J. Steroid Biochem. Molec. Biol. Vol. 41, No. 3-8, pp. 747-752, 1992 Printed in Great Britain. All rights reserved
0960-0760/92 $5.00 + 0.00 Copyright © 1992 Pergamon Press pie
TWO LIVER-ENRICHED TRANS-ACTING FACTORS SUPPORT THE TISSUE-SPECIFIC BASAL TRANSCRIPTION FROM THE RAT TYROSINE AMINOTRANSFERASE PROMOTER G. SCHWEIZER-GROYER, l A. GROYER,2. F. CADEPOND, l T. GRANGE, 2
E.-E. B^ULmU1and R. PICTET2 qNSERM U. 33, 80 Rue du G6n6ral Leclerc, 94276 Le Kremlin-Bic~tre-CEDEXand 2CNRS, Institut Jacques Monod and INSERM U. 257, Couloir 43-44, 2 Place Jussieu, 75251 Paris, France Summary--The rat tyrosine aminotransferase gene (TAT) is a glucocorticoid-inducible gene, specifically expressed in liver. Using gel retardation assays, we have shown that its promoter (nt + 1 to -350; TAT.35) binds a combination of both ubiquitous and liver-specific trans-acting factors. Cis-acfing sequences spanning: (i) nt - 6 5 to - 8 5 bound NF-Y, an ubiquitous "AACCAAT" box binding factor; (ii) nt -157 to -171 bound a liver-enriched member of the NFI gene family [NF I uver(NF1Lhereafter)]; (iii) nt - 266 to - 281 bound the liver specific factor HNF1; and (iv) nt -283 to -288 bound ubiquitous "CCAAT" box binding factor(s). Moreover, the TAT gene promoter was able to drive liver-specific basal transcription, even in an in vitro assay using TAT-expressing (liver) vs non-expressing (spleen) crude nuclear extracts (NEs). Competition studies in transcription with both unmutated and mutated ds-oligonucleotides (ds-oligos) demonstrated that NF1L and HNF1 supported approx. 60 and 25% of the basal transcriptional activity sustained by TAT.35in the liver, respectively. Neither of these oligos affected the very low level of transcription sustained by spleen NEs. This suggests a minor role for HNF1 in liver-specific basal TAT gene expression, consistent with previous observations with dedifferentiated C2 hepatoma cells (which does not express HNF1) [Deschatrette and Weiss. Biochimie 56 (1974) 1603-1611 and Cereghini et al. EMBO Jl 9 (1990) 2257-2263]. Competition studies in liver-specific in vitro transcription with ds-oligo - 2 6 5 / - 2 9 0 yielded a 90% inhibition, suggesting either that sequences spanning nt -283 to -288 sequester "CCAAT-box" binding factor(s) that may be relevant elsewhere for TAT promoter function (e.g. NF-Y which interacts with nt - 6 5 to -85), or that such a factor interacts functionally with HNFI.
INTRODUCTION
The rat tyrosine aminotransferase (TAT) gene belongs to the liver gene cluster activated in the neonatal period [1]. The liver-specific control of T A T gene basal and hormone (glucocorticoid, glucagon-via cAMP)-induced expression (i) relies on at least two trans-acting loci: hsdr/alf, a positive regulator [2, 3] and Tse-l, a tissue-specific extinguisher [4] and (ii) is primarily regulated at the transcriptional level through modulation of the gene transcription rate [5, 6]. Proceedings of the lOth International Symposium of the Journal of Steroid Biochemistry and Molecular Biology, Recent Advances in Steroid Biochemistry and Molecular Biology, Paris, France, 26--29 May 1991. *To whom correspondence should be addressed at: INSERM U.142, Bfitiment INSERM, H6pital SaintAntoine, 184, rue du Faubourg Saint-Antoine, 75571 Paris, Cedex 12, France.
In hepatocytes and hepatoma cell lines, ongoing expression of the TAT gene is accompanied by the occurrence of eight DNAse I hypersensitive sites (HSs) spread over 11 kbp of 5'-flanking sequences[7]. Among all these HSs, characteristic of the strongly active gene, only the promoter HS (up to nt - 2 0 0 ) is altered in a hepatoma cell line carrying the negative regulator Tse-1 [7]. This suggests that, although the sequences underlying promoter HS are not targets for T s e - l - m e d i a t e d extinction [8], they may play a role in the liver-specificity o f basal TAT gene expression. In the present paper, we provide evidence that TAT gene promoter sequences spanning nt + 1 to - 3 5 0 (TAT.35) were able to drive liverspecific in vitro transcription, and that this tissue-specificity was related to the binding of at least two liver-enriched trans-acting factors: NF1L [9] and to a lesser extent H N F I [10, 11]. 747
748
G. SCHWEIZER-GROYER et al. EXPERIMENTAL
Oligonucleotides
PE 56 and PE DS34 represent wild type and mutated HNFI binding sites from the rat albumin gene promoter, respectively [11]. NF 1L and mNFIL represent wild type and mutated NF1L binding sites from the mouse albumin gene promoter, respectively[9], NF-Y, CTF/NF1 (adenoviral and globin), and SV40 "core" enhancer sequences were as described in Raymonjean et al.[12]. Ds-oligos spanning TAT 5'-flanking sequences were as follows: II ( - 6 5 / -85), V ( - 1 4 0 / - 1 6 3 ) , VI ( - 1 6 3 / - 2 0 0 ) , VII ( - 155/- 176), VIII ( - 2 6 5 / - 290), IX ( - 266/ - 280) and X ( - 2 8 0 / - 300).
