The synergistic action of hepatocyte nuclear factor (HNF)-1α and HNF-4 plays an important role in expression of the α1-antitrypsin (α1-AT) gene in human hepatic and intestinal epithelial cells. Recent studies have indicated that the α1-AT gene is also expressed in human pulmonary alveolar epithelial cells, a potentially important local site of the lung antiprotease defense. In this study, we examined the possibility that α1-AT gene expression in a human pulmonary epithelial cell line H441 was also directed by the synergistic action of HNF-1α and HNF-4 and/or by the action of HNF-3, which has been shown to play a dominant role in gene expression in H441 cells. The results show that α1-AT gene expression in H441 cells is predominantly driven by HNF-1β, even though HNF-1β has no effect on α1-AT gene expression in human hepatic Hep G2 and human intestinal epithelial Caco-2 cell lines. Expression of α1-AT and HNF-1β was also demonstrated in primary cultures of human respiratory epithelial cells. HNF-4 has no effect on α1-AT gene expression in H441 cells, even when it is cotransfected with HNF-1β or HNF-1α. HNF-3 by itself has little effect on α1-AT gene expression in H441, Hep G2, or Caco-2 cells but tends to have an upregulating effect when cotransfected with HNF-1 in Hep G2 and Caco-2 cells. These results indicate the unique involvement of HNF-1β in α1-AT gene expression in a cell line and primary cultures derived from human respiratory epithelium.
- protease inhibitors
- tissue-specific gene expression
α 1 -antitrypsin (α1-AT) is an ∼55-kDa serum glycoprotein that inhibits the destructive neutrophil proteases elastase, cathepsin G, and proteinase 3 (28). It is the archetype of a family of serum proteins, many of which are serine protease inhibitors and are therefore called serpins (6, 18). Lack of elastase inhibitory activity in the lung is thought to be responsible for the predisposition of α1-AT-deficient individuals to destructive lung disease/emphysema (20). Moreover, functional inactivation of α1-AT by active oxygen intermediates released during cigarette smoking is believed to play a role in pulmonary emphysema in α1-AT-sufficient individuals (20). A subgroup of individuals with the classical form of α1-AT deficiency are predisposed to liver injury and hepatocellular carcinoma presumably because of the hepatotoxic effect of the mutant α1-AT molecule within liver cells (28).
Plasma α1-AT is predominantly derived from the liver, as shown by studies of changes in α1-AT allotypes after orthotopic liver transplantation (1, 16). Synthesis of α1-AT is abundant in human hepatoma cell lines and in human hepatocytes in primary culture, and α1-AT mRNA is extremely abundant in hepatocytes in human liver, as determined by in situ hybridization analysis (26). There is also evidence for extrahepatic sites of synthesis, including blood monocytes, tissue macrophages, and intestinal epithelial cells (26, 29). Our recent studies have indicated that expression of the α1-AT gene in hepatocytes and enterocytes is predominantly driven by the synergistic action of hepatocyte nuclear factor (HNF)-1α and HNF-4 (17). HNF-1α plays an important role in the increase in α1-AT gene expression that accompanies differentiation of enterocytes from crypt to villous tip. HNF-1β, in the absence or presence of HNF-4, has no effect on α1-AT gene expression in hepatocytes or enterocytes (17).
Several recent studies have indicated that α1-AT is also synthesized in human pulmonary airway epithelial cells, including the Calu-3 and A549 cell lines and normal bronchial epithelial cells in primary culture (9, 34). Previous studies have shown α1-AT mRNA in bronchial and bronchiolar epithelial cells of human fetal lung by in situ hybridization analysis (22). Expression of α1-AT by airway epithelial cells could potentially constitute an important component of the antiprotease defense system locally within the lung.
In this study, we used the human pulmonary epithelial cell line H441 as well as nuclear extracts and RNA from human respiratory epithelial cells in primary culture to examine the possibility that α1-AT gene expression in pulmonary epithelial cells is directed by the synergistic action of HNF-1α and HNF-4 and/or by the action of more pneumocyte-specific transcription factors such as HNF-3, which have been shown to play an important role in gene expression in pneumocytes (3-5, 30).
