THE JOURNAL OF BIOLOGICAL CHEMISTRY (0 1992 by The American Society for Biochemistry and Molecular Biology, Inc

Vol. 267, No. 17, Issue of June 15, pp. 12049-12054,1992 Printed in U.S.A.

Identification of the 12-0-Tetradecanoylphorbol-13-acetate- responsive Enhancer of the MS Gene of the Epstein-Barr Virus*

(Received for publication, January 10, 1992)

Qingyun LiuSB and William C. Summersll From the DeDartments of iMolecular BioDhysics and Biochemistry, Therapeutic Radiology, and Genetics, Yale University School of Medicine, -New Haven; Connecticut 06510

We previously located two 12-0-tetradecanoylphor- bol-13-acetate (TPA)-responsive enhancers, MSTRE- I and MSTRE-11, in the upstream sequence of the MS gene of Epstein-Barr virus (Liu, Q., and Summers, W. C. (1989) J. Virol. 63,5062-5068). The core sequence of the MSTRE-I enhancer is now determined to be between -718 and -708 of the upstream sequence of the MS gene. The activity of the enhancer is also sen- sitive to its immediate surrounding sequence on either side. A single copy of a 30-base pair (bp) fragment containing the MSTRE-I sequence was able to confer T P A responsiveness upon the MS promoter even in the absence of an AP-1 binding site. Multiple tandem cop- ies of this 30-bp fragment, regardless of their relative orientations to each other, could function synergisti- cally to enhance the MS promoter activity. At least two copies of the 30-bp fragment were required to bestow TPA induction upon the thymidine kinase gene pro- moter of herpes simplex virus type 1. The MSTRE-I sequence could also be bound by a Fos-GCN4 chimeric protein but with an affinity much lower than that between the chimeric protein and the AP-1 binding site. This MSTRE-I region has strong homology to one of the TPA-responsive elements (the ZII domain) in the upstream sequence of the EBV BZLFl gene. In addi- tion, a putative negative regulatory region or silencer was found immediately downstream of the MSTRE-I enhancer. This potential silencer region contains a 14- bp sequence that is homologous to the silencer consen- sus sequence of the BZLFl gene. Therefore, the regu- lation of the MS gene may share the same pathway with the immediate early gene BZLF1.

The regulation of the expression of the immediate-early genes of the human herpes virus, EBV’ (Epstein-Barr virus), is important in the control of the latent state of the virus in uiuo. The disruption of the latent state by the tumor promoter

* This work was supported by Grant CA39238 from the National Institutes of Health. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

§ Submitted in partial fulfillment of the requirements for a docto- rate of philosophy degree at Yale University. Present address: Dept. of Biological Chemistry and Molecular Pharmacology, Harvard Med- ical School, Boston, MA 02115.

2986. 7l To whom correspondence should be addressed. Tel.: 203.785-

’ The abbreviations used are: EBV, Epstein-Barr virus; TPA, 12- 0-tetradecanoylphorbol-13-acetate; HEPES, 4-(2-hydroxyethyl)-l- piperazineethanesulfonic acid; CAT, chloramphenicol acetyltransfer- ase; bp, base pair(s); kb, kilobase pair(s); HSV, herpes simplex virus; TBS, Tris-buffered saline; TK, thymidine kinase.

TPA (12-0-tetradecanoylphorboll3-acetate) suggests that at least part of the mechanisms involved in the maintenance of the latency involves known transcriptional control elements found in other systems. In addition to the BZLFl gene which has been shown to have a central role in the control of EBV latency, a few other EBV genes are also actively transcribed after TPA treatment. These genes may play accessory roles in disrupting viral latency. One of these genes which we have studied is designated the MS gene and encodes the BMLFl open reading frame for a polypeptide with a molecular mass of 51 kDa (1). The MS gene is expressed shortly after TPA induction even though its transcription is sensitive to potent inhibitors of de m u o protein synthesis (2-4). Two major protein species associated with the MS gene, with apparent molecular masses of 50 and 60 kDa each, have been detected in the nuclei of induced cells (5, 6). The heterogeneity of the BMLFl product is presumed to be the result of phosphoryl- ation and other post-translational modifications (5).

The function of the MS gene during lytic infection remains unclear, however. Early studies suggested the MS product was a “promiscuous” transactivator since it was able to increase the apparent activities of a wide variety of promoters when the bacterial chloramphenicol acetyltransferase (CAT) was used to monitor the assays (7-11). However, additional ex- periments found that those promiscuous transactivations were not the result of transcriptional regulation and seemed to be reporter gene-dependent (12). More evidence suggests that the MS protein may act by stabilizing mRNAs and increasing their translation (13). Since the MS gene product may play an unusual regulatory role in gene expression, the mechanism of its transcriptional control is of special interest.