5 mM MgC12, 7.5% v/v glycerol). The binding reaction was allowed to proceed at 0°C for 20 min, then the samples were made 0.25% with respect to Ficoll and loaded on 6% polyacrylamide gels made in 0.5 x TBE (1 x TBE: 0.9 M Tris-pH 8.0, 0.9 M Boric acid, 2 mM EDTA). Electrophoresis was performed at room temperature (150 V; 90 min). RESULTS AND DISCUSSION
TAT.ss drives liver-specific in vitro transcription
Crude rat liver NEs generated ~-amanitin (2#g/ml)-sensitive, correctly initiated transcripts of 383 and 373 bp from the ubiquitous Adenovirus Major Late promoter (pAdML plasmid) and from the first 350 bp of the TAT Transcription vectors gene promoter (pTAT plasmid), respectively. The promoterless p(C2AT)19 and No transcripts were generated from pTATi,v or pML(C2AT)19 (referred to herein as pAdML) p(CzAT)I 9 ([15] and submitted). We have furhave been described by Sawadogo and ther investigated the transcriptional efficiency Roeder [13]. Insertion of a SstI (cap site)/KpnI of pAdML and pTAT in several experiments ( n t - 3 5 1 ) fragment of TAT gene 5'-flanking (n = 2-7) (Fig. 1). When transcription was studsequences into the unique SstI site of p(C2AT)~9 ied as a function of template DNA concenyielded pTAT and pTATi.v, depending on tration, at a single concentration of liver NEs, whether the insertion was sense or antisense, similar bell-shaped curves were obtained with relative to the "G-free cassette", respectively. pAdML and pTAT. Although the optima were identical for both promoters, the amounts of Nuclear extract preparation and in vitro trantranscripts derived from pTAT represented scription 17-48% of those synthetized from pAdML, Liver and spleen nuclear extracts (NEs) from depending on template concentration [Fig. 1(A)]. 10-11-week-old male Sprague-Dawley rats In the presence of spleen NEs, a similar were prepared as described by Gorski et al. [14], bell-shaped curve was obtained with pAdML except that all buffers contained the protease [Fig. I(B)]. In contrast, when the latter extracts inhibitors aprotinin (0.3 TIU/ml), benzamidine were used, TAT35 driven transcription was very (50#M), leupeptin (0.5pg/ml) and pepstatin weak when compared to that sustained by the (1 pg/ml) in addition to PMSF (0.5 mM). AdML promoter: only slight amounts of tranTranscription mixtures (20#I) contained scripts were generated at the two highest con2.4 mg protein/ml of liver and spleen NEs, and centrations of template DNA (7.2 and 13.6% of 5-40pg/ml of template DNA. Transcription those synthetized from pAdML at identical was carried out according to Gorski et al. [14], template concentrations) [Fig. I(B)]. At temexcept that the nucleotide analogue 3'-O-methyl plate DNA concentration to nuclear protein GTP was added at 0.4 mM. Radioactive tran- concentration ratio optimal for pAdML driven scripts were separated in 4% acrylamide-7 M transcription with either liver or spleen NEs, the urea sequencing gels. The gels were autoradio- index of tissue specificity (TSindex) equalled 6.7, graphed and the areas corresponding to full identical to that obtained for the same promoter length transcripts were excised and counted in fragment in transient transfection experiments liquid scintillation. (not shown). Although still significant, it declined sharply (2.9-fold) at a higher ratio. The Gel mobility shift assays occurrence of significant TS~,d.... led us to pro1 ng of 32P-labelled synthetic ds-oligos and pose that the first 350 bps of TAT gene 5'-flank1 pg of ds-poly (dI:dC) were mixed with crude ing sequences closest to the cap site were able to liver or spleen NEs (0.15--0.4 mg protein/ml) in regulate transcription in a tissue-specific mana final volume of 20 #1 (25 mM HEPES-pH 7.6, ner, and thus to sustain at least part of liver60mM KCI, 60#M EDTA, 0.75mM DTT, specific basal TAT gene expression. Such
Tissue-specific basal transcription from the rat TAT promoter A
749
B s
o>.5
o d
b
/.
o
3/2~/ 10
20
3~0
40 ~ 10 Template (IJg/ml)
3~
2~)
0 30
40
Fig. 1. Comparative/n vitro transcription of pAdML and pTAT in liver (A) and spleen (B) NEs. The radioactivity of the transcripts synthesizedfrom either TAT3~ or AdML Y-flanking sequences was measured and standardized relative to the radioactivityof the transcripts synthesizedfrom 30#g/ml of pAdML in the same experiments,with the homologousNE (2.4mg protein/ml). Data collected from 2 to 7 independentexperiments(numbers quoted on the figureclose to each data point) were compiledand the mean values were plotted as function of template concentration. (O), pAdML; (©), pTAT; (*) significantdifferencebetweenthe amounts of transcripts synthetisedfrom pTAT and pAdML (P < 0.05; paired student t-test). conclusions (i) could also be inferred from transient transfection data reported by Nitsch et al. [7], provided that the results were standardized relative to pHDCAT (an ubiquitous expression vector) and (ii) are further supported by the observation that promoter-HS is altered in Tse-I containing hepatoma cells when compared either to their T s e - I - counterparts or to the liver [7]. Liver-specific transcription from TA 7"35is depen dent on the binding of at least two liver-enriched trans-acting factors That TAT.35 was able to drive liver-specific in vitro transcription might reflect the interaction of promoter elements with liver-specific trans-acting factors. In this regard, we have previously shown that sequences spanning nt - 143 to - 2 0 0 of TAT.35 were responsible for a liver-specific pattern of protection in in vitro D N A a s e I footprinting ([15] and submitted). When ds-oligo VII ( n t - 1 5 5 to -176) was incubated with either liver of spleen NEs, clearcut tissue-specific gel shifts were generated (Fig. 2). Liver NEs generated a wide retardation pattern that consisted of the combination of two retarded bands. The slower migrating band was thick and liver-specific, the faster migrating one thin and ubiquitous (lanes 1 and 2). In the presence of spleen NEs, only one heavy retarded band and a fainter smear were observed, the migration pattern of the major gel shift being identical to that of the faster migrating band generated by liver NEs. In both
cases, binding was competed by a 90-fold excess of homologous unlabelled ds-oligo VII (lane 4). Competition was also obtained with ds-oligos encompassing 5' or 3' ends of ds-oligo VII and additional TAT.35 surrounding sequences: it was complete for ds-oligo VI (5" end of VII, including a CCAAT homology) and only partial for ds-oligo V (3' end of VII) (lanes 5 and 6). Cross-competition experiments showed that neither SV40 "core enhancer" nor ds-oligos encompassing the recognition sequence of acknowledged "CCAAT box" binding factors (NF-Y, C T F / N F 1) (lanes 10-13) nor the H N F 1 recognition sequence of the albumin gene promoter (PE56) (not shown) was able to compete with the liver-specific (upper) band shift observed with 32p-labelled ds-oligo VII in the presence of the liver NEs. In contrast, a 90-fold excess of an oligo that spans wild type NFIL (q~NFI) sequences of the mouse albumin gene promoter [9], but not its mutated counterpart, completely abolished the binding of liver nuclear proteins to 32p-labeled ds-oligo VII (lanes 7 and 8), arguing for the interaction of this liver-enriched member of the NF1 gene family [16] with TAT.3s promoter elements. In addition, both the upper band shift obtained with ds-oligo VII and the bona fide albumin NF1 t ds-oligo displayed the same migration pattern in the presence of liver NEs (not shown). On the other hand, computer-assisted search of homology over TAT.3s pointed out the existence of a putative H N F I binding site in a remote position ( n t - 2 6 5 to -280). Accord-
