as well as non-transgenic littermates were anesthetized by an intraperitoneal injection of 2% Avertin (0.017 ml/g of body weight). The back hair from the mice was shaved in a region of 1 cm², and a full-thickness dermal wound (3-4 mm in diameter) was made using a pair of surgical scissors. Bactericidal ointment (Mycitracin, The Upjohn Co.) was applied to the wound, and the mice were allowed to recover under a warming lamp. The mice were sacrificed 1-3 days after surgery, and tissues were processed for -gal staining or histology. A control region of hair was shaved and/or wounded prior to sacrifice. In this case, tissues were excised immediately and processed along with experimental tissues.

Assay of Transgene Expression-- Embryos from staged matings between transgenic heterozygous mice and wild-type CD-1 mice were collected from sacrificed females, and tissues were fixed in buffered 4% paraformaldehyde for 25 min, washed three times in PBS, and stained as whole mounts in buffered 2% X-gal solution at 30 °C for 5-12 h (40). Embryos past embryonic day 15 (E15) were cut sagittally in half and processed. Adult organs or tissues were surgically removed from the mice, cut into smaller pieces, and fixed and stained as described above. After staining, the tissues were washed in PBS, refixed in 4% paraformaldehyde, and photographed immediately. Some tissues were then embedded in paraffin, sectioned (6 µm), and stained with hematoxylin and eosin.

Electrophoretic Mobility Shift Assay (EMSA)-- To collect tissue to prepare nuclear extracts, rabbits were anesthetized, transferred to a cold room, and sacrificed by intracardiac injection of phenobarbital. The corneal epithelium within a 9-mm centrally demarcated region from wound-healing (six) eyes or uninjured (six) eyes was removed by abrasion with a No. 10 Bard Parker surgical bade. Tissue was transferred to a 1.5-ml microcentrifuge tube containing 0.6 ml of ice-cold 10 mM HEPES, pH 7.9, 10 mM KCl, 0.1 mM EDTA, 0.1 mM EGTA, 5 mM NaF, and 1 mM each dithiothreitol, phenylmethylsulfonyl fluoride, benzamidine, aprotinin, spermidine, antipain, and leupeptin. The pooled wound-healing epithelial tissue and separately pooled control tissue were homogenized on ice with the aid of a microcentrifuge homogenizer. The remaining steps essentially followed the procedure of Dignam et al. (42). Equal aliquots of equal protein were frozen at 70 °C. Nuclear extracts were prepared from corneal epithelial cells in a similar manner.

Immunolocalization Studies-- Frozen serial sections of wound-healing mouse eyes were processed for immunostaining according to standard methods (44). Antibodies (Santa Cruz Biotechnology) were diluted in PBS (anti-Sp1, 1:500; anti-AP-2, 1:100; and anti-NF-B (p65), 1:500). An anti-fibronectin antibody, R61 (a gift from Dr. Richard Hynes, Massachusetts Institute of Technology), was diluted 1:400 in PBS. Sections were blocked in normal goat serum, and antibodies (or normal rabbit serum) were applied to the sections for a 3-h incubation at room temperature. After washing, sections were incubated with fluorescein isothiocyanate-conjugated goat anti-rabbit IgG (diluted 1:50 in PBS) for 45 min and then washed again. Slides were photographed by epifluorescence.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Development and Analysis of SIRC-43 and PAM-43, Two Stably Transfected Cell Lines Containing the gelB Promoter/lacZ Fusion Gene-- We recently demonstrated that the DNA sequence between bases 519 and +19 of the rabbit gelB gene promoter drives expression of a chloramphenicol acetyltransferase reporter gene transiently transfected into cultured cells in a manner similar to the endogenous gelB gene (37). The base position at 519 interrupts an NF-B-binding motif located between 522 and 514, which was shown to be important for human gene response to tumor necrosis factor- (36). Therefore, we designed a new reporter gene construct for the current studies, Pr43, which includes 3 additional bases up to 522 (Fig. 1). We also replaced the chloramphenicol acetyltransferase reporter gene with LacZ to facilitate localization of expressing cells and utilized a new cloning vector containing a neomycin aminoglucotransferase gene for selection of stably transformed cells. To analyze the relationship between endogenous gelB gene expression and promoter activity of Pr43 in parallel, we developed two stably transformed keratinocyte cell lines, SIRC-43 and PAM-43. When plated in culture wells at equal density, each cell line exhibited constitutive -gal activity, but at different levels. When treated with TPA, a 2-fold induction of -gal activity was observed in SIRC-43; a much larger 8-fold induction was observed in PAM-43 cells (Fig. 2, left panels). Zymographic analysis of conditioned media from the same cultures used for -gal analysis revealed gelB in the proenzyme form. In the rabbit corneal epithelial cell line SIRC-43, gelB ran as a doublet of glycosylated and non-glycosylated forms with the same relative mobility as the rabbit gelB standards. In the mouse epidermal cell line PAM-43, gelB ran as a single 105-kDa band, as expected for mouse gelB (Fig. 2, right panels). The endogenous gelB gene was induced only 2-fold in SIRC-43 cells, similar to the -fold induction of the -gal reporter gene. In contrast, gelB expression was induced dramatically by TPA in PAM-43 cells, from essentially undetectable levels to a level as high as in TPA-treated SIRC-43 cells. These data indicate that the 522 to +19 gelB promoter drives -gal reporter gene expression in cultured keratinocytes in a manner closely paralleling the endogenous gelB gene.