The transcription of the MS gene is controlled by at least two promoters, PM and PM1 (13). The PM promoter directs the synthesis of two abundant mRNA species with lengths of 1.9 and 2 kb. The PMl promoter, which lies about 2 kb upstream of the PM promoter, is responsible for the expres- sion of at least three mRNA species, 3.6, 4.0, and 4.4 kb. The transcription of the MS gene can be increased by the product of the BZLFl gene as well as TPA. Various transcriptional controlling elements have been discovered in the upstream sequence of the PM promoter (13-17). We previously reported that the upstream sequence between -726 and -540 of the PM promoter contained additional TPA-responsive en- hancers, which could be further subdivided into two domains: MSTRE-I and MSTRE-I1 (15). The two domains are able to function independently of the downstream AP-1 binding site. No obvious AP-1 homologous sequences can be found within the MSTREs. Furthermore, the MSTREs are also able to confer TPA responsiveness upon heterogeneous promoters, such as the thymidine kinase (TK) promoter of the human herpes simplex virus type 1 (HSV-1).

Since the regulation of the MS gene appears to be relevant

12049

12050 TPA Induction of EBV

to the control of EBV latency, a detailed understanding of this regulation may provide important insights into the path- ogenesis of EBV-related diseases. In this communication we have used deletion analysis to define a short EBV DNA sequence from MSTRE-I that is required for efficient tran- scription of the MS gene and is also required for full respon- siveness to TPA treatment. This sequence can function in conjunction with a constitutive heterogeneous promoter to both enhance gene expression and confer TPA inducibility on this promoter. Thus, this regulatory region from the EBV- MS gene has both enhancer and TPA response attributes.

EXPERIMENTAL PROCEDURES

Cell Lines and Transient Transfection-HepG2 cells, a line derived from a human hepatoma, were obtained from M. Karin (University o f California, San Diego) and grown in Dulbecco's modified Eagle's medium with 10% fetal calf serum. D98/HR-1 cells, a hybrid line between the epithelioid D98 line and the lymphoblastoid P"HR-1 line, were obtained from G. Miller (Yale University) and grown in minimum essential medium with 5% fetal calf serum. Both cell lines were transfected by the DEAE-dextran method (18). One day before transfection, cells were subcultured into 100-mm diameter tissue culture dishes. The cells appeared to be approximately 30% confluent a t the time of being exposed to DNA. TBS (0.6 ml) (Tris-buffered saline: 25 mM Tris-HC1, pH 7.4, 137 mM NaC1, 5 mM KCl, 0.6 mM Na,HPO,, 0.7 mM CaCII, 0.5 mM MgCl,) containing 10 pg of plasmid DNA was mixed with an equal volume of DEAE-dextran in TBS solution, resulting in the final concentration of 500 pg/ml DEAE- dextran. This 1.2-ml transfection mixture was equally divided into two dishes and incubated for 30 min a t room temperature. The cells were then washed twice with TBS and fed with complete medium a t :17 "C.

Plasmid Construction-DNA manipulations were performed as described before (19). The 1.6-kb BstEII-BamHI fragment in the JIHN-CAT plasmid (15), which contained the MS promoter and the complete CAT gene, was isolated and inserted into the SmaI site of the plasmid pGEMblue (Promega) by blunt-end ligation, resulting in the plasmid pGBMSC. Then pGBMSC was linearized by SphI- HnmHI double digestion (both sites are from the polylinker region of pGEMblue), followed by digestion with ExoIII to remove the AP-1 binding sequence from the MS promoter and religation, resulting in the plasmid pMSC. Sequencing the plasmid pMSC revealed that the sequence from the BstEII site (at -120) to -72 was deleted. pMSC was cut with HindIII and blunt-end-ligated to the 180-bp fragment containing the MS upstream sequence from -726 to -546, creating three plasmids pL-l80/MSC, pR-lSO/MSC, and p3x180/MSC, with one copy in the leftward orientation, one copy in the rightward orientation, and three copies all in the rightward orientation, respec- t ively.

Cloning of MSTRE-I into the upstream sequence of pTKCAT was conducted as follows. The BamHI fragment containing the TKCAT sequence was isolated from the plasmid pMSTKCAT (15) and was then inserted into the BamHI site of pUC13, creating the plasmid pTKCAT. A 31-bp fragment containing the MS sequence from -726 to -703 (MSTRE-I) was preligated with T4 DNA ligase to generate multimers, and then inserted into the filled-in HindIII site of this pTKCAT plasmid by blunt-end ligation, generating three plasmids pll-TKCAT, pl2-TKCAT, and p21-TKCAT (Table 11). The cloning of MSTRE-I into pMSC was performed in a similar way. Multimers of that 31-bp sequence containing MSTRE-I were inserted into the filled-in HindIII site of the plasmid pMSC by blunt-end ligation, generating four plasmids: pl-MSC, p21-MSC, p22-MSC, and p31- MSC (Fig. 3 ) . The copy number and orientation of the MSTRE-I sequence(s) within all the plasmids were then determined by DNA sequencing.