MATERIALS AND METHODS
Progressive deletions of the human α1-AT promoter region from −1951 to −2 were generated by PCR amplification using the human genomic α1-AT clone hAAT7zf (kindly provided by Dr. K. Ponder, St. Louis, MO) as template. These PCR fragments were subcloned into the KpnI/HindIII site of the pGL3 basic luciferase reporter vector (Promega, Madison, WI) to generate α1-AT promoter-luciferase fusion plasmids. The Mut-2 plasmid, which is an α1-AT (−137 to −2) promoter-luciferase chimera with a mutation in the HNF-1-binding site, has been previously described (17). Our previous electrophoretic mobility shift assay (EMSA) studies showed a marked decrease in binding of HNF-1α and HNF-1β to this plasmid compared with the plasmid without alterations in the HNF-1-binding region (17). The Mut-3 plasmid, which contains a mutation in HNF-3 binding, was generated for these studies. For this plasmid, we altered two nucleotides in the HNF-3-binding site at −101 to −93 (5′-TGTTTGCTC-3′) within the α1-AT (−137 to −2)-promoter luciferase chimera. The highly conserved thymidines at −101 and −97 were converted to adenines. Results of EMSA indicated that these mutations disrupted binding to nuclear proteins in extracts from Hep G2, Caco-2, and H441 cells (data not shown).
Murine HNF-1α and HNF-1β expression plasmids (25) provided by Dr. T. C. Simon (St. Louis, MO) were subcloned into the pSG5 plasmid (Stratagene, La Jolla, CA). Full-length HNF-4 cDNA subcloned into the pMT2 expression vector (31) was provided by Dr. J. Rottman (St. Louis, MO). HNF-3α and HNF-3β expression plasmids pPac HNF-3α and pPac HNF-3β (12) were provided by Dr. G. Suske (Marburg, Germany). The pRL-TK vector was purchased from Promega.
Cell culture, DNA transfections, and luciferase assay.
The Hep G2, HeLa, and Caco-2 cells were grown as previously described (26). H441 cells were purchased from American Type Culture Collection and maintained in RPMI 1640 medium with 2 mMl-glutamine, sodium bicarbonate (0.5 g/l), 10 mM HEPES, 1.0 mM sodium pyruvate, and 10% fetal bovine serum. This cell line is derived from a human pulmonary papillary adenocarcinoma. It expresses surfactant proteins A and B. Electron-microscopic analysis shows the presence of multilamellar bodies and cytoplasmic structures resembling Clara cell granules in these cells (27). The promoter-luciferase plasmids were cotransfected with the indicated amount of HNF-1α, HNF-1β, HNF-3α, HNF-3β, and/or HNF-4 expression vectors. A pGL3-basic vector was used as a negative control and a pGL3-control vector containing SV40 promoter and enhancer sequences was used as a positive control in all transfection experiments. The pRL-TK vector was also included in all transfections as an internal control for transfection efficiency as monitored by the Promega dual-luciferase reporter assay system. Cells (1–1.5 × 106) were plated on 60-mm tissue culture dishes or six-well plates and incubated for 24 h before transfection. Duplicate or triplicate dishes were transfected using the FuGENE 6 transfection reagent or the calcium phosphate-DNA coprecipitation method. For the calcium phosphate method, the cells were shocked with 15% glycerol 24 h after transfection. Cells were harvested 48 h after transfection. Luciferase activity was detected on the Turner Designs luminometer (model TD-120/20, Promega).
Preparation of double-strand oligonucleotides and labeling.