View larger version (9K): [in this window] [in a new window]   Fig. 1.   Schematic diagram of Pr43. The DNA between 522 and +19 of the rabbit gelB gene was fused to the lacZ gene in pcDNA3, which contains the neomycin aminoglucotransferase gene for G418 drug resistance. The functionally characterized transcription factor response elements in the promoter are shown; base numbering is with respect to the transcriptional start site (+1) of the rabbit gene. These elements were characterized in the following genes: AP-1, human and rabbit genes; PEA3, human gene; NF-B, human and mouse genes; proximal Sp1, mouse and human genes; distal Sp1, human gene; AP-2 sites (filled boxes), rabbit gene. Two AP-2-like sites (cross-hatched boxes) fit the binding sequence consensus, but have not been functionally characterized. The references for promoter characterization are as indicated: human gene (36, 38), mouse gene (35), and rabbit gene (37).

View larger version (69K): [in this window] [in a new window]   Fig. 2.   Comparison of gelB promoter activity and endogenous gelB expression in stably transfected keratinocyte cell lines. Rabbit SIRC-43 or mouse PAM-43 cell clones were cultured in equal number in the wells of a six-well plate. Cells were left untreated or treated with TPA for 24 h. Left panels, -gal activity in equivalent samples of cell lysates prepared from SIRC-43 cells (upper panel) and PAM-43 cells (lower panel). The bars indicate S.D. (n = 4). Right panels, gelatin zymography of conditioned media samples (equal volumes) from the cell cultures that were used to make lysates. Rabbit gelB standards, as well as reduced molecular mass standards, were run in parallel for sizing of the zymogram bands (not shown). Rabbit gelB migrates at 92 kDa, which is observed as a doublet of glycosylated and non-glycosylated forms, whereas mouse gelB migrates at 105 kDa.

Embryonic Expression Pattern of the gelB Promoter/lacZ Fusion Gene in the Mouse Line 3445-- We next tested the activity of the 522 to +19 gelB promoter fragment cloned into Pr43 in transgenic mice. Injection of the excised EcoRI-HindIII fragment from Pr43 containing the gelB promoter/lacZ fusion gene (Fig. 1) into the pronuclei of fertilized mouse eggs resulted in five transgenic founders. Founders were mated with wild-type mice, and the resulting F1 embryos were analyzed for developmental -gal expression patterns by histochemical staining of whole mounts. Genetics predicts that approximately half of these embryos would be of the heterozygote genotype and half of the wild type; the latter provided an internal negative control for assessment of endogenous -gal activity (not due to the transgene). Two of the five transgenic lines were found to express the transgene. Southern blotting revealed that both mouse lines contained between 1 and 10 copies of the transgene (data not shown). In Figs. 3 and 4, we show examples of the expression pattern of line 3445, which was the strongest; expression of line 3433 was similar but weaker.

View larger version (100K): [in this window] [in a new window]   Fig. 3.   Developmental pattern of gelB promoter activity in embryos of transgenic mouse line 3445. gelB promoter activity was detected by histochemical localization of -gal expression in whole mounts of E11 (a-c) and E13 (d) embryos. a, strong -gal expression was observed in the ventral regions and developing mesencephalon of the brain (arrowhead) and in neural tube and schlerotomic fissures of the somites (arrow) at E11. b, shown is a non-transgenic littermate showing the complete lack of -gal activity. c, illustrated is a dorsal view of the embryo shown in a. d, strong staining occurs in limb bones (arrow) and along vertebrae (arrowhead) at E13.

