Cis-active elements controlling lung cell-specific expression of human pulmonary surfactant protein B gene.

Human surfactant protein B (SPB) is a 79-amino-acid hydrophobic protein that enhances the surface active properties of pulmonary surfactant. SPB is expressed in nonciliated bronchiolar and alveolar type II cells of the respiratory epithelium, and its expression increases markedly late in gestation. In the present study, a human pulmonary adenocarcinoma cell line, H441, was used in both functional and biochemical assays to identify DNA sequences controlling lung cell-specific expression of the SPB gene. DNase I hypersensitive studies demonstrated two distinct regions of lung cell-specific hypersensitivity located proximal to the SPB promoter and within the eighth intron of the gene. To functionally define these DNA sequences, a series of plasmid vectors were constructed in which segments of the human SPB gene and 5'-flanking sequence were linked to a CAT reporter gene and assayed for expression in lung and nonlung cell lines. Whereas far upstream and intronic sequences did not contain enhancer-like elements, a 259-base pair DNA segment (base pair -218 to +41) was sufficient to support lung cell-specific expression. DNase I footprinting demonstrated that this pulmonary epithelial cell-specific promoter fragment contained five nuclear protein-binding sites, two of which bound lung cell-specific nuclear protein complexes. These results suggest that the pulmonary epithelial cell-specific expression of SPB is determined, in part, by both ubiquitous and cell type-specific protein-DNA interactions within the proximal promoter region.

functional components of pulmonary surfactant (for reviews, see Refs. 1 and 2). The surfactant protein genes are expressed selectively in epithelial cells of the lung. Recently, molecular studies have provided much insight into the spatial, temporal, and hormonal regulation of surfactant protein synthesis during lung development (3-6). However, little is known about the cis-acting sequences and trans-acting factors that regulate surfactant protein gene expression. In particular, it is unclear whether gene expression in distinct cell types of the lung is regulated by common, overlapping, or unique sets of cis-active elements and trans-acting factors. SPB, a low molecular weight hydrophobic protein released by proteolytic processing of a preproprotein (7,8), interacts strongly with surfactant phospholipids to enhance the surface active properties of surfactant mixtures (9,10). SPB is expressed selectively in epithelial cells of the lung and has been detected in both nonciliated bronchiolar cells and alveolar type I1 cells in human, mouse, and rat lung tissue (11)(12)(13)(14).
SPB gene expression increases with advancing gestation and is influenced by a variety of humoral and cellular factors (for review, see Ref. 1). SPB synthesis is enhanced by glucocorticoids in human fetal lung explant tissue, fetal rat lung in uitro, and in two distinct human adenocarcinoma cell lines (14)(15)(16)(17). Other effector molecules, such as phorbol esters and tumor necrosis factor a, inhibit SPB synthesis (18,19). Thus, precise molecular mechanisms have evolved to regulate SPB gene expression.
In previous studies from this laboratory, we described a human pulmonary adenocarcinoma cell line, H441, that expressed SPB mRNA and preproprotein (20,21). In this report, transient expression assays as well as DNase I hypersensitivity and DNase I footprinting were used to define cis-active sequences that regulate SPB gene expression in H441 cells. These studies demonstrate that SPB gene transcription in H441 cells is regulated by both ubiquitous and cell typespecific DNA-binding proteins.