Oligonucleotides were synthesized on an Applied Biosystems DNA synthesizer in the Nucleic Acid Chemistry Laboratory, Biotechnology Center, Washington University School of Medicine. The potential HNF-1-binding sequence corresponds to the α1-AT promoter sequence from −87 to −52 (5′-ATAACTGGGGTGA-CCTTGGTTAATATTCACCAGCAG-3′). The potential HNF-4-binding sequence corresponds to the sequence from −128 to −93 (5′-ATCCAGCCAGTGGACTTAGCCCCGTTTG-3′) with respect to the downstream transcriptional start site. The HNF-3 consensus sequence corresponds to 5′-GCCCATTGTTTGTTTTAAGCC-3′, and the HNF-4 consensus sequence corresponds to 5′-GGAAAGGTCCAAAGGGCGCCTTG-3′. The oligonucleotides were purified by denaturing PAGE, annealed at 65°C for 1 h, allowed to cool slowly at room temperature, and then labeled using [α-32P]dCTP and the Klenow fragment of DNA polymerase I. Radiolabeled oligonucleotides were separated from free nucleotide by nondenaturing PAGE.
Complementary oligonucleotide probes for the HNF-1 and HNF-4 sequences in the α1-AT promoter as well as consensus HNF-1, HNF-3, and HNF-4 sequences were labeled with [32P]dCTP by the 3′-end filling reaction, and 2 × 104 counts/min were incubated for 20–40 min at room temperature with nuclear extracts from Hep G2, Caco-2, H441, and human respiratory epithelial cells (25). The nuclear extracts from H441, Hep G2, and Caco-2 cells were prepared according to the protocol described by Dignam et al. (11). Nuclear extracts from HeLa cells (HeLa Scribe nuclear extract, in vitro transcription grade) were purchased from Geneka Biotechnology (Montreal, PQ, Canada). Nuclear extracts from human respiratory epithelial cells in primary culture, also prepared by the method of Dignam et al., were kindly provided by Dwight Look (St. Louis, MO). Nuclear extracts were resuspended in buffer so that a 10-μl reaction volume contained a final concentration of 60 mM KCl, 25 mM HEPES, pH 7.6, 1 mM dithiothreitol, 0.1 mM EDTA, 7.5% glycerol, 1 μg of poly(dI-dC), and 2% polyvinyl alcohol. Unlabeled oligonucleotide in molar excess was used in designated experiments. Nuclear extract dialysis buffer and BSA were used as negative controls. Rabbit polyclonal anti-HNF-1α TC 284 (kindly provided by M. Yaniv and M. Pontoglio, Paris, France), anti-HNF-1β [kindly provided by G. Ryffel, Essen, Germany (35)], anti-HNF-3α and HNF-3β (kindly provided by R. H. Costa, Urbana, IL), and anti-HNF-4 (kindly provided by I. Talianidis, Crete) were used in designated experiments by incubation with nuclear extracts for 2 h at 4°C before addition of radiolabeled oligonucleotide probe. The products were analyzed on 5% polyacrylamide-2.5% glycerol gels cast in 0.5× Tris-borate-EDTA.
RT-PCR for analysis of RNA levels.
Total RNA from H441, Hep G2, and Caco-2 cells was isolated with the RNeasy kit (Qiagen, Hilden, Germany) and then digested with RQ1 RNase-free DNase (Promega). Poly(A)+ RNA was isolated using oligo(dT) cellulose column chromatography (32). Poly(A)+ RNA from human respiratory epithelial cells was kindly provided by Drs. Theresa Joseph and Dwight Look. The RNA was then subjected to RT-PCR using the Access RT-PCR system (Promega). The primers for amplification were based on previous studies (10,12): 5′-GAAAGCAACGGGAGATCCTCCGAC-3′ (sense) and 5′-CCTCCACTAAGGCCTCCCTCTCTTCC-3′ (antisense) for HNF-1β, 5′-GTAGACAGTAGGGGCTC-3′ (sense) and 5′-GGGGAATCCTTTAAACGG-3′ (antisense) for HNF-3α, 5′-GCCTGAGCCGCGCTCGGGAC-3′ (sense) and 5′-GGTGCAGGGTCCAGAAGGAG-3′ (antisense) for HNF-3β, 5′-CTTCCTTCTTCATGCCAG-3′ (sense) and 5′-ACACGTCCCCATCTGAAG-3′ (antisense) for HNF-4, 5′-TCACGTCTAGAACAGTGAATCGAC-3 (sense) and 5-GTGGGCTGCAGTACCAGCTCAACC-3′ (antisense) for α-AT, and 5′-AGGGCTGAGTGTTCTGGGATTTC-3′ (sense) and 5′-GGTTACGGCAGCACTTTTATTTTT-3′ (antisense) for β-actin. Minimal modifications were applied to the manufacturer's instructions for the conditions: annealing was done at 60–65°C depending on melting temperature of oligonucleotides, the concentration of MgSO4 was 1 mM, and 35 cycles were used for amplifications. PCR products were separated on 1% agarose gels and visualized by ethidium bromide staining.
