- Email: [email protected]
A ubiquitous CCAAT factor is required for efficient in vitro transcription from the mouse albumin promoter
J. Mol. Biol. (1990) 214, 865-874
A Ubiquitous CCAAT Factor is Required for Efficient irz Vitro Transcription from the Mouse Albumin Promoter ExBme Wuarin, Department Quai
Christopher
Mueller
and Ueli Schibler
of Molecular Biology, Sciences II, University of Geneva Ernest Ansermet 30, CH-1211 Geneva-4, Switzerland
(Received 13 March
1990; accepted 2 Nay
1990)
Among the various factors binding to DNA elements within the mouse albumin promoter, NF-Y is the only one present at identical concentrations in the nuclei of all examined tissues. NF-Y binds to albumin promoter element C, which contains the sequence CCAAT. To determine whether this factor augments in vitro transcription from the albumin promoter, an extensive point-mutation analysis was performed wit’hin the promoter element C. In liver extracts, six out of the ten mutations result in a strong inhibition of NF-Y decrease in promoter activity. Two mut,ations that increase binding and in a concomitant the affinity of the C-element for NF-Y also augment, the transcription efficiency from the albumin promoter. A similarly strong correlation of NF-Y-binding with transcription efficiency has also been observed in spleen nuclear extracts. The liver-enriched CCAAT and enhancer binding factor C/EBP also recognizes the C element. In contrast to NF-Y; no correlation between the affinity of mutant C-elements for C/EBP and transcript’ional activity could be observed in liver nuclear extracts.
reached by Heard et al. (1987) on the basis of DNA transfection studies. This promoter region contains six distinct elements (A to F) that are protected from DNase I digestion by proteins present in a rat et al., 1987). Most liver nuclear extract (Lichtsteiner of the factors interacting with the various albumin promoter elements have been identified (Lichtsteiner et al., 1987; Schorpp et al., 1988: Courtois et al., 1987; Baumhueter et al., 1988; Babiss et al., 1987; Costa et al., 1988a,b; Cereghini et al.: 198’7; Raymondjean et al., 1988; Lichtsteiner & Schibler, 1989; Mueller et al., 1990). Figure 1 shows the inventory of transcription factors with affinity to the various k-acting elements found within t’he albumin promoter. C/EBP (Graves et al., 1986; Johnson et al., 1987; Landschulz et al., 1988a,b; Vinson et al., 1988) has a high affinity for site D, lower affinities for sites A and F, and an even lower but readily measurable affinity for site C (see below; and see Lichtsteiner et al., 1987; Mueller et al., 1990). Site E can be filled with several members of the CTF/NFl factor family, NF-Y (Dorn et al., 1987; Raymondjean et al., 1988), one of several CAAT factors, recognizes site C which bears the common promoter motif CCAAT. Finally, site B is a highaffinity site for HNFl, a factor binding to promoter elements of a number of liver-specific genes (Lichtsteiner & Schibler, 1989; Cereghini et al., 1988; Courtois et al., 1988). Most of these factors, such as HNFl (Courtois et al., 1987; Baumhueter et al.; 1988; Lichtsteiner &. Schibler, 1989), C-EBP
1. Introduction The tissue-specific activation of many genes is controlled primarily at the level of transcription initiation (e.g. see Derman et al., 1981; Schibler et al., 1983; Tilghman & Belayew, 1982; Shaw et al., 1985). For several cases; it has been suggested that the combinatorial interplay of trans-acting transcription factors with &s-acting promoter elements plays an important role in controlling differential transcription (for a review, see Maniatis et al., 1983). Therefore, a detailed knowledge of these interactions is essential for understanding the molecular basis of cell-type specific gene expression. The developmentally regulat’ed expression of the serum albumin gene provides an excellent model system for studies of this nature (Tilghman & Belayew, 1982; Krumlauf et al., 1985; Godbout et al., 1986; Ott et al., 1984; Heard et aZ., 1987; Gorski et al., 1986). Furthermore, the elaboration in our laboratory of a, tissue-specific in vitro transcription system differential accurately reproduces the that, expression of the albumin gene renders this system amenable to biochemical dissection (Gorski et al., 1986: Lichtsteiner et al., 1987; Lichtsteiner & Schibler, 1989; Maire et al., 1989). Previously reported studies from our laboratory indicated that the minimal albumin promoter sequences required for optimal in, vitro transcription span about’ 170 nucleotides of 5’ flanking region et al., 1986). A similar conclusion was (Gorski
865 002%2836/90/160865-10
$03.00/O
0
1990 Academic
Press Limited
J. Wuarin et al
'\
..." :
/.’ ..’
E
-130
\ D
-100
....’
c
-80
i
B
;
‘i
-50
k
~I~
A
““, h,<,, -30
A AG TATGGTTAATGATCT A CAGTT ATTGGTTAAAG T TC ATACCAATTACTAGA T GTCAA TAACCAATTTC
Figure I. Schematic representation of the factors interacting with the albumin promoter. The different fact,ors that have been shown to interact with the 6 binding sites (A, B, C. D; E and F) of the albumin promoter (Lichtsteiner el’ aE.. 1987; Schorpp et al., 1988; Mueller et al.. 1990; S. Lichtsteiner and U. Schibler, unpublished results) are depicted. Projected onto this schema is the nucleotide sequence of the albumin promoter from - 166 to - 27 with the sequences protected by each factor from DNase I digestion appearing within the boxed regions. C/EBP, a protein with relaxed sequence specificity binds strongly to its primary site D, but more weakly to its secondary sites A, C and F. Pseudo NF- 1 appears to recognize the same site as NF-I but exhibits greater sensitivity to competition as well as showing some tissue specificity. NF-Y is a ubiquit,ous CCAAT binding protein (Lichtst,einer et al., 1987; Dorn et al., 1987; Raymondjean el al.,
1988). (Lichtsteiner et al., 1987; Costa et al., 19886; S. McKnight, personal communication) and some members of CTF/NFl related proteins (Lichtsteiner et al., 1987; Paonessa et al., 1988), are highly enriched in hepatocytes. These factors may be responsible for the liver-specific expression of the albumin gene. The CCAAT-binding factor NF-Y is the only factor w&h affinity to an aibumin promoter element that is present at similar concentrations in all examined cell types (Lichtsteiner et al., 1987; Hooft van Huijsduijnen et al., 1987). Since its recognition site C can also be recognized by the alternative, liverenriched CCAAT factor C/EBP, we decided to identify the positive effector molecule acting via this element. Our experiments show that NF-Y, and not C/EBP enhances albumin in vitro transcription via the CCAAT motif.
2. Materials (a) Nuclear
and Methods
extract preparation
und in vitro tra,nscription
Liver and spleen nuclear extracts were prepared according to Lichtsteiner et al. (1987) and Maire et al. (1989). In vitro transcription reactions were performed as described by Gorski et al. (1986) using a final protein concent#ration of 3 to 5 mg/ml. (b) Fractionation
of a liver nuclear
extract
A liver nuclea,r extract (180 mg in 12 ml) was adjusted to 1 M-NaCl to release DNA-binding proteins from any contaminating DNA, chromatographed on a BioGel Al.5 M column (bed volume 200 ml) that had been equilibrated with running buffer (10 mivr-Hepes (pH 7.6); 1 mM-EDTA, 100 miv-Nacl. 10% (v/v) glycerol, 61 y0 NP40. 1 mnr-dithiot,hreitol. 0.1 mivi-phenyl(v/v) met,hylsulphonyl fluoride, 61 O/c(v/v) Trasylol). Fractions of 10 ml were collected and kept frozen in liquid nitrogen. The purification of C/EBP by the use of gel filtration. heat trea,tment and DNA-cellulose affinity chromatography will be reported elsewhere.
