Tissue-specificity of liver gene expression: a common liver-specific promoter element

Several liver specific genes contain a sequence similar to the hepatocyte- specific promoter element (HP1) identified in the Xenopus albumin gene. By competition experiments in gel retardation assays we demonstrate that the albumin, the a1-antitrypsin and the 3-fibrinogen promoter elements bind an identical factor, whereas the HPl-related sequence in the transferrin gene has different binding properties. The elements of all four genes are able to confer increased transcription in nuclear extracts from rat liver. The HP1 found in the albumin, a1-antitrypsin and 1-fibrinogen promoter bind identical transcription factors, since in vitro transcription can be specifically inhibited by the addition of an excess of the corresponding oligonucleotide. The cis-element HP1 and its transacting factor have been conserved during evolution, since elements derived from frog, rat and human genes function in rat liver nuclear extracts.


INTRODUCTION
A large part of tissue-specific protein synthesis in higher eukaryotes is achieved by selective transcription of a limited set of genes. Gene transfer experiments have revealed tissue-specific enhancer and promoter elements in many genes (1). One would speculate that different genes active in a given tissue contain common regulatory elements that allow coordinate expression of various genes using the same trans-acting factor. Liver cells express a series of well characterized genes e.g. albumin (2,3), a1-antitrypsin (4,5), 1-fibrinogen (6), and transferrin (7), which are predominantly active in hepatocytes and are therefore a good system to search for regulatory elements common for different genes. We have recently identified in the Xenopus albumin gene (8) a hepatocyte-specific promoter element (HP1) which has some similarity with promoter elements identified in the human alantitrypsin (4) and the rat 1-fibrinogen gene (6). In this report we demonstrate by competition experiments using a binding assay as well as in vitro transcription, that these 3 different genes contain a common liverspecific promoter element. Syn 0-TG was constructed as described previously (8) and the oligonucleotides AT, FB, TF (see Fig.l) were used to replace syn 0 in the transcription vector syn 0-TG. As a negative control (PL) we replaced syn 0 by the polylinker sequence (HindIII-BglII fragment) of pBL CAT3 (13). As a positive control (AML) we used clone pML (C2AT)19 as described by Sawadogo and Roeder (9). As an internal control we used a derivative of clone pML (C2AT)19 i.e. pML cas 9, with a shortened -G-cassette of 190 nt. Band shift assays Nuclear extracts were essentially prepared from rat liver as described by Gorski et al. (10). 5 jg nuclear proteins were incubated in 20 Al binding buffer (11) containing 250 ng of unspecific competitor poly (dI-dC) poly (dI-dC) and 104 cpm of 32P-labelled oligonucleotides as given in the figure legend. After 30 min incubation at room temperature, the resulting complexes were resolved on a 4% polyacrylamide gel as described by Sing et al., (11). The gel was fixed in 10% acetic acid, dried and exposed for autoradiography. Preparation of labelled oligonucleotides The oligonucleotides were synthesized with the Pharmacia Gene Assembler and purified on a preparative 20% polyacrylamide gel in 8M urea as described (8).

In vitro transcription assays
The in vitro transcription was done as described (8) with 7 gl nuclear extract in a total volume of 20 3l, using 35 ng of the corresponding transcription vector and as internal control 17 ng pML cas 9. For competition experiments 100 ng of the indicated oligonucleotide (see Fig.3c) was included in the reaction. The transcription products were separated on a sequencing gel and exposed for autoradiography (3 b, 40 hrs; 3 c, 16 hrs).
The transcription activity of the various constructs was determined by densitometry of the bands using the pML cas 9 signal as reference. The values were standardized to the activity found without added competitor.