Expression of α1-AT in the human pulmonary epithelial cell line H441.
RT-PCR analysis showed the presence of α1-AT RNA in the H441 cell line as well as in Hep G2 and Caco-2 cells (Fig.1). This data indicated that we could use the H441 cell line as a model for expression of the α1-AT gene in pulmonary epithelium.
To determine whether there were cis-acting regulatory elements within the upstream flanking region for expression of the α1-AT gene in H441 cells and whether these differed from those responsible for expression in Caco-2 and Hep G2 cells, we used seven α1-AT-luciferase fusion plasmids differing in the length of the α1-AT 5′-flanking sequence upstream of the luciferase coding sequence (Fig. 2). We examined the expression of these plasmids in H441, Caco-2 and Hep G2 cells. The results show that the general localization ofcis-acting elements in H441 cells is similar to that in Caco-2 and Hep G2 cells. There is a drop in expression on deletion from −991 to −661, −490 to −270, −270 to −137, and −137 to −2, indicating the presence of cis-acting elements in these regions. The major positive elements were in the two most proximal regions. We decided to first examine the most proximal region, −137 to −2, in H441 cells in more detail. Previous studies showed that this region contains most of the elements responsible for tissue-specific regulation of transcription (17 and references therein). Moreover, all the potential HNF-1-, HNF-4-, and HNF-3-binding sequences are found in this proximal promoter region.
Binding of nuclear proteins from H441 cells to the proximal α1-AT promoter.
Because our previous studies indicated that HNF-1 and HNF-4 andcis-acting HNF-1- and HNF-4-binding elements in the proximal upstream flanking region of the α1-AT gene play a major role in expression of α1-AT in one extrahepatic site, intestinal epithelium, we first used EMSA to determine whether HNF-1 and HNF-4 are expressed in H441 cells and whether they bind to the HNF-1 (−87 to −52)- or HNF-4 (−128 to −93)-binding regions of the proximal α1-AT promoter (Fig.3). The results show that the HNF-1-binding region (−87 to −52) forms a single complex with nuclear extracts from H441 cells compared with two complexes present in Hep G2, undifferentiated Caco-2 (day 1), and differentiated Caco-2 (day 4) cells. Our previous studies showed that the more slowly migrating complex corresponds to HNF-1α and the more rapidly migrating complex corresponds to HNF-1β (17). These previous studies also showed an increase in the relative proportion of HNF-1α compared with HNF-1β during differentiation of Caco-2 cells. The single complex present in H441 cells appears to correspond to HNF-1β. These data also show a marked increase in the relative proportion of HNF-1α compared with HNF-1β in Hep G2 compared with Caco-2 cells and a trace of the HNF-1β complex in HeLa cells that do not express α1-AT.
The HNF-4-binding region (−128 to −93) forms a single complex with Hep G2 and Caco-2 cells but does not form a complex with nuclear proteins from H441 or HeLa cells (Fig. 3). This complex is much more abundant in Caco-2 than in Hep G2 cells and decreases during differentiation of Caco-2 cells. Because it comigrates with the complex formed with the HNF-4 consensus sequence and migrates faster than the complex formed with the HNF-3 consensus sequence, this complex is probably formed by HNF-4, not HNF-3.