750
G. SCHWEIZER-GROYERet al.
A
a2p ds o l i g o
at
Competitor
oligo
VII
: -
3
Competitor
1 Nuclear
extract
2
VII
V
VI
4
5
6
NF1L mNF1L
7
8
CTF/NF1 NF-Y SV40 Ad. QIo.
oligo:
9
10
:LIV. SPL.
11
12
13
LIVER
B
3Zp ds o l i g o Competitor
VIII
oligo
VIII SV40 PE56 DS34
!i
:2
14
15
16
2: '
17
18
19
20
LIV. SPL. LIVER Fig. 2. Ds-oligos VII and VIII interact with liver specifictrans-acting factors. End labeled ds-oligos VII (A) and VIII (B) were incubated with liver (lanes 1, and 3-13, 14, 16-20) or spleen (lane 2, 15) NEs in the absence (lanes 1-3, 9, 14-16) or in the presence (lanes 4-8, 10-13, 17-20) of a 90-fold molar excess of the indicated unlabeled competiting oligo (see above each slot). Nuclear
extract
ingly, 3~P-labeled ds-oligo VIII ( n t - 2 6 5 to - 2 9 0 ) generated tissue-specific gel shifts when incubated with liver or spleen NEs, the slower migrating band (upper) being liver-specific [Fig. 2(B), lanes 14, 15]. Since this latter band was completely abolished by a 90-fold excess of wild type (PE56), but not of mutated (DS34) HNF1 binding site, it was tentatively identified as authentic ds-oligo V I I I - H N F I complexes
[Fig. 2(B), compare lane 16 with lanes 19 and 20]. In an attempt to define the relative contributions of NF1L and HNF1 in liver-specific gene expression from TAT35 promoter elements, we have first carried out competition experiments in in vitro transcription with dsoligos V and VI that encompass the recognition sequence of NFI L on the one hand, and with
Tissue-specific basal transcription from the rat TAT promoter
Nuclear Competitor
:
extract oligo
~
LIVER
:
"e-
II
v
vI
2
3
4
751
" S P L E E N " II
V
VI
6
7
8
501 " 489
"
343 -
M
1
5
B Competitor
oligo
-
PE56
DS34 v l t t
-
PE56 DS34 V I I I
501 489 343 -
M
9
10
11
12
13
14
t5
16
Fig. 3. Factor binding to ds-oligos II, V, VI (A) and VIII (B) behave as transcriptional activators. Nuclear proteins (2.4 mg/ml) prepared from liver (lanes I-4, 9-12) or spleen (lanes 5-8, 13-16) tissues were
preincubated at 0°C for 10rain in the absence(lanes 1, 5, 9-13) or in the presence(lanes 2-4, 6-8, 10-12, 14-16) of competitords-oligo(quotedaboveeach lane). 270 fmolof pTAT (30/~g/ml, finalconcentration) were then added to each reaction mixtureand transcriptionwas allowedto proceed at 30°C for 45 rain. Since a 18-19h exposure the filmwas insufficientto detect TAT.35driven transcripts when transcription was performed with spleen NEs only, 72-120h overexposureswere displayed for the latter. ds-oligos VIII and PE56, that bind HNF1, on the other hand (Fig. 3). When transcription experiments were performed with liver NEs in the presence of a 50-fold molar excess of either ds-oligos V or VI in the reaction mixture, the amounts of transcripts obtained were decreased by 33 and 58%, respectively [Fig. 3(A), compare lane 1 with lanes 3 and 4]. As expected, transcription carried out with spleen nuclear proteins generated a faint band after a 18 h exposure of the sequencing gel, but a longer exposure (72 h) allowed us to determine that a 50-fold molar excess of either ds-oligos V or VI were without effect on the very low amount of transcription [Fig. 3(A), compare lane 5 with lanes 7 and 8]. This lack of inhibition was not due to the inability of the transcription mixtures containing spleen NEs to be inhibited by any oligo in excess. Indeed, ds-oligo II (nt - 6 5 to - 8 5 of TAT.35), an oligo that encompasses a binding site for the ubiquitous trans-acting factor NF-Y ([15] and submitted), reduced transcription by ~ 50% in the presence of both liver and spleen NEs [Fig. 3(A), lanes 2 and 6]. Altogether, these results argue for the involvement of NF1L in the transcription process. Likewise, the patterns of gel retardation were similar when 32p-labeled ds-oligo VI was incubated in the presence of crude NEs prepared
from either non-expressing hepatoma x fibroblast hybrid cells or from spleen (not shown). Our observation thus favors the conclusion that the absence o f N F l Lmay be responsible for the lack of TAT35 driven transcription in the presence of spleen NEs, and are consistent with the observed absence of the DNAse I hypersensitive cut at - 2 0 0 bp in Tse-I + hepatoma cells [7]. This suggests an additional hypothesis for Tse-l-induced alterations in promoter-HS: Tse-I should exert its effect through the disappearance or the lack of posttranslational modification (e.g. phosphorylation) of a liver-enriched trans-acting factor (NFlliver is a good candidate). Similarly a 90-fold excess of PE56, but not of its mutated counterpart (DS34) reduced transcription only slightly ( ~ 2 5 % ) in the presence of liver NEs [Fig. 3(B), compare lane 9 with lanes 10 and 11], consistent with previous observations that in C2 dedifferentiated hepatoma cells, a FaO derivative which does not express HNF1 [17], basal TAT gene expression was not extinguished but was only decreased by 50-60% and was still inducible by glucocorticoids [18]. Surprisingly, the same molar excess of ds-oligo VIII abolished TAT a5 driven transcription by at least 90% [Fig. 3(B), lane 12]. This could be explained either by functional interactions be-