(c) DNase I footprintinq DNase I footprinting (Gallas 85 Schmitz, 1978) was performed according to Lichtsteiner et al. (1987). except that the binding reaction time on ice was reduced from 99 min to 5 min. (d) Gel retardation
assay
Gel retardation experiments were carried out as described by Lichtsteiner et al. (1987), except for the binding reaction time, which was reduced from 90 to 10 min. The double-stranded oligonucleotides used as probes span the sequence from -95 to -73 of the albumin promoter. For the mutation analysis, a 23-mer oligonucleotide corresponding tfo t’he wild-type sequence and 10 23-mer oligonucleotides, each containing a different point mutation were synthetized. To render these oligonucleot,ides double-stranded, a small primer, complementary to the unmodified sequence; was annealed to the 11 oligonucleotides and extended with Klenow enzyme in the presence of unlabelled dGTP; dCTP. dTTI’ and [a-32P]dATP (Mania.tis et ul., 1982). The 2nd strand synthesis was performed in a volume of 20 ~1 at 20°C for 30 min in a cocktail containing 50 m&r-Tris. NC1 (pH 7.5), 10 rnx-MgCl,, 50 mvr-NaCl, 1 mM-dit’hiothreitol, unlabelled dNTP (1 mM each), 20 (*Ci of d.ATP (3000 Ci/mmol), 5 units of Klenow enzyme, 200 ng of oligonucleotide and 200 ng of primer. To complete the synthesis of the 2nd strand. 10 mmol of cold dATP was added, and the reaction incubated for 10 min more at 20°C. The sequences of the oligonucleotides are given below. 1 , TTACGCTCC-5’ 5’.GTAGGAACCAATGAAATGCGAGG-3’ 5’-GTAGc AACCAATGAAATGCGAGG-3’ 5’-GTAGGcACCAATGAAATGCGAGG-3’ 5’.GTAGCA c C C_4ATGAAATGCGSGG-3’ a’-GTAGCAAg CAATGAAATGCGAGG-3’ 5GTAGGAAC a AATGAAATGCGAGG-3’ 5’.GTAGGAACC c ATGAAATGCGAGG-3’ 5’.GTAGGAACCA c TGAAATGCGAGG-3’ 5’.GTAGGAACCAAgGAAATGCGAGG-3’ 5’-GTAGGAACCAAT c AAATGCGAGG-3’ S-GTAGGAACCBATG c AATGCGAGG-3’
Primer Wild-type sequence Mutant 1 Mutant 2 Mut,ant 3 Mutant 1 Mutant 5 Mut,ant 6 Mmant 7 Mutant, 8 Mutant 9 Mutant’ IO
Ubiquitous
Factor for Transcription
(e) Site-directed mutagenesis using Ml3 single-stra,nded templates The EcoRIIHindIII fragment of recombinant plasmid Alb400/ - 170 (Gorski et al., 1986) containing the albumin promoter (- 172 to + 23) fused to the G-less cassette of Sawadogo & Roeder (1985), was subcloned by insertion into EcoRI and Hind111 sites of the vector M13mp19. The construction and selection of the mutants templates were carried out as described by Lichtsteiner et al. (1987). All modified sequences were verified by nucleotide sequencing. (f) Other techniques Plasmid DNA was purified as described (Gorski et al.; 1986). End-labelling by filling in 5’ protruding ends with Klenow enzyme and [a-32P]dNTPs was carried out according to Maniatis et al. (1982).
3. Results (a) Element C is a high-a&&y low-affinity
site for NF- Y and a site for CJEBP
Several groups have reported on the existence of multiple distinct CCAAT binding proteins (Dorn et al., 1987; Raymondjean et al., 1988; Chodosh et al., 1988a,b). Unfortunately, there is no unified terminology for these factors and, for reasons of clarity, we have used the terminology of Dorn et al. (1987) here. The CCAAT binding proteins include NFljCTF (Jones et al., 1987; Lichtsteiner et al., 1987; Paonessa et al., 1988), NF-Y (Dorn et al., 1987) and C/EBP (Graves et al., 1986; Johnson et al., 1987; Landschulz et al., 1988a,b; Vinson et al., 1988). To separate these different activities, a liver nuclear extract was fractionated by gel filtration using BioGel Al.5M. Each fraction was then assayed for site C binding activities by gel retardation assay using a synthetic oligonucleotide corresponding to this site. As shown in Figure 2(c), most of the protein binding to the CCAAT motif (site C) elutes in fractions 11 to 15. This activity corresponds to a ubiquitous transcription factor (Lichtsteiner et al., 1987) identified by Raymondjean et al. (1988) as NF-Y. Some protein-DNA complexes are also detected in fraction 18. These complexes show a migration pattern different from NF-Y. Fraction 18 also contains the majority of the D-binding activity (Fig. 2(b)), which includes another CCAAT factor, the enhancer and CCAAT binding factor C/EBP (Graves et al., 1986; Johnson et al., 1987; Mueller et aZ., 1990; Lichtsteiner et aZ., 1987). That C/EBP indeed has affinity for the C-element was confirmed by a gel shift assay with recombinant C/EBP (Fig. 3(d)). In conclusion then, the CCAAT motif of the albumin promoter has a strong affinity for NF-Y, a lower yet readily measurable affinity for C/EBP. (b) NF-Y
binding correlates with the transcriptional activity of mutant promoters
To identify the factor that activates transcription via binding to the CCAAT-containing element C of
867
the albumin promoter, an extensive point mutation analysis was performed. The objective of these experiments was to find mutant sites with high affmity for one factor and low affinity for the other factor. Ideally, the transcriptional activity of mutant templates should correlate with the affinity of one, but not the other site C-binding activity. Ten different oligonucleotides, each containing a single nucleotide change in the sequences around and including the consensus CCAAT motif, were synthetized (see Materials and Methods). The affinity of C/EBP and NF-Y for each of the mutant binding sites was then tested by a gel shift assay. As shown in Figure 3(b), the binding of NF-Y is very sensitive to any modification within the “CCAAT box”. In contrast, only two single point mutations (-84, T to G, and -83, G to C) decrease the affinity of purified or recombinant C/EBP for site C (Fig. 3(c) and (d)). Interestingly, one of the down mutations for C/EBP (-83) brings the albumin C-motif closer to the NF-Y consensus sequence (Dorn et al., 1987) and indeed increases the affinity of NF-Y significantly, when compared to the wildtype sequence. The converse is observed with the A to C mutation at -82. This modified site binds C/EBP more strongly than the wild-type sequence but shows a greatly diminished affinity for NF-Y. Clearly, these two point mutations are of particular value in determining which of the two factors, NF-Y or C/EBP, acts as a positive effector molecule when bound to site C. After having determined the affinities of NF-Y and C/EBP for the different mutant binding sites, these same sequence changes were introduced into the plasmid Alb 400/- 170 by oligonucleotidedirected, site-specific mutagenesis (Gorski et al., 1986). This plasmid contains the essential albumin promoter region from - 172 to +22 fused to a G-free cassette of 380 base-pairs (Sawadogo & Roeder, 1985). The ten resulting point mutant templates, as well as the parent plasmid bearing the wild-type albumin promoter, were then tested for their transcriptional activity in liver and spleen nuclear extracts. Each experiment contains, as an internal reference, a template bearing a short G-free cassette under the control of the adenovirus major late promoter (Sawadogo & Roeder, 1985). As shown in Figure 3(a) and (b), there is an excellent correlation between the affinities of the mutant templates for NF-Y and their transcriptional activity in the liver nuclear extract. All the mutants with decreased affinities for NF-Y are transcribed considerably less efficiently than the wild-type template. In fact, the mutations that lower the binding of NF-Y to an undetectable level decrease the transcription efficiency to the same level as a linker scanning mutation that removes site C altogether (Fig. 4). Moreover, the two mutants with increased affinity for this factor (- 91, G to C, and -83,
G to C) are more effcient
in vitro transcription
templates than the plasmid containing the wildtype promoter (Alb 400/ - 170). In contrast to the results obtained with
T
8
9
IO
II
12
13
14
IS
!6
17
18
19
-
Free
‘i “3
Figure 2. Separation of C/EKE’ and KF-Y activities. (a) A liver nuclear extract (i&J mg) \*vas fractionat& 1): 2~’ filtration on a KioQel A1.5M column and lo-ml fractions were collect.ed. The absorbance of each of the fractions was measured and is depicted schema.tically. (b) Gel retardat,ion assay of binding site D with the fractions described in (a). A radiolabelled oligonucleotide 1) (1 ng) was incubated with each of the different fractions (2 kd), and the reaction was separated on a 60/O non-denaturing polyacrylamide gel. (c) Gel retardation assay with binding site C. Reactions are as in (b), but. with radiolabelled oligonucleotlde C. The numbers of the fractions are given on t,op of each gel retardation assay. T: gel retardat’ion assay performed with unfractionated liver nuclear extract.