TPA Treatment and CAT Assay-After culturing the transfected cells in complete medium for 4 h, TPA (250 pg/ml in Me,SO) was added to the medium to a final concentrat,ion of 100 ng/ml. Control cells received the same amount of Me2S0. Approximately 40 h later, cells were harvested and subsequently lysed in 0.1 ml of 250 mM Tris- (:1 (pH 7.8) by freezing (in dry ice) and thawing (at 37 "C) for four cycles. The lysate was heated for 10 min a t 60 "C to inactivate endogenous acetylases and was cleared by centrifugation at 12,000 x ,q for 5 min a t room temperature. The resulting supernatant was assayed for chloramphenicol acetyltransferase activity as previously

described (20). Transfection efficiency varied slightly from one ex- periment to another, but the relative results within one single exper- iment were reproducible. The standard assay condition was incuba- tion for 2 h at 37 "C in a total volume of 100 p1 with 0.5 pCi of ["C] chloramphenicol and 1 mM acetyl-coenzyme A and was within the linear range of the assay. Another 20 pl of acetyl-coenzyme A was added after each additional hour of incubation. The acetylation of chloramphenicol was determined by thin-layer chromatography fol- lowed by autoradiography and scintillation counting. Most of the transient transfection assays were repeated a t least twice, and the mean CAT activity was calculated. Even with prolonged assay time and repeated experiments, the ratios of induced uersus uninduced CAT activity could be quite variable from time to time for some plasmids. This phenomenon resulted from the fact that the uninduced CAT activity was always very low so that slight variations of the uninduced activity led to great changes in the values of the induction ratio. Therefore, the value of fold-induction can only be regarded as an qualitative indication of inducibility. In recent experiments an alternative method of measuring CAT activity was employed (21). This method used ["C]CoASAc. The results from the two different methods were comparable.

Gel Mobility Shift Assay-An Escherichia coli protein extract con- taining a Fos-GCN4 hybrid protein was obtained from N. Taylor (Yale University). The binding reactions (20 pl) were in a mixture of 75 mM KC1, 15 mM HEPES, pH 7.9, 1 mM dithiothreitol, 0.2 mM EDTA, 7.5% glycerol, -5 pg of the crude extract proteins, 1 pg of poly(dI.dC), and 20,000 cpm (about 0.2 ng) of DNA probe which was labeled by [a-:"P]dATP with the Klenow fragment of E. coli DNA polymerase I. For competition assays, 100-fold excess of unlabeled probe was added 5 min before the labeled probe was added. The binding mixture was incubated at 0 "C for 30 min and subject to electrophoresis a t 8 "C in a 4% polyacrylamide gel with 0.5 X TBE (45 mM Tris, 45 mM boric acid, 0.5 mM EDTA, pH 8.3). The gel was then dried under vacuum and autoradiographed a t -70 "C with an intensifying screen.

RESULTS AND DISCUSSION

Our previous experiments demonstrated that the 180-bp sequence from -726 to -546 in the upstream region of the EBV MS promoter displayed TPA-responsive activity. To define the minimum sequence for this TPA responsiveness, we cloned this 180-bp fragment into vector pMSC, which had a truncated 81-bp MS promoter (from -71 to +lo) linked to the reporter gene CAT (Table I). The backbone of the plasmid pMSC was derived from the pUC19 plasmid, which does not contain any known TPA responsive elements. A single copy of the 180-bp MS TPA-responsive sequence was inserted immediately upstream of the MS promoter at the HindIII site in either orientation to create plasmids pL-l80/MSC and pR- 18O/MSC, and a three-copy tandem repeat of this sequence exists in the plasmid p3x180/MSC. These plasmids were

TABLE I TPA remonse of the DIBO-MSC plasmids

Transfected cell line and plasmid

HepG2 pMSC pL-lSO/MSC

P3x180/MSC pR-l80/MSC

D98/HR-1 pMSC pL-lSO/MSC pR-lBO/MSC u3x180/MSC

Relative CAT activities'

-TPA +TPA

ND' ND 2 23 6 (4, 8) 100

58 1170

2 (1, 2) 2 (2, 1) l ( 1 , 1) 12 (10, 14) 3 (2,3) 100 6 (8. 4) 435 (436,434)

Inductionh

12 16 20

1 12 33 73

CAT level for the plasmid pR-180-MSC. The two original values are " Expressed as the mean ( n = 2) of the percentage of the induced

given in parentheses. A single value indicates that it was only meas- ured once.