The HNF-3 consensus sequence forms a single complex in H441, Hep G2, Caco-2, and HeLa cells. These data indicate that HNF-3 is present in H441, Hep G2, and Caco-2 cells but does not bind to the only two potential sites for HNF-3 binding in the proximal promoter of the α1-AT gene. Together, these initial EMSA studies suggest that HNF-1β, but not HNF-3, binds to the proximal α1-AT promoter in H441 cells. There is no evidence for HNF-4 in H441 cells.
To determine the specificity of each of these complexes, EMSA was done in the absence or presence of unlabeled specific and unrelated oligonucleotide competitors and antibodies. The single complex formed in H441 cells with the HNF-1-binding region of the α1-AT promoter (−87 to −52) is blocked by unlabeled −87 to −52 oligonucleotide but not by unlabeled glucocorticoid response element (GRE) oligonucleotide (Fig.4 A). It is also blocked by antibody to HNF-1β (Fig. 4 B). Our previous studies showed that the more slowly migrating complex formed with the HNF-1-binding region of the α1-AT promoter in Hep G2 and Caco-2 cells is supershifted by antibody to HNF-1α (17). In Fig.4 B, we also examined nuclear extracts from primary cultures of human respiratory epithelial cells and found a single complex that was supershifted by antibody to HNF-1β. These results indicated that HNF-1β also binds to the proximal α1-AT promoter in human airway epithelium in vivo. It is not clear to us at this time why the single complex in human respiratory epithelial cells is slightly slower in migration than that in H441 cells.
Next, we examined the specificity of complexes formed with the HNF-4-binding region of the α1-AT promoter. The single complex formed with this region in Hep G2 and Caco-2 (18) cells is blocked by unlabeled −128 to −93 oligonucleotide, but not by unlabeled GRE oligonucleotide, and is supershifted by antibody to HNF-4 (Fig. 4 C). A complex is not formed with this region by nuclear proteins from H441 cells.
The single complex formed by nuclear proteins from H441 cells with the HNF-3 consensus sequence is blocked by antibody to HNF-3α and antibody to HNF-3β but not by antibody to α1-AT (Fig.4 D). It is also blocked by unlabeled HNF-3 but not by unlabeled GRE oligonucleotide. This shows the identity and specificity of the complex formed with the HNF-3 consensus sequence by HNF-3 in H441 cells. Together with the results in Fig. 1, these data show that HNF-3 is present in H441 cells but does not bind to the only potential HNF-3-binding sites in the proximal α1-AT promoter.
Role of HNF-1 and HNF-4 in expression of α1-AT in H441 cells.
H441 cells were compared with Hep G2 and Caco-2 cells for luciferase activity after cotransfection of the α1-AT promoter (−137 to −2)-luciferase reporter plasmid and HNF-1β expression plasmid (Fig. 5 A). The results show that HNF-1β mediates a concentration-dependent increase in luciferase activity in H441, but not in Hep G2 or Caco-2, cells. In contrast, HNF-1α mediates increases in luciferase activity in all three cell types, albeit to a significantly greater extent in H441 than in the other two cell types (Fig. 5 B). The role of HNF-1 in expression of α1-AT in H441 cells was also shown by comparing the luciferase activity of the α1-AT promoter (−137 to −2)-luciferase reporter plasmid with that of the corresponding Mut-2 plasmid, mutated in the HNF-1-binding site. There was a marked decrease in the luciferase activity of the Mut-2 plasmid (data not shown).
Next, we examined the effect of HNF-1β and HNF-4 together. Our previous studies showed that HNF-1α and HNF-4 have a synergistic effect on α1-AT expression in Hep G2 and Caco-2 cells (17). The results in Fig. 6are very similar to those in Fig. 5, showing that HNF-1α alone mediates an increase in luciferase activity in all three cell types but HNF-1β alone mediates a significant increase only in H441 cells. HNF-4 by itself only mediates a two- to threefold increase in the three cell types. The combination of HNF-1α and HNF-4 has an additive effect on luciferase activity in Hep G2 cells, a marked synergistic effect in Caco-2 cells, but neither additive nor synergistic effects in H441 cells. The combination of HNF-1β and HNF-4 is also synergistic in Caco-2 cells but neither additive nor synergistic in Hep G2 or H441 cells. These data show that HNF-1β plays a unique role in α1-AT expression in H441 cells and that HNF-4 has minimal activity in H441 cells in the absence or presence of HNF-1α or HNF-1β.