Ubiquitous
In vifro
Factor for Transcription
transcription
869
(e) 200
WT G AAC GA CAAT + + + + + + + + CCCG ACCG
CCCGACCGCC t?tttttttt GAACCAATGA
WT-91 -90-89 -88-87 -86 -85 -8.4 -83 -82
+ i
Bound
Free
Gel retardation
(site Cl
Figure 3. P;F-Y is required to activate transcription via the element C. (a) Transcription with albumin promoter templates containing the point mutation within site C. The promoter (- 170 to +22), fused to a long G-free cassette (380 base-pairs), directs the synthesis of the 400
Affinity + for NF-Y +
i+
+ +
+ +
-
+
Affinity for EBP
+ +
-I+
+ +
+ +
+++ ++++_
+ +
-
-
-
nucleotide long transcript (Alb 400). Each reaction contains, a,s an internal reference, a plasmid bearing the adenovirus major late promoter (-404 to + 10) fused to a 190 nucleotide long G-free cassette, resulting in a 200 nucleotide long transcript (AdML 200). Each reaction (20 ~1) contains about 100 /lg of nuclear protein and an equimolar mixture of the 2 templates (400 ng/eaeh). The change made and nucleotide number of each of the point mutants is indicated below the appropriate lane. (b) Affinity of NF-Y for the various site C mutants. The gel filtration fractions containing NF-Y (Fig. Z(a), fractions 12 and 13) were incubated with 23 base-pair oligonucleotides (-95 to - 73) containing each of the point mutations indicated or the wild type sequence (WT). The reactions were separated on a 6% non-denaturing gel. (c) Affinity of C/EBP for the various site C mutants. Reactions are as in (b!, but with affinity-purified C/EBP (see Materials and Methods). (d) Affinity of C/EBP for the various site C mutants. Reactions as in (b); but with recombinant C/EBP (a gift from Steven McKnight). (e) The regions containing the 2 radiolabelled transcripts were excised from dried gels and counted in Aquasol (Amersham) in a liquid scintillation counter. The transcriptional signals corresponding to the 400 nucleotide transcript (albumin promoter) were normalized with respect to the 200 nucleotide t~ranscript (adenovirus major late promoter), assuming that transcription directed by the viral promoter was equally efficient in all reactions. The values, expressed as a percentage of the wild-type (WT) template, are averages of 3 independent experiments with 3 different nuclear extracts and 3 different plasmid preparations. The percentage values obtained in the 3 independent experiments are given within the respective bars for each mutation. The transcription signal obtained with the wild-type promoter was arbitrarily set as 100%.
870
J. Wuarin
e EXP.
et al. ext,ra.cts. As shown in Figure 3, this mutation for NF-Y while increasing the decrea,ses the afinity affinity for C/EBP. In liver nuclear extracts, this mutation reduces transcription significantly, but has less of an effect in spleen nuclear extracts. Conceivably, C/EBB, present in hepatocytes but, absent in spleen cells, competes successfully for this mutant site in liver but does not activate transcription via this element. (c) NF- Y ‘may act synergistically with liver-enriched transcription factors binding to the albumin promoter
e Ref
Figure 4. Comparison of the transcription from 91b400 templates containing either a point mutation or a linker scanning mutation in site C. Transcriptions were performed with (lane l), the Alb400/-170 plasmid (WT), (lane 2), mutant’ C’ (Lichtsteiner et al., 1987) and (lane 3), mutant -84 (T to 0: see Fig. 3). Mutant C’ contains a substitution in site C in which the 11 nucleotide sequence CCAATGAAATG has been replaced by the unrelated 11 nucleotide sequence agatctgact,ct. Comparison of the transcription signals obtained with these 3 templates revealed the following relative transcription efficiencies: WT. 100%: mutant’ C’, 25%; mutant 8; 23%.
mutations affecting KF-Y binding, the mutations affecting C/EBP binding do not correlate with the relative in vitro transcription efficiencies of these templat~es. These results strongly suggest that in liver nuclear extracts the binding of KF-Y, and not C/EBP, to site C is required for optimal transcription efficiency of the albumin promoter. As mentioned above, NF-Y is t’he only factor binding to the albumin promoter with a similarly high concentration in expressing and non-expressing tissues (Lichtsteiner et al., 1987). This suggested that this factor may be mainly responsible for the residual in vitro transcription from the albumin promoter observed in spleen nuclear extracts. As seen in Figure 5, the strong down mutations for NF-Y binding result in a, marked decrease of in vitro transcription in spleen nuclear extracts. Indeed, the transcription of the templat,es mutated anywhere in the core CCAAT sequence are transcribed at the same level as a core promoter truncated at position -35, which is devoid of all upstream elements. Transversion of the A t’o C at position -82 is the only mutation exhibiting a slightly different effect, on in vitro transcription in liver and spleen nuclear
Albumin transcripts are undetectable in spleen. as measured by Northern blot or nuclear run-on transcription experiments (Ca,rneiro, 1985). Yet, the ubiquitous CCAAT factor XF-Y can clearly activate albumin in v&o transcription in spleen nuclear extracts. Conceivably, in the context of nuclear ehromatin, NF-Y requires co-operative interactions with proteins binding to other promoter or enhancer elements to successfully recognize its target DNA sequence (see Discussion). Ss reported previously (Lichtsteiner et al., 1987; Lichtsteiner 62 Schibler: factors 1989; Cereghini et aZ., 1988), most other binding to albumin promoter elements are more concentrated in liver then in spleen nuclear extracts. To examine whether these liver-enriched fact,ors may stabilize the interaction of NF-Y with its cognate site, in vitro transcription with liver and spleen nuclear extracts were performed in the presence of increasing amounts of binding site 6. Since NF-Y is present at similar concentrat,ions in the two extracts (Lichtsteiner et al., 1987), this competition experiment should provide information on the relative stability of NF-Y within the transcription complexes formed in liver and spleen nuclear extracts. As shown in Figure 6: the in vitro tran.scription efficiency of the albumin promoter is considerably reduced in spleen extracts by the addition of NF-Y binding sites. In contrast, addition of identical amounts of this DKA to liver nuclear extracts had no measurable effect on transcription from the albumin promoter. Transcription from the adeno major late promoter was not changed either in spleen or liver nuclear extracts by the addition of PJF-Y binding sites. In a cont,rol experiment with an artificial promoter consisting of a TATA motif and seven C-elements, addit,ion of competitor C-oligonucleotides resulted in decreased transcription signals in both liver and spleen nuclear extracts (J. W. unpublished results). Thus, the differential sensitivit’y of in vitro transcription to the added competitor DNA is specific for the albumin promoter. On the basis of these results, we propose that KF-Y is more firmly bound to the albumin promoter in the transcription complexes formed in liver nuclear extract than in those formed in spleen nuclear extracts. We do not know whether this increased binding stability is due to direct or indirect co-operative interactSions (Ptashne, 1988) with other transcription factors present in liver but absent in spleen nuclear extracts.
Ubiquitous
Factor for Transcription
In vitro
871
transcription
AdML ( 200 1
;$;+G$CCGCC WT (b)
G A -91 -90
A C C : 1 f -89 -88 -87 -86 -85 -84
z i -83 -82
Bound
Free
Gel retardation
(site
C)
Figure 5. NF-Y is mainly responsible for the residual activity of the albumin promoter in spleen nuclear extract. (a) Transcription in spleen extract of albumin promoter templates bearing the point mutations within the site C. The reactions, containing the same templates as depicted in Fig. 3(a), are performed with 100 pg of spleen nuclear extract. This experiment has been performed twice with independently isolated nuclear extracts and qualitatively similar results were obtained. Due to the low transcription signals observed in spleen nuclear extracts with templates containing NF-Y site down mutations, no attempts have been made to quantify these signals. (b) Affinity of spleen NF-Y for the point mutated sites C. Spleen nuclear extract (5 pg) was incubated with the point mutated oligonucleotides, as described in Fig. 3(b).
J. Wuarin
872
Liver
0
30
et a!.
Splee
6Q
0
30
60 n
Alb (400
9
dML i 200 1
Figure 6. Oligonucleotide competition of transcription in liver or spleen nuclear extracts. in V&O franscriptiom performed in the presence of increasing amounts of the oligonucleotide for site C. Each rea~ction (20 ~1) contains 100 pg of spleen or liver nuclear extracts, 400 ng of Mb400 and AdMLZOO template, Q and increasing amounts (0 to 60 ng) of multimerized site C: oligonucleotides. (a) The dried film was exposed for 6 h at -70°C: with an intensif3;ing screen (DuPont Lightning Plus). (b) As in (a), but, with short exposure (60 min).