Expressed as the ratio of the mean of the TPA-treated levels to that of untreated CAT levels.

' Non-detectable (<l).

TPA Induction of EBV 12051

transiently transfected by the DEAE-dextran method (15,18) into D98/HR-1 cells and HepG2 cells, and the levels of CAT activities expressed from these plasmids are presented in Table I. In HepG2 cells, the CAT activity expressed by pMSC which lacked the EBV enhancer sequence was too low to be determined accurately under either induced or uninduced conditions. In contrast, the three plasmids with the 180-bp enhancer fragment all displayed TPA-responsive CAT activi- ties. Furthermore, the induced CAT activity from the plasmid p3x180/MSC was 10 times more than that from the plasmid pR-l80/MSC, suggesting this EBV enhancer sequence is able to function synergistically. The results in D98/HR-l cells were similar except that the synergism of multiple copies of ths sequence is not so striking as in HepG2 cells (Table I) . Since pR-lSO/MSC better simulates the physiological genetic structure of EBV than pL-lSO/MSC, it was used for additional dissection of the MSTRE-I region.

A series of fine deletion mutants from pR-lSO/MSC was constructed as described in the legend to Fig. 1. These plas- mids were tested in both D98/HR-1 and HepG2 cells. Only the results from D98/HR-1 cells are presented since the data in HepG2 cells are qualitatively the same (22).

Since our previous experiments have determined that MSTRE-I is located between -726 and -690, we first tested a group of mutants (p4-5, p4-7, p6-31, p6-32) in which the MSTRE-I sequence was progressively deleted in the 5’ -3’ direction. Deletion of one base pair (p4-5) did not affect the expression pattern of the enhancer (Fig. 1). However, remov- ing the sequence between -726 and -718 (Fig. 1, p4-7) reduced the induced CAT activity by 75% and the basal activity by 80%. TPA responsiveness, on the other hand, was still obviously exhibited by this plasmid. Additional deletions (p6-31, p6-32, and other mutants not shown here) made the plasmids unable to express more than 1% of the induced wild type activity, which was too low to be accurately determined under the experimental conditions. These data indicate that

the 5’ boundary of MSTRE-I is located between -725 and

The second group of plasmids (pl-4,p2-3, pl-7, p2-8, pl- 8, and pl-25) were selected to determine the 3’ boundary of the enhancer since their MSTRE-I regions have identical downstream sequences except pl-8 and pl-25. Their CAT activities and TPA responsiveness are presented in Fig. 1. Deleting the sequence between -703 and -698 (pl-4) retained the TPA responsiveness and full level CAT activities. Remov- ing three more base pairs (p2-3) decreased the induced CAT activity by 33% without changing the magnitude of TPA induction. Additional deletions (pl-7, p2-8) made the plas- mids express only about 10% of the wild type activity under induced conditions. The magnitude of TPA induction also diminished with the mutant pl-7. These results suggested that the 3‘ border of MSTRE-I is located between -706 and -713.

In these deletion experiments we used a vector that did not contain any known TPA-responsive elements, and the results demonstrated that the sequence between -725 and -708 was required for the complete function of the enhancer while the core sequence of MSTRE-I consists of the somewhat smaller 12-bp region from -718 to -706. Even though deletion of the sequence between -725 and -718 maintained the TPA re- sponsiveness, the induced CAT activity was only about one- third of the wild type level. Therefore, the function of MSTRE-I is very sensitive to its surrounding sequence, which has also shown to be true with other enhancers (reviewed in Ref. 23). The TPA responsiveness exhibited by the two dele- tion mutants pl-7 and p2-8 that lacked MSTRE-I may be accounted for by the presence of MSTRE-I1 located about 90 bp downstream. In addition, the lack of a detectable activity shown by the two mutants p6-30 and p6-32 was likely to be the results of the low transfection efficiency observed when those plasmids were analyzed. An alternative hypothesis is that the MSTRE-I region can be further divided into two

-718.

