Proximal Promoter of the Surfactant Protein D Gene

Surfactant protein D (SP-D) plays roles in pulmonary host defense and surfactant homeostasis and is increased following lung injury. Because AP-1 proteins regulate cellular responses to diverse environmental stimuli, we hypothesized that the conserved AP-1 motif (at −109) and flanking sequences in the human SP-D promoter contribute to the regulation of SP-D expression. The AP-1 sequence specifically bound to fra-1, junD, andjunB in H441 lung adenocarcinoma nuclear extracts. Mutagenesis of the AP-1 motif in a chloramphenicol acetyltransferase reporter construct containing 285 base pairs of upstream sequence nearly abolished promoter activity, and co-transfection of junD significantly increased wild type but not mutant promoter activity. The sequence immediately downstream of the AP-1 element contained a binding site for HNF-3 (FOXA), and simultaneous mutation of this site (fox-d) and an upstream FoxA binding site (−277,fox-u) caused a 4-fold reduction in chloramphenicol acetyltransferase activity. Immediately upstream of the AP-1-binding site, we identified a GT box-containing positive regulatory element. Despite finding regions of limited homology to the thyroid transcription factor 1-binding site, SP-D promoter activity did not require thyroid transcription factor 1. Thus, transcriptional regulation of SP-D gene expression involves complex interactions with ubiquitous and lineage-dependent factors consistent with more generalized roles in innate immunity.

Pulmonary surfactant protein D (SP-D) 1 is a member of the collectin (collagenous lectin) subfamily of mammalian C-type lectins, which includes pulmonary surfactant protein A (SP-A), serum mannose-binding lectin, and at least two bovine serum lectins related to SP-D, conglutinin and CL-43 (1,2). The genes for both lung collectins and mannose-binding lectin are encoded in close proximity on human chromosome 10q. The pulmonary collectins, like the homologous serum proteins, are believed to contribute to innate (nonclonal) immunity and the host response to microorganisms (2). In addition, both SP-A and SP-D may contribute to the regulation of surfactant lipid homeostasis under certain circumstances in vivo (3)(4)(5).
SP-A and SP-D are secreted into the distal airways and pulmonary alveoli by Clara cells and type II pneumocytes. The expression of SP-A and SP-D by these cells is increased following many forms of pulmonary injury (2), and the rapid increase in SP-A and SP-D accumulation following intratracheal instillation of bacterial endotoxin suggests that they contribute to a pulmonary acute phase response (6). However, the regulation of these responses is not understood. We have previously shown that DNA sequences within 285 bp of the start site of transcription of the human SP-D promoter are able to confer glucocorticoid-responsive gene expression in H441 human lung adenocarcinoma cells but not liver HepG2 cells (7), suggesting that the proximal sequence contains information sufficient to direct lung-restricted expression. Sequence analysis of the proximal promoter suggested potential regulatory roles for a variety of ubiquitous and lung-restricted transcription factors.
In considering the potential contributions of these regulatory motifs to SP-D gene regulation, we initially focused our attention on the AP-1 consensus (5Ј-TGAGTCA-3Ј) at Ϫ109 bp relative to the start site of transcription (7). Preliminary footprinting assays showed preferential protection of this motif and its contiguous flanking sequences by nuclear extracts from H441 cells. In addition, an identical AP-1 motif is spatially conserved in the rat and mouse promoters (7,8) and in the promoter of the homologous bovine conglutinin gene (9, 10) ( Fig. 1). Because alterations in the composition of the AP-1 complex regulate a wide variety of transcriptional events associated with cellular differentiation and the response to environmental stimuli including cell injury, inflammation, and the systemic acute phase response, we hypothesized that the AP-1 motif plays important roles in the regulation of SP-D expression.
The flanking regions of the conserved AP-1 motif also contain sequences resembling binding sites for factors known to regulate respiratory epithelial cell differentiation and the expression of various secreted lung proteins including Clara cellspecific protein (CCSP) and SP-A, -B, and -C (11). These include potential binding sites for the homeodomain-containing protein, thyroid transcription factor 1 (TTF-1), and the forkhead/winged helix transcription factors, such as HNF-3 (FOXA). Previous studies have shown that TTF-1 and HNF-3 can act in combinatorial fashion with other ubiquitous and cell-specific transcription factors to regulate lung-specific genes (12,13). Because pulmonary cells known to express TTF-1 and forkhead box proteins secrete SP-D in vivo, we hypothesized that these nuclear factors similarly contribute to the regulation of SP-D expression.
To characterize the regulatory role(s) of the putative AP-1 element and other potential regulatory motifs in its conserved flanking sequences, we examined the interactions of oligomers containing these sequences with H441 nuclear proteins using electrophoretic mobility shift and antibody supershift assays. We also selectively mutated these sequences and compared the activity of wild type and mutant constructs in transient transfection assays. These studies are the first to demonstrate specific cis-acting elements in the human SP-D gene. The regulatory profile is consistent with more generalized roles in pulmonary and nonpulmonary host defense.

EXPERIMENTAL PROCEDURES
Genomic Clones and Sequencing-An approximately 7-kilobase EcoRI fragment of the previously described human genomic clone (H5), designated H5E7, containing 5Ј hSP-D sequence was isolated and subcloned into pGEM3Z as described previously (7). Most studies employed a XbaI/SacI fragment of the hSP-D gene containing 285 bp immediately upstream of the start site of transcription (XS285 or XS) (Fig. 1).
Cells-NCI-H441 human lung adenocarcinoma cells were obtained as a gift from Dr. A. Gazdar (University of Texas Medical Center, Dallas, TX) and propagated as described previously (7). HeLa (CCL-2), A549 (CCL-185), and HepG2 (HB-8065) cell lines were obtained from the American Type Culture Collection and maintained under recommended conditions of cell culture.