View larger version (103K): [in this window] [in a new window]   Fig. 4.   Localization of gelB promoter activity to highly vascularized areas of developing cartilage, bone, and neural tissue in embryos of mouse line 3445. gelB promoter activity was detected by histochemical localization of -gal expression in whole mounts of E15 embryos (a, b, and d) and thin sections cut from E15 (c, e, f, and h) and E13 (g, i, and j) -gal-stained whole mounts counterstained with eosin or with hematoxylin and eosin. a, whole mount cut in half sagittally reveals staining in ventricular aspects of the brain and in the rib bones (arrow), nasal cavity, jaw structures, and developing vertebrae (arrowhead). b, a higher magnification of the snout portion of the embryo shown in a reveals intense staining localized to developing bone surrounding the emerging incisors (white and black arrows). c, sagittal section through the snout area reveals -gal expression in osteoclasts around the incisor (arrow) and in mesenchymal tissue (arrowheads). d, close view of whole mount reveals -gal staining in the metatarsi (white arrow) and in the tips of the digits (black arrowhead) of the paw. e, sagittal section through the forelimb shown in d reveals strong -gal activity in vascularized regions around the developing bones of the fingers (arrowhead). f, tangential section through the digit of the paw shown in d reveals -gal staining in developing muscle (arrowhead) and in mesenchymal tissue (arrow) of the digits. g, sagittal section reveals -gal expression in the mesenchymal cells (arrowheads) of the fissures surrounding the posterior somites (arrow) of the embryo. h, section through the umbilical vein shows -gal staining of vascular endothelial cells (arrowhead) and smooth muscle cells (arrows). i, sagittal section through the brain shows strong -gal expression in the modified ependymal cells (arrow) in the choroid plexus of the fourth ventricle and in mesenchymal cells of developing vasculature. A red blood cell is indicated by the arrowhead. j, sagittal section through choroidal fissures (arrowheads) extending to the hippocampal area of the brain. i, incisor; pc, perichondrium; mt, metatarsus; cb, cerebellum; ch, choroid plexus; t, tongue; mes, mesenchyme.

Comparison of gelB Promoter/lacZ Fusion Gene Activity in Migrating Epithelial Cells in Culture and in Mice-- MMPs are expressed in the migrating sheet of the epithelium during wound healing in skin or cornea. Migrating epithelial sheets can be modeled in cell culture by plating cells at cloning density; as the colony expands, the cells migrate outward as a sheet. We stained single colonies of SIRC-43 and PAM-43 cells to learn whether the gelB promoter activity shows differential expression at the edge of the colony (where cells are migrating outward) versus the center of the colony (where cells are stationary). However, staining was found to be uniform in all cells of the colonies examined, with no specific up-regulation at the migrating edge or down-regulation near the center. An example of this result using the SIRC-43 cells is shown in Fig. 5a.

View larger version (78K): [in this window] [in a new window]   Fig. 5.   Localization of gelB promoter activity in migrating colonies of SIRC-43 cells and in injured tissues of mouse line 3445. gelB promoter activity was detected by histochemical localization of -gal expression. a, shown is a portion of a SIRC-43 cell colony in culture. The arrow indicates the migrating cell front of the colony. b, shown is a full-thickness skin wound after 2 days of healing. The circular pink region is the exposed granulation tissue. The arrowhead indicates the migrating epithelium at the wound edge, which stains for -gal. The arrows indicate -gal staining in hair follicles within the area shaved before surgery was performed. c, non-transgenic mouse skin healing for 3 days after wounding fails to show -gal staining at the wound edge or in the muscle layer under the skin (arrowheads). d, thin section of X-gal-stained 3-day-old skin wound reveals specific -gal expression in the migrating basal cells of the epidermis, in wound fibroblasts beneath the remodeling basement membrane, in sebaceous glands (sg), and in hair follicles (hf). The arrow in d indicates the edge of the migrating epidermal sheet. e, illustrated is an section adjacent to the one shown in d, counterstained with hematoxylin and eosin. The arrow indicates the edge of the migrating epidermal sheet.