752
G. SCHWEIZER-GRo~Ret al.
tween HNF1 and "CCAAT box" binding factor(s) that may interact with cis-acting sequences spanning nt - 2 8 3 to -288, located 5' to the HNFI binding site within TAT as, or by the sequestration by ds-oligo VIII of "CCAAT box" binding factor(s) that may be relevant elsewhere for TAT promoter functions. The latter hypothesis was supported by two additional observations: (i) 90-fold excess of dsoligos VIII and X did compete the binding of nuclear proteins to 32P-labeled ds-oligo II (NFY binding sequence) in the presence of either liver or spleen NEs (not shown), and (ii) dsoligo VIII (but not of PE56) inhibited in vitro transcription by 71% when it was added in excess in transcription experiments performed with spleen NEs.
7.
8.
9. 10.
1I.
12. REFERENCES
1. Grcengard O.: The developmental formation of enzymes in liver. In Biochemical Action of Hormones (Edited by G. Litwack). Academic Press, New York, Vol. 1 (1970) pp. 53-87. 2. Mcknight S. L., Lane M. D. and Giuecksobn-Waeisch S.: Is CCAAT/enhancer-binding protein a central regulator of energy metabolism? Genes Dev. 3 (1989) 2021-2024. 3. Ruppert S., Boshart M., Bosch F. X., Schmid W., Fournier R. E. K. and Schiitz G.: Two genetically defined trans-acting loci coordinately regulated overlapping sets of liver-specific genes. Cell 61 (1990) 895-904. 4. Thayer M. J. and Fournier R. E. K.: Hormonal regulation of TSEl-repressed genes: evidence for multiple genetic controls in extinction. Molec. Cell. Biol. 9 (1989) 2837-2846. 5. Scherer G., Schmid W., Strange C., Rowekamp W. and Schiitz G.: Isolation of cDNA clones coding for rat tyrosine arninotransferase. Proc. Natn. Acad. Sci. U.S.A. 79 (1982) 7205-7208. 6. Schmid E., Sehmid W., Jantzen W., Mayer D., Jastorff B. and Schiitz G.: Transcription activation of tyrosine aminotransferase gene by glucocorticoids and cyclic
13.
14. 15.
16.
17.
18.
AMP in primary hepatocytes Eur. J. Biochem. 165 (1987) 499-506. Nitsch D., Stewart A. F., Boshart M., Mestril R., Weih F. and Schiitz G.: Chromatin structures of the rat tyrosine aminotransferase gene relate to the function of its c/s-acting elements. Molec. Cell. Biol. 10 (1990) 3334-3342. Boshart M., Weih F., Schmid A., Fournier R. E. K. and Sehiitz G.: A cyclic AMP response element mediates repression of tyrosine aminotranferase gene transcription by the tissue-specific extinguisher locus Tse-1. Cell 61 (1990) 905-916. Lichtsteiner S., Wuarin J. and Schibler U.: The interplay of DNA-binding proteins on the promoter of the mouse albumin gene. Cell 51 (1987) 963-973. Courtois G., Morgan J. G., Campbell L. A., Fourel G. and Crabtree G. R.: Interaction of a liver-specific nuclear factor with fibrinogen and u 1-antitrypsin promoters. Science 238 (1987) 688-692. Cereghini S., Blumenfeld M. and Yaniv M.: A liverspecific factor essential for albumin transcription differs between differentiated and dedifferentiated rat hepatoma cells. Genes Dev. 2 (1988) 957-974. Raymondjean M., Cereghini S. and Yaniv M.: Several distinct "CCAAT" box binding proteins coexist in eucaryotic cells. Proc. Natn. Acad. Sci. U.S.A. 85 (1988) 757-761. Sawadogo M. and Roeder R. G.: Factors involved in specific transcription by human RNA polymerase lI: analysis by a rapid and quantitative in vitro assay. Proc. Natn. Acad. Sci. U.S.A. 82 (1985) 4394-4398. Gorski K., Carneiro M. and Schibler U.: Tissue-specific in vitro transcription from the mouse albumin promoter. Cell 47 (1986) 767-776. Groyer A., Schweizer-Groyer G., Cadepond F., Grange T., Baulieu E. E. and Pictet R.: Liver-specific transcription of tyrosine aminotransferase (TAT) is related the binding of at least one liver specific transcription factor. J. Cell. Biochem. 1411 (1990) (Abstr.) 202. Paonessa G., Gounari F., Frank R. and Cortese R.: Purification of NFl-like DNA-binding protein from rat liver and cloning of the corresponding cDNA. EMBO JI 7 (1988) 3115-3123. Cereghini S., Yaniv M. and Cortese R.: Hepatocyte dedifferentiation and extinction is accompanied by a block in the synthesis of mRNA coding for the transcription factor HNFI/LFB1. EMBO JI 9 (1990) 2257-2263. Deschatrette J. and Weiss M. C.: Characterization of differentiated and dedifferentiated clones from rat hepatoma. Biochimie 56 (1974) 1603-1611.
