BAP, a rat liver protein that activates transcription through

Nucleic Acids Research, Vol. 18, No. 23 6943

BAP, a rat liver protein that activates transcription througha promoter element with similarity to the USF/MLTFbinding site

Wilfried Kugler+, Michael Kaling§, Katrin Ross, Ulrike Wagner and Gerhart U.Ryffell *Kernforschungszentrum Karlsruhe, Institut fOr Genetik und Toxikologie, Postfach 36 40, D-7500Karlsruhe and 'Institut fOr Zellbiologie (Tumorforschung), Universitatsklinikum Essen, Hufelandstrasse55, D-4300 Essen 1, FRG

Received August 23, 1990; Revised and Accepted October 23, 1990

ABSTRACT

The vitellogenin genes of Xenopus are liver-specificallyexpressed. An in vitro transcription system derivedfrom rat liver nuclei allowed us to define the cis-elementBABS (B-activator binding site) in the promoter of theBi vitellogenin gene. An oligonucleotide encompassingthe region from - 53 to - 44 linked to a TATA box issufficient for a tenfold increase of the transcriptionalactivity. Gel retardation assays with nuclear rat liverproteins reveal two DNA-protein complexes: Complex 1can be competed by the USF/MLTF binding site of theadeno major late promoter whereas complex 2 is adistinct protein we refer to as BAP (B-activator protein).In vitro transcription experiments in the presence ofUSF/MLTF binding site as competitor show that BAPis an efficient transcription factor. Based on UV cross-linking we estimate that BAP has a molecular weightof 58 kd. Phosphatase treatment reveals that DNAbinding of BAP requires phosphorylation. BABS is alsopresent in the hepatitis B virus enhancer suggestingthat it might play a role in the tumorigenic potential ofthe virus.

INTRODUCTIONTranscriptional control of gene expression is the majormechanism involved in regulating gene activities in eukaryotes.This process is largely based on the interaction of transcriptionfactors with short stretches of DNA sequences in the promotersand enhancers (1, 2). It is concievable that the activity of a givengene is regulated by the presence of various cis-elements andthe availability of transcription factors interacting with theseregulatory elements. To elucidate regulatory processes it isessential to define the various cis-elements and theircorresponding trans-acting factors.

The four vitellogenin genes Al, A2, BI and B2 of Xenopusprovide a unique system for studying cis-elements and trans-actingfactors involved in estrogen inducible, liver-specific anddevelopmentally regulated gene expression (3, 4). As it has wellbeen established that trans-acting factors and the correspondingcis-elements have been conserved during evolution (5), regulatoryelements can be described in heterologous systems. Using suchan approach, we have identified in frog genes regulatory elementsfunctioning in mammalian cells: The albumin promoter containsHPl, a regulatory element conferring liver-specific geneexpression (6) and this regulatory element corresponds to theLFBl/HNFI binding site identified in mammalian genesspecifically expressed in hepatocytes (7, 8, 9, 10). Thevitellogenin genes contain an estrogen responsive element (1 1,12) and a cell-specific activator binding site (11, 13) functioningin mammalian cells. The cell-specific activator binding site hasbeen identified in the promoter of the A2 vitellogenin gene. Asthe A and B vitellogenin genes of Xenopus are coordinatelyexpressed (14), we have looked for a corresponding regulatoryelement in the other vitellogenin genes.

In this report we describe the characterization of atranscriptional regulatory element (BABS) of the Xenopus Blvitellogenin gene which is distinct from previously describedregulatory elements. Using extracts from rat liver nuclei we havecharacterized the corresponding trans-acting factor, BAP.

MATERIALS AND METHODS

Crude rat liver nuclear extract was prepared from maleSprague-Dawley rats as described by Gorski et al. (15) withthe following modifications: (1) Just prior to use, the proteaseinhibitors pepstatin A and leupeptin (1 pg/ml each) as well asbenzamidine and PMSF (phenylmethylsulfonyl fluoride; 0.1 mMeach) were added to all buffers used in the nuclear isolation and

* To whom correspondence should be addressed

Present addresses: +Sektion Molekularbiologie, Abt. Kinderheilkunde II, Helmholtzstr. 10, D-7900 Ulm and §Pharmakologisches Institut, UniversitiatHeidelberg, Im Neuenheimer Feld 366, D-6900 Heidelberg, FRG

kQ-D 1990 Oxford University Press

6944 Nucleic Acids Research, Vol. 18, No. 23

protein purification procedure. (2) The tissue homogenate wascentrifuged only once at 24,000 rpm for 60 min in an SW 27rotor. Protein concentration was determined according to Lowry(16).Nuclear extract from 20 rat livers (about 200 mg protein) was

applied to a Q-Sepharose FF matrix column (1.6 x 12 cm)preequilibrated with dialysis buffer (buffer D = 25 mM HEPES,40 mM NaCl, 0.1 mM EDTA, 10% glycerol, 1 mM DTT,pH 7.6) with 20 mM sodium molybdate and 0.1 % Nonidet P40.Fractions were analyzed for BAP activity by a gel shift assay.The flowthrough, containing the BAP activity, was precipitatedwith ammonium sulfate (0.3 g/ml) and applied to a heparin-Sepharose CL-6B matrix column (1.6 x 8 cm) in buffer D. Thecolumn was washed with 100 ml of buffer D and proteins elutedwith a 200 ml step gradient of NaCl (0.3, 0.6, 0.9, 1.2 M). BAPactivity eluting at 0.6 M was dialyzed against buffer Z (25 mMHEPES, 0.1 M NaCl, 12.5 mM MgCl2, 1 mM DTT, 20%glycerol and 0.1 % Nonidet P40, pH 7.8), containing 20 mMsodium molybdate and chromatographed on a DNA affinitycolumn. The DNA affinity column was prepared as describedby Kadonaga and Tjian (17), using CNBr-activated SepharoseCL-4B and multimers of a synthetic B-activator binding site,prepared by annealing and ligating the two oligonucleotides:5' AGCTTGGTGCACATGCGCCA 3' and 5' GAT-CTGGCGCATGTGCACCA 3'. Heparin-Sepharose-purifiedBAP was incubated with sonicated salmon sperm DNA (5 Ag/ml)and loaded onto the DNA affinity column equilibrated withbuffer Z. Proteins were step eluted with buffer Z containingincreasing amounts of NaCl. Active fractions were pooled andreapplied twice to the same affinity column without adding anyunspecific competitor DNA. Proteins were analyzed on 10%SDS-polyacrylamide gels according to Laemmli (18) and silverstained as described by Wray et al. (19).