Sequence Requirements for gelB Promoter/lacZ Promoter Activity in Epithelial Cell Culture and in Vivo-- We previously examined the effects of progressive 5' to 3' cutback of the DNA sequences between 519 and 92 on promoter activity in cultured cells (37). Deletion to base 438, removing regulatory elements up to and including the distal AP-1-binding site, resulted in a dramatic loss of promoter activity. However, continued deletion to base 92 restored promoter activity. This suggested that DNA sequences between 438 and 92 might bind factors that inhibit the activity of the gelB promoter in cultured cells. However, we also showed that this DNA contains binding sites that enable positive response to transcription factor AP-2.

View larger version (21K): [in this window] [in a new window]   Fig. 6.   Characterization of functional sequences in gelB gene promoter in vitro and in vivo by deletion analysis. The gelB/lacZ constructs analyzed in this study are depicted to the left, with the approximate positions of known transcription factor-binding sites indicated schematically. The columns labeled In Vitro tabulate results of transient transfection analysis in untreated (Basal) cultures of SIRC and HIG-82 cells and in HIG-82 cells treated with TPA. The values indicate -gal expression relative to the expression measured for Pr22, which was arbitrarily set to 1. *, to maintain uniformity of plasmids for transient transfection analysis, construct Pr22 was used instead of Pr43. Values represent the mean of results obtained from three replicate culture wells. The columns labeled In Vivo summarize expression characteristics of the gelB/-gal constructs in transgenic mice. The ratio of expressing to non-expressing mouse lines is indicated.

Transcription Factor Activity in Epithelial Cell Cultures and in Vivo-- gelB is constitutively expressed in cultures of corneal epithelium, but its expression in adult normal corneal epithelium is absent (14, 25, 26). To learn whether differences in specific transcription factor activities could help explain this finding, we compared nuclear extracts from primary cultures of rabbit corneal epithelial cells with extracts from corneal epithelial tissues taken directly from rabbit eyes.

View larger version (69K): [in this window] [in a new window]   Fig. 7.   DNA binding characteristics of AP-2 and NF-B transcription factors from corneal epithelial cells and corneal epithelium. EMSA was performed using lysates prepared from cultures of rabbit corneal epithelial cells (Epi Cell Culture) or from epithelia isolated directly from rabbit corneas (Epi Tissue). A, analysis with canonical AP-2 or NF-B probes. The arrows indicate specific DNA-binding complexes. The arrowheads indicate the positions of the antibody-supershifted complexes. The asterisks indicate nonspecific complexes that are not competed well with unlabeled probe or supershifted with antibody. An overexposure of the autoradiogram (Epi Cell Culture, left panel) was necessary to display the AP-2 supershifted bands. B, analysis with two probes derived from the gelB gene (RM70/71 and RM72/73) that contain binding sites for AP-2. The arrows identify the major AP-2·DNA complex formed; the minor complex identified by asterisks is nonspecific. The arrowheads indicate the positions of the antibody-supershifted complexes. 1X = 200 ng of antibody. Additional minor nonspecific bands were seen in EMSA with RM72/73 because the autoradiogram was exposed longer to film.

Transcription Factor Activity in Normal and Migrating Epithelia in Cell Culture and in Vivo-- As a first step to elucidate mechanisms of wound-induced transcriptional activation of gelB expression, we compared the activity, expression, and localization of specific transcription factors involved in gelB promoter regulation in normal rabbit corneal epithelium and in epithelium migrating to heal a wound. Analysis was done on rabbit corneas (rather than mouse) to take advantage of the larger tissue area. EMSAs of nuclear lysates derived from normal and migrating corneal epithelia revealed similar banding patterns for transcription factors Sp1, AP-2, NF-B, and AP-1 in both normal and migrating corneal epithelia (Fig. 8A). In the case of Sp1, a new minor band appeared when analysis was done with nuclear lysate from migrating tissue that was not seen when nuclear lysate from normal tissue was used. The overall level of DNA-binding activity was quite striking for the Sp1 probe. DNA-binding activity for transcription factors AP-2 and NF-B was also up-regulated, although not to as dramatic a level as for Sp1. The DNA-binding activity of AP-1 was roughly equivalent in both wound-healing and normal corneal epithelia.