0960-0760/92 $5.00 + 0.00 Copyright © 1992 Pergamon Press pie
TWO LIVER-ENRICHED TRANS-ACTING FACTORS SUPPORT THE TISSUE-SPECIFIC BASAL TRANSCRIPTION FROM THE RAT TYROSINE AMINOTRANSFERASE PROMOTER G. SCHWEIZER-GROYER, l A. GROYER,2. F. CADEPOND, l T. GRANGE, 2
E.-E. B^ULmU1and R. PICTET2 qNSERM U. 33, 80 Rue du G6n6ral Leclerc, 94276 Le Kremlin-Bic~tre-CEDEXand 2CNRS, Institut Jacques Monod and INSERM U. 257, Couloir 43-44, 2 Place Jussieu, 75251 Paris, France Summary--The rat tyrosine aminotransferase gene (TAT) is a glucocorticoid-inducible gene, specifically expressed in liver. Using gel retardation assays, we have shown that its promoter (nt + 1 to -350; TAT.35) binds a combination of both ubiquitous and liver-specific trans-acting factors. Cis-acfing sequences spanning: (i) nt - 6 5 to - 8 5 bound NF-Y, an ubiquitous "AACCAAT" box binding factor; (ii) nt -157 to -171 bound a liver-enriched member of the NFI gene family [NF I uver(NF1Lhereafter)]; (iii) nt - 266 to - 281 bound the liver specific factor HNF1; and (iv) nt -283 to -288 bound ubiquitous "CCAAT" box binding factor(s). Moreover, the TAT gene promoter was able to drive liver-specific basal transcription, even in an in vitro assay using TAT-expressing (liver) vs non-expressing (spleen) crude nuclear extracts (NEs). Competition studies in transcription with both unmutated and mutated ds-oligonucleotides (ds-oligos) demonstrated that NF1L and HNF1 supported approx. 60 and 25% of the basal transcriptional activity sustained by TAT.35in the liver, respectively. Neither of these oligos affected the very low level of transcription sustained by spleen NEs. This suggests a minor role for HNF1 in liver-specific basal TAT gene expression, consistent with previous observations with dedifferentiated C2 hepatoma cells (which does not express HNF1) [Deschatrette and Weiss. Biochimie 56 (1974) 1603-1611 and Cereghini et al. EMBO Jl 9 (1990) 2257-2263]. Competition studies in liver-specific in vitro transcription with ds-oligo - 2 6 5 / - 2 9 0 yielded a 90% inhibition, suggesting either that sequences spanning nt -283 to -288 sequester "CCAAT-box" binding factor(s) that may be relevant elsewhere for TAT promoter function (e.g. NF-Y which interacts with nt - 6 5 to -85), or that such a factor interacts functionally with HNFI.
INTRODUCTION
The rat tyrosine aminotransferase (TAT) gene belongs to the liver gene cluster activated in the neonatal period [1]. The liver-specific control of T A T gene basal and hormone (glucocorticoid, glucagon-via cAMP)-induced expression (i) relies on at least two trans-acting loci: hsdr/alf, a positive regulator [2, 3] and Tse-l, a tissue-specific extinguisher [4] and (ii) is primarily regulated at the transcriptional level through modulation of the gene transcription rate [5, 6]. Proceedings of the lOth International Symposium of the Journal of Steroid Biochemistry and Molecular Biology, Recent Advances in Steroid Biochemistry and Molecular Biology, Paris, France, 26--29 May 1991. *To whom correspondence should be addressed at: INSERM U.142, Bfitiment INSERM, H6pital SaintAntoine, 184, rue du Faubourg Saint-Antoine, 75571 Paris, Cedex 12, France.
In hepatocytes and hepatoma cell lines, ongoing expression of the TAT gene is accompanied by the occurrence of eight DNAse I hypersensitive sites (HSs) spread over 11 kbp of 5'-flanking sequences[7]. Among all these HSs, characteristic of the strongly active gene, only the promoter HS (up to nt - 2 0 0 ) is altered in a hepatoma cell line carrying the negative regulator Tse-1 [7]. This suggests that, although the sequences underlying promoter HS are not targets for T s e - l - m e d i a t e d extinction [8], they may play a role in the liver-specificity o f basal TAT gene expression. In the present paper, we provide evidence that TAT gene promoter sequences spanning nt + 1 to - 3 5 0 (TAT.35) were able to drive liverspecific in vitro transcription, and that this tissue-specificity was related to the binding of at least two liver-enriched trans-acting factors: NF1L [9] and to a lesser extent H N F I [10, 11]. 747
748
G. SCHWEIZER-GROYER et al. EXPERIMENTAL
Oligonucleotides
PE 56 and PE DS34 represent wild type and mutated HNFI binding sites from the rat albumin gene promoter, respectively [11]. NF 1L and mNFIL represent wild type and mutated NF1L binding sites from the mouse albumin gene promoter, respectively[9], NF-Y, CTF/NF1 (adenoviral and globin), and SV40 "core" enhancer sequences were as described in Raymonjean et al.[12]. Ds-oligos spanning TAT 5'-flanking sequences were as follows: II ( - 6 5 / -85), V ( - 1 4 0 / - 1 6 3 ) , VI ( - 1 6 3 / - 2 0 0 ) , VII ( - 155/- 176), VIII ( - 2 6 5 / - 290), IX ( - 266/ - 280) and X ( - 2 8 0 / - 300).