DNase I Hypersensitiuity-H441 and RAJI cells were disrupted by
Dounce homogenization in polyamine buffers modified from that of Hewish and Burgoyne (22). The use of the polyamine buffer was critical in that DNA purified from nuclei that contained calcium exhibited substantial cleavage at the typical hypersensitive sites even in the absence of added DNase I. The polyamine buffer contained 0.34 M sucrose, 53 mM KCl, 13 mM NaCl, 2 mM EDTA, 0.5 mM EGTA, 0.13 mM spermine, 0.5 mM spermidine, 14 mM freshly prepared 2-mercaptoethanol, 0.1% Triton X-100, 13 mM Tris-HC1, pH 7.4, 3 mM MgCI,, and 1 mM freshly prepared phenylmethylsulfonyl fluoride. Nuclei were prepared from the homogenates and centrifuged at 2400 x g for 30 min over a cushion of 1.2 M sucrose in polyamine buffer. The nuclear pellet was washed twice in polyamine buffer without sucrose and detergent and resuspended in a DNase I digestion buffer that contained 60 mM KC1, 5 mM MgClz, 0.1 mM EGTA, 0.5 mM dithiothreitol, 5% glycerol, and 15 mM Tris-HC1, pH 7.5. Nuclei were resuspended at a concentration of 1.25 X lo7 to 3.5 X lo7 nuclei/ ml, and gentle DNase I digestions were carried out in a volume of 0.2 11160 ml with 7 units of DNase I (Boehringer Mannheim) at 30 "c for 1, 2.5, 5, 10, and 15 min. Zero time points were not subjected to DNase I. DNA was prepared from nuclei treated or untreated with DNase I by the addition of an equal volume of a buffer that contained 0.6 M NaC1, 20 mM EDTA, 20 mM Tris-HC1, pH 7.5, and 0.5% SDS. The nuclear lysates were digested with 40 pg/ml of heat-treated RNase A for 2 h at 50 "C followed by 300 pg/ml of proteinase K overnight at 37 "C. DNA was purified by phenol extraction and ethanol precipitation and quantitated spectrophotometrically. DNA samples were digested with HindIII, electrophoresed through agarose gels, blotted to Nytran, and hybridized to probe radiolabeled by means of random primers. The probe was a 1044-bp PCR subfragment of the SPB genomic clone XPG13-2 (bp 6053-7096) and is shown to scale in Fig. 1c.
Plasmids-The isolation and cloning of the entire SPB gene has been reported elsewhere (23). Clone XPG13-2 contains the entire SPB gene and more than 2.2 kb of 5"flanking sequence (23). XPG13-2 was used to clone sequence for all SPB constructions.
Plasmids pSVO-CAT, pRSV-CAT, and pCMV-@gal have been described elsewhere (24, 25). p2244/436-CAT contains SPB genomic sequence from -2244 to +436 in the HindIII site of pSVO-CAT and was constructed in three steps. First, the 2.2-kb SalI-KpnI SPB genomic fragment (bp -2244 to -4) was subcloned into the corresponding sites of pUC-19. Second, these sequences were liberated from the polycloning site of pUC-19 by digestion with HindIII and EcoRI and introduced into the HindIII site of pSVO-CAT with HindIII linkers in a 5' to 3' orientation with respect to the CAT gene to give plasmid p2.2-CAT. Sequences downstream of the KpnI site (-5 to +436) were amplified from XPG13-2 using the PCR to generate a KpnI-HindIII-linkered fragment containing a single base pair substitution at +15 (A to T). This fragment was cloned into the KpnI and downstream HindIII site of p2.2-CAT to give p2244/436-CAT. The single base pair change at +15 alters the translation start signal encoded in SPB exon I (AUG to UUG) and was necessary to prevent the generation of an SPB-CAT fusion protein (26).
Cell Culture-Human lung adenocarcinoma cell line NCI-H441 was maintained in RPMI medium containing 10% fetal bovine serum. Human lung adenocarcinoma cell line A549 and HeLa cells were maintained in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum. GM 4671 (RAJI) is a human B-lymphoid cell line and was maintained as described (27). All cell lines were cultured at 37 "C and 5% CO,.