View larger version (56K): [in this window] [in a new window]   Fig. 8.   Distribution, expression, and activity of transcription factors in wound-healing corneal epithelium. A, EMSA showing the complexes formed between the canonical probes indicated and lysates of epithelial tissue derived from control, uninjured rabbit eyes (C) or lysates of migrating epithelial tissue from wounded eyes (W). The position of the minor Sp1 complex that appears in EMSA performed with extracts of migrating tissue is denoted by the arrow. B, immunofluorescent localization of transcription factors Sp1, AP-2, and NF-B in migrating epithelia from healing mouse corneas. panel a, Sp1; panel b, AP-2; panel c, NF-B; panel d, fibronectin (defines the wound bed). The arrows point to the tip of the leading edge of the resurfacing epithelium, and the arrowheads indicate the site of the original wound margin. Note that the migrating epithelium shown in panel c detached from the basement membrane during tissue processing.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

In this paper, we show that the DNA sequences contained between 522 and +19 of the rabbit gelB gene constitute a minimal in vivo promoter that drives appropriate reporter gene expression in stably transfected keratinocyte cell lines and in transgenic mice. The pattern of -gal reporter gene expression captured in the transgenic mouse line 3445 faithfully depicts the spatial and temporal developmental pattern of endogenous gelB gene expression as demonstrated by in situ hybridization analyses reported by several different laboratories (17-20, 22). The mouse line 3445 also displays appropriate injury-specific induction of reporter gene expression (23-27, 45).

Transgenic mice allow the study of promoter characteristics in situations for which cell culture models clearly cannot substitute, such as developmental regulation in the complex environment of the embryo (46). However, cell culture models are often used to represent aspects of wound healing such as keratinocyte migration (47). Here, we show some of the deficiencies of cell culture models for characterization of MMP promoter activity. First, we could not create a situation in cell culture in which the gelB promoter was entirely inactive, as it is in intact corneal epithelium or skin epidermis. The activity of the 522 to +19 promoter in the stably transfected keratinocyte lines SIRC-43 and PAM-43 very closely paralleled the activity of the endogenous gelB gene, but this gene was expressed constitutively, as occurs in keratinocytes migrating to heal a corneal (25-27) or skin (23, 24) wound in vivo. Second, analysis at the single cell level revealed that gelB promoter activity was uniform throughout cultures; it was not increased at the edges of migrating colonies or reduced at their centers. Yet the same promoter was found to drive -gal expression specifically in the migrating epithelium during re-epithelialization of injured skin and corneas of line 3445 transgenic mice, whereas the gelB promoter remained inactive behind the migrating cells in the region of uninjured epidermis or the corneal epithelium. Third, transient transfection analysis revealed that much of the DNA between 522 and +19 of the promoter could be eliminated without compromising basal promoter activity, whereas all of the DNA was required for promoter activity in vivo. These differences in promoter activity when analyzed in cell culture and transgenic mice may reflect differences in transcription factor expression and activity. In fact, we show that the corneal epithelium differs from cultured epithelial cells with respect to the types of complexes that form on an AP-2 DNA-binding probe. The differences in complex size may indicate different protein composition, which might be important in determining the activity of the gelB promoter. Overall, this work emphasizes the drawbacks of cell culture for analysis of the factors that control promoter activity in vivo.

Binding sites for transcription factor AP-1 are essential for gelB promoter activity in cultured cells (35, 37, 38), as is true for several other MMP genes. In addition, binding sites for transcription factors NF-B, PEA3, and Sp1 determine promoter response to specific stimuli (35, 36, 38). All of these elements are clustered within two regions at the proximal or distal ends of the 522 to +19 promoter region characterized in this study. The stretch of DNA between these clusters (438 to 92) is not essential for promoter activity in cell culture (37); nevertheless, much of the sequence is highly conserved among humans, mice, and rabbits, suggesting functional importance. In fact, we previously showed that DNA sequences within a portion of this region (423 to 330) confer selectivity of promoter activity for cultured corneal epithelial cells as opposed to corneal fibroblasts, and we localized this activity to three response elements for AP-2 (37). More important, we show that these internal sequences are essential for the activity of the promoter in transgenic mice. Furthermore, this requirement is not confined to the 423 to 330 internal sequences that contain the characterized AP-2 sites: all of the internal promoter region is required. The concept that emerges from this work is that the in vivo situation imposes more stringent requirements; multiple cis-acting elements in the gelB promoter are therefore required to provide sufficient impetus for promoter induction.