5 mM MgC12, 7.5% v/v glycerol). The binding reaction was allowed to proceed at 0°C for 20 min, then the samples were made 0.25% with respect to Ficoll and loaded on 6% polyacrylamide gels made in 0.5 x TBE (1 x TBE: 0.9 M Tris-pH 8.0, 0.9 M Boric acid, 2 mM EDTA). Electrophoresis was performed at room temperature (150 V; 90 min). RESULTS AND DISCUSSION
TAT.ss drives liver-specific in vitro transcription
Crude rat liver NEs generated ~-amanitin (2#g/ml)-sensitive, correctly initiated transcripts of 383 and 373 bp from the ubiquitous Adenovirus Major Late promoter (pAdML plasmid) and from the first 350 bp of the TAT Transcription vectors gene promoter (pTAT plasmid), respectively. The promoterless p(C2AT)19 and No transcripts were generated from pTATi,v or pML(C2AT)19 (referred to herein as pAdML) p(CzAT)I 9 ([15] and submitted). We have furhave been described by Sawadogo and ther investigated the transcriptional efficiency Roeder [13]. Insertion of a SstI (cap site)/KpnI of pAdML and pTAT in several experiments ( n t - 3 5 1 ) fragment of TAT gene 5'-flanking (n = 2-7) (Fig. 1). When transcription was studsequences into the unique SstI site of p(C2AT)~9 ied as a function of template DNA concenyielded pTAT and pTATi.v, depending on tration, at a single concentration of liver NEs, whether the insertion was sense or antisense, similar bell-shaped curves were obtained with relative to the "G-free cassette", respectively. pAdML and pTAT. Although the optima were identical for both promoters, the amounts of Nuclear extract preparation and in vitro trantranscripts derived from pTAT represented scription 17-48% of those synthetized from pAdML, Liver and spleen nuclear extracts (NEs) from depending on template concentration [Fig. 1(A)]. 10-11-week-old male Sprague-Dawley rats In the presence of spleen NEs, a similar were prepared as described by Gorski et al. [14], bell-shaped curve was obtained with pAdML except that all buffers contained the protease [Fig. I(B)]. In contrast, when the latter extracts inhibitors aprotinin (0.3 TIU/ml), benzamidine were used, TAT35 driven transcription was very (50#M), leupeptin (0.5pg/ml) and pepstatin weak when compared to that sustained by the (1 pg/ml) in addition to PMSF (0.5 mM). AdML promoter: only slight amounts of tranTranscription mixtures (20#I) contained scripts were generated at the two highest con2.4 mg protein/ml of liver and spleen NEs, and centrations of template DNA (7.2 and 13.6% of 5-40pg/ml of template DNA. Transcription those synthetized from pAdML at identical was carried out according to Gorski et al. [14], template concentrations) [Fig. I(B)]. At temexcept that the nucleotide analogue 3'-O-methyl plate DNA concentration to nuclear protein GTP was added at 0.4 mM. Radioactive tran- concentration ratio optimal for pAdML driven scripts were separated in 4% acrylamide-7 M transcription with either liver or spleen NEs, the urea sequencing gels. The gels were autoradio- index of tissue specificity (TSindex) equalled 6.7, graphed and the areas corresponding to full identical to that obtained for the same promoter length transcripts were excised and counted in fragment in transient transfection experiments liquid scintillation. (not shown). Although still significant, it declined sharply (2.9-fold) at a higher ratio. The Gel mobility shift assays occurrence of significant TS~,d.... led us to pro1 ng of 32P-labelled synthetic ds-oligos and pose that the first 350 bps of TAT gene 5'-flank1 pg of ds-poly (dI:dC) were mixed with crude ing sequences closest to the cap site were able to liver or spleen NEs (0.15--0.4 mg protein/ml) in regulate transcription in a tissue-specific mana final volume of 20 #1 (25 mM HEPES-pH 7.6, ner, and thus to sustain at least part of liver60mM KCI, 60#M EDTA, 0.75mM DTT, specific basal TAT gene expression. Such
Tissue-specific basal transcription from the rat TAT promoter A
749
B s
o>.5
o d
b
/.
o
3/2~/ 10
20
3~0
40 ~ 10 Template (IJg/ml)
3~
2~)
0 30
40
Fig. 1. Comparative/n vitro transcription of pAdML and pTAT in liver (A) and spleen (B) NEs. The radioactivity of the transcripts synthesizedfrom either TAT3~ or AdML Y-flanking sequences was measured and standardized relative to the radioactivityof the transcripts synthesizedfrom 30#g/ml of pAdML in the same experiments,with the homologousNE (2.4mg protein/ml). Data collected from 2 to 7 independentexperiments(numbers quoted on the figureclose to each data point) were compiledand the mean values were plotted as function of template concentration. (O), pAdML; (©), pTAT; (*) significantdifferencebetweenthe amounts of transcripts synthetisedfrom pTAT and pAdML (P < 0.05; paired student t-test). conclusions (i) could also be inferred from transient transfection data reported by Nitsch et al. [7], provided that the results were standardized relative to pHDCAT (an ubiquitous expression vector) and (ii) are further supported by the observation that promoter-HS is altered in Tse-I containing hepatoma cells when compared either to their T s e - I - counterparts or to the liver [7]. Liver-specific transcription from TA 7"35is depen dent on the binding of at least two liver-enriched trans-acting factors That TAT.35 was able to drive liver-specific in vitro transcription might reflect the interaction of promoter elements with liver-specific trans-acting factors. In this regard, we have previously shown that sequences spanning nt - 143 to - 2 0 0 of TAT.35 were responsible for a liver-specific pattern of protection in in vitro D N A a s e I footprinting ([15] and submitted). When ds-oligo VII ( n t - 1 5 5 to -176) was incubated with either liver of spleen NEs, clearcut tissue-specific gel shifts were generated (Fig. 2). Liver NEs generated a wide retardation pattern that consisted of the combination of two retarded bands. The slower migrating band was thick and liver-specific, the faster migrating one thin and ubiquitous (lanes 1 and 2). In the presence of spleen NEs, only one heavy retarded band and a fainter smear were observed, the migration pattern of the major gel shift being identical to that of the faster migrating band generated by liver NEs. In both
cases, binding was competed by a 90-fold excess of homologous unlabelled ds-oligo VII (lane 4). Competition was also obtained with ds-oligos encompassing 5' or 3' ends of ds-oligo VII and additional TAT.35 surrounding sequences: it was complete for ds-oligo VI (5" end of VII, including a CCAAT homology) and only partial for ds-oligo V (3' end of VII) (lanes 5 and 6). Cross-competition experiments showed that neither SV40 "core enhancer" nor ds-oligos encompassing the recognition sequence of acknowledged "CCAAT box" binding factors (NF-Y, C T F / N F 1) (lanes 10-13) nor the H N F 1 recognition sequence of the albumin gene promoter (PE56) (not shown) was able to compete with the liver-specific (upper) band shift observed with 32p-labelled ds-oligo VII in the presence of the liver NEs. In contrast, a 90-fold excess of an oligo that spans wild type NFIL (q~NFI) sequences of the mouse albumin gene promoter [9], but not its mutated counterpart, completely abolished the binding of liver nuclear proteins to 32p-labeled ds-oligo VII (lanes 7 and 8), arguing for the interaction of this liver-enriched member of the NF1 gene family [16] with TAT.3s promoter elements. In addition, both the upper band shift obtained with ds-oligo VII and the bona fide albumin NF1 t ds-oligo displayed the same migration pattern in the presence of liver NEs (not shown). On the other hand, computer-assisted search of homology over TAT.3s pointed out the existence of a putative H N F I binding site in a remote position ( n t - 2 6 5 to -280). Accord-