Oligomers-Several wild type and mutant oligomers were synthesized (DNA International) for this study (see Table I). In addition, oligomers corresponding to the CCSP upstream and/or downstream HNF-3-binding sites (14), FREAC2 (15), TTR-S (16), and wild type and mutant TTF-1 (17) were synthesized based on published sequences. The GT box oligomer corresponds to the GT box-binding site in human SP-A (18). Commercial consensus oligomers to AP-1 and AP-2 were purchased from Promega, and commercial Sp1, E box, and GATA (1-6) oligomers were obtained from Santa Cruz. The oligomers and their reverse complements were annealed and used in electrophoretic mobility shift assays (EMSAs) as described below.
Mutagenesis by Overlap Extension-XS285 ( Fig. 1) was subcloned into pGEM-3Z (Promega) and used for thermal cycling-coupled mutagenesis. Forward and reverse directed oligomers were synthesized, each containing a mutated consensus sequence. pXS was linearized outside the multiple cloning site by digestion with ScaI and used as template for thermal cycling reactions. Approximately 200 ng of template DNA, 200 ng of forward or reverse mutagenesis oligomer, and 200 ng of an oligomer directed to the appropriate SP6 or T7 RNA polymerase site in pGEM were combined with 200 M dNTPs (Roche Molecular Biochemicals) and 1 unit of Taq polymerase (Fisher) in buffer supplied with the enzyme. 20 -25 cycles were performed, each consisting of 1 min at 95°C (denaturing), 1 min at 45°C (annealing), and 2 min at 70°C (extension). Resultant DNA fragments were gel purified using the QIA Quick Gel extraction kit (Qiagen). The 5Ј and 3Ј fragments of the mutated promoter DNA were joined together by extension thermal cycling, using an overlapping internal oligomer sequence and oligomers to the flanking SP6 and T7 sites for amplification. The wild type or final mutated fragment was subcloned into pSK-CAT as described previously (7) or pBLCAT3 (plasmid backbone of pTK-CAT), and the orientation was verified by restriction digestion and sequencing. All SP-D sequences terminated at a SacI site within the untranslated first exon and were numbered from the start site of transcription (7).
Nuclear Extracts-Nuclear extracts were prepared from cultured cell lines using a rapid mini-extraction technique (19). Briefly, cells were scraped into 7 ml of Tris-buffered saline, pH 7.5, and centrifuged. The pellet was resuspended in 1 ml of Tris-buffered saline, transferred to a 1.5-ml microcentrifuge tube, and recentrifuged. The resultant pellet was thoroughly resuspended in 400 l of ice-cold Buffer A (10 mM HEPES, pH 7.8, 10 mM KCl, 0.1 mM EDTA, 0.1 mM EGTA, 1 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride, 4 g/ml leupeptin, and 2 g/ml aprotinin) and incubated on ice for 15 min. Nonidet P-40 (Sigma) was then added to a final concentration of 0.625%, and the sample was vortexed for 10 s prior to microcentrifugation at 1350 rpm in a Hermle Z233M centrifuge for 30 s at 4°C. The resultant nuclear pellet was resuspended in 50 l of cold Buffer C (20% glycerol, 400 mM NaCl, 10 mM HEPES, pH 7.8, 1 mM EDTA, 1 mM EGTA, 1 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride, 4 g/ml leupeptin, and 2 g/ml aprotinin). The pellet was vigorously agitated on a rocker platform for 15 min at 4°C and then clarified by centrifugation for 15 min at 4°C in a microcentrifuge (12,000 rpm). The supernatant was transferred to a chilled fresh tube, and the protein content was analyzed by dye binding assay. Yields were in the range of 300 -500 g/75cm 2 flask. Once quantified, the extract was frozen in liquid nitrogen and stored at Ϫ70°C.
EMSAs-Gel retardation assays were performed by a modification of a method employed by Bingle and co-workers (20). Briefly, 5-10 g of nuclear extract in the absence or presence of 50 -100-fold excess of competitive inhibitor was incubated for 10 min at 18°C in a buffer consisting of 1 mM MgCl 2 , 20 mM HEPES, pH 7.8, 40 mM KCl, 1 mM dithiothreitol, 100 mM EGTA, 4% (w/v) Ficoll (Sigma), and 50 g/ml poly(dI-dC) (Sigma). Approximately 1.75 pmol of kinase-labeled doublestranded oligomer were then added, and the incubation was continued for 10 min. For supershift experiments purified antibodies were then added to samples at the optimal concentration and incubated for an additional 10 min. Antibodies to jun or fos proteins, GATA-6, Sp1, and Sp3 were from Santa Cruz Biotechnology; antibodies to HNF-3␣ and HNF-3␤ were from Dr. Robert Costa (University of Illinois, Chicago, IL), and the antibody to TTF-1 was provided by Dr. Roberto Di Lauro (Stazione Zoologica "Anton Dohrn," Naples, Italy). Prior to loading on polyacrylamide gel electrophoresis, 1 ⁄10 volume of loading buffer (250 mM Tris-HCl, pH 7.5, 0.2% bromphenol blue, 0.2% xylene cyanol, and 40% glycerol) was added to each sample. Complexes were resolved by polyacrylamide gel electrophoresis (2.5-3 h, 250 V, 4°C) using nondenaturing 4% bisacrylamide, 2.5% glycerol gels, and 0.5ϫ TBE running buffer. Gels were dried to 3MM paper (Whatman) and exposed to X-Omat AR film (Eastman Kodak) for a period of 1 h to 1 week with an intensifying screen.
Transient Transfection-In brief, for experiments characterizing the promoter activity of 5Ј deletion or mutant constructs, H441 target cells (5 ϫ 10 5 ) were plated on 35-mm plates in RPMI medium (Life Technologies, Inc.) supplemented with 10% (v/v) newborn calf serum (Life Technologies, Inc.), allowed to attach overnight, and washed twice with RPMI devoid of phenol red (7). The cells were transfected with 2 g of CAT reporter construct and 1 g of pCMV-␤-gal (Promega) using Lipo-fectAMINE (Life Technologies, Inc.) and incubated for 2 h at 37°C in the absence of serum. Parallel transfections were performed using pSK-CAT vector controls and/or a glucocorticoid-responsive control promoter, pMSG-CAT (Amersham Pharmacia Biotech). Concentrated serum-containing medium was then added, and the cells were incubated overnight. Cells were harvested at various times up to 48 h with one medium change at 24 h when needed. Where indicated, a previously optimized concentration of dexamethasone (Dex; final concentration, 50 nM) was added to the concentrated medium and with the succeeding medium change. All assays were performed on duplicate or triplicate plates. For some experiments, similar studies were performed using HeLa or HepG2 cells.