Plasmid

WT

p4-5

p4-7

p6-30

p6-32

p1-4

p2-3

p l - 7

p2-8

p1-8

p l - 2 5

p l - 1 6

p2-6

p1-10

Remaining DNA Sequence

-726 -716 -706 -696

CC CGTCACAAGA TCAAAGAGAT TGGGGCCCCG CTTTTTGACA G GG GCAGTGTTCT AGTTTCTCTA ACCCCGGGGC GAAAAACTGT C

AG GTCACAAGA TCAAAGAGAT TGGGGCCCCG CTTTTTGACA G

AG GA TCAAAGAGAT TGGGGCCCCG CTTTTTGACA G

AG CAAAGAGAT TGGGGCCCCG CTTTTTGACA G

AG GAGAT TGGGGCCCCG CTTTTTGACA G

CC CGTCACAAGA TCAAAGAGAT TGG CG CTTTTTGACA G

CC CGTCACAAGA TCAAAGAGAT CG CTTTTTGACA G

CC CGTCACAAGA TCA CG CTTTTTGACA G

CC CGTCA CG CTTTTTGACA G

CC CGTCACAAGA TCAAAGAG CA G

cc c TTTTTGACA G

CC CGTCACAAGA TCAAAGAGAT TGG CCG CTTTTTGACA G

CC CGTCACAAGA TCAAAGAGAT TGG TTGACA G

CC CGTCACAAGA TCAAAGAGAT TGG G

Relative CAT activities Induction

-TPA

5 13, 61

8 18, 8 1

1 (1, 1 1

N . D .

N . D .

4 ( 5 , 31

3 ( 3 , 2 )

2 (1, 2 )

N.D.

6 ( 6 , 5 )

N . D .

3 12, 4)

14 ( 1 9 , 8 )

1 2 115, 91

tTPA

100 20

121 1112, 1421 16

3 5 (45, 241 35

N . D .

N . D .

1 3 4 1 1 5 7 , 112) 34

67 172 , 611 22

11 1 1 4 , I) 6

8 16, 91

140 (141, 138) 2 3

N . D .

112 1100, 124) 31

2 3 9 1245. 2 3 3 ) 17

368 ( 2 4 0 , 491) 31

FIG. 1. TPA responsiveness of the fine deletion mutants. The 43-bp double-stranded DNA sequence is shown for the wild type plasmid ( W T ) . The sequence is numbered relative to the transcription initiation site. For all the deletion mutants, their remaining DNA sequences were continuous with the space being used to indicate the missing sequence. CAT activities are relative to the induced activity ( 100%) of the wild type plasmid. The two original values are given in the parentheses. Induction is the ratio of induced to uninduced level. N.D., non-detectable ( 4 ) . For the construction of fine deletion mutants, pR-l80/MSC was completely digested with SphI and BamHI in the polylinker regions flanking the 180-bp fragment) and treated with EroIII, which only acts at the BamHI site, for different time intervals. The digested DNA was trimmed with nuclease S1 and closed with T4 DNA ligase by blunt-end ligation. Deletion mutants were selected by choosing those with the appropriately shortened PuuII fragments. Internal deletion mutants were made by Bal31 digestion of pR-l80/MSC that was linearized at the ApaI site located a t -703 of the MS sequence, and then closed by blunt-end ligation. Each mutant was sequenced to determine precisely the deleted sequence.

120.52 TPA Induction of ERV

domains that are closely located and function cooperatively Lvhile MSTRE-I1 was not functional in the pl80-MSC struc- ture. Xevertheless, the sequence between -718 and -706 fhrms the most critical domain for the TPA induction and full promoter function and thus is called MSTRE-I.

The last group of mutants (pl-10, p2-6, pl-16) were dis- covered in screening for the second group of deletion mutants. These mutants were initially included as positive control plasmids for the second group in transient transfections since they all have intact MSTRE-I but different downstream flanking sequence. The mutant pl-16 performed as the wild type. However, two additional deletion mutant,s (p2-6 and 111-10) were not only fully T P A responsive but also displayed significantly higher CAT activities t,han the wild t-ype, or any other deletion mutants, under both the induced and unin- duced conditions. This observation tentatively suggests that the sequence following base pair -699 has a negative regula- tory effect on MSTRE-I. Similar negative regulatory regions in other genes have been termed silencers since they have opposite effects of enhancers. The BZLFl gene contains four copies of the sequence (2%) which has both positive and negative effects (24). Interestingly, the sequence immediately downstream of base pair 698 has striking homology with the ZIs of the BZLFl gene (Fig. 44). A common feature of the ZIs and the putative MS silencers is their AT-rich sequence composition. Such AT-rich silencers have also been found in t h e regulatory regions of several cellular genes (24, 25). In this way, EBV may have exploited a general cellular regula- tory system to maintain its own latency.

So far, the deletional analysis had been conducted from the plasmid pR-l80/MSC which contained not only MSTRE-I but also MSTRE-I1 located downstream. To test whether the MSTRE-I sequence alone is able to bestow TPA inducibility upon target promoters, we inserted the 25-bp fragment from -726 to -701, which harbored the 19-bp MSTRE-I into the vector pMSC in different copy numbers and orientations, as shown at the top of Fig. 2. The target MS promoter consisted of the sequence from -71 to f l l , excluding its original upstream AP-1 site. Six plasmids were transfected into D98/ HR-1 cells and HepG2 cells. The results are consist.ent in both cell lines, and one representative experiment in D98/ H R - 1 cells is shown in Fig. 2. pMSC showed no detectable amount of CAT activity under either induced or uninduced conditions. The plasmid pl-MSC displayed TPA induction