4. Discussion Several distinct fact,ors have been identified that bind to promoter elements containing CCAAT motifs. These include C/EBP (Johnson et al., 1987): NF-Y (Darn et al.: 1987); CTF/NF-1 (Jones et al., 1987): CP-1 and CP-2 (Chodosh et al., 1988a). Apparent,ly the sequences surrounding the canonical CCAAT sequence determine which of the multiple CCAAT binding factors has the highest affinity for a particular k-acting element. The CCAAT element centred around -90 in t’he albumin promoter is occupied most readily wit,h KF-Y, a protein found at indistinguishable levels in nuclei of liver, spleen,
thymus and brain (Lichtsteiner et tel.. l987; U. Schibler, unpublished S. Licht,steiner and results). In addition, gel shift experiments with fractionated liver nuclear extracts (Figs 2 and 3) and competition experiments (Lichtsteiner et al., 1987) also revealed some a%nity of this element for C/EBP. While NF-Y was found to be the major site C-binding act~ivity, this by itself did not identify this factor as a transcriptional act,ivator acting through this DNA element. In fact, high-affinity CCAAT factors have been proposed as competitive repressors in at least two different systems. In the sea urchin hist.one gene, a CCAAT displacements factor was shown to prevent binding of another,
Ubiquitous
Factor for Transcription
perhaps positively acting protein to the CCAAT element (Barberis et al., 1987). In another example, a naturally occurring point mutation in a CCAAT element of a human embryo-specific globin gene has been suggested to be the cause of its continued expression during adulthood (Gelinas et al., 1985). To examine which of the two alternative factors binding to the albumin C-element, NF-Y or C/EBP, is the relevant activator protein, a systematic point mutation analysis was performed. Our in vitro transcription studies with templates that were mutated in site C strongly suggest that NF-Y, and not C/EBP, activates transcription via this CCAAT boxcontaining element. Indeed, a strong correlation between the transcriptional activity of a given mutant
and its affinity
for NF-Y
could be observed
in both liver and spleen nuclear extract. In contrast, no such relationship exists for mutations in site C that either augmented or decreased the relative affinity for C/EBP. While C/EBP does not appear to activate albumin transcription via the low-affinity site C, cotransfection experiments suggest that it can do so by binding to its high-affinity sites D (Friedman et al., 1989; P. Maire and U.S., unpublished results). The existence of multiple alternative proteins recognizing a common regulatory DNA element appears to be the rule rather than the exception, at least in higher eukaryotes. Thus, among the six albumin promoter elements, four have been shown to bind alternative factors: element E is recognized by a series of different NF-l-related activities (Licht)steiner et al., 1987), element D is a strong affinity site for three distinct proteins C/EBP, DBP-1 (Mueller et al., 1990) and DBP-2 (Descombes et all.: unpublished results) element C is recognized by the two CCAAT factors (NF-Y and C/EBP), as shown in this paper, and element A; another lowet al., affinity C/EBP binding site (Lichtsteiner 1987) binds in addition a ubiquitous, yet unidentified factor (S. Lichtsteiner and U.S., unpublished results). Similar examples have been reported in a number of other systems, including the two octamer-binding proteins OTF-1 and OTF-2, the receptors for several steroid hormones (Green & Chambon, 1988), NF-KB and HTF-2 (Baldwin & Sharp, 1988) and several homeobox-containing proteins involved in Drosophila development (Jaynes & O’Farrell, 1988). The precise role of the multiple possible interactions of alternative proteins with a single DNA
regulatory
sequence remains
to be eluci-
dated for eukaryotic systems. Probably the best studied example of such competitive interactions in prokaryotes
is the bacteriophage
lambda.
In this
case, two phage-encoded regulatory proteins, Cro and repressor, are involved in deciding whether the phage enters replication and phage production (lysis) or whether it becomes integrated into the bacterial chromosome (lysogeny). Both of these proteins bind with different affinities to a series of related
operator
sequences
Ptashne, 1986). While the albumin
(for
a
review,
promoter is transcriptionally
see
873
much more active in liver than in spleen nuclear extract (Fig. 6), it is clearly stimulated in both by the ubiquitous factor NF-Y. Yet, the endogenous albumin gene appears to be entirely silent in spleen cells (Carneiro; 1985). Clearly, some other mechanism must be operating in viwo that eliminates the effect of NF-Y. Two dissimilarities between in viva and in vitro conditions could account for the different activation potential of NF-Y in cell-free extracts and in the nucleus of living cell. In the latter, transcription occurs in the presence of an exceedingly high concentration of DNA (about 50 mg/ml) and the assembly of pre-initiation complexes has to compete with the formation of inactive chromatin (nucleosomes or higher-order structures: Workman et al., 1988). The high concentration of DNA reduces the free factor concentration by non-specific and specific competition and t,he formation of chromatin reduces the accessibility of a given &s-acting element for its cognate tra,nscription factor. Thus, the successftll assembly of initiation complexes may require multiple co-operative interactions between a series of factors on a given promoter and/or enhancer sequence. Indeed, many more proteins bind to the albumin promoter in li-ver as compared to spleen nuclear extract (Lichtsteiner et al., 1987). The competition experiment shown in Figure 6 supports the importance of co-operative interactions for the stability of promoter complexes containing NF-Y. When albumin in vitro transcription in liver and spleen nuclear extract, which contain very similar concentrations of NF-Y, is challenged with equivalent amounts of NF-Y binding sites, transcriptional inhibition is only observed in the latter extract. Thus. it appears that in the more elaborate RNA polymerase II initiation complex formed with liver nuclear proteins, NF-Y is more stably bound than in the simpler complex assembled in spleen nuclear extract. This co-operative action may be accomplished directly by contacts between different transcription factors and/or indirectly by common contacts with a third component, such as the TATA factor TFIID or the RNA polymerase II (Ptashne, 1988). The co-operative action of ubiquitous and tissue-enriched transcriptional activators could explain how even generic regulatory molecules could enhance gene expression in a tissue-specific fashion. We thank 0. Jenni and F. Ebener for preparing the Figures. We thank Steven McKnight for providing the C/EBP recombinant plasmid, and Serge Lichtsteiner for preparing the recombinant C/EBP protein. This research was supported by the State of Geneva and a grant from the Swiss National Science Foundation. C.M. acknowledges a post-doctoral fellowship from the National Cancer Institute of Canada, which he obtained during the initial phase of this work.
References Babiss, L. E., Herbst, R. S.; Bennett, 8. L. & Darnell, J. E. (1987). Genes Develop. 1, 256-257.
715
720
721
722
1514
1525
1528
1530
-
2546
2547
795
2534 [2537] 12535 1
(52)
5394
5396
1890
5369 [538] [53.7]
pts1
3 identical Xanthosine Regulatory
tRNA%c tRNAzc phosphorylase gene for zapA
Nucleoside transport system Phosphohistidinoprotein-hexose phosphotransferase HPr Phosphotransferase system enzyme
2 tRNAr, 2 tRNA7’” 2 tRNA7’” and 1 tRNAL’” N-Acetylglucosamine6.phosphate deacetylase GlucosamineB-phosphate deaminase Enzyme IInag
I
Sulphate permease Cysteine synthase o-Acetylserine sulphydrylase B Regulatory gene for deo operon containing deoD encoding purine-nucleoside phosphorylase Glucokinase Glutaminyl-tRNA synthetase Glutamyl-tRNA synthetase DNA ligase 2 identical tRNA&
B
Gene product
2 identical tRNA$$ Asparagine synthetase Enzyme I@’
tRNAb&,
Table 1
be related
same tRNA
as 1ysT
as 1ysV
(in min) genes.
are derived
from
with
with
physical
position
N99 DG2 N99 DG2 CA273 w3110 N99
§
0
§
E. coli B § N99 C600 CA273 w3110 AU2547 N99 5
§ N99 DG2 XPh43 N99 XPh43
Strainx
Reference
1987
1987
(in kh) by multiplying
1986 by 47.2. Brackets
Bachmann, 1987 De Reuse et al., 1984 Saffen et al., 1987 De Reuse et al., 1984 Saffen et al., 1987 Yoshimura et al., 1984 Komine et al., 1990 This work Buxton et al., 1980 Buxton et al., 1980 Kocharyan & Melkumyan,
Plumbridge,
Plumbridge,
Fukuda et al., 1983 Plumbridge, 1987 Breton et al., 1986 Ishino et al., 1986 Yoshimura et al., 1984 Komine et al., 1990 Uemura et al., 1985; this work Nakajima et al., 1981 Plumbridge, 1987 Plumbridge, 1987
This work Plumbridge, 1987 De Reuse et al., 1984 Saffen et al., 1987 Sirko et al., 1987 De Reuse et al., 1984 Sirko et al., 1987 Bachmann, 1987
in this study, and strains
the exact
as valT
In vallJ operon same tRNA Could be related to deoD Could be related to deoR
positions mapped
as valU
operon
same tRNA
In lysl’
Probably shares a common ancestor enzymes IIp’c/IIIp’c (ptsG/crr) -
operon
In valU
to gltX to glnS
same tRNA
related related
to xapR
In 1ysT operon,
Evolutionarily Evolutionarily
Could
-
Located between nagA and metTct Probably shares a common ancestor enzyme IInag (nag& -
Comments
gene products and possible relationships for genes mentioned whose DNA was used for the restriction mapping data
1987). Other t Parentheses indicate position from the E. eoli K-12 linkage map (Bachmann, indicate positions deduced for genes known or strongly suspected to lie between 2 physically 1 Strains from which the DNA used for the restrictlon mapping data was cloned. 4 Strain from which the DNA was cloned was not indicated in the reference.
xapR
xapA
val uclpy
nupc
ptsH
W&E
W/B
nagA
(operon)
metT
2534
53.69
gztx lig
glnS
1ysV (supN)
723 2533 2540 795
2551 2545 2551
[7171 2548
2531
W)
lysT@
(19)
5404 5391 5404
53.62 [152] 5398
position?