CAT Assays-Cell layers were harvested, and transient transfection assays were performed using protein equivalent amounts of cell extract using and CAT promoter constructs as described previously (7). Prior to each transfection the quality of the plasmids was reassessed by gel electrophoresis. CAT activity was measured by phase partitioning and thin layer chromatography, followed by autoradiography. To quantify relative acetylation, gels were exposed to a phosphorimaging screen (Storage phosphor Screen GP; Eastman Kodak Co.) and scanned using a STORM image reader (Molecular Dynamics). To prepare figures the data files were imported to ImageQuant (Molecular Dynamics) and transferred to Illustrator (Adobe) as unmanipulated TIFF files. Each assay was performed in duplicate, and each figure is representative of at least three experiments. When indicated, conversion data were normalized to ␤-galactosidase activity as described previously (7).
Cotransfections with Nuclear Factor Expression Vectors-The junD expression plasmid (HA1-junD) was a gift of Dr. Lester Lau (University of Illinois). The fra-1 cDNA was kindly provided by Dr. Thomas Curran (St. Jude's Hospital, Memphis, TN). Expression plasmids for c-fos and c-jun were a gift of Dr. G. Doyle (University of Wisconsin, Madison, WI). These were constructed by inserting the nuclear factor cDNA in the pcDNA3 vector (Invitrogen, Carlsbad, CA; contains the cytomegalovi-rus immediate-early promoter, a polylinker, and the bovine growth hormone polyadenylation sequence). The pCMV-human HNF-3␣ cDNA (21) and TTF-1 cDNA (20) were gifts from Dr. Colin Bingle (University of Sheffield Medical School, Sheffield, UK). Dr. James Darnell (Rockefeller University, New York, NY) kindly provided the rat HNF-3␣ and HNF-3␤ cDNAs. Transfections were performed using up to 1 g of pcDNA3 containing the desired cDNA or an equivalent weight of the pcDNA3 vector. For some experiments, expression of mRNA or protein encoded by the cotransfected plasmid was confirmed by Northern hybridization or gel supershift assays, respectively.

H441 Nuclear Proteins Bind to the AP-1 Motif in SP-D-A
labeled SP-D oligomer (20-mer) containing the AP-1 sequence (Table I, Oligo1) showed binding to proteins in H441 nuclear extracts ( Fig. 2A, lane 2). The major complex was specific as demonstrated by competition with unlabeled Oligo1 ( Because surrounding sequences can influence the binding of specific AP-1 proteins (22), we also examined the interactions of H441 nuclear proteins with a commercial AP-1 consensus oligomer (pAP-1) (Fig. 2B). Binding was competed by Oligo1 or pAP-1 (Fig. 2B, lanes 3 and 4), but not by Oligo1m or AP-2 (Fig. 2B, lanes 5 and 6). Binding was also competed with a smaller SP-D AP-1 oligomer (Oligo 2, Table I) (data not shown). Thus, the data demonstrate that the SP-D AP-1 motif can specifically bind to nuclear proteins expressed by H441 cells. Complexes of comparable mobility were also identified using nuclear extracts of an SP-D producing lung adenocarcinoma cell line (NCI-H969) (23) and freshly isolated rat type II cells (data not shown).
AP-1 Proteins Bind to the AP-1 Sequence-AP-1 complexes consist of homodimers of two jun family members (c-jun, junB, and junD), or heterodimers of one jun family member with c-fos or one of the fos-related proteins (fosB, fra-1, and fra-2) (24). Supershift assays using pan-fos and pan-jun antibodies and radiolabeled Oligo1 incubated with H441 nuclear extracts demonstrated binding of components immunologically related to jun and fos family members (Fig. 3, lanes 2 and 6). In other experiments, stronger supershift bands were obtained with the pan-jun antibody (data not shown). Supershift assays with antibodies for specific fos and jun proteins demonstrated relatively strong and specific supershift bands with antibodies to junB, junD, and fra-1 (lanes 3, 5, and 9) and much weaker bands for c-jun, c-fos, or fra-2 (lanes 4, 7, and 10), and no detectable signal for fosB (lane 8). The fra-1, junD, and junB supershift bands comigrated with the major bands detected

SP-D AP-1 and forkhead box Downstream of AP-1 and forkhead
fox-d

5Ј-GGTGGGGgatccAGTGAG
Oligo4 with mutated GATAA using the pan-fos and pan-jun antibodies but had a reproducibly lower mobility than the faint complexes formed with antibodies to c-fos and c-jun. Consistent with the finding shown in Fig. 2, binding was blocked by competition with the commercial AP-1 oligomer or Oligo1 but not by Oligo1m (data not shown).
Thus, the data indicate that selected AP-1 family members are binding to the SP-D sequence and suggest that the binding of fra-1, junD, and junB reflects the predominating species of AP-1 proteins in H441 cells rather than an intrinsic property of the AP-1-binding sequence. Because prior studies have described fra-1 and junB but not junD in H441 cells (25), we performed Northern hybridization assays to further confirm the presence of junD. Although there was a predominance of junB message, junD mRNA was also identified (data not shown).
Site-directed Mutagenesis of the Conserved AP-1 Consensus Decreases SP-D Promoter Activity-To study the potential functional consequences of AP-1 binding, we employed a transient transfection assay utilizing H441 lung adenocarcinoma cells in conjunction with wild type and mutant CAT reporter con-structs. Mutagenesis of the conserved AP-1 motif in pXS (pXSm) resulted in a marked decrease in both basal and promoter activity (Fig. 4A). The reduction in normalized basal activity was 3.9 Ϯ 0.9 in five separate experiments. Decreases in promoter activity were also observed when the mutation was examined within the context of two shorter restriction fragments, pPS and pFS ( Fig. 1; data not shown).