MSTRE-I 4 +4 +*4 4-4-

Plasmid pl-HSC pHSC pZ1-HSC p3-HSC pZZ-HSC ~TK-CRT

n n n n n n CATAct. 0.2. 1.1 m, m 0.2. 6 . 6 1.9, 5 1 0 . 4 . 11 2 . 5 , 3.9

FIG. 2. Autoradiograph of CAT assays measuring the TPA responsiveness of "STRE-I with the MS promoter in D98/ I-IK-1 cells. Each orrow at the top of' the f'i:wre represents one \ l S ' ~ l ~ l ~ - l fragment as well as its orientation relative to the M S 1)rolnoter. I'lasmicls p\ISC and pTK-CAT do not contain MSTIIE-I. ( ' : \ ' / 'Act . . C A T activity expressed in percentage of acetylation of the t o t a l substrate as determined hy scintillation counting. For all the l)lasnlids, the C A T activity value on the Irft is without T P A treatment while the value on the right is with TPA treatment. ,VI). non- tlctcct:\hle (<().I ).

though p22-MSC is more active. Furthermore, the induced CAT activity of the dual-MSTRE-I plasmids is at least 6 times that of the single enhancer plasmid. The plasmid p3- MSC has three copies of MSTRE-I and expressed much higher activity both at the basal level and at the induced level. Its induced activity is almost 50 times that of pl-MSC. As a control, the plasmid pTK-CAT showed little TPA induction. These data strongly support the hypothesis that MSTRE-I is a TPA-responsive enhancer capable of functioning synergis- tically when present in multiple tandem arrangements.

All the experiments above support the h-ypothesis that MSTRE-I requires the sequence between -725 and -706. T h e deletion mutants that lacked MSTRE-I showed either no detectable basal act.ivity or much weaker but still meas- urable TPA responsiveness. Thus, in these constructions it was difficult to separate the enhancing effects from the TPA- responding effects. Unless one thinks this point is only se- mantic, we believe it is logical to conclude that the MSTRE- I element, while it is a general enhancer, also functions to confer TPA- responsiveness beyond that observed in its ab- sence. This is because the inducibility ratio incrpases with increasing enhancement. of transcription rather than remain- ing constant.

To further test the TPA-inducing activity of MSTRE-I, we placed this element into the upstream sequence of the HSV- 1 T K promoter. The HSV-1 TK promot.er alone is not signif- icantly TPA-responsive and, most importantly for this study, is capable of expressing CAT activities which can be deter- mined accurately (Fig. 2, pTK-CAT lanes). The construction of pTKCAT was described above. A 31-bp fragment contain- ing the 23-bp sequence between -726 and -703 was inserted, either singly or in two copies, upstream of the TK promoter . The orientation and copy number of the MSTRE-I fragment in the four tested plasmids are presented in Table 11. These plasmids were transfected into D98/HR-1 and HepG2 cells, and the results of the CAT assays are presented in Table IT. In D98/HR-1 cells, pTKCAT expressed easily det.ectable CAT activity and showed an average 2.5-fold TPA induction. The two plasmids with a single copy of MSTRE-I, p11-TKCAT and pl2-TKCAT, did not show more TPA responsiveness than pTKCAT alone. p12-TKCAT, which had MSTRE-I in the physiological orientation, had increased CAT activities both at the basal level and induced level but no improved TPA induction. In contrast., p21-TKCAT, which contained two copies of MSTRE-I, displayed an average 4.7-fold T P A induction and higher CAT activities under both induced and uninduced conditions. In HepG2 cells, none of the three plasmids, pTK-CAT, pll-TKCAT, or pl2-TKCAT, showed T P A responsiveness. However, p21-TKCAT showed an av- erage %fold TPA induction and increased basal activity. Therefore, the data indicate that. at least two copies of MSTRE-I seem to be required to bestow TPA responsiveness upon HSV-1TK promoter. While we cannot rule out that the inducibility results from the tail-to-tail palindromic construc- tion of plasmid, this possiblity seems unlikely since, in gen- eral, a palindrome is insufficient to create a new TPA respon- sive site.

Even though the MSTRE-I region does not. display obvious

homology to the AP-1 binding site (TGACTCA), it is still

possible that the two distinct sequences may interact with the same proteins. The AP-1 binding site is strongly bound by the Fos-Jun protein complex and by the yeast transcriptional regulatory factor GCN4. A Fos-GCN4-hybrid protein ex- pressed in E. coli also specifically binds to the AI'-1 binding site (26). We used such an extract to test whether the Fos-

G

TPA Induction of ERV 12053

' I ' h e insert includes the sequence from -726 to 703. In the EHV M S gene. the sequence runs in the orientation from -526 to -703. Ilspressed a s the mean ( r 7 = 2 ) of'the percentage of the induced CAI' level for the plasmid pTKCAT. The two original values are given

Ihpressed a s the ratio o f the mean o f the TI'A-induced level to that of the untreated CAT level. i n parentheses. A single value means the experiment \vas only done once.