750
G. SCHWEIZER-GROYERet al.
A
a2p ds o l i g o
at
Competitor
oligo
VII
: -
3
Competitor
1 Nuclear
extract
2
VII
V
VI
4
5
6
NF1L mNF1L
7
8
CTF/NF1 NF-Y SV40 Ad. QIo.
oligo:
9
10
:LIV. SPL.
11
12
13
LIVER
B
3Zp ds o l i g o Competitor
VIII
oligo
VIII SV40 PE56 DS34
!i
:2
14
15
16
2: '
17
18
19
20
LIV. SPL. LIVER Fig. 2. Ds-oligos VII and VIII interact with liver specifictrans-acting factors. End labeled ds-oligos VII (A) and VIII (B) were incubated with liver (lanes 1, and 3-13, 14, 16-20) or spleen (lane 2, 15) NEs in the absence (lanes 1-3, 9, 14-16) or in the presence (lanes 4-8, 10-13, 17-20) of a 90-fold molar excess of the indicated unlabeled competiting oligo (see above each slot). Nuclear
extract
ingly, 3~P-labeled ds-oligo VIII ( n t - 2 6 5 to - 2 9 0 ) generated tissue-specific gel shifts when incubated with liver or spleen NEs, the slower migrating band (upper) being liver-specific [Fig. 2(B), lanes 14, 15]. Since this latter band was completely abolished by a 90-fold excess of wild type (PE56), but not of mutated (DS34) HNF1 binding site, it was tentatively identified as authentic ds-oligo V I I I - H N F I complexes
[Fig. 2(B), compare lane 16 with lanes 19 and 20]. In an attempt to define the relative contributions of NF1L and HNF1 in liver-specific gene expression from TAT35 promoter elements, we have first carried out competition experiments in in vitro transcription with dsoligos V and VI that encompass the recognition sequence of NFI L on the one hand, and with
Tissue-specific basal transcription from the rat TAT promoter
Nuclear Competitor
:
extract oligo
~
LIVER
:
"e-
II
v
vI
2
3
4
751
" S P L E E N " II
V
VI
6
7
8
501 " 489
"
343 -
M
1
5
B Competitor
oligo
-
PE56
DS34 v l t t
-
PE56 DS34 V I I I
501 489 343 -
M
9
10
11
12
13
14
t5
16
Fig. 3. Factor binding to ds-oligos II, V, VI (A) and VIII (B) behave as transcriptional activators. Nuclear proteins (2.4 mg/ml) prepared from liver (lanes I-4, 9-12) or spleen (lanes 5-8, 13-16) tissues were
preincubated at 0°C for 10rain in the absence(lanes 1, 5, 9-13) or in the presence(lanes 2-4, 6-8, 10-12, 14-16) of competitords-oligo(quotedaboveeach lane). 270 fmolof pTAT (30/~g/ml, finalconcentration) were then added to each reaction mixtureand transcriptionwas allowedto proceed at 30°C for 45 rain. Since a 18-19h exposure the filmwas insufficientto detect TAT.35driven transcripts when transcription was performed with spleen NEs only, 72-120h overexposureswere displayed for the latter. ds-oligos VIII and PE56, that bind HNF1, on the other hand (Fig. 3). When transcription experiments were performed with liver NEs in the presence of a 50-fold molar excess of either ds-oligos V or VI in the reaction mixture, the amounts of transcripts obtained were decreased by 33 and 58%, respectively [Fig. 3(A), compare lane 1 with lanes 3 and 4]. As expected, transcription carried out with spleen nuclear proteins generated a faint band after a 18 h exposure of the sequencing gel, but a longer exposure (72 h) allowed us to determine that a 50-fold molar excess of either ds-oligos V or VI were without effect on the very low amount of transcription [Fig. 3(A), compare lane 5 with lanes 7 and 8]. This lack of inhibition was not due to the inability of the transcription mixtures containing spleen NEs to be inhibited by any oligo in excess. Indeed, ds-oligo II (nt - 6 5 to - 8 5 of TAT.35), an oligo that encompasses a binding site for the ubiquitous trans-acting factor NF-Y ([15] and submitted), reduced transcription by ~ 50% in the presence of both liver and spleen NEs [Fig. 3(A), lanes 2 and 6]. Altogether, these results argue for the involvement of NF1L in the transcription process. Likewise, the patterns of gel retardation were similar when 32p-labeled ds-oligo VI was incubated in the presence of crude NEs prepared
from either non-expressing hepatoma x fibroblast hybrid cells or from spleen (not shown). Our observation thus favors the conclusion that the absence o f N F l Lmay be responsible for the lack of TAT35 driven transcription in the presence of spleen NEs, and are consistent with the observed absence of the DNAse I hypersensitive cut at - 2 0 0 bp in Tse-I + hepatoma cells [7]. This suggests an additional hypothesis for Tse-l-induced alterations in promoter-HS: Tse-I should exert its effect through the disappearance or the lack of posttranslational modification (e.g. phosphorylation) of a liver-enriched trans-acting factor (NFlliver is a good candidate). Similarly a 90-fold excess of PE56, but not of its mutated counterpart (DS34) reduced transcription only slightly ( ~ 2 5 % ) in the presence of liver NEs [Fig. 3(B), compare lane 9 with lanes 10 and 11], consistent with previous observations that in C2 dedifferentiated hepatoma cells, a FaO derivative which does not express HNF1 [17], basal TAT gene expression was not extinguished but was only decreased by 50-60% and was still inducible by glucocorticoids [18]. Surprisingly, the same molar excess of ds-oligo VIII abolished TAT a5 driven transcription by at least 90% [Fig. 3(B), lane 12]. This could be explained either by functional interactions be-