Dex increases the production of SP-D in lung tissue in vivo and in vitro, and these effects are mediated at the level of transcription (7). However, the stimulation of pXS activity achieved with Dex is indirect (e.g. requires 24 -48 h for maximal stimulation) and does not involve direct interactions of glucocorticoid receptor with classical response elements in the proximal promoter (7). We therefore sought to determine whether mutation of the AP-1 could inhibit the 2-3-fold higher levels of promoter activity observed in the presence of glucocorticoids. As shown in Fig. 4A, mutation of the AP-1 element (pXSm) markedly decreased promoter activity in the absence or presence of Dex. Although the residual activity of pXSm showed a detectable increase in the presence of Dex, the pSK vector control often showed a similar increase (Fig. 4A, compare pSK with pXSm). We occasionally observed a slight decrease in mobility and intensity of the complexes formed on Oligo1 in the presence of Dex; however, there were no major or reproducible effects on AP-1 binding (data not shown). Thus, the effects of Dex do not appear to be secondary to increased occupancy of the AP-1 site.
Overexpression of AP-1 Proteins Modulates SP-D Promoter Activity-To further examine the potential modulatory roles of specific AP-1 proteins we performed cotransfection studies using the selected CAT reporter constructs and jun and fos ex-  , Table I) bound to proteins in H441 nuclear extracts. The single major complex is specific as demonstrated by competition with Oligo1 or a commercial AP-1 consensus oligomer (pAP-1). There was no competition by an unrelated oligomer (AP-2), and Oligo1 with a mutated AP-1 consensus (Oligo1m) was ineffective as a competitor. B, parallel experiments were performed using a radiolabeled commercial AP-1 oligomer (pAP-1).
FIG. 3. Representative supershift assays using antibodies for specific fos and jun proteins. Strong supershift bands are observed in the presence of antibodies to fra-1, junB, and junD (arrow). Although the pan-fos antibody gave a similarly strong band, the pan-jun antibody gave a much weaker signal in this particular experiment. Binding to H441 nuclear proteins was blocked by competition with the commercial AP-1 oligomer or Oligo1, but not by Oligo1m (not shown).

FIG. 4. Mutagenesis of the conserved AP-1 sequence decreases basal and junD-stimulated promoter activity.
A, mutagenesis of the conserved AP-1-binding site in pXS (pXSm) resulted in a marked decrease in basal and Dex-stimulated promoter activity in transient transfection assays using H441 cells. The activity of pXS-CAT and pXSm-CAT are compared in this representative experiment. Dex significantly increased the activity of the pMSG-CAT, which contains known glucocorticoid response elements, but not the activity of the vector control, pSK-CAT. B, H441 cells were cotransfected with the CAT reporter construct and a cDNA encoding junD in the absence or presence of Dex. A representative experiment is shown. Co-transfection with junD increased pXS-CAT activity, and this effect was markedly decreased with pXSm-CAT. pression constructs. The activities of pXS were increased 3-4fold by cotransfection with junD cDNA in the absence or presence of Dex, and mutation of the AP-1 site markedly decreased basal and glucocorticoid stimulated expression (Fig.  4B). The effects of junD and Dex appeared additive rather than synergistic. Cotransfection with junD did not alter the activity of a pMSG-CAT control plasmid, which lacks a known AP-1 site, consistent with specific stimulation. Slightly lower specific stimulation was observed with fra-1 cotransfection, but inhibition was reproducibly observed with c-fos or c-jun (data not shown).
Plasmids containing the mutated AP-1 showed some residual modulation by co-transfected AP-1 proteins under conditions where there was no significant alteration in the activity of control plasmids (Fig. 4B and data not shown). However, the only other sequence in the proximal promoter with significant homology to an AP-1 consensus (Ϫ215, TGAGTTCA) (7) did not bind AP-1 proteins in supershift assays using either pan-jun or pan-fos antibodies (data not shown), suggesting that the residual modulatory effects are secondary to interactions of AP-1 proteins with components of the transcriptional initiation complex (26) or other potential regulatory factors such as C/EBP␤ (27).
Because phorbol esters such as 12-O-tetradecanoylphorbol-13-acetate (TPA) activate protein kinase C and can modulate the expression of AP-1 family members, we examined the effects of TPA. Interestingly, TPA showed no reproducible effect on the proximal promoter activity of SP-D constructs in H441 cells and no obvious alteration in AP-1 DNA binding as detected by EMSA (data not shown). This is consistent with the failure of phorbol to stimulate SP-D production in human fetal lung explants (28). Thus, protein kinase C-independent signal transduction pathways appear to regulate the AP-1-dependent promoter activity.
The Proximal Promoter Contains a HNF-3 (FOXA)-binding Site Downstream of the AP-1 Element-The AP-1 element partially overlaps with a downstream sequence (fox-d) that shows homology to the forkhead-binding site consensus, RARY-MAAYAWT ( Fig. 1) (29). Close inspection reveals that this region contains two partially overlapping regions of forkhead box binding motif homology: cAATAAAgAAg (8/11), which begins at Ϫ104, and AAGaAAATtgc (7/11), which begins at Ϫ96. The latter sequence is spatially conserved between the known SP-D and conglutinin promoter sequences (Fig. 1).