The cornea offers an unusually accessible model to study transcription factor activity in wound healing, as the epithelium can be rapidly and easily isolated as a pure tissue from both normal and healing corneas. We utilized this model to begin analysis of factors that could determine gelB promoter induction in the migrating epithelium. By EMSA, we demonstrated constitutive DNA-binding activity in nuclear extracts from the normal, uninjured corneal epithelium for four of the defined regulatory elements in the gelB promoter: AP-1, NF-B, Sp1, and AP-2. In the migrating epithelium, the amount of binding activity for the AP-1 probe present in nuclear extracts was the same as in the stationary epithelium. However, binding activity for NF-B, Sp1, and AP-2 DNA probes was clearly elevated. For NF-B, increased nuclear translocation of the existing p65 subunit appeared to occur in the migrating epithelium, explaining the induction of binding activity. Sp1 and AP-2 were already localized to nuclei; however, induced binding activity could be explained by a striking increase in the amount of these proteins in nuclei of migrating cells. The composition of the DNA-binding complexes and their relative capacity to stimulate transcription still needs to be examined. Nevertheless, these results suggest that the combined induction of several transcription factors necessary for gelB promoter activity occurs in the migrating epithelium, and this provides a focus for further experiments.

GelB promoter activity is induced in migrating epithelial cells in vivo, but only in the basal cell layer. Our immunolocalization studies revealed that NF-B and Sp1 are present in all layers of the corneal epithelium, and these factors could therefore not determine such localized expression. In contrast, we confirm reports that AP-2 is found predominantly in the basal cell layer of the corneal epithelium (48),⁴ similar to its localization in another stratified epithelium, the skin epidermis (50). AP-2 is a tissue-specific transcription factor confined to tissues derived from embryonic ectoderm and neural crest (51). AP-2 controls expression of the keratins, which are cytoskeletal proteins expressed by the keratinocytes composing tissues of body surface epithelia such as the skin and cornea (52). Recent evidence has been published (53) suggesting that preferential stimulation of AP-2 DNA-binding activity in the basal layers of the corneal epithelium (or skin epidermis) is an important determinant of keratin gene switching during differentiation of the keratinocytes. A requirement for localized transcription factors such as AP-2 along with more widespread transcription factors such as Sp1 might similarly confine expression of gelB to the basal cell layers of corneal and skin epithelia migrating to cover a wound.

What determines the localized induction of Sp1 and AP-2 expression to the epithelium migrating over the wound bed? As the epithelium migrates outward from the edge of a wound, it loses contact with its basement membrane and moves across a provisional matrix deposited from the tear film in the cornea (54) and the vasculature in skin (55). We showed that the location of fibronectin deposited from the tear film beneath the migrating epithelium precisely defines the realm of Sp1/AP-2 up-regulation. This invites us to speculate that reciprocal interaction between keratinocytes and their ECM might control the expression of Sp1 and AP-2. The control by the ECM on signaling the activity of transcription factors has recently emerged as a novel concept (56). Additional experiments to address the role of ECM components in modulating Sp1 and AP-2 DNA-binding activities and distribution in epithelial cells during wound healing need to be performed to address this idea.

Attenuation of gene expression through the use of transcriptional inhibitors is a novel strategy to control inflammatory and stress-related disorders. Transcription factors AP-1 and NF-B have attracted much interest for drug targeting because they are so widely involved in these pathologies (57). Currently, transcriptional inhibitors are being tested on artificial gene promoters that contain single or multiple copies of response elements for a particular transcription factor (49). However, it is necessary to extend testing to response elements in their natural context because transcription factors are known to cross-couple their activities by interacting among themselves and with other transcriptional co-activators; this can significantly modulate gene expression profiles (32). We propose the gelB gene promoter as a model for the extension of drug tests into a physiological context. Like other MMPs, it is inducible by a variety of stress stimuli (4). Its advantage over the promoters of other MMP genes lies in the diversity of characterized stress-activated response elements that it harbors. Thus, whereas AP-1 response elements have been identified in most of the MMP genes, gelB contains an important NF-B DNA-binding site that has not been defined in other MMP promoters to date (4). Like other MMP genes, gelB also contains an important transcription factor ets response element (PEA3), but it also contains AP-2 and Sp1 response elements, which are more specific for this particular MMP gene. Therefore, we propose the SIRC-43 and PAM-43 cell lines and the transgenic mouse line 3445 as useful tools for identifying transcriptional inhibitors of gelB gene expression, which in some cases may be extended to other MMPs.

In conclusion, we present the first study on the activity of a MMP promoter in transgenic mice. We suggest that line 3445 transgenic mice, which express -gal under control of the gelB promoter, provide a model to study regulation of gelB gene expression in development, tissue repair, and disease. In addition, these mice will be a useful tool for the in vivo testing of drugs to control MMP activity in pathology.