(52) 1532 53.67 53 81 1690
dk
CySA cysK eysM df2OR
-
Map (min)
Gene name, ‘map position,
A Ubiquitous CCAAT Factor is Required for Efficient irz Vitro Transcription from the Mouse Albumin Promoter ExBme Wuarin, Department Quai
Christopher
Mueller
and Ueli Schibler
of Molecular Biology, Sciences II, University of Geneva Ernest Ansermet 30, CH-1211 Geneva-4, Switzerland
(Received 13 March
1990; accepted 2 Nay
1990)
Among the various factors binding to DNA elements within the mouse albumin promoter, NF-Y is the only one present at identical concentrations in the nuclei of all examined tissues. NF-Y binds to albumin promoter element C, which contains the sequence CCAAT. To determine whether this factor augments in vitro transcription from the albumin promoter, an extensive point-mutation analysis was performed wit’hin the promoter element C. In liver extracts, six out of the ten mutations result in a strong inhibition of NF-Y decrease in promoter activity. Two mut,ations that increase binding and in a concomitant the affinity of the C-element for NF-Y also augment, the transcription efficiency from the albumin promoter. A similarly strong correlation of NF-Y-binding with transcription efficiency has also been observed in spleen nuclear extracts. The liver-enriched CCAAT and enhancer binding factor C/EBP also recognizes the C element. In contrast to NF-Y; no correlation between the affinity of mutant C-elements for C/EBP and transcript’ional activity could be observed in liver nuclear extracts.
reached by Heard et al. (1987) on the basis of DNA transfection studies. This promoter region contains six distinct elements (A to F) that are protected from DNase I digestion by proteins present in a rat et al., 1987). Most liver nuclear extract (Lichtsteiner of the factors interacting with the various albumin promoter elements have been identified (Lichtsteiner et al., 1987; Schorpp et al., 1988: Courtois et al., 1987; Baumhueter et al., 1988; Babiss et al., 1987; Costa et al., 1988a,b; Cereghini et al.: 198’7; Raymondjean et al., 1988; Lichtsteiner & Schibler, 1989; Mueller et al., 1990). Figure 1 shows the inventory of transcription factors with affinity to the various k-acting elements found within t’he albumin promoter. C/EBP (Graves et al., 1986; Johnson et al., 1987; Landschulz et al., 1988a,b; Vinson et al., 1988) has a high affinity for site D, lower affinities for sites A and F, and an even lower but readily measurable affinity for site C (see below; and see Lichtsteiner et al., 1987; Mueller et al., 1990). Site E can be filled with several members of the CTF/NFl factor family, NF-Y (Dorn et al., 1987; Raymondjean et al., 1988), one of several CAAT factors, recognizes site C which bears the common promoter motif CCAAT. Finally, site B is a highaffinity site for HNFl, a factor binding to promoter elements of a number of liver-specific genes (Lichtsteiner & Schibler, 1989; Cereghini et al., 1988; Courtois et al., 1988). Most of these factors, such as HNFl (Courtois et al., 1987; Baumhueter et al.; 1988; Lichtsteiner &. Schibler, 1989), C-EBP
1. Introduction The tissue-specific activation of many genes is controlled primarily at the level of transcription initiation (e.g. see Derman et al., 1981; Schibler et al., 1983; Tilghman & Belayew, 1982; Shaw et al., 1985). For several cases; it has been suggested that the combinatorial interplay of trans-acting transcription factors with &s-acting promoter elements plays an important role in controlling differential transcription (for a review, see Maniatis et al., 1983). Therefore, a detailed knowledge of these interactions is essential for understanding the molecular basis of cell-type specific gene expression. The developmentally regulat’ed expression of the serum albumin gene provides an excellent model system for studies of this nature (Tilghman & Belayew, 1982; Krumlauf et al., 1985; Godbout et al., 1986; Ott et al., 1984; Heard et aZ., 1987; Gorski et al., 1986). Furthermore, the elaboration in our laboratory of a, tissue-specific in vitro transcription system differential accurately reproduces the that, expression of the albumin gene renders this system amenable to biochemical dissection (Gorski et al., 1986: Lichtsteiner et al., 1987; Lichtsteiner & Schibler, 1989; Maire et al., 1989). Previously reported studies from our laboratory indicated that the minimal albumin promoter sequences required for optimal in, vitro transcription span about’ 170 nucleotides of 5’ flanking region et al., 1986). A similar conclusion was (Gorski
865 002%2836/90/160865-10
$03.00/O
0
1990 Academic
Press Limited
J. Wuarin et al
'\
..." :
/.’ ..’
E
-130
\ D
-100
....’
c
-80
i
B
;
‘i
-50
k
~I~
A
““, h,<,, -30
A AG TATGGTTAATGATCT A CAGTT ATTGGTTAAAG T TC ATACCAATTACTAGA T GTCAA TAACCAATTTC
Figure I. Schematic representation of the factors interacting with the albumin promoter. The different fact,ors that have been shown to interact with the 6 binding sites (A, B, C. D; E and F) of the albumin promoter (Lichtsteiner el’ aE.. 1987; Schorpp et al., 1988; Mueller et al.. 1990; S. Lichtsteiner and U. Schibler, unpublished results) are depicted. Projected onto this schema is the nucleotide sequence of the albumin promoter from - 166 to - 27 with the sequences protected by each factor from DNase I digestion appearing within the boxed regions. C/EBP, a protein with relaxed sequence specificity binds strongly to its primary site D, but more weakly to its secondary sites A, C and F. Pseudo NF- 1 appears to recognize the same site as NF-I but exhibits greater sensitivity to competition as well as showing some tissue specificity. NF-Y is a ubiquit,ous CCAAT binding protein (Lichtst,einer et al., 1987; Dorn et al., 1987; Raymondjean el al.,
1988). (Lichtsteiner et al., 1987; Costa et al., 19886; S. McKnight, personal communication) and some members of CTF/NFl related proteins (Lichtsteiner et al., 1987; Paonessa et al., 1988), are highly enriched in hepatocytes. These factors may be responsible for the liver-specific expression of the albumin gene. The CCAAT-binding factor NF-Y is the only factor w&h affinity to an aibumin promoter element that is present at similar concentrations in all examined cell types (Lichtsteiner et al., 1987; Hooft van Huijsduijnen et al., 1987). Since its recognition site C can also be recognized by the alternative, liverenriched CCAAT factor C/EBP, we decided to identify the positive effector molecule acting via this element. Our experiments show that NF-Y, and not C/EBP enhances albumin in vitro transcription via the CCAAT motif.
2. Materials (a) Nuclear
and Methods
extract preparation
und in vitro tra,nscription
Liver and spleen nuclear extracts were prepared according to Lichtsteiner et al. (1987) and Maire et al. (1989). In vitro transcription reactions were performed as described by Gorski et al. (1986) using a final protein concent#ration of 3 to 5 mg/ml. (b) Fractionation
of a liver nuclear
extract
A liver nuclea,r extract (180 mg in 12 ml) was adjusted to 1 M-NaCl to release DNA-binding proteins from any contaminating DNA, chromatographed on a BioGel Al.5 M column (bed volume 200 ml) that had been equilibrated with running buffer (10 mivr-Hepes (pH 7.6); 1 mM-EDTA, 100 miv-Nacl. 10% (v/v) glycerol, 61 y0 NP40. 1 mnr-dithiot,hreitol. 0.1 mivi-phenyl(v/v) met,hylsulphonyl fluoride, 61 O/c(v/v) Trasylol). Fractions of 10 ml were collected and kept frozen in liquid nitrogen. The purification of C/EBP by the use of gel filtration. heat trea,tment and DNA-cellulose affinity chromatography will be reported elsewhere.