Phosphatase treatment was done as described by Boxer et al.(20). Where indicated Na2MoO4 at a final concentration of20 mM was added before addition of potato acid phosphatase.

In vitro transcription reactions were performed as previouslydescribed (13, 21), using nuclear extract from rat liver. Thetranscription templates were generated by insertion ofHindHI/BamHI fragments into PL-TG (6). Bal 31 deletions wereconstructed as described (11). Oligonucleotides synthesized ona Pharmacia Gene Assembler were annealed and cloned withoutfurther purifications. All plasmid constructs were verified bysequencing of the relevant portion using the method describedby Chen et al. (22).

For band shift experiments, the probes were labeled by fillin using E. coli polymerase I (Klenow) with the appropriateradioactive deoxynucleotide triphosphates. 3-5 jig nuclearproteins from rat liver were incubated in 15 Al of 10 mM HEPES,60 mM KCI, 1 mM EDTA, 1 mM DTTF, 4% Ficoll, pH 7.9 with200 ng of sonicated salmon sperm DNA as unspecific competitor.In competition experiments specific oligonucleotides were addedprior to the addition of the extract. When affinity-purifiedfractions were analyzed, 50 ng of unspecific competitor and0.1 mg/ml bovine serum albumin were added. The mixture waspreincubated at room temperature for 15 min before 5 ,ul of 32plabeled oligonucleotides (104 cpm) were added. The incubationwas continued for another 15 min and the resulting complexeswere electrophoretically resolved at 100 V for 90 min at roomtemperature on a 4% nondenaturing polyacrylamide gel in0.25xTBE (1xTBE = 90mM Tris base, 90)mM boric acid,

2.5 mM EDTA, pH 8.3). Before being exposed to X-ray films,the gels were fixed in 10% acetic acid and dried.

For methylation interference assay on the labeled upper strandof BABS, 50 Ag of the deletion plasmid [B lwt (- 95/ - 37)-TG]was linearized with Hindlll which cuts at position -95. Thedigested DNA was treated with 24 U of calf intestinal alkalinephosphatase for 30 min at 370 C, 10 min at 750 C and isolatedby phenol/chloroform extraction and ethanol precipitation. TheDNA was then cleaved with Xho I which cuts in the polylinker20 bp downstream of position -37. For the lower strand ofBABS, 30 yg of the deletion plasmid was linearized with Eco RIwhich cuts in the polylinker 26 bp downstream of position - 37.After dephosphorylation of the DNA the fragment was cleavedwith Hindlll. The dephosphorylated DNA fragments wereisolated by gel electrophoresis. 10 pl of each fragment (about50 ng) was treated with 10 U of T4-polynucleotide kinase in a20 yl reaction containing 50 itCi of ty-32P-ATP for 30 min at370 C and then ethanol precipitated. A portion of the restrictionfragments (5 x 105-106 cpm) were treated with dimethylsulfate (DMS) for 3 min at room temperature (23). Themethylated fragments were recovered by two precipitations withethanol and used in a preparative binding reaction with 30 ygof affinity purified protein (one cycle over an affinity column),500 ng of salmon sperm DNA and I05 cpm of partiallymethylated DNA. Free and complexed DNA were recoveredfrom a polyacrylamide gel, treated with piperidine and thecleavage products analyzed on a sequencing gel (23).

For UV cross-linking a mixture as used for gel retardation assaycontaining 9 yg protein was spotted onto a piece of parafilm,put on ice and photo-cross-linked under a UV lamp (X emission= 254 nm) for 20 min at a distance of 6 cm from the UV source.After UV irradiation the probes were mixed with an equal volumeof loading buffer, boiled for 3 min and electrophoresed on a 10%SDS-polyacrylamide gel (18). The gel was fixed in 10% aceticacid, dried and exposed for autoradiography.

RESULTSBABS, a Regulatory Element of the Bi Vitellogemin GeneConferring High in vitro Transcriptional ActivityVarious 5' deletions ( -240 to -48) of the B 1 vitellogenin genewere tested in vitro for their transcriptional efficiency with ratliver nuclear extract. These fragments were inserted at theirnatural BglII site (ending at position -37 of the B1 vitellogeningene) in front of a TATA box present in the vector PL-TGpreviously used to identify regulatory elements by in vitrotranscription assays. This vector contains a G-free cassette of400 bp that allows easy detection of correctly initiated transcripts(6). In all reactions a construct containing the adeno major latepromoter linked to a shortened G-free cassette of 200 bp wasincluded as an internal control to standardize the transcriptionalactivity of various reactions. Fig. 1 a illustrates that in rat livernuclear extracts 5' deletion constructs extending to -55 yieldhigh positive transcriptional activity comparable to the adenomajor late promoter in clone pML (C2AT)19 (24). Deletion from-240 to -55 gives an overall reduction of activity by a factorof 3 whereas a further deletion of 7 bp to position -48 reducesthe activity at least lOfold, to the level observed by a constructcontaining a polylinker sequence in place of the Bi vitellogeningene insert (PL).As the sequence between -55 to -37 of the B1 vitellogenin


-i D m 0

a V.- r-e

.jX I I I- I-

<L

exp l_ i

W. ....... R:.-.

cont on_

b

a): c LoX r o LOe t

I I

Blwt

-

:

Figure 1. In vitro transcription of various BI vitellogenin gene constructs revealsa major regulatory element. a) Various 5' deletion mutants of the BI vitellogeningene promoter were cloned into the transcription vector PL-TG (6), replacingthe polylinker sequence in front of a TATA box linked to a 400 bp G-free cassette.In vitro transcription products obtained in assays with rat liver nuclear extractswere derived from the experimental plasmids (exp) and the internal control (cont)with a shortened G-free cassette under the control of the adeno major late promoter(pML cas 9, 15). For comparison we used either the adeno major late promoter(AML) linked to the G-free cassette, or the vector PL-TG (PL) lacking any

regulatory element (6). The 5' end point of each deletion mutant is indicated abovethe lanes. The 3' end point of all the inserts is the natural Bglll site at -37.The autoradiogram is an overexposure to reveal the basic transcription of thepolylinker construct (PL, contains no regulatory element) and the -48 deletions.In such an overexposure minor partial transcripts are overestimated in reactionswith very high transcriptional activity. b) Sequence of the BI vitellogenin promoterupstream of the TATA box extending to the BglII site used for cloning as publishedby Walker et al. (41). The 5' end points of BI vitellogenin gene constructs are

indicated and the palindromic sequence of the B-activator binding site is markedby arrows.

gene mediates high in vitro transcriptional activity, we analyzedthe sequence specific interactions at this position. In a firstapproach we incubated the labeled oligonucleotide extending from-55 to -37 (Blwt) with the rat liver nuclear extract used forthe in vitro transcription. In a gel retardation assay complexformation is seen (Fig. 2 a) that is competed by a 250fold excess

of the unlabeled Blwt oligonucleotide (lane 5), but is resistantto the same amount of heterologous probe (lane 6). At low inputof homologous competitor the complexes can be resolved in twospecies (see also Fig. 4 b). This suggests that formation ofcomplex 1 requires higher concentration of oligonucleotides. Thegel retardation assay establishes that the region -55 to -37 ofthe B1 vitellogenin gene is specifically recognized by nuclearproteins present in the rat liver. We refer to the recognizedsequence as the B-activator binding site (BABS). To furtherdocument sequence specific interaction at BABS we made a

methylation interference assay to identify guanines and adeninesin the DNA binding site that, when methylated, interfere with

1 2 3 4 5 6

b)

(-52) G

(-50) G*

(-44) Go

lower upper

I

G B F F B G

) G (-45)0 G (-47)

* G (-53)

-53 -47 45

-65 CCACTTCTGTGTGCACATGCGCCAGATCT TTC_34GGTGAAGACACACGTGTACGCGGTCTAGAAAG

I I

C. * 0-52 50 - 4 4

Figure 2. BABS specifically interacts with proteins from rat liver nucleia) Sequence specificity of the complexes 1 and 2. Three 1tg of rat liver nuclearprotein were incubated with 32P-labeled Blwt oligonucleotide and 100 ng ofsalmon sperm DNA (lane 1). In addition lanes 2-5 contain the indicated foldmolar excess of unlabeled homologous Blwt oligonucleotide as competitor(COMP), whereas in lane 6 a 250fold molar excess of an unrelated oligonucleotidecontaining the ERE (11) was included. The retarded complexes, 1 and 2, are

indicated. b) Methylation interference assay. Preparative binding reactions werecarried out with partially methylated DNA fragments containing BABS labeledeither on the lower, non-coding strand (lanes 1 to 3) or on the upper, codingstrand (lanes 4-6). The partially methylated DNA was cleaved either prior tothe binding reaction (G, lanes 1 and 6) or after binding recovered from the bound(B, lanes 2 and 5) and free (F, lanes 3 and 4) fraction. The result of the assayis summarized on the sequence of the 5' flanking region of the Bl vitellogeningene containing BABS. Methylated guanine residues interfering strongly ( 0) and

weakly (0) with proteins from rat liver and their location in the BABS are marked.Brackets indicate the region protected from DNase I digestion (footprint analysis,data not shown).

IIf 0 0 a I IN In 0 I0 1i

r4 eq e-

ERE COMP

EXCESS

minmi~~~~~~~.AW- md

'P..bo

i :pJI.

123456:

lo: :..

imio 10 :,:::. 21


protein binding (23). Partially methylated DNA either labeledat the lower or upper strand was incubated with rat liver nuclearproteins and separated in free and complex bound DNA by gelretardation analysis. The DNA recovered was treated withpiperidine to visualize positions of methylated purines. Fig. 2 bdemonstrates that the guanines in the upper strand at -53, -47and -45 (lanes 4 to 6) and in the lower strand at -52, -50and -44 (lanes 1 to 3) are not or less methylated in thecomplexed compared to the free DNA. This is clear evidencethat methylation of these sites interferes with binding.To further prove sequence specific recognition of BABS we

made a footprint analysis in the 5' flanking region. The dataobtained agreed with a recent report (25) and therefore are justincluded as a summary in the sequence shown in Fig. 2 b. Theprotected area includes the guanines shown to be important inthe methylation interference assay and corresponds to the A sitedescribed by Corthesy et al. (25).Based on these binding studies, we conclude that the sequence

between -55 and -37 is recognized specifically by nuclearproteins from rat liver.

Two Single Point Mutations Introduced into BABS Destroyits Functional Activity in vitroExamination of the sequence from -55 to -37 of the B1vitellogenin gene reveals a palindromic sequence from -53 to-44 (Fig. 1 b). As many transcription factors recognizepalindromic sequences (2, 26), we speculated that the sequencefrom -53 to -44 is sufficient to confer positive transcriptionalactivation. To verify this assumption, we inserted thecorresponding oligonucleotide into PL-TG, and tested itsactivating potential in vitro using nuclear extract from rat liver.The data given in Fig. 3 a show that this sequence mediates hightranscriptional activity (lane 1) identical to the construct extendingfrom -55 to -37 (Fig. 1, comparison not shown directly).To define the sequence requirements for a functional B-

activator binding site (BABS), we introduced single pointmutations, i. e. transitions in BABS, and analyzed thetranscriptional activity of these mutants in a rat liver nuclearextract. As illustrated in Fig. 3 a, two point mutations, mt 7 andmt 9, abolish B-activator function. Quantification by densitometry(Fig. 3 b) revealed that these two point mutations reduce theactivity at least Sfold, whereas mutations at other positions, i. e.mt 2, 4, 5, 6 and 8 interfere with the function at most twofold.