When radiolabeled oligomers containing the entirety of the forkhead box sequence (fox-d, Table I) were incubated with nuclear extracts from H441 cells, a single complex was observed (Fig. 5A, left panel, lane 2). The binding is specific as shown by competition with unlabeled oligomer (left panel, lanes 3 and 5) but not by oligomers with a mutated forkhead motif (lane 4A, Table I). The binding was similarly competed by an upstream HNF-3 sequence from CCSP (CCSPu, lane 7), as well as a downstream CCSP site and a FREAC2 (FOXF2) consensus oligomer (15) (data not shown). The complex formed using H441 nuclear extracts was supershifted with anti-HNF-3␣ (Fig. 5B, lane 4) but not control Igs (Fig. 5B, lane 3) and was competed by unlabeled fox-d (Fig. 5B, lane 5). To further confirm these results, HeLa cells, which do not express significant amounts of HNF-3␣, were co-transfected with HNF-3␣ cDNA. A specific supershift complex with the same mobility was observed in the co-transfected but not the mock co-transfected cells (data not shown).
Oligo1, which contains the AP-1 sequence and 8 residues from the 5Ј end of the region of forkhead homology, did not compete for binding to fox-d (Fig. 5A, right panel, lane 5). This is consistent with the absence of a second (i.e. forkhead con-taining) complex in mobility shift assays using this oligomer (Fig. 2). However, a 26-mer that included the AP-1 element and the entirety of the downstream forkhead box (AP-1/fox-d, Table  I) showed at least three specific complexes (Fig. 6, lane 2) that were efficiently competed with unlabeled oligomer (lane 3). Complexes 1 and 2 were also competed with the SP-D AP-1 (Oligo1) (lane 4), and complex 3 was competed with fox-d but not by Oligo1 (lane 5). However, only complex 1 was competed with commercial AP-1 consensus oligomer or Oligo 2 or was altered in the presence of pan-jun or pan-fos antibodies ( lanes  6 and 7). None of the complexes were competed by both Oligo1 and fox-d, and the same three complexes were observed even when much larger amounts of nuclear extract were incubated with the smallest possible concentration of a high specific activity probe (data not shown). Thus, the AP-1 and fox-d sites each bind a unique protein or protein complex, and the sites are not simultaneously occupied under the conditions of assay. In addition, complexes 2 and 4 contain as yet unidentified nuclear proteins.
The Proximal Promoter Also Contains an Upstream HNF-3 (FOXA1)-binding Site-The proximal promoter contains an additional forkhead motif upstream at Ϫ277 (fox-u) that matches the forkhead box consensus at 7 of 11 positions (5Ј-ctATA-AATAca). Electrophoretic mobility shift assays using labeled FIG. 5. Nuclear proteins and HNF3-␣ bind to downstream forkhead motifs. A, a labeled oligomer containing the downstream forkhead-binding site (fox-d) gave a single specific complex when incubated with H441 nuclear extracts. Two separate experiments are shown in the left and right panels. The complex is efficiently competed by unlabeled oligomer but not by oligomer containing a mutated forkhead site (fox-d-m). Binding was not competed by Oligo1, which contains the 5Ј end of the region of forkhead homology. B, a specific supershift band was visualized using antibody to HNF-3␣ using H441 nuclear extracts (lane 4) or extracts from HeLa cells co-transfected with HNF-3␣ (data not shown).
fox-u oligomer (Table I) gave two specific complexes that were competed by unlabeled fox-u (Fig. 7A). The larger of the two complexes was efficiently competed by CCSP oligomers containing HNF-3 forkhead motifs (lanes 4 -6, left; CCSPu, CCSPd, combined) but not by a mutant oligomer (fox-um, Table  I) (data not shown). Interestingly, the downstream and upstream SP-D sites were ineffective cross-competitors (e.g. Fig.  5A, left panel, lane 6). Similar to the downstream site, radiolabeled fox-u showed a specific supershift complex when incubated with H441 nuclear extracts and antibody to HNF-3␣ (Fig. 7B, arrow).
The Upstream and Downstream Forkhead Sites May Interact with Other Forkhead Box Proteins-Although HNF-3␣ binds to the fox-d and fox-u sites, several additional observations suggest that binding is not specific for HNF-3␣ (or HNF-3␥) and that the complexes contain other forkhead proteins. For example, nuclear extracts from A549 type II-like cells, which predominantly express HNF-3␤, give specific complexes of identical mobility (data not shown). In addition, TTR-S, which corresponds to a high affinity HNF-3 site in the transthyretin promoter that binds with high affinity to HNF-3␣, -3␤, and -3␥ (16), only weakly inhibits the binding of H441 or A549 nuclear proteins to fox-u and fox-d, and the SP-D sequences are inefficient competitors of the complexes formed on radiolabeled TTR-S (data not shown).
Mutagenesis of the Forkhead Sites Decreases the Activity of pXS-CAT in H441 Cells-Mutagenesis of downstream site within the context of pXS and the same residue substitutions used for the mutated oligomer (Table I, fox-dm) gave a slight but detectable inhibition in transient transfection assays using H441 cells (Fig. 8). By contrast, selective mutagenesis of the upstream site gave a slight but reproducible stimulation suggesting an associated inhibitory element. Notably, simultaneous mutation of both sites resulted in a significant, 3-4-fold decrease in CAT activity (Fig. 8).
HNF-3 cDNAs Do Not Transactivate SP-D Promoter Constructs in H441 Cells-Numerous attempts to specifically transactivate pXS-CAT with human HNF-3␣ or rat HNF-3␣ or -␤ cDNAs were unsuccessful. No activation was observed over a wide range of plasmid concentrations, with both cytomegalovirus and SV40-driven expression vectors, and in H441 or HeLa cells (data not shown). Although suppression was sometimes observed, consistent with the stimulation of promoter activity observed following mutagenesis of the upstream forkhead box, there was no reproducible inhibition under conditions where the activity of control promoters was unaltered. Similar results were obtained with pPS-CAT or pFS-CAT, which lack fox-u (Fig. 1).