752
G. SCHWEIZER-GRo~Ret al.
tween HNF1 and "CCAAT box" binding factor(s) that may interact with cis-acting sequences spanning nt - 2 8 3 to -288, located 5' to the HNFI binding site within TAT as, or by the sequestration by ds-oligo VIII of "CCAAT box" binding factor(s) that may be relevant elsewhere for TAT promoter functions. The latter hypothesis was supported by two additional observations: (i) 90-fold excess of dsoligos VIII and X did compete the binding of nuclear proteins to 32P-labeled ds-oligo II (NFY binding sequence) in the presence of either liver or spleen NEs (not shown), and (ii) dsoligo VIII (but not of PE56) inhibited in vitro transcription by 71% when it was added in excess in transcription experiments performed with spleen NEs.
7.
8.
9. 10.
1I.
12. REFERENCES
1. Grcengard O.: The developmental formation of enzymes in liver. In Biochemical Action of Hormones (Edited by G. Litwack). Academic Press, New York, Vol. 1 (1970) pp. 53-87. 2. Mcknight S. L., Lane M. D. and Giuecksobn-Waeisch S.: Is CCAAT/enhancer-binding protein a central regulator of energy metabolism? Genes Dev. 3 (1989) 2021-2024. 3. Ruppert S., Boshart M., Bosch F. X., Schmid W., Fournier R. E. K. and Schiitz G.: Two genetically defined trans-acting loci coordinately regulated overlapping sets of liver-specific genes. Cell 61 (1990) 895-904. 4. Thayer M. J. and Fournier R. E. K.: Hormonal regulation of TSEl-repressed genes: evidence for multiple genetic controls in extinction. Molec. Cell. Biol. 9 (1989) 2837-2846. 5. Scherer G., Schmid W., Strange C., Rowekamp W. and Schiitz G.: Isolation of cDNA clones coding for rat tyrosine arninotransferase. Proc. Natn. Acad. Sci. U.S.A. 79 (1982) 7205-7208. 6. Schmid E., Sehmid W., Jantzen W., Mayer D., Jastorff B. and Schiitz G.: Transcription activation of tyrosine aminotransferase gene by glucocorticoids and cyclic
13.
14. 15.
16.
17.
18.
AMP in primary hepatocytes Eur. J. Biochem. 165 (1987) 499-506. Nitsch D., Stewart A. F., Boshart M., Mestril R., Weih F. and Schiitz G.: Chromatin structures of the rat tyrosine aminotransferase gene relate to the function of its c/s-acting elements. Molec. Cell. Biol. 10 (1990) 3334-3342. Boshart M., Weih F., Schmid A., Fournier R. E. K. and Sehiitz G.: A cyclic AMP response element mediates repression of tyrosine aminotranferase gene transcription by the tissue-specific extinguisher locus Tse-1. Cell 61 (1990) 905-916. Lichtsteiner S., Wuarin J. and Schibler U.: The interplay of DNA-binding proteins on the promoter of the mouse albumin gene. Cell 51 (1987) 963-973. Courtois G., Morgan J. G., Campbell L. A., Fourel G. and Crabtree G. R.: Interaction of a liver-specific nuclear factor with fibrinogen and u 1-antitrypsin promoters. Science 238 (1987) 688-692. Cereghini S., Blumenfeld M. and Yaniv M.: A liverspecific factor essential for albumin transcription differs between differentiated and dedifferentiated rat hepatoma cells. Genes Dev. 2 (1988) 957-974. Raymondjean M., Cereghini S. and Yaniv M.: Several distinct "CCAAT" box binding proteins coexist in eucaryotic cells. Proc. Natn. Acad. Sci. U.S.A. 85 (1988) 757-761. Sawadogo M. and Roeder R. G.: Factors involved in specific transcription by human RNA polymerase lI: analysis by a rapid and quantitative in vitro assay. Proc. Natn. Acad. Sci. U.S.A. 82 (1985) 4394-4398. Gorski K., Carneiro M. and Schibler U.: Tissue-specific in vitro transcription from the mouse albumin promoter. Cell 47 (1986) 767-776. Groyer A., Schweizer-Groyer G., Cadepond F., Grange T., Baulieu E. E. and Pictet R.: Liver-specific transcription of tyrosine aminotransferase (TAT) is related the binding of at least one liver specific transcription factor. J. Cell. Biochem. 1411 (1990) (Abstr.) 202. Paonessa G., Gounari F., Frank R. and Cortese R.: Purification of NFl-like DNA-binding protein from rat liver and cloning of the corresponding cDNA. EMBO JI 7 (1988) 3115-3123. Cereghini S., Yaniv M. and Cortese R.: Hepatocyte dedifferentiation and extinction is accompanied by a block in the synthesis of mRNA coding for the transcription factor HNFI/LFB1. EMBO JI 9 (1990) 2257-2263. Deschatrette J. and Weiss M. C.: Characterization of differentiated and dedifferentiated clones from rat hepatoma. Biochimie 56 (1974) 1603-1611.