Effects of HNF-3 on α1-AT gene expression in H441 cells.
Previous studies indicated that HNF-3 plays a dominant role in gene expression in H441 and pulmonary epithelial cells (3-5,28). Moreover, HNF-3 has been shown to interact with HNF-1 and HNF-4 and to bind to a cis-acting sequence element that overlaps with the element that binds HNF-1 and HNF-4 (8,13-15, 22, 23, 33). Therefore, we examined the role of HNF-3 in α1-AT gene expression in H441 compared with Hep G2 and Caco-2 cells (Fig. 7). The results show that HNF-3 by itself has no effect on luciferase activity in H441 cells. Moreover, HNF-3 does not mediate a significant effect on α1-AT gene expression in H441 cells when added together with HNF-1α or HNF-1β. HNF-3 also has no effect by itself on luciferase activity in Hep G2 and Caco-2 cells but does mediate a modest upregulatory effect when added together with HNF-1α in Hep G2 and Caco-2 cells. The lack of effect could not be attributed to saturation by endogenous HNF-3 in the nuclear extract, because there was no significant reduction in luciferase activity in H441 cells transfected with the α1-AT (−137 to −2)-luciferase promoter mutated in the HNF-3-binding site compared with the same plasmid without the mutation (data not shown). These data indicate that HNF-3 has no effect on α1-AT gene expression in pulmonary epithelial cells. HNF-3 does have an additive effect with HNF-1α in Hep G2 and Caco-2 cells, even though it does not bind to the proximal α1-AT promoter, implying that this effect involves an interaction with HNF-1α or with a cofactor necessary for HNF-1 activity.
Expression of α1-AT and HNF-1β in primary cultures of human respiratory epithelial cells.
To examine the possibility that HNF-1β is also involved in expression of the α1-AT gene in another respiratory epithelial cell system, we used RT-PCR analysis of human respiratory epithelial cells in primary culture (Fig. 8). The results show the presence of α1-AT and HNF-1β RNA in the cultured cells, as well as in H441, Hep G2, and Caco-2 cells. The results also show HNF-3α and HNF-3β, but not HNF-4, RNA in primary cultures, demonstrating that the repertoire of these transcription factors expressed in H441 cells faithfully reflects respiratory epithelium in vivo. Taken together with data from the EMSA (Fig.4 B), these results provide further evidence for the unique involvement of HNF-1β in α1-AT gene expression in respiratory epithelial cells.
Although α1-AT has always been considered a hepatic-derived plasma protein, there is abundant evidence that it is synthesized in extrahepatic sites by mononuclear phagocytes and epithelial cells (reviewed in Ref. 28). Interestingly, extrahepatic synthesis of α1-AT is particularly characteristic of the human compared with the mouse and rat. In fact, early studies of transgenic mice, engineered to express human α1-AT by using as transgene a genomic fragment encompassing the coding region and most of its 5′-flanking region, showed expression of human α1-AT, but not endogenous mouse α1-AT, in many extrahepatic tissues (19). More recent studies have shown that α1-AT is synthesized in human airway epithelial cells in primary culture, cell lines, and human lung tissue in situ (9,21, 22). Airway epithelium is likely to be a particularly important site of synthesis for α1-AT, because it is the major physiological inhibitor of neutrophil elastase, cathepsin G, and proteinase 3, neutrophil proteases that can degrade the connective tissue matrix of the lung in vivo, and because α1-AT deficiency predisposes to destructive lung disease/emphysema (20). However, the detection of α1-AT gene expression in a cell line and primary cultures derived from respiratory epithelium here and in previous studies (9, 22) does not by itself indicate that this expression is physiologically relevant. Even the detection of α1-AT mRNA in human respiratory epithelium by in situ hybridization analysis (21) does not ensure that a physiological function for α1-AT derived from this source as opposed to the liver, in which expression levels are much higher. Previous studies showed that the allotype of α1-AT in plasma converts to that of the donor after orthotopic liver transplantation (1, 16), indicating that plasma α1-AT is predominantly derived from liver. Whether extrahepatic expression is physiologically relevant will be definitively addressed only if there is correction of a functional abnormality by tissue-specific transgenic expression of α1-AT in an α1-AT-knockout mouse.