A G/C-rich Sequence Upstream of the AP-1 Element Is a Functional GT Box-like Element-Sequences immediately upstream of the AP-1 element in the human SP-D gene contain several potential cis-acting elements. These include at least  7. Nuclear proteins and HNF3-␣ bind to upstream forkhead motifs. A, a labeled oligomer containing the upstream forkheadbinding site (fox-u) gave two specific complexes when incubated with H441 nuclear extracts. The upper complex (arrow) is efficiently competed by unlabeled oligomers containing forkhead-binding sites from CCSP. B, a specific supershift band (arrow) was visualized using antibody to HNF-3␣ using H441 nuclear extracts. three partially overlapping motifs that approximate the most 5Ј protected region identified in Fig. 1. The first is an E box motif (CANNTG), which is a recognition site for basic helixloop-helix-zipper transcription factors; E boxes have been identified as core sequences in the upstream and downstream regulatory elements of the rabbit SP-A gene (30,31). Immediately downstream of the E box is a potential GATA-binding motif (gGATAA) homologous to those recently implicated in the regulation of TTF-1, SP-A, and SP-C expression by GATA-6 (32). Overlapping both of these potential binding sites is a conserved GT box, which has been identified as an important regulatory element in the human SP-A2 gene and potential binding site for zinc finger transcription factors such as lung Kruppel-like factor and Sp1-related proteins (18).
To determine whether these sites bind to nuclear proteins, we examined the interactions of H441 nuclear proteins with two nested oligomers. The first, designated Oligo3, contains the overlapping E box, GT box, and GATA motifs and sequences immediately upstream of the AP-1 element (Table I). The second, designated Oligo4, contains only the GATA motif and sequence contiguous with the AP-1 (Table I). EMSAs using the labeled Oligo3 oligomer showed a single, major specific complex (Fig. 9, lane 2) that was efficiently competed by unlabeled oligomer (lane 3). Binding was also competed by a GT box oligomer (lanes 6 and 10) or by Oligo3 oligomers containing site-specific substitutions in the E box or GATA motifs (data not shown). This complex was not efficiently competed by Oligo3 m1, which contained site-directed substitutions in the GT box motif (lane 4), by consensus oligomers containing the GATA (lane 7) or E box (lane 8) motifs, or by Oligo4 (data not shown), and there was only partial competition by an SP-1 consensus oligomer (lane 9). Antibodies to Sp1 and Sp3 specifically supershifted complexes formed on radiolabeled Sp1 oligomer but not the major complex formed on Oligo3 (data not shown).
Consistent with these results, no major specific complex was observed using labeled Oligo3 m1 (mutated GT box motif, lane 11). In addition, assays using a labeled GT box consensus oligomer showed a single major complex that was efficiently competed by unlabeled probe or by Oligo3 (data not shown). Mobility shift assays using the labeled Oligo4 showed only a single faint specific complex that migrated much more rapidly than the specific complexes observed with Oligo3 (data not shown). This complex was not competed by the commercial GATA oligomer or other available oligomers. Together the data strongly suggest that nuclear proteins are binding to the GT box but not the GATA motif.
The GT Box Is a Positive Regulatory Element-The functional role of protein binding to the GT motif was examined in transient transfection assays using H441 cells. Site-directed mutagenesis of the GT box using the same substitutions as in the Oligo3 m1 showed a significant decrease in promoter activity as compared with a wild type pXS-CAT (Fig. 10). The decrease, although reproducible, was approximately 2-fold less than observed for pXSm, which contained substitutions in the AP-1 element. Site-directed mutagenesis of the GATA motif gave an apparent but insignificant decrease in promoter activity (not shown). Thus, the combined data indicate a specific and functional interaction of H441 nuclear proteins with the GT box motif.

TTF-1 Is Not Required for SP-D Gene Expression in Cell
Lines-Cell-specific gene expression in some lung epithelial cells is subject to regulation by TTF-1 (11). TTF-1-binding sites, which can show considerable degeneracy, are usually present in multiple copies within a few hundred base pairs of the transcription start site of TTF-1-responsive genes (13). Although the near distal promoter in the region of 300 -700 bp upstream of the start site of transcription contains several regions of limited TTF-1 homology, only a single region was identified by computer-assisted analysis of the proximal promoter fragment, using consensus sequences compiled for thyroid-specific genes (CCACTCAAGTG) and SP-B (GCNCT-NNAG) (17). This sequence (5Ј-GAACGCAGGTGGGG-3Ј; Fig.  1) shows limited homology (7 of 14), with a known TTF-1binding site in SP-Bf1 (5Ј-GCCTCCAGGTGCTT-3Ј), which contains a core CAGG-binding motif (17). Consistent with the absence of dyad symmetry or contiguous regions of TTF-1 homology, there was no evidence of TTF-1 binding to the SP-D sequence by EMSA using H441 nuclear extracts, cells known to express TTF-1 (data not shown). More significantly, when TTF-1 cDNA was overexpressed in HeLa cells, the resulting nuclear extracts gave specific complexes that were efficiently competed with oligomers containing TTF-1-binding sites from the thyroglobulin gene and CCSP (Fig. 11A, lanes 3 and 6) and FIG. 9. Nuclear proteins bind to a GT box motif upstream of the AP-1 element. Labeled Oligo3, which contains the E box, GT box, GATA motif, and sequence upstream of the AP-1 (Table I), gave a specific complex when incubated with H441 nuclear extracts (lane 2, arrow) and a prominent nonspecific band (NS). The specific complex was efficiently competed by unlabeled Oligo3 (lane 3) and a GT box consensus oligomer (lanes 6 and 10) but not by Oligo3 m1, which contains substitutions in the GT box motif (Table I). Oligo3 m2, which contains substitutions in the GATA motif but partially overlaps with the GT box, was a less efficient competitor (lane 5). The complex was not competed by the commercial GATA consensus oligomer (lane 7), a commercial E box containing sequence (lane 8), and was minimally competed with a commercial Sp1 oligomer (lane 9). In a parallel experiment radiolabeled Oligo3 m1 (substitutions in the GT box) was incubated with H441 nuclear extracts; no specific complex was identified (lane 11). The figure is a composite from three separate experiments: lanes 1-6, lanes 7-10, and lane 11, respectively. that were supershifted with antibody to TTF-1 (data not shown). By contrast, oligomers containing the putative SP-Dbinding site (lane 5) or mutated TTF-1-binding sites showed no evidence of competition (lanes 4 and 8).