DNA Elements Similar to BABS are Present in AnimalVirusesComparing BABS with known regulatory sequences we observedthat BABS is similar to the USF/MLTF binding site of theadenovirus major late promoter (27, 28, 29) since the CACGTGcore sequence of the USF/MLTF site corresponds in BABS toCACATG (see Fig. 4a). Furthermore, a BABS related sequencewas also detected in the hepatitis B virus enhancer (Fig. 4a).To investigate whether these BABS related DNA sequences

interact with the same proteins, we compared the ability of theoligonucleotides listed in Fig. 4 a to compete binding to BABSin rat liver nuclear extract. Fig. 4 b illustrates a gel retardationassay using BABS present in the B1 vitellogenin gene as 32plabeled oligonucleotide (Blwt). The analysis revealed theexistence of two DNA-protein complexes, 1 and 2 (Fig. 4 b,lane 1). Both these complexes can be competed by the additionof an excess of unlabeled oligonucleotide Blwt (lane 2 and 3)or unlabeled fragment of the HBV enhancer (lane 4 and 5). Using

'4b 4 :mw 1 .1as -sa. 4*g

Figure 3. Mutational analysis of BABS demonstrates nucleotide positions essentialfor function. a) The oligonucleotide B1-0 encompassing BABS of the BIvitellogenin gene (-53 to -44) is given with the single point mutations introducedto generate the indicated mutants. For cloning purposes the oligonucleotide containsa HindlIl and BglII site at the 5' end and 3' end, respectively. The in vitrotranscription products of BI -O and the various mutant constructs (mtl to mtlO)were analyzed in a rat liver nuclear extract with the internal control pML cas 9(see Fig. 1). For comparison the transcription vector PL-TG (PL) was used. b)The transcriptional activity of the various constructs was compared to B1-0 andquantified by densitometry using the internal control pML cas 9 as standard. Thedata are the mean of at least 4 independent experiments. The broken lines indicatethe standard deviations.

limiting amounts of competitor complex 2 is preferentiallydisappearing. At high competitor input both complexes disappear(data not shown). In contrast the USF/MLTF binding site derivedfrom the adeno major late promoter selectively competescomplex 1 whereas the intensity of complex 2 is unaffected (lane6 and 7). The factor bound in complex 2 we refer to as the B-activator protein (BAP).These binding data establish that the elements found in the

Xenopus B 1 vitellogenin gene and the hepatitis B virus enhancerbind the same factors whereas the USF/MLTF binding site isable to bind a subset of these factors. These results wereconfirmed by gel retardation competition assays containing eitherthe hepatitis B virus or the adenovirus element as labeled probesin the binding assay (data not shown).To analyze whether the BABS related sequences as found in

the hepatitis B virus enhancer and the adeno major late promoter


F IBi v.t e nc-e.ni- gerne nroenoter AGCTTGTGCACATGCGCCAGATC

1t-Ar d -

henat. t is Vi rlis enlnanll,er

a -ie-o ma onr .tiee pr l-rno t e r{....

b

I vAGCTTGCGCATGCGdAC,ATC

a a Ba

AGCTTGGCCCACGACCGCTAGAGGATC

d4ind --- -K

Bl wt HBV USF- 100 250 100 250 100 250

COMPEXCESS

1 2 3 4 5 6 7

Figur 4. BABS related sequences and their binding properties. a) The nucleotidesequence of synthetic oligonucleotides are given, that were, used for gel retardation(Fig. 4 b) and in the in vitro transcription assays (Fig. 5). The HindIll and Bglllor BamHI sites were added to allow cloning in PL-TG. The numbers refer tothe position in the promoter or the enhancer. The sequences were taken fromref. 41, 35, 29 for the BlIwt, HBV and USF oligonucleotides, respectively. Theblack dots denote nucleotides, that do not agree with the sequence found in BlIwt.The core sequence in the USF/MLTF binding site is underlined. b) Nuclearproteins prepared from rat liver were incubated with 32P-labeled BlIwtoligonucleotide, separated on a polyacrylamide gel and the retardedoligonucleotides, forming complexes 1 and 2 visualized by autoradiography(lane 1). Lanes 2-7 contain the indicated unlabeled oligonucleotides as competitor.Numbers indicate the molar excess of competitor versus probe used.

can also increase the transcriptional activity, we replaced theXenopus sequence in the transcription vector with thesehomologues and analyzed their transcriptional potential. Usinga rat nuclear liver extract, we observed that both sequencesmediate increased transcriptional activity. The HBV elementincreased the activity 27fold above the level observed with thepolylinker construct, whereas the USF/MLTF binding sitecontaining template gave a lOfold increase, similar to the l3foldstimulation observed by BABS of the vitellogenin Bi gene. Tocompare these regulatory elements we added the differentoligonucleotides as competitors into the in vitro transcriptionreaction. The quantification of these competition assays revealedthat the activity of the transcription vector containing the XenopusB-activator binding site (Fig. 5) is inhibited by 90% upon theaddition of the homologous oligonucleotide as well as of theBABS homologue as found in the hepatitis B virus enhancer. Noinhibition but rather an increase by a factor of three is observed

Figure 5. Competition experiments in the in vitro transcription assay distinguishBABS from the USF/MLTF binding site. The oligonucleotides Blwt, HBV andUSF as given in Fig. 4 a were inserted into PL-TG and used in a rat liver nuclearextract as transcriptional template (TEMP). The activity measured by densitometryin assays as given in Fig. 1 was defined as 1. Parallel incubations contained 100 ng

of the indicated oligonucleotides as competitors (COMP). As internal control a

plasmid consisting of HP1 in front of a shortened G-free cassette was used (HP1casl00, ref. 47).

with the USF/MLTF binding site as competitor. Using thetemplate with the HBV element, the homologous oligonucleotideand the one containing the Xenopus element show efficientinhibition, whereas the USF/MLTF binding site is unable tocompete. Experiments with the template containing theUSF/MLTF binding site, revealed that the homologousoligonucleotide does inhibit the transcription Sfold, whereastranscriptional activity is reduced only 2fold upon adding thevitellogenin or the HBV containing oligonucleotide as competitor.