In this study, we examined the possibility that expression of α1-AT in human pulmonary epithelial cells involved transcriptional mechanisms similar to those at play in intestinal epithelial cells and/or hepatocytes. Our previous studies showed that the synergistic action of HNF-1α and HNF-4 plays a prominent role in α1-AT transcription in enterocytes and hepatocytes and in the mechanism by which the α1-AT gene is upregulated during differentiation of enterocytes from crypt to villous tip (26). Previous studies from other laboratories have implicated HNF-3 and thyroid transcription factor-1 in gene expression in pulmonary epithelium (3-5, 30). To our surprise, the results showed that α1-AT gene expression in pulmonary epithelial cells is predominantly driven by HNF-1 and not at all by HNF-3. Even more surprising, HNF-1β is responsible for activating α1-AT gene expression in pulmonary epithelial cells, even though it has no effect on the α1-AT gene in intestinal epithelial cells or hepatocytes. HNF-1β is highly homologous to HNF-1α, particularly in the NH2-terminal DNA-binding region, but diverges within the COOH-terminal activation domain (7). There is relatively limited information about the function of HNF-1β, except that it is expressed in dedifferentiated cells and somatic cell hybrids that have lost the ability to express liver-specific genes and lack the transcription factors HNF-1α and HNF-4 (2). HNF-1β is also known to be expressed in some tissues that do not express HNF-1α, including thymus, testis, ovary, and lung (2, 7), but this is the first report that we can find in the literature of a specific transcriptional role for HNF-1β in one of these tissues. It is also noteworthy that HNF-1β activates the α1-AT gene, a gene characteristic of the differentiated hepatocyte, in lung cells, but not in hepatocytes or enterocytes. These data imply that there are cell type-specific mechanisms for the action of HNF-1β and militate against it being merely a part of dedifferentiation from the hepatic phenotype. The fact that HNF-4 does not appear to mediate a synergistic effect with HNF-1β on α1-AT gene expression in lung cells, even though it does mediate a synergistic effect with HNF-1α in liver and intestinal cells, also implies cell type specificity in the role of HNF-1β on the α1-AT gene in lung cells.
The results also indicate that HNF-3α and HNF-3β do not directly activate α1-AT gene expression in lung, intestinal, or liver cells. This is somewhat surprising, because HNF-3 appears to play a major role in gene expression in differentiated pulmonary epithelial cells, and there are two potential HNF-3-binding sites within the proximal promoter of the α1-AT gene. We did find, however, that HNF-3α and HNF-3β have an additive effect with HNF-1α on α1-AT gene expression in Hep G2 and Caco-2 cells. Previous work has shown that HNF-3 interacts with HNF-1α in regulation of the liver-specific trans-activation of aldolase-β, but in this case the interaction is antagonistic (14). Because HNF-3α and HNF-3β do not bind to the proximal promoter of the α1-AT gene, their additive effect must involve interaction with HNF-1α or a cofactor necessary for transcriptional activation by HNF-1α.
The authors are indebted to M. Maksin for preparation of the manuscript.
The studies were supported in part by National Institutes of Health Grants HL-37784 and DK-52526.
Address for reprint requests and other correspondence: D. Perlmutter, Dept. of Pediatrics, University of Pittsburgh School of Medicine, Children's Hospital of Pittsburgh, 3705 Fifth Ave., Pittsburgh, PA 15213-2583 (E-mail:).
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2002 the American Physiological Society