Transient Transfection-A mixture of 5 pmol of test plasmid was mixed with 2.5 pmol of the internal control plasmid pCMV-Bgal and coprecipitated by the calcium phosphate procedure. Precipitates (1 ml) were added directly to the tissue culture medium. Eighteen to 24 h subsequent to transfection the cells were washed and the medium was changed to RPMI with 10% fetal bovine serum. Cells were harvested by scraping 24 or 48 h later. Assays for 0-galactosidase were performed according to Miller (25). CAT assays were performed as described by Gorman et al. (24). Chloramphenicol, [dichloroacetyl-1,2-"C], and its derivatives were separated by thin layer chromatography. The percent acetylation was quantitated using a Molecular Dynamics PhosphorImager. To ensure linearity of the assay, data were quantitated from CAT assays in which less than 20% conversion had occurred. Relative CAT activities were calculated by comparing the activities of the promoter-containing plasmids with the activity 1993), SUUI (pA5'-1552), BstEII (pA5'-1414), StuI (pA5"900), pSVO-CAT. of pSVO-CAT (which produced 0.082% acetylationhnit of P-galactosidase activity/h in H441 cells and 0.018% acetylation/unit of 8galactosidase activity/h in HeLa cells) within each cell line following correction for transfection efficiency. Although transfection efficiencies (units &galactosidase activity/pg protein) and absolute CAT conversion varied between experiments (approximately 2-10-fold), relative CAT activities were similar between experiments.
DNA probes for footprint analysis were prepared by using the PCR and 32P-end-labeled synthetic oligonucleotide primers. The SPB genomic clone, XPG13-2, was used as template for the amplification of downstream primers used were (5"CAGGAACATGGGAGTCTGG sequence between base pairs -221 and +81. The upstream and G) and (5'-CAGTGCCTGGGCCACAGAGC), respectively. The upstream or downstream primer (3 pmol) was 32P-end-labeled in a 20-p1 kinase reaction mixture containing 30 pmol of [r3'P]ATP as described (31). Kinase reactions were terminated by incubation at 65 "C for 10 min and added directly to a standard 100-p1 PCR reaction mixture containing 3 pmol of unlabeled primer oligonucleotide and 500 ng of template DNA. PCR products were isolated using Promega PCR Preps DNA Purification System.
The DNase I protection assay was performed in a 50-pl reaction.
DNA binding reactions were carried out in a mixture containing 10 mM Tris, pH 7.5, 0.5 mM dithiothreitol, 5 mM MgC12, 0.1 mM EDTA, 75 mM KC1,0.2 mM phenylmethylsulfonyl fluoride, and 12% glycerol. Nuclear proteins were incubated with 2 pg of poly(d1. dC) competitor DNA for 15 min at 0 "C prior to the addition of 20,000 counts/min of labeled DNA (about 0.3 ng). After another 60-min incubation at 0 "C, the samples were set at room temperature and after 5 min digested with DNase I (Promega) for 2 min. The reactions were stopped by the addition of 350 pl of stop buffer containing 230 mM NaCl, 17 mM EDTA, 1.14% SDS, 11.4 mM Tris, pH 7.8, and 230 pug/ ml proteinase K. DNA was purified by phenol extraction and ethanol precipitation. DNA samples were fractionated on 6% polyacrylamide, 7 M urea sequencing gels.

Identification of DNase I hypersensitive Sites
Flanking the SPB Promoter-Because many enhancer-like elements and other functional regions are associated with perturbations of chromatin structure, DNase I hypersensitivity (DH) assays were used to evaluate the SPB gene and 5'-flanking DNA. A 12.1-kb HindIII fragment was used to map DH sites. This fragment contained over 5 kb of 5"flanking sequence and over 8 kb of intragenic sequence extending to a HindIII site in intron 10. Autoradiograms of the indirectly end-labeled fragments that were generated by DNase I treatment of nuclei are shown in Fig exists in a unique structure which is sensitive to DNase I a n d indicates that important regulatory regions may lie in close proximity to the promoter or within the gene. Sequences Flanking the SPB Promoter Direct Lung Cellspecific Expression-To determine if sequences encompassing DNase I hypersensitive sites I and I1 were associated with functional transcriptional regulatory domains, 2.7 kb of sequence (-2244 to +436) was linked to a CAT reporter gene.

FIG.