In summary, these transcription experiments confirm theconclusions drawn from the binding studies. Both approachesdemonstrate that the Xenopus Bl vitellogenin gene and thehepatitis B virus enhancer contain BABS as a common regulatoryelement, recognized by the same transcription factors and thiscis-element is distinct from the USF/MLTF binding site foundin the adenovirus major late promoter.

Phosphorylation of the B-activator Protein, BAP, isAbsolutely Required for DNA BindingIt has been reported that transcription factors may be subjectedto posttranslational modification, e. g. glycosylation or

phosphorylation (2). To analyze whether the B-activator protein(BAP) binding to BABS is glycosylated, rat liver nuclear extractwas chromatographed on a wheat germ lectin agarose column.All the BAP activity as measured by gel retardation assay was

in the flow-through fraction, whereas HNF1 known to beglycosylated (10) was efficiently retained on the column (datanot shown) as measured by HP1 binding (6). This implies thatBAP is not glycosylated.To examine the inportance of phosphorylation we treated crude

rat liver nuclear extract with potato acid phosphatase and assayedfor DNA binding activity by gel retardation assay. As Fig. 6

a

3

-W

u0

._._

Eu I

._

0

4.' > ea' > I. J. LL. > U. I

BI-wt HBV USF

COMP

TEMP


131 -w'v US'f.+ - :+ + -,+ + ;

_+ -++- +

-~.inIIIPIIIII

::-

Figure 6. Dephosphorylation destroys DNA binding activity of BAP. Rat livernuclear extract was incubated with or without potato acid phosphatase prior toDNA binding assay using the Blwt oligonucleotide (Blwt), the USF/MLTFbinding site (USF) or HP1 oligonucleotide (synO) as labeled probe. 20 mM sodiummolybdate (Na2MoO4) was added as indicated to inhibit phosphatase.

illustrates, binding activity is completely lost upon phosphatasetreatment (compare lane 1 with lane 2). This effect is specificsince in the presence of 20 mM sodium molybdate, a specificphosphatase inhibitor, binding activity is retained (lane 3).Identical results were obtained by using the oligonucleotide probefrom the hepatitis B virus enhancer (data not shown). In contrast,the binding of USF/MLTF to the adeno major late promoter isnot affected by phosphatase treatment (lanes 4 to 6). This holdsalso true for the binding of the liver-specific transcription factorHNF1 that was assayed as an additional control in the same

extract Oanes 7 to 9). We conclude that phosphorylation of BAPis absolutely required for its DNA binding activity.

Molecular Weight and Purification of the B-activator Protein(BAP)UV cross-linking was used to estimate the molecular mass ofthe B-activator protein (BAP) directly in a crude nuclear extract.In contrast to several other procedures for UV cross-linking, wefound it unnecessary to incorporate BrdU for thymidine inpreparing the probe. Rat liver nuclear extract was incubated withthe labeled oligonucleotide Blwt and irradiated with UV lightto generate covalent links between the DNA and BAP. Analysison an SDS polyacrylamide gel revealed three polypeptides withapparent molecular weights of 73 kd, 51 kd and 39 kd (Fig. 7).If a 250fold molar excess of unlabeled B-activator oligonucleotidewas added, the 73 kd band disappeared specifically, whilecompetition with the same amount of heterologousoligonucleotides had no effect. This demonstrates that the 73 kdband represents a specific interaction. This finding was confirmedby using a highly purified preparation of BAP, whose UV cross-

linking revealed only the 73 kd band (Fig. 7), whereas the 51 kdand 39 kd band corresponding to nuclear proteins interactingnonspecifically with the probe were absent. Again, the interactionwith the 73 kd protein is specific as it is competed exclusivelywith the homologous probe. As we did not observe a cross-linkedprotein that might represent USF/MLTF, we assume that underthe conditions used this factor cannot be covalently bound to itstarget sequence. By assuming that the cross-linked Blwt

Figure 7. Molecular weight of BAP as determined by UV cross-linking to BABS.A crude rat liver nuclear extract (NE) and a highly purified BAP preparation(A3, three cycles of specific DNA affinity chromatography) were incubated withlabeled Blwt oligonucleotide, UV cross-linked and separated on an SDSpolyacrylamide gel. The competitor oligonucleotides Blwt (Blwt, see Fig. 4 a),A-activator binding site (A2), corresponding to the sequence - 121 to -87 ofthe A2 vitellogenin gene (TGGTGAGGTAATTGTTTACACAACCTGATAA-CAGT in ref. 13) and the USF/MLTF binding site (USF, see Fig. 4 a) wereincluded in a 250fold molar excess of the labeled probe as indicated. The molecularweights of radioactive marker proteins (M) are given. The position of the complexinteracting specifically with Blwt oligonucleotide is indicated by an arrow.