A. 1 2

- e

4 Bound

R 3 4 5 6

Bound

C Free

I>IG. 3. A u t o r a d i o g r a p h of gel mobi l i ty shift assays for the i n t e r a c t i o n b e t w e e n "STRE-I and Fos-GCS.1. Notrnd indicates 1)rotein-DSA complex. tvhile I.'rcc means unhound probe. 1,nnc.s I , 3 , i ~ n d .I contain the A P l prohe: Inr7c3s 2. 5. and 6 contain the MSTRE- I pro1)e. I,nr7cs .I and 6 also had 100-fold excess of unlabeled AI)-I 1)rnl)e. In Inr7c.s :I and 4. the unbound AI'-1 prohe migrated off the gel.

GCN4 hybrid protein was able to bind specifically to MSTRE- I by employing the gel mobility shift assay. The DNA probe representing MSTRE-I was a 59-bp fragment while the probe for the AP-1 binding site was a 20-bp double-stranded oligo- nucleotide. An E. coli protein extract wit,hout the Fos-GCN4 polypeptide did not change the mobility of the AP-1 probe (26). As shown in Fig. 3A, both the AP-1 and the MSTRE-I probes could be bound by some factors in the extract (Fig. :<A, lanes 1 and 2 ) . The two DNA-protein complexes migrated with almost the same mobility even after prolonged electro- phoresis (Fig. 3R, lanes 3 and 5 ) though the AP-1 probe was only about one third the size of the MSTRE-I probe. In addition, the amount of bound MSTRE-I probe is a t least 10

A .

c o n s e n s u s Y 7 A T T T T A G A C A C T

2.A " " " - T " " - C

z;a ZIC " C " " " - A " -

ZID G T G T G

R I

MS-SLR

"""""""

- - - - - - - - -

- A - A - - G - T T C " -

- G C - - - - T - " - G C

0 .

c- jun T G A C A T C A

pz 7.11 C C A T G A C A T C A C A G A G G A

FSTRE-I T - - C A - G - " - A - " - A T

FIC. 4. Homologies between the MSTRE-I r e g i o n a n d t h e ZI/ZII e l e m e n t s (13). A , alignment of the potential MS silencer (M.S-.SI,II) sequence (from -698 to -684) with the ZI element. H. homologies among the TPA-responsive elements of the c-jun gene, the F:HY ZII region, and the EHV MS gene. I h h e s represent identical nucleotides.

times less than that of the AP-1 probe complex in the presence of the same quantity of Fos-GCN4 (Fig. 3, lane 1 uersus lane 2, lane 3 versus lane 5 ) . Furthermore, both complexes were competed by a 100-fold excess of unlabeled AP-1 probe (Fig. 3R). However, neither complex was competed by 100-fold excess of unlabeled MSTRE-I probe.' Since there are no homologies with the AP-1 sequence inside the flanking and joining regions of the MSTRE-I probe, these observations suggest that the Fos-GCN4 protein can bind to the MSTRE- I sequence but with an affinity significantly lower than that for the AP-1 binding site.

The immediate-early gene RZLFl has been shown to con- tain multiple TPA-responsive elements (24, 27). The primary TPA-responsive region defined by Flemington and colleagues (24, 27) is designated ZII and consists of the sequence TGAGCTCA. Interestingly, the MSTRE-I region had striking homology to the ZII domain plus its flanking sequence. The sequences of these two TPA-responsive elements plus the AP- 1 binding site of the c-jun gene are aligned in Fig. 4R. ZII has a perfect 8-bp homology to the AP-1 binding site of the c-jun gene. On the other hand, the 15-bp sequence between -723 and -708 of the MSTRE-I region shares a 74% homology with the ZII domain (Fig. 4H). Even though the jun-fos complex can bind to the ZII element in uitro, it was not able to transactivate the ZII element in lymphoid cells (24). MSTRE-I may also be weakly bound by the jun-fos complex, as suggested from the discussion above. Given the homology between MSTRE-I and the ZII element, these two enhancers

' 1,iu. Q., and Summers, W. C., unpublished results.

12054 TPA Induction of EB V

may be bound by the same or similar transcriptional factors of the AP-1 family in lymphoid cells.