1. Cell type-specific DNase I hypersensitivity in the SPB gene and promoter region. A and B, nuclei were isolated from lung (H442) and nonlung (RAJI) cell lines and exposed to DNase I for 0, 1, 2.5, 5, 10, and 15 min (indicated 1-6 for each cell line). The sizes of the parental fragment and the fragments liberated by DNase I digestion are indicated in kilobases within parentheses. The sizes of each fragment were determined by comparison to known standards in each gel. Each distinct DNase I-liberated fragment is indicated by a Roman numeral (I-IV). In preparations of H441 cell nuclei, it was not possible to completely eliminate endogenous DNase I activity, and daughter fragments were detected even in the absence of added DNase I ( h e 1 ) . C, a schematic of the SPB gene and flanking sequence is shown. The transcriptional activity of this construction (~22441436-CAT) in the indicated cell lines was determined by transient transfection. Increased transcription of the CAT reporter was observed only in H441 cells, where an approximate 10-fold increase in expression relative to promoterless vector pSV0-CAT was observed ( Fig. 2A, lanes 1 and 2). Transfection of ~22441436 into A549 cells, a human pulmonary adenocarcinoma cell line that does not express SPB, or HeLa cells did not support CAT transcription above promoterless vector ( Fig. 2, B and C, lanes 1 and 2). This result indicated that a human lung adenocarcinoma cell line, H441, was capable of expressing chimeric SPB-CAT genes and that the human SPB gene promoter and flanking sequences contained within -2244 to +436 was transcriptionally active in a cell typespecific manner.
To determine if sequence encompassing hypersensitive sites I11 and IV contained additional regulatory elements, a genomic subfragment spanning intron eight was subcloned into the BamHI site downstream of the CAT reporter gene and SPB promoter and flanking sequence (-2244 to +436). The transcriptional activity of this construction was similar to p22441436-CAT (data not shown). This result suggested that DHIII and DHIV were not associated with a typical enhancer element.
Deletion Analysis of Sequence Flanking the SPB Promoter-To better delineate cis-acting sequences that regulate SPB transcription in H441 cells, a series of 5'-flanking deletions t +.a36 of SPB sequence were analyzed in transient expression assays. Each 5' deletion mutant had the same 3' end point at +436, containing sequence into SPB exon 2. A summary of the results obtained from transfection of these CAT reporter constructs is shown in Fig. 3. Each construction was assayed for expression in both H441 and HeLa cell lines. CAT activity varied in H441 cells with deletion of 5'-flanking DNA to -218 (pA5'-218), but there was no loss of activity relative to ~22441 436-CAT and no construction expressed above the level of pSVO-CAT in HeLa cells. However, deletion of sequence to -80 (pA5'-80) resulted in an 82% reduction in reporter activity compared to p2244/436-CAT, suggesting that a positive cis-active element was located between -218 and -80.
To determine if additional regulatory elements were located downstream of the SPB transcription start site, we constructed a series of 3' intragenic deletion mutants. Each 3' deletion mutant had the same 5' end point at -2244 bp. A summary of the results obtained from transient expression of these CAT reporter constructs in H441 and HeLa cells is shown in Fig. 4.

FIG. 2. Cell-specific function of the SPB promoter region.
The SPB promoter-CAT construction, p2244/436-CAT, contains bp -2244 to +436 and is described under "Materials and Methods." Each plasmid (5 pmol) containing the CAT reporter gene was cotransfected along with 2.5 pmol of pCMV-@gal into the indicated cell lines. CAT activity was measured 48 h later and normalized to @galactosidase activity. The activity in each cell line is compared to the promoterless vector pSVO-CAT as described under "Materials and Methods." pRSV-CAT is an external positive control for CAT activity. The results shown are representative of a t least three independent transfections.
suggests the existence of a second positive regulatory element located between +8 and +38. Finally, the deletion of both 5'flanking DNA to -218 and adjacent intragenic DNA to +41 (p218/41) demonstrated that a 259-bp promoter fragment was sufficient to support a level of cell type-specific CAT expression similar to p2244/436-CAT.