Figure 8. BAP and USF/MLTF are separated along the purification protocol.A fraction from the heparin-Sepharose eluate (HS) and from the second DNAaffinity column (AC2) were used in gel retardation assays using the Blwtoligonucleotide and the USF/MLTF binding site as probes. For competition(COMP), a 250 fold excess of Blwt oligonucleotide (Blwt), of USF/MLTFbinding site (USF) or of the A-activator binding site (A2) was used as indicated.The sequences of the oligonucleotides can be taken from Figure 4 a (Blwt, USF)and the legend to Figure 7 (A2).

oligonucleotide contributes a molecular weight of 15 kd, weestimate that BAP has a molecular weight of 58 kd.As the B-activator binding site binds BAP as well as

USF/MLTF (see above), it was important to analyze whether

W 'i.-} - e s - - l E-~~~owmw

_. y , :..,

I


these two proteins can be separated during the purification. Gelretardation assays with the Blwt oligonucleotide as a probereveals in a highly purified BAP preparation that has beenfractionated by 3 cycles of DNA affinity columns with BABSas ligand, complex 2 exclusively (Fig. 8, lane 2). This DNA-protein interaction can specifically be competed by Blwtoligonucleotide (lane 3) but not by the USF/MLTF binding site(lane 4) or an unrelated oligonucleotide (lane 5). This suggeststhat we succeeded to separate BAP from USF/MLTF. The lossof USF/MLTF upon purification of BAP is also evident in gelretardation experiments using the USF/MLTF binding site asprobe: High specific binding is found in the heparin-Sepharosefraction (lanes 6-9) whereas no distinct complexes can be foundin the highly purified BAP preparation (lanes 10-13).

DISCUSSIONThe B-activator Binding Site, BABS, a Novel PromoterElement Increasing Transcription in vitroIn the present study, we have used an in vitro transcription systemderived from rat liver nuclei (15) using a G-free reporter cassette(24), to establish functional promoter regions of the Xenopus BIvitellogenin gene. Using this heterologous approach we haveshown that a promoter segment, the B-activator binding site(BABS), extending from position -55 to -37 of the Bi gene,confers most of the activity observed in in vitro transcription.The same region has very recently been identified as the majorpositive acting promoter element, VA element, in transfectionsof Bi vitellogenin constructs into Xenopus cells (30). Thishomologous assay proves that BABS is of functional importancein Xenopus cells in vivo. BABS can further be reduced to 10 bpthat are sufficient in the in vitro transcription assay if insertedin front of the TATA box derived from the Xenopus albumingene (6). As we have not systematically mutated the regionsurrounding these 10 bp, we cannot rigorously exclude that othersequences flanking the element are important. In accordance withour recent experiments analyzing HP1, a regulatory elementconferring liver-specific transcriptional activity (31), we assumethat BABS just requires the TATA box as an obligatory elementfor its activity.

Mutational analysis of BABS (Fig. 3) has shown that at leasttwo positions in the 10 bp element are essential for its functionin the in vitro transcription system. This is a minimum estimate,as we have introduced only transitions that might be less harmfulthan changes from pyrimidines to purines, i. e. transversions.This may also explain why some guanines found to be importantin the methylation interference assay (Fig. 2 b) can be replacedby adenines without loss of transcriptional activity. Theinterpretation of the mutational analysis is further complicatedby our finding that BABS as found in the Bi vitellogenin geneis also recognized by the transcription factor USF/MLTF.Therefore, some of the transcriptional activity mediated by BABSmay be influenced by USF/MLTF binding (see below). Thisimplies that some point mutations may lead to a change in thebalance of BAP and USF/MLTF binding. This can best beexemplified by mutant 5 which generates the perfect coresequence of the USF binding site, CACGTG.The sequence as found in BABS has palindromic properties

with 5' GC 3' on its ends and the palindromic CATG internalregion (Fig. 1). However, BABS function is not mediated by aperfect palindrome as mutations at corresponding sites of thepalindrome, behave differently: Changes as made in mutants 7

and 9 are critical for function, whereas those present in mutant 2and 4 hardly interfere with the activity.Gel retardation experiments have shown that the B-activator

binding site specifically interacts with proteins from rat livernuclei. We have confirmed these data by footprint experiments(data not shown) and these findings agree with a recent report(25) which describes three protected sites (A, B and C) in theBi vitellogenin promoter. Site A, containing BABS we haveidentified in this paper, did not mediate any functional activityin these experiments whereas the other sites play a crucial rolein in vitro transcription. Our data give direct evidence that site Aconfers increased transcriptional activity. The fact that theaddition of an oligonucleotide containing the BABS sequenceinhibits specifically the in vitro transcription implies that theinteraction with BABS linked to the promoter is an essential stepin transcriptional activation.A comparison of the BABS sequence with the recognition sites

of known transcription factors (32,33) revealed some homologybetween BABS and the USF/MLTF binding site of the adenomajor late promoter as well as a binding site in the hepatitis Bvirus enhancer. Competition experiments using the gel retardationassay (Fig. 4 b) as well as the in vitro transcription (Fig. 5),established that the hepatitis B virus (HBV) enhanceroligonucleotide binds similar if not identical transcription factorsas BABS found in the Bi vitellogenin gene. Since the preciserole of HBV in hepatocellular tumorigenesis is unclear (34), itis of particular interest to elucidate regulatory elements that mightcontrol the HBV life cycle and regulate its gene expression.Recently, Ben-Levy et al. (35) were able to show that a varietyof cellular factors interact with the HBV enhancer, a possiblecandidate for the activation of host protooncogenes, therebyinducing tumor formation. Interestingly, in the HBV enhancerthe BABS element we have identified is located close to a NF-Ibinding site (35), a comparable situation as reported for the Bivitellogenin gene (25).Our data establish that BABS is related but distinct from the

USF/MLTF binding site found in the adenovirus major latepromoter. As USF/MLTF binding sites have also been detectedin the promoters of cellular genes, e. g. the mousemetallothionein I gene (36), the rat -y fibrinogen gene (37, 38)and the Xenopus gene coding for TFIHA (39, 40) we speculatethat some of these sites might also interact with BAP. Analyzingthe promoters of the other three vitellogenin genes of Xenopus,we note that the B2 gene which is most closely related to theBi gene (41) contains the identical sequence present in BABSof the B1 gene, whereas related sequences are present in the Aland A2 genes. Recent experiments have shown that the sequence-57 to -40 of the A2 vitellogenin gene inserted into PL-TGconfers high in vitro transcription, that can specifically becompeted by a BABS oligonucleotide (unpublished data). Thissuggests that all 4 Xenopus vitellogenin genes known to becoordinately expressed contain BABS as common promoterelement.