(c) DNase I footprintinq DNase I footprinting (Gallas 85 Schmitz, 1978) was performed according to Lichtsteiner et al. (1987). except that the binding reaction time on ice was reduced from 99 min to 5 min. (d) Gel retardation
assay
Gel retardation experiments were carried out as described by Lichtsteiner et al. (1987), except for the binding reaction time, which was reduced from 90 to 10 min. The double-stranded oligonucleotides used as probes span the sequence from -95 to -73 of the albumin promoter. For the mutation analysis, a 23-mer oligonucleotide corresponding tfo t’he wild-type sequence and 10 23-mer oligonucleotides, each containing a different point mutation were synthetized. To render these oligonucleot,ides double-stranded, a small primer, complementary to the unmodified sequence; was annealed to the 11 oligonucleotides and extended with Klenow enzyme in the presence of unlabelled dGTP; dCTP. dTTI’ and [a-32P]dATP (Mania.tis et ul., 1982). The 2nd strand synthesis was performed in a volume of 20 ~1 at 20°C for 30 min in a cocktail containing 50 m&r-Tris. NC1 (pH 7.5), 10 rnx-MgCl,, 50 mvr-NaCl, 1 mM-dit’hiothreitol, unlabelled dNTP (1 mM each), 20 (*Ci of d.ATP (3000 Ci/mmol), 5 units of Klenow enzyme, 200 ng of oligonucleotide and 200 ng of primer. To complete the synthesis of the 2nd strand. 10 mmol of cold dATP was added, and the reaction incubated for 10 min more at 20°C. The sequences of the oligonucleotides are given below. 1 , TTACGCTCC-5’ 5’.GTAGGAACCAATGAAATGCGAGG-3’ 5’-GTAGc AACCAATGAAATGCGAGG-3’ 5’-GTAGGcACCAATGAAATGCGAGG-3’ 5’.GTAGCA c C C_4ATGAAATGCGSGG-3’ a’-GTAGCAAg CAATGAAATGCGAGG-3’ 5GTAGGAAC a AATGAAATGCGAGG-3’ 5’.GTAGGAACC c ATGAAATGCGAGG-3’ 5’.GTAGGAACCA c TGAAATGCGAGG-3’ 5’.GTAGGAACCAAgGAAATGCGAGG-3’ 5’-GTAGGAACCAAT c AAATGCGAGG-3’ S-GTAGGAACCBATG c AATGCGAGG-3’
Primer Wild-type sequence Mutant 1 Mutant 2 Mut,ant 3 Mutant 1 Mutant 5 Mut,ant 6 Mmant 7 Mutant, 8 Mutant 9 Mutant’ IO
Ubiquitous
Factor for Transcription
(e) Site-directed mutagenesis using Ml3 single-stra,nded templates The EcoRIIHindIII fragment of recombinant plasmid Alb400/ - 170 (Gorski et al., 1986) containing the albumin promoter (- 172 to + 23) fused to the G-less cassette of Sawadogo & Roeder (1985), was subcloned by insertion into EcoRI and Hind111 sites of the vector M13mp19. The construction and selection of the mutants templates were carried out as described by Lichtsteiner et al. (1987). All modified sequences were verified by nucleotide sequencing. (f) Other techniques Plasmid DNA was purified as described (Gorski et al.; 1986). End-labelling by filling in 5’ protruding ends with Klenow enzyme and [a-32P]dNTPs was carried out according to Maniatis et al. (1982).
3. Results (a) Element C is a high-a&&y low-affinity
site for NF- Y and a site for CJEBP
Several groups have reported on the existence of multiple distinct CCAAT binding proteins (Dorn et al., 1987; Raymondjean et al., 1988; Chodosh et al., 1988a,b). Unfortunately, there is no unified terminology for these factors and, for reasons of clarity, we have used the terminology of Dorn et al. (1987) here. The CCAAT binding proteins include NFljCTF (Jones et al., 1987; Lichtsteiner et al., 1987; Paonessa et al., 1988), NF-Y (Dorn et al., 1987) and C/EBP (Graves et al., 1986; Johnson et al., 1987; Landschulz et al., 1988a,b; Vinson et al., 1988). To separate these different activities, a liver nuclear extract was fractionated by gel filtration using BioGel Al.5M. Each fraction was then assayed for site C binding activities by gel retardation assay using a synthetic oligonucleotide corresponding to this site. As shown in Figure 2(c), most of the protein binding to the CCAAT motif (site C) elutes in fractions 11 to 15. This activity corresponds to a ubiquitous transcription factor (Lichtsteiner et al., 1987) identified by Raymondjean et al. (1988) as NF-Y. Some protein-DNA complexes are also detected in fraction 18. These complexes show a migration pattern different from NF-Y. Fraction 18 also contains the majority of the D-binding activity (Fig. 2(b)), which includes another CCAAT factor, the enhancer and CCAAT binding factor C/EBP (Graves et al., 1986; Johnson et al., 1987; Mueller et aZ., 1990; Lichtsteiner et aZ., 1987). That C/EBP indeed has affinity for the C-element was confirmed by a gel shift assay with recombinant C/EBP (Fig. 3(d)). In conclusion then, the CCAAT motif of the albumin promoter has a strong affinity for NF-Y, a lower yet readily measurable affinity for C/EBP. (b) NF-Y
binding correlates with the transcriptional activity of mutant promoters
To identify the factor that activates transcription via binding to the CCAAT-containing element C of
867
the albumin promoter, an extensive point mutation analysis was performed. The objective of these experiments was to find mutant sites with high affmity for one factor and low affinity for the other factor. Ideally, the transcriptional activity of mutant templates should correlate with the affinity of one, but not the other site C-binding activity. Ten different oligonucleotides, each containing a single nucleotide change in the sequences around and including the consensus CCAAT motif, were synthetized (see Materials and Methods). The affinity of C/EBP and NF-Y for each of the mutant binding sites was then tested by a gel shift assay. As shown in Figure 3(b), the binding of NF-Y is very sensitive to any modification within the “CCAAT box”. In contrast, only two single point mutations (-84, T to G, and -83, G to C) decrease the affinity of purified or recombinant C/EBP for site C (Fig. 3(c) and (d)). Interestingly, one of the down mutations for C/EBP (-83) brings the albumin C-motif closer to the NF-Y consensus sequence (Dorn et al., 1987) and indeed increases the affinity of NF-Y significantly, when compared to the wildtype sequence. The converse is observed with the A to C mutation at -82. This modified site binds C/EBP more strongly than the wild-type sequence but shows a greatly diminished affinity for NF-Y. Clearly, these two point mutations are of particular value in determining which of the two factors, NF-Y or C/EBP, acts as a positive effector molecule when bound to site C. After having determined the affinities of NF-Y and C/EBP for the different mutant binding sites, these same sequence changes were introduced into the plasmid Alb 400/- 170 by oligonucleotidedirected, site-specific mutagenesis (Gorski et al., 1986). This plasmid contains the essential albumin promoter region from - 172 to +22 fused to a G-free cassette of 380 base-pairs (Sawadogo & Roeder, 1985). The ten resulting point mutant templates, as well as the parent plasmid bearing the wild-type albumin promoter, were then tested for their transcriptional activity in liver and spleen nuclear extracts. Each experiment contains, as an internal reference, a template bearing a short G-free cassette under the control of the adenovirus major late promoter (Sawadogo & Roeder, 1985). As shown in Figure 3(a) and (b), there is an excellent correlation between the affinities of the mutant templates for NF-Y and their transcriptional activity in the liver nuclear extract. All the mutants with decreased affinities for NF-Y are transcribed considerably less efficiently than the wild-type template. In fact, the mutations that lower the binding of NF-Y to an undetectable level decrease the transcription efficiency to the same level as a linker scanning mutation that removes site C altogether (Fig. 4). Moreover, the two mutants with increased affinity for this factor (- 91, G to C, and -83,
G to C) are more effcient
in vitro transcription
templates than the plasmid containing the wildtype promoter (Alb 400/ - 170). In contrast to the results obtained with
T
8
9
IO
II
12
13
14
IS
!6
17
18
19
-
Free
‘i “3
Figure 2. Separation of C/EKE’ and KF-Y activities. (a) A liver nuclear extract (i&J mg) \*vas fractionat& 1): 2~’ filtration on a KioQel A1.5M column and lo-ml fractions were collect.ed. The absorbance of each of the fractions was measured and is depicted schema.tically. (b) Gel retardat,ion assay of binding site D with the fractions described in (a). A radiolabelled oligonucleotide 1) (1 ng) was incubated with each of the different fractions (2 kd), and the reaction was separated on a 60/O non-denaturing polyacrylamide gel. (c) Gel retardation assay with binding site C. Reactions are as in (b), but. with radiolabelled oligonucleotlde C. The numbers of the fractions are given on t,op of each gel retardation assay. T: gel retardat’ion assay performed with unfractionated liver nuclear extract.