Because such studies cannot exclude TTF-1 binding to other degenerate or more upstream sites, we also co-transfected TTF-1 cDNA with SP-D promoter reporter constructs containing up to 700 bp of upstream sequence in H441 or HeLa cells.
Although initial experiments using pSK-CAT showed apparent activation, increased activity was also observed with a minimal SP-D promoter or promoterless controls. The same promoter sequences in pBLCAT3 (the backbone of pTK-CAT) showed no increase in activity in the presence of TTF-1 (Fig. 11B). While performing these experiments, we also observed that HeLa cells, which lack TTF-1, could support the activity of pXS (data not shown). Thus, the preferential expression of SP-D promoter constructs by H441 cells, as compared with HepG2 cells, is not dependent on transactivation by TTF-1, and TTF-1 is not required for proximal promoter activity in this system. DISCUSSION Little is known about the transcriptional regulation of members of the collectin family of host defense proteins. These studies are the first to identify specific cis-acting elements and trans-regulatory proteins that may contribute to the regulation of the SP-D gene and its response to environmental stimuli. Our previous studies have shown that sequences within 285 bp of the transcription start site support basal, glucocorticoidactivated, and cell type-restricted activity. Here, we focus on a highly conserved region of the proximal promoter (Ϫ85 to Ϫ135) that contains a dense cluster of protected, protein-binding sites. We demonstrate that this region interacts with ubiquitous and lineage-dependent, but not lung-specific, transcription factors that are required for SP-D promoter activity.
Our studies indicate that a conserved, canonical AP-1 sequence within the proximal promoter of the human SP-D gene is a functional AP-1 element. Oligomers containing the SP-D AP-1 sequence formed specific complexes containing fra-1, junD, and junB consistent with the binding of fra-1/junD or fra-1/junB heterodimers (33) and/or jun homodimers. Mutation of the AP-1 element greatly decreased the ability of junD and other co-transfected AP-1 proteins to modulate promoter activity. This strongly suggests that AP-1 proteins play important roles in the transcriptional regulation of SP-D gene expression in lung epithelial cells, consistent with recent studies that have implicated a more upstream and nonconserved AP-1 sequence in the basal regulation of the bovine conglutinin gene (34).
Relatively little is known about the regulation of respiratory epithelial cell proteins, such as surfactant proteins or CCSP, by AP-1. An AP-1 element is present downstream of the start site of transcription of the human SP-B gene (35), and AP-1 motifs or binding sites have been identified in the promoters for rat CCSP (25) and SP-A. Recent studies have demonstrated two functionally distinct AP-1 sites at Ϫ18 and Ϫ370 within the mouse SP-B gene (36). The proximal sequence of SP-B binds to junB and junD in MLE-15 mouse epithelial cell nuclear extracts. Interestingly, co-transfection with junD stimulated proximal promoter activity by approximately 2-fold, whereas c-jun decreased promoter activity, in part through interactions with the distal binding site. We observed similar specific inhibition by c-jun and c-fos. Thus, the composition of the AP-1 complex may determine the functional activity of protein binding.
Mutagenesis of the AP-1 element markedly decreased glucocorticoid-stimulated and basal promoter activity. However, we observed no increase in occupancy of the AP-1 element using nuclear extracts from Dex-treated H441 cells. Although more subtle alterations in the composition or state of posttranslational phosphorylation of AP-1 proteins could contribute to the stimulatory effects of glucocorticoids, we have no evidence to support this possibility. Instead, we speculate that glucocorticoids exert their indirect effects through the expression of other transactivating factors, such as C/EBP␤ (37), that may bind to the proximal promoter and/or modulate AP-1-dependent promoter activity.
Immediately downstream to the AP-1 element was a HNF3binding motif. Proteins related to the Drosophila homeotic gene, forkhead, and the rat HNF-3 family share a homologous DNA-binding domain (forkhead box) and are now designated by standardized genetic nomenclature as fox (f box) genes. Fox proteins contribute to the regulation of the expression of lungrestricted proteins and to the expression of proteins by nonpulmonary epithelial cell types, particularly cells of endodermal  8). B, H441 cells were cotransfected with XS in pSK-CAT or pBLCAT3 vector (pTK-CAT) reporter constructs and 0.5 or 1.0 g/ml of TTF-1 cDNA. A representative experiment with XS is shown; identical results were obtained with a construct containing 700 bp of upstream sequence, pSS700 (data not shown). Although the pSK-CAT constructs showed apparent transactivation (top panel), the increase in activity was not dependent on SP-D promoter sequence (data not shown). There was no evidence of transactivation when the corresponding SP-D promoter fragments were inserted in pTK-CAT (bottom panel). derivation (11,12,29,38). At least two forkhead-binding elements were identified in the proximal promoter, fox-d and fox-u. Although mutation of the individual sites gave slight inhibition or stimulation, respectively, mutagenesis of both sites decreased promoter activity by approximately 4-fold. Similar complex, modulatory effects of HNF-3 and interactions among separate HNF-3-binding sites have been described for a number of promoters, including the rat CCSP gene (39).
Our many attempts to activate or specifically inhibit transcription by cotransfecting HNF-3 cDNAs were unsuccessful, using either H441 or HeLa cells. Some investigators, but not all, have observed transactivation or specific inhibition of gene expression in cotransfection experiments using H441 cells (25, 39 -41). In other promoters, HNF-3 has been observed to decrease gene expression by blocking the binding of other transactivating proteins (42). Thus, the activity of the expressed proteins could be gene-dependent and/or could be regulated by the availability of specific forkhead proteins or other transactivating factors. We considered the possibility that our studies were confounded by a superimposition of the stimulatory and inhibitory effects mediated by the upstream and downstream binding sites; however, cotransfection with 5Ј deletion mutants lacking the upstream element ( Fig. 1; pPS or pFS) did not show activation. Another possibility is that the net stimulatory effect of the interacting sites is masked by competition of HNF-3␣ for AP-1 occupancy of the AP-1/fox-d site, which could be consistent with our inability to identify simultaneous occupancy of the AP-1 and fox-d sites in the AP-1/fox-d oligomer.