BAP, the Factor Recognizing the B-activator Binding SiteClearly, BAP is a transcription factor distinct from USF/MLTF.This is proven by our purification data as well as by the variouscharacteristics that distinguish BAP from USF/MLTF. Thesedifferences include the molecular weight, 58 kd versus 43-46 kd(29, 42) and the need of phosphorylation for DNA binding forBAP but not for USF/MLTF. The distinction between BAP andUSF/MLTF can also be deduced from a recent analysis of the


BI vitellogenin promoter (25): The protein binding to site Acontaining BABS (see above) cannot be competed by aUSF/MLTF binding site. This observation is consistent with ourfinding that BABS binds two distinct proteins in the gel retardationassay (Fig. 4 b), only one of which can be competed byUSF/MLTF binding site competition. Thus a footprint is expectedto persist using USF/MLTF binding site as competitor.The vitellogenin genes were shown to be active only in liver

cells (3, 4), raising the possibility that the B-activator bindingsite carries the recognition sequence for a liver-specific trans-acting factor, like HPl (6), i. e. the HNFI/LFBl binding site(7, 8, 9, 10). In vitro transcription experiments revealed thatBABS mediates transcriptional activation not only in extracts fromhepatic cells but also in extracts from many other sourcesincluding HeLa, MCF-7 and Ltk- cells (data not shown). Gelretardation assays allowed us to detect the formation of variousdifferent complexes in all these extracts (data not shown).Therefore, we conclude that BABS is recognized by factors thatform a family of related proteins. This family includesUSF/MLTF known to be heterogeneous (42, 43). Furthercharacterization is needed to decide whether BAP as found inrat liver nuclear extract is a hepatocyte-specific or a ubiquitoustranscription factor.A common feature of many transcription factors is that they

exist as phosphoproteins or glycoproteins. Therefore, theirfunction could be conceivably regulated by phosphorylation-dephosphorylation or glycosylation-deglycosylation events. Here,we provide evidence that phosphorylation is absolutely necessaryfor DNA binding activity of BAP. We speculate that BAP isactivated by protein kinases that might participate in controllinggenes containing BABS in their promoter. Phosphorylationdependent DNA-binding has also been reported for thetranscription factor CREB (44) and the serum response factor(20, 45).One essential approach towards understanding the molecular

regulation of the vitellogenin genes is to purify the trans-actingfactors interacting with promoter elements of these genes. In thispaper we describe the properties of the trans-acting factor BAPwhich allow its purification. We show that BAP can efficientlybe separated from USF/MLTF mainly by using 3 cycles ofchromatography on DNA affinity columns with BABS as ligand(Fig. 8).Recognition of the Same Regulatory Element by DifferentTranscription FactorsThe B-activator binding site, BABS, is recognized by twodifferent proteins, BAP and USF/MLTF. This conclusion is basedprimarily on binding assays: Gel retardation experiments indicatethat both BAP and USF/MLTF are able to bind to BABS, therebygenerating DNA-protein complexes of different mobilities(complex 2 and 1 in Fig. 4 b). Both these complexes arecompeted by adding the BABS containing oligonucleotide, butupon adding the USF/MLTF binding site complex 1 is competedexclusively. As the mobility of the remaining complex in thecompetition experiments is not changed, we assume that BAPand USF/MLTF bind in a mutually exclusive manner.

In agreement with the binding studies, an USF/MLTF bindingsite oligonucleotide cannot compete the transcriptional activityof a template containing BABS as regulatory element. In fact,in all experiments an increase in activity was observed. This maysuggest that USF/MLTF binding leads to a less efficienttranscriptional complex than the binding of BAP. It was quite

surprising that BABS was not sufficient to compete efficientlythe transcription of a USF/MLTF binding site template even whenup to 500 ng of oligonucleotide per reaction were used. This maybe explained by the observation that USF/MLTF is heterogeneous(42, 43) and possibly only some forms are competed by BABS.This agrees with the observation that in gel retardation assaysthe USF/MLTF binding site forms complexes that cannot becompeted by the BABS of the BI vitellogenin gene (Fig. 8 lanes6 and 7).

Binding data suggest that the regulatory element found in thehepatitis B virus enhancer is equivalent to BABS as found in theBi vitellogenin gene. Obviously, it also binds USF/MLTF andBAP. However, the in vitro transcription assays show somedistinct features compared to BABS as found in the Bivitellogenin gene: The USF/MLTF binding site as competitordoes not lead to an increase in in vitro transcription of the HBVtemplate as found for the Blwt template. A further distinctionbetween the HBV element and BABS as found in the vitellogeningene is given by the fact that the HBV template is 3-fold moreactive in the in vitro transcription assay as compared to thecorresponding vitellogenin gene construct.

In conclusion, our data show that a short DNA stretch of theBI vitellogenin promoter has the potential to interact with varioustranscription factors. Similar situations have been reported forother promoter and enhancer elements (2, 46). The balance inbinding of the various factors may allow the complex regulatorypattern.

ACKNOWLEDGEMENTSWe thank R.Burkhardt for the synthesis of the oligonucleotidesand L. Klein-HitpaB for critical reading of the manuscript. Thiswork was supported by the Deutsche Forschungsgemeinschaft.

REFERENCES1. Maniatis, T., Goodboum, S., and Fischer, J.A. (1987).