Identification and Cellular Specificity of Nuclear Proteinbinding Sites within the SPB Promoter-To identify nuclear
protein-binding sites within the SPB promoter and flanking sequence, DNase I footprinting experiments were performed using extract prepared from lung (H441) and nonlung (HeLa) cell lines. Five nuclear protein-binding sites were identified using H441 nuclear extracts on both the coding and noncoding DNA strands (single and double lines, Fig. 5, A and B ) . In addition, multiple DNase I hypersensitive sites, reflected as more intense bands of digestion, were observed between and within some of the footprinted regions (arrowheads, Fig. 5, A  and B ) . This type of DNase I footprint has been described previously for complex promoters and enhancers containing multiple closely spaced cis-active elements and may reflect the bending of DNA adjacent to these sites (32, 33). Two footprinted regions, designated S P B factor 1 (SPB-fl; bp -107 to -93) and SPB factor 2 (SPB-f2; bp -90 to -73), were protected only with H441 cell nuclear extract (double lines, Fig. 5, A and B ) . The 5'-most binding site, SPB-fl, did not contain any previously identified enhancer or promoter motif. SPB-f2 contained a sequence motif for hepatocyte nuclear factor 5 (HNF-5; TGTTTGT), a transcription factor previously described in liver (3435). Three additional nuclear protein-binding sites were identified in both H441 and HeLa cell nuclear extracts (single lines, Fig. 5, A and B ) and designated SPB factor 3 to 5 (SPB-f3 to SPB-f5). SPB-f3 contained a six of nine match to the consensus CAAT box. SPB-f4 contained a TATA box and Spl-binding site motif. Notably, SPB-f5 was located entirely within the protein coding region of the gene and encompassed a consensus AP1-binding site motif (5"TGAGTCA). The locations of protected sequences and binding site motifs are summarized in Fig. 6.
Comparison of the human SPB promoter proximal region to the corresponding murine sequence' revealed uninterrupted conservation of 11 (TGGAGGGCTCT) and 12 (CAAACACT-GAGG) nucleotides in the SPB-fl-and SPB-E-binding sites, respectively. Much less conservation was found in regions protected by both H441 and HeLa cell nuclear extract. Only 4 of 16, 6 of 24, and 15 of 19 nucleotides were conserved in the SPB-f3-, SPB-f4-, and SPB-f5-binding sites, respectively. Within SPB-f4, the murine sequence did not contain an Spl motif, however, a 7-bp TATA box element was conserved.
Although an AP1-binding site motif was not identified within the murine sequence corresponding to SPB-f5 in exon 1, this motif was identified 7 bp downstream of the murine TATA box. Taken together, these experiments demonstrate that the SPB promoter proximal region contains five nuclear proteinbinding sites, two of which bind novel lung cell-specific nuclear protein complexes. In particular, with the exception of the HNF-5 motif in SPB-f2, the sequence of the DNase I footprints specifically protected in H441 cells does not correspond to any known promoter or enhancer binding site motif and was conserved between the human and murine genes, suggesting that these elements represent novel lung cell-specific transcriptional regulatory pathways.

DISCUSSION
There is currently very little information on the DNA regulatory elements or transcription factors that direct lungspecific gene expression. The distinct regulated expression of SPB in both nonciliated bronchiolar and alveolar type I1 cells of the respiratory epithelium suggested that studies of the transcriptional regulation of this gene may provide important insights into lung-specific gene expression. This report demonstrates that the lung cell-specific transcription of the SPB gene is dependent on a 259-bp promoter fragment. This region is associated with a prominent domain of lung cell-specific DNase I hypersensitivity and interacts with both ubiquitous and lung cell-specific nuclear DNA-binding proteins.