On the other hand, there is increasing evidence that the forkhead box proteins are not necessarily acting as classical transregulatory proteins but as mediators of chromatin remodeling with reversal of chromatin-mediated repression. Chromatin repression occurs in vivo but can also occur in vitro in association with transiently expressed DNA (43). A role of HNF-3 in chromatin remodeling is consistent with its role in nucleosome positioning over the albumin enhancer (44,45) and the structural homology of HNF-3 with linker histone molecules (46). Thus, the SP-D promoter fragment may be "constitutively derepressed," obviating at least part of the necessity for forkhead binding. The inhibition in promoter activity observed following mutagenesis of the forkhead-binding sites may reflect more subtle effects on DNA structure such as DNA bending (15,47) and the binding of other regulatory molecules, consistent with the observed "interaction" between the upstream and downstream sites.
Epithelial cell types known to synthesize SP-D express several forkhead proteins from at least three subfamilies. Nonciliated bronchiolar cells (Clara cells) express HNF-3␣, HNF-3␤, HNF-3␥, HFH-11 (38), and lun (human homolog of FREAC2) (48). Of these, HNF-3␤, HFH-11, and lun are expressed by human type II pneumocytes. The fox-u and fox-d sites bind to HNF-3␣ in H441 nuclear extracts, which contain HNF-3␣ and HNF-3␥ and little if any HNF-3␤ (25,49). However, our data suggest that these cell lines have even greater heterogeneity in forkhead box proteins. In this regard, the HNF-3 homolog HFH-11 (FoxM1A) was recently identified in HeLa, H441, and A549 cells (38). Consistent with differences in the core and flanking sequences of the upstream and downstream forkheadbinding sequences, fox-u oligomers did not efficiently crosscompete for nuclear protein binding to fox-d (Fig. 6A), even though both oligomers participate in the formation of complexes that are competed with FREAC2 and CCSP oligomers and contain HNF-3␣. This could reflect differences in the affinity of binding of HNF-3␣ to the two sites, interactions with other forkhead proteins, or interactions with other nuclear proteins.
The close apposition of AP-1 and forkhead box-binding elements is of particular interest. Although a similar arrangement of these motifs is present in the rat CCSP gene (25), the AP-1 motif is not conserved in the human CCSP gene, and mutation of the site did not inhibit promoter activity. On the other hand, the promoter of the hepatic transthyretin gene contains a nearly identical AP-1/forkhead motif. Here, the AP-1 and forkhead proteins bind independently to the overlapping AP-1 and HNF-3 sites, and there is circumstantial evidence that changes in the relative abundance of HNF-3␣ and -3␥ modulate AP-1 binding and contribute to the regulation of the acute phase response (16). There is also ample evidence that forkhead proteins are regulated post-developmentally in the setting of inflammation or tissue injury. For example, partial hepatectomy or systemic endotoxin administration alters the hepatic expression of HNF-3 proteins (16), and ␥-interferon increases the transcription of HNF-3␤ mRNA in the mouse mtCC1-2 Clara cell line (50). Thus, modulation of AP-1 and forkhead box protein binding may lead to alterations in SP-D expression in the setting of lung injury.
Upstream of the AP-1 element we identified a G/C-rich sequence that is important for promoter activity in H441 cells. Nuclear protein binding to this region does not involve interactions with the E box or GATA motifs, consistent with their lack of conservation among the conglutinin and SP-D genes ( Fig. 1) and supporting a role for regulation by the Cys 2 -His 2 (Sp) family of zinc finger transcription factors (51). The CCSP/ rabbit uteroglobin genes have functional binding sites for Sp1 and Sp3 within the proximal promoter (20,41). However, binding to Oligo3 was inefficiently competed with a GC box-containing Sp1 oligomer, and the complexes were not supershifted with antibodies to Sp1 or Sp3, suggesting that other family members are involved. A conserved GT box contributes to the regulation of SP-A2 promoter activity by type II pneumocytes (18), and there is circumstantial evidence that the SP-A2 GT box binds to lung Kruppel-like factor (52). Interestingly, Sp family members interact combinatorially with HNF-3␣-and -3␤-binding sites in the rabbit uteroglobin/CCSP gene (41), and interactions between Sp1 and binding sites for AP-1 have been described (53).
In contrast to other surfactant proteins and CCSP, the activity of SP-D promoter constructs does not require TTF-1. Promoter fragments containing up to 700 bp of upstream sequence are not transactivated by co-transfection of TTF-1 cDNA in H441 or HeLa cells. Although initial studies using pSK-CAT reporter constructs suggested transactivation by TTF-1 cDNA, this proved to be mediated by the plasmid backbone. There was no effect of co-transfection on corresponding SP-D constructs in pTK-CAT, despite transactivation with junD, suggesting that TTF-1 binding is not required for promoter activity. Furthermore, the SP-D promoter constructs showed significant activity in HeLa cells, which lack TTF-1 and do not support significant activity of reporter constructs for other surfactant protein or CCSP in the absence of cotransfected TTF-1 cDNA (54 -57). In summary, the proximal promoter supports cell-restricted but not lung-specific activity in transient transfection assays, and the findings are consistent with the lack of dependence of SP-D promoter activity on TTF-1 binding.
Despite limitations inherent in deducing gene regulatory mechanisms based on in vitro studies using reporter plasmids, these data support the conclusion that the regulation of SP-D is distinct from that of other surfactant proteins. Although the lung appears to be a major site of SP-D expression, there is growing evidence that SP-D is also synthesized by epithelial cells in the upper respiratory tract, gastrointestinal tract, gen-itourinary system, and various mucosal sites (58). Thus, the potential regulatory mechanisms identified in these studies are consistent with pulmonary and extrapulmonary expression and a more generalized role in innate mucosal immunity.