Science 236, 1237-12452. Mitchell, P.J., and Tjian, R. (1989). Science 245, 371-3783. Wahli, W., and Ryffel, G.U. (1985). Oxford Sur. Eukar. Genes 2, 96- 1204. Wahli, W. (1988). TIG 4, 227-2325. Struhl, K. (1989). TIBS 14, 137-1406. Schorpp, M., Kugler, W., Wagner, U., and Ryffel, G.U. (1988). J. Mol.

Biol. 202, 307-3207. Cereghini, S., Blumenfeld, M., and Yaniv, M. (1988). Genes and

Development 2, 957-9748. Courtois, G., Baumhueter, S., and Crabtree, G.R. (1988). Proc. Natl. Acad.

Sci. 85, 7937-79419. Frain, M., Swart, G., Monaci, P., Nicosia, A., Stiipfli, S., Frank, R.,

and Cortese, R. (1989). Cell 59, 145-15710. Lichsteiner, S., and Schibler, U. (1989). Cell 57, 1179-118711. Klein-Hitpass, L., Schorpp, M., Wagner, U., and Ryffel, G.U. (1986). Cell

46, 1053-106112. Seiler-Tuyns, A., Walker, P., Martinez, E., Merillat, A.M., Givel, F., and

Wahli, W. (1986). Nucleic Acids Res. 14, 8755 -877013. D6bbeling, U., Ross, K., Klein-Hitpass, L., Morley, C., Wagner, U., and

Ryffel, G.U. (1988). EMBOJ. 7, 2495-250114. May, F.E.B., Ryffel, G.U., Weber, R., and Westley, B.R. (1982). J. Biol.

Chem. 257, 13919-1392315. Gorski, K., Carneiro, M., and Schibler, U. (1986). Cell 47, 767-77616. Lowry, O.H., Rosebrough, N.J., Farr, A.L., and Randall, R.J. (1951).

J. Biol. Chem. 193, 265-27517. Kadonaga, J.T., and Tjian, R. (1986). Proc. Natl. Acad. Sci. 83, 5889-589318. Laemmli, U.K. (1970). Nature 227, 680-68519. Wray, W., Boulikas, T., Wray, V.P., and Hancock, R. (1981). Analytical

Biochemistry 118, 197-203


20. Boxer, L.M., Prywes, R., Roeder, R.G., and Kedes, L. (1989). Mol. Cell.Biol. 9, 515-522

21. Kugler, W., Wagner, U., and Ryffel, G.U. (1988). Nucleic Acids Res. 16,3165 -3174

22. Chen, E.Y., and Seeburg, P.H. (1985). DNA 4, 165-17023. Ausubel, F.M., Brent, R., Kingston, R.E., Moore, D.D., Seidman, I.G.,

Smith, I.A., and Struhl, K. (1987). Current Protocols in Molecular Biology.John Wiley & Sons, New York

24. Sawadogo, M., and Roeder, R.G. (1985). Proc. Natl. Acad. Sci. 82,4394-4398

25. Corthesy, B., Cardinaux, J.R., Claret, F.X., and Wahli, W. (1989). Mol.Cell. Biol. 9, 5548-5562

26. Evans, R.M. (1988). Science 240, 889-89527. Carthew, R.W., Chodosh, L.A., and Sharp, P.A. (1985). Cell 43, 439-44828. Sawadogo, M., and Roeder, R.G. (1985). Cell 43, 165-17529. Chodosh, L.A., Carthew, R.W., and Sharp, P.A. (1986). Mol. Cell. Biol.

6, 4723-473330. Chang, T.-C., and Shapiro, D. J. (1990). J. Biol. Chem. 265, 8176-8182.31. Ryffel, G.U., Kugler, W., Wagner, U., and Kaling, M. (1989). Nucleic

Acids Res. 17, 939-95332. Jones, N.C., Rigby, P.W.J., and Ziff, E.B. (1988). Genes and Development

2, 267-28133. Wingender, E. (1988). Nucleic Acids Res. 16, 1879-190234. Blum, H.E., Gerok, W., and Vyas, G.N. (1989). TIG 5, 154-15835. Ben-Levy, R., Faktor, O., Berger, I., and Shaul, Y. (1989). Mol. Cell. Bio.

9, 1804-180936. Carthew, R.W., Chodosh, L.A., and Sharp, P.A. (1987). Genes &

Development 1, 973-98037. Chodosh, L.A., Carthew, R.W., Morgan, J.G., Crabtree, G.R., and Sharp,

P.A. (1987). Science 238, 684-68838. Morgan, J.G., Courtois, G., Fourel, G., Chodosh, L.A., Campbell, L.,

Evans, E., and Crabtree, G.R. (1988). Mol. Cell. Biol. 8, 2628-263739. Hall, R.K., and Taylor, W.L. (1989). Mol. Cell. Biol. 9, 5003-501140. Scotto, K.W., Kaulen, H., and Roeder, R.G. (1989). Genes & Development

3, 651-66241. Walker, P., Germond, J.E., Brown-Luedi, M., Givel, F., and Wahli, W.

(1984). Nucleic Acids Res. 12, 8611-862642. Sawadogo, M., Van Dike, M.W., Gregor, P.D., and Roeder, R.G. (1988).

J. Biol. Chem. 263, 11985-1199343. Sawadogo, M. (1988). J. Biol. Chem. 263, 11994-1200144. Yamamoto, K.K., Gonzalez, G.A., Biggs, W.H., and Montminy, M.R.

(1988). Nature 334, 494-49845. Prywes, R., Dutta, A., Cromlish, J.A., and Roeder, R.G. (1988). Proc.

Natl. Acad. Sci. USA 85, 7206-721046. Berk, A.J., and Schmidt, M.C. (1990). Genes & Development 4, 151-15547. Kaling, M., Weimar-Ehl, T., Kleinhans, M., and Ryffel, G.U. (1990). Mol.

Cell. Endocrinol. 69, 167-178

BAP, a rat liver protein that activates transcription through

Documents

Transcript of BAP, a rat liver protein that activates transcription through