In summary, the data presented here together with those of Flemington et al. (24,27) suggest that the BZLFl gene and the MS gene may be regulated by the same pathway. Like the promoter of the BZLFl gene, the promoter of MS gene may also be repressed by a silencer in its upstream sequence. Treatment with TPA or other inducing agents activates the TPA-responsive elements and consequently overcomes the negative effect of the silencer sequence. Since the BZLFl product is also a transcriptional activator of its own promoter, the BRLFl promoter, and the MS promoter (24), the expres- sion of the BZLFl gene may further increase its own expres- sion and activate the transcription of the BRLFl gene and the MS gene. Then, the products of these three genes may activate other early and delayed-early genes, triggering the cascade of gene expression of the lytic infection.

with the cell culture work. We thank Naomi Taylor for the gift of the Achnowledgments-We are grateful to Wilma Summers for help

b,’. coli extract containing the Fos-GCN4 hybrid protein. We also thank Siu-long Yao for valuable discussions and for providing some of‘the competent E. coli cells.

REFERENCES

1. Baer, R., Bankier, A. T., Biggin, M. D., Deininger, P. L., Farrel, P. J., Gibson, T. J., Hatfull, G., Hudson, G. S., Satchwell, S. C., Seguin, S., Tuffnell, P. S., and Barrell, B. (1984) Nature

2. Biggin, M., Bodescot, M., Perricaudet, M., and Farrell, P. (1987)

3 . Cho, M. S., Jeang, K., and Hayward, S. D. (1985) J. Virol. 56,

4. Sample, J., Lancz, G., and Nonoyama, M. (1986) J. Virol. 57,

5 . Marschall, M., Leser, U., Seibl, R., and Wolf, H. (1989) J. Virol.

310, 207-211

J . Virol. 61, 3120-3132

852-859

145-154

63,938-942

6. Wong, K., and Levine, A. J. (1989) Virology 168, 101-111 7. Chevallier-Greco, A,, Manet, E., Chavrier, P., Mosnier, C., Daillie,

J., and Sergeant, A. (1986) EMBO J . 5, 3243-3249 8. Lieberman, P., O’Hare, P., Hayward, G. S., and Hayward, S. D.

(1986) J. Virol. 60, 140-148 9. Oguro, M. O., Shimizu, N., Ono, Y., and Takada, K. (1987) J.

Virol. 61, 3310-3313 10. Sample, J., Tanaka, A., Lancz, G., and Nonoyama, M. (1984)

Virology 139, 1-10 11. Wong, K., and Levine, A. J . (1986) J. Virol. 60, 149-156 12. Kenney, S., Kamine, J., Holley-Guthrie, E., Mar, E. C., Lin, J.

3877 C., Markovitz, D., and Pagano J. (1989) J. Virol. 63, 3870-

13. Buisson, M., Manet, E., Trescol-Biemont, M., Gruffat, H., Dur- and, B., and Sergeant, A. (1989) J. Virol. 63, 5276-5284

14. Kenny, S., Holley-Guthrie, E., Mar, E. C., and Smith, M. (1989) J. Virol. 63, 3878-3883

15. Liu, Q., and Summers, W. C. (1989) J. Virol. 63, 5062-5068 16. Rooney, C. M., Rowe, D. T., Ragot, T., and Farrel, P. J . (1989)

17. Urier, G., Buisson, M., Chambard, P., and Sergeant, A. (1989)

18. Banerji, J., Olson, L., and Schaffner, W . (1983) Cell 33, 729-740 19. Maniatis, T., Fritsch, E., and Sambrook, J. (1982) Molecular

Cloning: A Laboratory Manual, pp. 363-401, Cold Spring Har- bor Laboratory, Cold Spring Harbor, NY

20. Gorman. C. M.. Moffat. L. F.. and Howard, B. H. (1982) Mol.

J . Virol. 63, 3109-3116

EMRO J. 8, 1447-1453

Cell. Biol. 2, 1044-1051 21. Sambrook. J.. Fritsch. E.. and Maniatis, T. (1990) Molecular

Cloning’A Laboratory Manual, 2nd Ed., pp. 16.64-16.65, Cold Spring Harbor Laboratory, Cold Spring Harbor, NY

22. Liu, Q. (1990) Novel ~2-0-tetrudecunoylphorbol-I3-ncetate~TPAI~ Responsive Enhancers in the Upstream Sequence of the M S Gene Promoter of the Epstein-Barr Virus. Ph.D. thesis, Yale University

23. Dynan, W. S. (1989) Cell 58, 1-4 24. Flemington, E., and Speck, S. H. (1990) J . Virol. 64, 1217-1226 25. Arizumi, K., Takahashi, H., Nakamura, N., and Ariga, H. (1989)

26. Taylor, N., Flemington, E., Kolman, J. L., Baumann, R. P.,

27. Flemington, E., and Speck, S. H. (1990) J. Virol. 64, 1227-1232

Mol. Cell. Biol. 9, 4032-4037

Speck, S. H., and Miller, G. (1991) J. Virol. 65, 4033-4041