Ubiquitous
In vifro
Factor for Transcription
transcription
869
(e) 200
WT G AAC GA CAAT + + + + + + + + CCCG ACCG
CCCGACCGCC t?tttttttt GAACCAATGA
WT-91 -90-89 -88-87 -86 -85 -8.4 -83 -82
+ i
Bound
Free
Gel retardation
(site Cl
Figure 3. P;F-Y is required to activate transcription via the element C. (a) Transcription with albumin promoter templates containing the point mutation within site C. The promoter (- 170 to +22), fused to a long G-free cassette (380 base-pairs), directs the synthesis of the 400
Affinity + for NF-Y +
i+
+ +
+ +
-
+
Affinity for EBP
+ +
-I+
+ +
+ +
+++ ++++_
+ +
-
-
-
nucleotide long transcript (Alb 400). Each reaction contains, a,s an internal reference, a plasmid bearing the adenovirus major late promoter (-404 to + 10) fused to a 190 nucleotide long G-free cassette, resulting in a 200 nucleotide long transcript (AdML 200). Each reaction (20 ~1) contains about 100 /lg of nuclear protein and an equimolar mixture of the 2 templates (400 ng/eaeh). The change made and nucleotide number of each of the point mutants is indicated below the appropriate lane. (b) Affinity of NF-Y for the various site C mutants. The gel filtration fractions containing NF-Y (Fig. Z(a), fractions 12 and 13) were incubated with 23 base-pair oligonucleotides (-95 to - 73) containing each of the point mutations indicated or the wild type sequence (WT). The reactions were separated on a 6% non-denaturing gel. (c) Affinity of C/EBP for the various site C mutants. Reactions are as in (b!, but with affinity-purified C/EBP (see Materials and Methods). (d) Affinity of C/EBP for the various site C mutants. Reactions as in (b); but with recombinant C/EBP (a gift from Steven McKnight). (e) The regions containing the 2 radiolabelled transcripts were excised from dried gels and counted in Aquasol (Amersham) in a liquid scintillation counter. The transcriptional signals corresponding to the 400 nucleotide transcript (albumin promoter) were normalized with respect to the 200 nucleotide t~ranscript (adenovirus major late promoter), assuming that transcription directed by the viral promoter was equally efficient in all reactions. The values, expressed as a percentage of the wild-type (WT) template, are averages of 3 independent experiments with 3 different nuclear extracts and 3 different plasmid preparations. The percentage values obtained in the 3 independent experiments are given within the respective bars for each mutation. The transcription signal obtained with the wild-type promoter was arbitrarily set as 100%.
870
J. Wuarin
e EXP.
et al. ext,ra.cts. As shown in Figure 3, this mutation for NF-Y while increasing the decrea,ses the afinity affinity for C/EBP. In liver nuclear extracts, this mutation reduces transcription significantly, but has less of an effect in spleen nuclear extracts. Conceivably, C/EBB, present in hepatocytes but, absent in spleen cells, competes successfully for this mutant site in liver but does not activate transcription via this element. (c) NF- Y ‘may act synergistically with liver-enriched transcription factors binding to the albumin promoter
e Ref
Figure 4. Comparison of the transcription from 91b400 templates containing either a point mutation or a linker scanning mutation in site C. Transcriptions were performed with (lane l), the Alb400/-170 plasmid (WT), (lane 2), mutant’ C’ (Lichtsteiner et al., 1987) and (lane 3), mutant -84 (T to 0: see Fig. 3). Mutant C’ contains a substitution in site C in which the 11 nucleotide sequence CCAATGAAATG has been replaced by the unrelated 11 nucleotide sequence agatctgact,ct. Comparison of the transcription signals obtained with these 3 templates revealed the following relative transcription efficiencies: WT. 100%: mutant’ C’, 25%; mutant 8; 23%.
mutations affecting KF-Y binding, the mutations affecting C/EBP binding do not correlate with the relative in vitro transcription efficiencies of these templat~es. These results strongly suggest that in liver nuclear extracts the binding of KF-Y, and not C/EBP, to site C is required for optimal transcription efficiency of the albumin promoter. As mentioned above, NF-Y is t’he only factor binding to the albumin promoter with a similarly high concentration in expressing and non-expressing tissues (Lichtsteiner et al., 1987). This suggested that this factor may be mainly responsible for the residual in vitro transcription from the albumin promoter observed in spleen nuclear extracts. As seen in Figure 5, the strong down mutations for NF-Y binding result in a, marked decrease of in vitro transcription in spleen nuclear extracts. Indeed, the transcription of the templat,es mutated anywhere in the core CCAAT sequence are transcribed at the same level as a core promoter truncated at position -35, which is devoid of all upstream elements. Transversion of the A t’o C at position -82 is the only mutation exhibiting a slightly different effect, on in vitro transcription in liver and spleen nuclear
Albumin transcripts are undetectable in spleen. as measured by Northern blot or nuclear run-on transcription experiments (Ca,rneiro, 1985). Yet, the ubiquitous CCAAT factor XF-Y can clearly activate albumin in v&o transcription in spleen nuclear extracts. Conceivably, in the context of nuclear ehromatin, NF-Y requires co-operative interactions with proteins binding to other promoter or enhancer elements to successfully recognize its target DNA sequence (see Discussion). Ss reported previously (Lichtsteiner et al., 1987; Lichtsteiner 62 Schibler: factors 1989; Cereghini et aZ., 1988), most other binding to albumin promoter elements are more concentrated in liver then in spleen nuclear extracts. To examine whether these liver-enriched fact,ors may stabilize the interaction of NF-Y with its cognate site, in vitro transcription with liver and spleen nuclear extracts were performed in the presence of increasing amounts of binding site 6. Since NF-Y is present at similar concentrat,ions in the two extracts (Lichtsteiner et al., 1987), this competition experiment should provide information on the relative stability of NF-Y within the transcription complexes formed in liver and spleen nuclear extracts. As shown in Figure 6: the in vitro tran.scription efficiency of the albumin promoter is considerably reduced in spleen extracts by the addition of NF-Y binding sites. In contrast, addition of identical amounts of this DKA to liver nuclear extracts had no measurable effect on transcription from the albumin promoter. Transcription from the adeno major late promoter was not changed either in spleen or liver nuclear extracts by the addition of PJF-Y binding sites. In a cont,rol experiment with an artificial promoter consisting of a TATA motif and seven C-elements, addit,ion of competitor C-oligonucleotides resulted in decreased transcription signals in both liver and spleen nuclear extracts (J. W. unpublished results). Thus, the differential sensitivit’y of in vitro transcription to the added competitor DNA is specific for the albumin promoter. On the basis of these results, we propose that KF-Y is more firmly bound to the albumin promoter in the transcription complexes formed in liver nuclear extract than in those formed in spleen nuclear extracts. We do not know whether this increased binding stability is due to direct or indirect co-operative interactSions (Ptashne, 1988) with other transcription factors present in liver but absent in spleen nuclear extracts.
Ubiquitous
Factor for Transcription
In vitro
871
transcription
AdML ( 200 1
;$;+G$CCGCC WT (b)
G A -91 -90
A C C : 1 f -89 -88 -87 -86 -85 -84
z i -83 -82
Bound
Free
Gel retardation
(site
C)
Figure 5. NF-Y is mainly responsible for the residual activity of the albumin promoter in spleen nuclear extract. (a) Transcription in spleen extract of albumin promoter templates bearing the point mutations within the site C. The reactions, containing the same templates as depicted in Fig. 3(a), are performed with 100 pg of spleen nuclear extract. This experiment has been performed twice with independently isolated nuclear extracts and qualitatively similar results were obtained. Due to the low transcription signals observed in spleen nuclear extracts with templates containing NF-Y site down mutations, no attempts have been made to quantify these signals. (b) Affinity of spleen NF-Y for the point mutated sites C. Spleen nuclear extract (5 pg) was incubated with the point mutated oligonucleotides, as described in Fig. 3(b).
J. Wuarin
872
Liver
0
30
et a!.
Splee
6Q
0
30
60 n
Alb (400
9
dML i 200 1
Figure 6. Oligonucleotide competition of transcription in liver or spleen nuclear extracts. in V&O franscriptiom performed in the presence of increasing amounts of the oligonucleotide for site C. Each rea~ction (20 ~1) contains 100 pg of spleen or liver nuclear extracts, 400 ng of Mb400 and AdMLZOO template, Q and increasing amounts (0 to 60 ng) of multimerized site C: oligonucleotides. (a) The dried film was exposed for 6 h at -70°C: with an intensif3;ing screen (DuPont Lightning Plus). (b) As in (a), but, with short exposure (60 min).