T o identify putative distal regulatory elements, we have exploited the DNase I hypersensitivity assay (36,37). This method has provided consistent correlation between the location of DNA regulatory elements, such as enhancers or silencers, and the occurrence of DNase I hypersensitive sites (36,37). The most striking finding in examining the DNase I hypersensitivity pattern of the SPB gene and 5"flanking region was the cellular specificity of DH sites found close to or within the SPB promoter region and the lack of additional hypersensitivity within 5 kb of additional upstream sequence. Because those enhancers which have been examined are associated with DH sites (36, 37), this result suggested that sequence far upstream of DHI and DHII did not contain characteristic enhancer domains. In agreement with this finding, deletion of sequence between -2241 and -218 did not significantly alter the maximal transcriptional activity of the SPB promoter in transient expression assays. Taken together, * M. D. D'Amore-Bruno and J. A. Whitsett, manuscript in preparation. 3. 5' deletion analysis of the SPB promoter region. Each 5' deletion mutant was constructed from p2244/436 as described under "Materials and Methods." The 5"terminal nucleotide of the deletions is indicated. Each plasmid was cotransfected with pCMV-@gal into H441 and HeLa cells, and CAT activity was normalized to P-galactosidase activity. Relative CAT activities were calculated by comparing the activities of the SPB promoter containing plasmids with those of pSVO-CAT as described under "Materials and Methods." The lower line shows the location of consensus binding site motifs found within the SPB promoter region. Each determination of CAT activity represents the average of three independent transfections whose standard deviation was less than 25%. The construction of each downstream deletion mutant is described under "Materials and Methods." The extent of each deletion is shown relative to p2244/436 by broken lines. Each plasmid was cotransfected with pCMV-pgal into H441 and HeLa cells,and CAT activity was normalized to @galactosidase activity. Relative CAT activities were calculated by comparing the activities of the SPB promoter containing plasmids with those of pSVO-CAT as described under "Materials and Methods." Each determination of CAT activity represents the average of three independent transfections whose standard deviation was less than 25%. these data demonstrate that sequences sufficient to direct lung cell-specific expression of S P B reside within the proximal promoter region.

SPB Pmmofer Downstream Deletions
The failure of DHIII and DHIV to alter CAT activity in transient expression experiments indicates that these sites do not correspond to typical enhancer elements. Although it has been emphasized that not all DH sites denote transcriptionally functional domains (36, 37), the function of these sites may not be discernible outside of their genomic context or by of DNA required for tissuespecific expression in transgenic mice (39). This requirement for stable integration is also true for the hemoglobin locus control region and the recently identified distal regulatory region of the mouse MyoD gene (40,41). In none of these cases is there an understanding of the role integration plays in regulating the activity of these regions. To determine if higher order structure influences the regulatory properties of the SPB gene promoter and flanking sequence, we are now examining the spatial and temporal expression conferred by these sequences in transgenic mice and stable transfection ~~ experiments. Preliminary results3 indicate that the SPB promoter proximal region (bp -218 to +41) is sufficient to direct lung-specific expression of a CAT reporter gene in transgenic mice; however, the level of expression from this transgene is low and may reflect a species-specific difference or, alternatively, a lack of appropriate regulatory elements. If additional elements exist, they may contain integration-dependent enhancer elements similar to the locus control region and the distal regulatory region or elements like the A element of the lysozyme gene (42) and DHII of the adenosine deaminase gene (43) which give copy number-dependent expression but lack enhancer activity. Alternatively, additional regulatory regions may be located either far upstream or downstream of the SPB gene locus; a muscle-specific enhancer is located more than 24 kb downstream of the myosin light-chain 1/3 gene locus (44).