RESULTS
The promoter region of the albumin. a-fetoprotein al-antitrvpsin. 1fibrinoaen and transferrin aene contains a similar secuence element Transfection of various Xenopus albumin-CAT genes into mammalian cells allowed us to identify the promoter element HP1 which confers selective expression in hepatoma cells and mediates high in vitro transcription in rat  (12) is given at the top as it is present in the synthetic oligonulceotide syn 0. The boxed area is the functional element we have defined and the three positions of point mutations that destroy the function are indicated by a dot (8) The oligonucleotides corresponding to regulatory elements as found in the genes coding for human a1-antitrypsin (AT), human a1-antitrypsin mutant (AT-mt), rat 1-fibrinogen (FB), human transferrin (TF), mouse albumin (Alb) and mouse a-fetoprotein (AFB) are aligned in such a way to attain maximal homology to HP1 in syn 0. The position relative to the cap site is given and the HindIII and BglII site used for cloning are included.
liver nuclear extracts (8). This hepatocyte-specific promoter element is defined as a 13 bp sequence (syn 0 in Fig.l), which is recognized by a hepatocyte-specific transcription factor (8). The fact that this regulatory element is derived from a frog gene and functions in mammalian cells demonstrates that it has been conserved during evolution. This interpretation is supported by our observation that sequences similar to HP1 are also found in the promoter of the mammalian albumin and a-fetoprotein (AFP) genes (8). The   b) In vitro transcription products were analyzed on a sequencing gel. The various constructs included in the reactions are indicated above the lanes (AT, FB, TF were used to replace syn 0 in syn 0-TG, see Fig.1 for abbreviations used) . Additionally we used as a negative control a polylinker sequence (PL) and as a positive control the adeno major late promoter (AML). HaeIII digested pBR322 is used as size marker (in). Signals of the transcription vector and of the internal control are marked by exp. and contr., respectively. c) As competitor 100 ng of the oligonucleotide FB or TF (see Fig.1) is added in the reactions as indicated.
migrating complex represents a distinct protein-DNA interaction. In summary, the binding assays establish that the elements found in the albumin, AFP, a1-antitrypsin and 13-f ibrinogen gene bind the same factor. We therefore consider all these elements as HPl homologues. The identity of the binding at the a1-antitrypsin and 13-fibrinogen binding site confirms earlier data (6).
The nromoter element EIPl of the albumin. a-i-antitrvpsin and 13-f ibrinocren gfene interacts with the same transcription factor Our previous experiments have demonstrated that HPl as found in the Xenopus albumin gene, i.e. syn 0, is able to stimulate in vitro transcription in rat liver nuclear extracts (8). To analyze whether the HPl homologues found in lhe other genes can also increase the transcriptional activity, we replaced the Xenopus HPl sequence in syn 0-TG (see Fig.3a) with these homologues. The various constructs were tested in the in vitro transcription system using  were used, respectively. The data were quantified by densitometry of the autoradiograms using the internal control pML cas 9 for reference. The values were standardized to the activity found without (-) added competitor.
the -G-cassette of about 400 bp (9). In all assays the adeno major late promoter with a shortened -G-cassette of 200 bp was used as an internal standard to quantitate the in vitro transcripts. Figure 3b demonstrates that all four oligonucleotides confer a positive transcriptional activation to the TATA box in the rat liver nuclear extract. The B-fibrinogen element (FB) is as active as the Xenopus albumin element (syn 0), whereas the a1antitrypsin (AT) and transferrin (TF) element give approximately 3-fold lower activity which is still 5-10 fold higher than the activity observed with a control construct containing no regulatory sequence (PL). We have previously shown that transcription of the syn 0-TG construct containing the Xenopus albumin HP1 can be specifically inhibited by an excess of added oligonucleotide syn 0 (8). This observation allows us to compare the regulatory elements by the addition of various oligonucleotides as competitor. Fig.3c illustrates

DISCUSSION
Taken together our functional competition assays confirm the binding studies, that the albumin, a1-antitrypsin and 1-fibrinogen oligonucleotide bind identical proteins. Clearly, the albumin, a1-antitrypsin and the 1fibrinogen gene contain a common promoter element HP1, recognized by the same transcription factor and this cis-element is distinct from that found in the transferrin gene. We speculate that the regulatory sequence in the transferrin gene represents another hepatocyte-specific promoter element.
We have previously observed by the introduction of point mutations into HP1 of the Xenopus albumin gene that the first, third and 13th position of the element is crucial for its function in vivo (8). Based on this definition the B-fibrinogen element would not be functional as it is mutated at the third position compared to the Xenopus albumin element. On the other hand the transferrin element would fit to the albumin HP1. Obviously we are as yet unable to predict all the sequences that may form a functional HP1.
unambiguous identification is possible.
Based on the present finding that HP1 is also present in the 1-fibrinogen gene, it is most likely that the hepatocyte nuclear factor 1 (HNF1) identified as a binding protein in the 1-fibrinogen promoter region -102 to -75 (6) is identical to the factor we have identified in this paper.
However, we are not sure whether alTF-B described by Cortese's group (4) as component binding to the region homologous to HP1 in the a1-antitrypsin gene, represents the same component, since our gel retardation experiments always reveal two complexes with the a1-antitrypsin sequence (see Fig.2b) and only the slower migrating form can be competed with the other HP1 elements. In the case of the mouse albumin gene several DNA-binding proteins have been found to bind to the promoter (2,3). From these footprint studies we assume that the factor binding to site B (2) and to the proximal element which contains HPl (3), is identical to the transcription factor recognizing HP1 in our experiments.
In conclusion our data confirm for the first time that 3 different genes specifically expressed in the same tissue contain a common tissue-specific promoter element which is recognized by the same transcription factor.
Clearly, the cis-element HP1 and the corresponding trans-acting factor have been conserved during evolution, since elements derived from frog, rat and human genes function in rat liver nuclear extracts. The simplest interpretation we would provide to these data is the assumption that a single protein is involved in the interaction with HP1 and that the complex as seen in the gel retardation assay leads to an increase in tissue-specific transcription. Our previous experiments established that HP1 is recognized by a nuclear protein specific for hepatic cells (8). This suggests that the gene coding for this protein controls the expression of several liver specific-genes. It will be of interest to determine at which stage of embryogenesis and organogenesis this gene is activated.