4. Discussion Several distinct fact,ors have been identified that bind to promoter elements containing CCAAT motifs. These include C/EBP (Johnson et al., 1987): NF-Y (Darn et al.: 1987); CTF/NF-1 (Jones et al., 1987): CP-1 and CP-2 (Chodosh et al., 1988a). Apparent,ly the sequences surrounding the canonical CCAAT sequence determine which of the multiple CCAAT binding factors has the highest affinity for a particular k-acting element. The CCAAT element centred around -90 in t’he albumin promoter is occupied most readily wit,h KF-Y, a protein found at indistinguishable levels in nuclei of liver, spleen,
thymus and brain (Lichtsteiner et tel.. l987; U. Schibler, unpublished S. Licht,steiner and results). In addition, gel shift experiments with fractionated liver nuclear extracts (Figs 2 and 3) and competition experiments (Lichtsteiner et al., 1987) also revealed some a%nity of this element for C/EBP. While NF-Y was found to be the major site C-binding act~ivity, this by itself did not identify this factor as a transcriptional act,ivator acting through this DNA element. In fact, high-affinity CCAAT factors have been proposed as competitive repressors in at least two different systems. In the sea urchin hist.one gene, a CCAAT displacements factor was shown to prevent binding of another,
Ubiquitous
Factor for Transcription
perhaps positively acting protein to the CCAAT element (Barberis et al., 1987). In another example, a naturally occurring point mutation in a CCAAT element of a human embryo-specific globin gene has been suggested to be the cause of its continued expression during adulthood (Gelinas et al., 1985). To examine which of the two alternative factors binding to the albumin C-element, NF-Y or C/EBP, is the relevant activator protein, a systematic point mutation analysis was performed. Our in vitro transcription studies with templates that were mutated in site C strongly suggest that NF-Y, and not C/EBP, activates transcription via this CCAAT boxcontaining element. Indeed, a strong correlation between the transcriptional activity of a given mutant
and its affinity
for NF-Y
could be observed
in both liver and spleen nuclear extract. In contrast, no such relationship exists for mutations in site C that either augmented or decreased the relative affinity for C/EBP. While C/EBP does not appear to activate albumin transcription via the low-affinity site C, cotransfection experiments suggest that it can do so by binding to its high-affinity sites D (Friedman et al., 1989; P. Maire and U.S., unpublished results). The existence of multiple alternative proteins recognizing a common regulatory DNA element appears to be the rule rather than the exception, at least in higher eukaryotes. Thus, among the six albumin promoter elements, four have been shown to bind alternative factors: element E is recognized by a series of different NF-l-related activities (Licht)steiner et al., 1987), element D is a strong affinity site for three distinct proteins C/EBP, DBP-1 (Mueller et al., 1990) and DBP-2 (Descombes et all.: unpublished results) element C is recognized by the two CCAAT factors (NF-Y and C/EBP), as shown in this paper, and element A; another lowet al., affinity C/EBP binding site (Lichtsteiner 1987) binds in addition a ubiquitous, yet unidentified factor (S. Lichtsteiner and U.S., unpublished results). Similar examples have been reported in a number of other systems, including the two octamer-binding proteins OTF-1 and OTF-2, the receptors for several steroid hormones (Green & Chambon, 1988), NF-KB and HTF-2 (Baldwin & Sharp, 1988) and several homeobox-containing proteins involved in Drosophila development (Jaynes & O’Farrell, 1988). The precise role of the multiple possible interactions of alternative proteins with a single DNA
regulatory
sequence remains
to be eluci-
dated for eukaryotic systems. Probably the best studied example of such competitive interactions in prokaryotes
is the bacteriophage
lambda.
In this
case, two phage-encoded regulatory proteins, Cro and repressor, are involved in deciding whether the phage enters replication and phage production (lysis) or whether it becomes integrated into the bacterial chromosome (lysogeny). Both of these proteins bind with different affinities to a series of related
operator
sequences
Ptashne, 1986). While the albumin
(for
a
review,
promoter is transcriptionally
see
873
much more active in liver than in spleen nuclear extract (Fig. 6), it is clearly stimulated in both by the ubiquitous factor NF-Y. Yet, the endogenous albumin gene appears to be entirely silent in spleen cells (Carneiro; 1985). Clearly, some other mechanism must be operating in viwo that eliminates the effect of NF-Y. Two dissimilarities between in viva and in vitro conditions could account for the different activation potential of NF-Y in cell-free extracts and in the nucleus of living cell. In the latter, transcription occurs in the presence of an exceedingly high concentration of DNA (about 50 mg/ml) and the assembly of pre-initiation complexes has to compete with the formation of inactive chromatin (nucleosomes or higher-order structures: Workman et al., 1988). The high concentration of DNA reduces the free factor concentration by non-specific and specific competition and t,he formation of chromatin reduces the accessibility of a given &s-acting element for its cognate tra,nscription factor. Thus, the successftll assembly of initiation complexes may require multiple co-operative interactions between a series of factors on a given promoter and/or enhancer sequence. Indeed, many more proteins bind to the albumin promoter in li-ver as compared to spleen nuclear extract (Lichtsteiner et al., 1987). The competition experiment shown in Figure 6 supports the importance of co-operative interactions for the stability of promoter complexes containing NF-Y. When albumin in vitro transcription in liver and spleen nuclear extract, which contain very similar concentrations of NF-Y, is challenged with equivalent amounts of NF-Y binding sites, transcriptional inhibition is only observed in the latter extract. Thus. it appears that in the more elaborate RNA polymerase II initiation complex formed with liver nuclear proteins, NF-Y is more stably bound than in the simpler complex assembled in spleen nuclear extract. This co-operative action may be accomplished directly by contacts between different transcription factors and/or indirectly by common contacts with a third component, such as the TATA factor TFIID or the RNA polymerase II (Ptashne, 1988). The co-operative action of ubiquitous and tissue-enriched transcriptional activators could explain how even generic regulatory molecules could enhance gene expression in a tissue-specific fashion. We thank 0. Jenni and F. Ebener for preparing the Figures. We thank Steven McKnight for providing the C/EBP recombinant plasmid, and Serge Lichtsteiner for preparing the recombinant C/EBP protein. This research was supported by the State of Geneva and a grant from the Swiss National Science Foundation. C.M. acknowledges a post-doctoral fellowship from the National Cancer Institute of Canada, which he obtained during the initial phase of this work.
References Babiss, L. E., Herbst, R. S.; Bennett, 8. L. & Darnell, J. E. (1987). Genes Develop. 1, 256-257.
715
720
721
722
1514
1525
1528
1530
-
2546
2547
795
2534 [2537] 12535 1
(52)
5394
5396
1890
5369 [538] [53.7]
pts1
3 identical Xanthosine Regulatory
tRNA%c tRNAzc phosphorylase gene for zapA
Nucleoside transport system Phosphohistidinoprotein-hexose phosphotransferase HPr Phosphotransferase system enzyme
2 tRNAr, 2 tRNA7’” 2 tRNA7’” and 1 tRNAL’” N-Acetylglucosamine6.phosphate deacetylase GlucosamineB-phosphate deaminase Enzyme IInag
I
Sulphate permease Cysteine synthase o-Acetylserine sulphydrylase B Regulatory gene for deo operon containing deoD encoding purine-nucleoside phosphorylase Glucokinase Glutaminyl-tRNA synthetase Glutamyl-tRNA synthetase DNA ligase 2 identical tRNA&
B
Gene product
2 identical tRNA$$ Asparagine synthetase Enzyme I@’
tRNAb&,
Table 1
be related
same tRNA
as 1ysT
as 1ysV
(in min) genes.
are derived
from
with
with
physical
position
N99 DG2 N99 DG2 CA273 w3110 N99
§
0
§
E. coli B § N99 C600 CA273 w3110 AU2547 N99 5
§ N99 DG2 XPh43 N99 XPh43
Strainx
Reference
1987
1987
(in kh) by multiplying
1986 by 47.2. Brackets
Bachmann, 1987 De Reuse et al., 1984 Saffen et al., 1987 De Reuse et al., 1984 Saffen et al., 1987 Yoshimura et al., 1984 Komine et al., 1990 This work Buxton et al., 1980 Buxton et al., 1980 Kocharyan & Melkumyan,
Plumbridge,
Plumbridge,
Fukuda et al., 1983 Plumbridge, 1987 Breton et al., 1986 Ishino et al., 1986 Yoshimura et al., 1984 Komine et al., 1990 Uemura et al., 1985; this work Nakajima et al., 1981 Plumbridge, 1987 Plumbridge, 1987
This work Plumbridge, 1987 De Reuse et al., 1984 Saffen et al., 1987 Sirko et al., 1987 De Reuse et al., 1984 Sirko et al., 1987 Bachmann, 1987
in this study, and strains
the exact
as valT
In vallJ operon same tRNA Could be related to deoD Could be related to deoR
positions mapped
as valU
operon
same tRNA
In lysl’
Probably shares a common ancestor enzymes IIp’c/IIIp’c (ptsG/crr) -
operon
In valU
to gltX to glnS
same tRNA
related related
to xapR
In 1ysT operon,
Evolutionarily Evolutionarily
Could
-
Located between nagA and metTct Probably shares a common ancestor enzyme IInag (nag& -
Comments
gene products and possible relationships for genes mentioned whose DNA was used for the restriction mapping data
1987). Other t Parentheses indicate position from the E. eoli K-12 linkage map (Bachmann, indicate positions deduced for genes known or strongly suspected to lie between 2 physically 1 Strains from which the DNA used for the restrictlon mapping data was cloned. 4 Strain from which the DNA was cloned was not indicated in the reference.
xapR
xapA
val uclpy
nupc
ptsH
W&E
W/B
nagA
(operon)
metT
2534
53.69
gztx lig
glnS
1ysV (supN)
723 2533 2540 795
2551 2545 2551
[7171 2548
2531
W)
lysT@
(19)
5404 5391 5404
53.62 [152] 5398
position?
(52) 1532 53.67 53 81 1690
dk
CySA cysK eysM df2OR
-
Map (min)
Gene name, ‘map position,