DNase I footprint analysis of the human SPB promoter revealed five nuclear protein-binding sites between bp -102 and +32. The two 5'-most binding sites, SPB-fl and SPB-D, interacted with nuclear proteins present only in H441 cells, and deletion of these sites resulted in significant reduction in the transcriptional activity of the SPB promoter. With the exception of an HNF5 motif identified in SPB-f2, the se-R. J. Bohinski, unpublished observations. quence of SPB-fl and SPB-f2 did not contain significant homology to more than 150 functional elements for vertebrate genes (45). A search of the 5'-flanking regions of genes that are expressed in the lung, including human and murine surfactant proteins A and C, and rat Clara cell secretory protein, did not reveal elements with significant homology to SPB-fl or SPB-f2. However, it is possible that once important bases for binding are identified and/or transcriptional proteins are isolated or cloned, binding sites in these or other lung genes will become evident. Comparison of the human and murine SPB 5"flanking sequence demonstrated that SPR-fl and SPB-f2 were evolutionarily conserved in spite of sequence divergence outside of this region. The final indication that SPB-fl and SPB-f2 are important to the lung cell specificity of S P B gene regulation was the low promoter activity in HeLa cells which lacked SPB-fl and SPB-f2 binding activity but contained SPB-f3 to SPB-f5 binding activity. The HNF5 motif in SPB-f2 is noteworthy for several reasons. First, the interaction of HNF5 with its binding site within the tyrosine aminotransferase (TAT) 5"flanking DNA results in an in vitro DNase I footprint characterized by the presence on each strand of a DNase I-hypersensitive site that is situated between the fifth and sixth base of the motif (34,35). This unique pattern of protection and hypersensitivity is similar to the footprint associated with the HNF5 motif within SPB-f2. Second, HNF5 is involved in a complex mechanism of TAT gene activation involving the concerted action of HNF5 and a glucocorticoid receptor (34,35). Interestingly, SPB gene expression has been shown to respond dramatically to corticosteroids in both fetal lung explants and two distinct pulmonary adenocarcinoma cell lines (14-13, and part of this response has been attributed to modulation of SPB transcription (16). Third, several hepatocyte-enriched transcription factors have also been identified in the lung, including hepatocyte nuclear factor 3 and CCAAT/enhancer-binding protein family members (46). Recently, Sawaya et aL4 have described a functional hepatocyte nuclear factor 3-binding site within the rat Clara cell secretory protein promoter, a lung cellspecific gene with expression that overlaps but is distinct from SPB (30). Although the tissue distribution of HNF5 and the precise cellular distribution and transcriptional role of hepatocyte nuclear factor 3 and CCAAT/enhancer-binding protein family members within the lung is not currently known, it is possible that transcription factors specifying hepatocyte-specific gene expression have also been adapted to direct gene expression in the lung.
In addition to 5"flanking sequences, an intragenic DNA sequence (bp +8 to +38) was critical to SPB transcriptional activity. DNase I footprinting demonstrated that this site (SPB-f5) interacted with nuclear protein(s) from both H441 and HeLa cell nuclear extracts. Sequence analysis of this region identified a consensus DNA-binding site for the AP1 transcription factor (47). Interestingly, this site was contained entirely within the protein coding region of the gene. A similarly positioned cis-active element has previously been reported within the first translated exon of the glial fibrillary acidic protein and has been shown to function as an initiator element (48). Initiator elements have been proposed to act independently (49) or in conjunction with additional elements (50) to accurately direct the initiation of transcription. This process is distinct from that proposed for elements, including AP1-, Spl-, and CAAT box-binding protein, that modulate the rate of transcription (51, 52). Although the sequence motifs associated with SPB-f5 are not homologous to any previously described initiator elements, additional studies will be necessary to define the role of this binding site in SPB promoter function.
Finally, these results demonstrate that the positive regulatory influence of SPB-fl and SPB-f2 or SPB-f5 cannot function independently in the context of the SPB promoter. Deletion of either site resulted in a low level of expression, and preliminary data from heterologous gene systems indicate that these elements alone will not support the normal regulatory activity associated with SPB. Rather, it is likely the P. L. Sawaya, B. R. Stripp The transcription start site is indicated by an arrow and labeled +l.
concerted action of each of these regions that contributes to the expression of this gene. Current studies are investigating whether individual or multiple lung specific and/or ubiquitous trans-factors must interact within this region to produce appropriate expression. The finding that the SPB promoter region contains two evolutionarily conserved and previously undescribed nuclear protein-binding sites and that at least one of these sites is not related to any previously described lung regulatory region or to other consensus sites, strongly suggests the existence of novel lung cell-specific transcription factors. These results should facilitate studies designed to elucidate the mechanisms of cell type-specific gene expression within the lung.