Evaluation and mapping of the DNA binding and oligomerization domains of the IE2 regulatory protein of human cytomegalovirus using yeast one and two hybrid interaction assays

Gene 210 (1998) 25–36

Evaluation and mapping of the DNA binding and oligomerization domains of the IE2 regulatory protein of human cytomegalovirus using yeast one and two hybrid interaction assays Jin-Hyun Ahn a, Chuang-Jiun Chiou b, Gary S. Hayward a,b,* a The Molecular Virology Laboratories, Department of Pharmacology and Molecular Sciences, Johns Hopkins University School of Medicine, 725 N. Wolfe Street, WBSB 317, Baltimore, MD 21205, USA b The Molecular Virology Laboratories, Department of Oncology, Johns Hopkins University School of Medicine, 725 N. Wolfe Street, WBSB 317, Baltimore, MD 21205, USA Received 12 September 1997; accepted 17 November 1997; Received by C.M. Kane

Abstract The 86-kDa IE2 nuclear phosphoprotein encoded by the human cytomegalovirus (HCMV ) major immediate-early (MIE ) gene behaves as both a non-specific transactivator of viral and cellular gene expression and as a specific DNA-binding protein targeted to the cis-repression sequence (CRS ) at the cap site of its own promoter/enhancer region. Although the IE2 protein produced in bacteria has been shown to bind to the 14-bp palindromic CRS motif and IE2 synthesized in vitro forms stable dimers in solution through the conserved C-terminus of the protein, there is no direct evidence as yet that the intracellular mammalian forms of IE2 do so. Here, we show that the intact HCMV IE2 protein both binds to CRS DNA and dimerizes in yeast cells. In a one-hybrid assay system, a GAL4/IE2 fusion protein expressed in yeast cells activated target HIS3 expression only when CRS sites were located upstream of the GAL1 minimal promoter, but failed to do so on mutant CRS sites, demonstrating a requirement for sequence-specific DNA-binding by IE2. Examination of a series of deletion and triple amino acid point mutations in the C-terminal half of IE2 mapped the domains required for DNA-binding in yeast to the entire region between codons 313 and 579, whereas in the previous in vitro study with truncated bacterial GST fusion proteins, it was mapped to between codons 346 and 579. Transient co-transfection assays with deleted IE2 effector genes in Vero cells showed that the extra segment of IE2 between codons 313 and 346 is also required for both autoregulation and transactivation activity in mammalian cells. In a two-hybrid assay to study IE2 self-interations, we generated both GAL4 DNA-binding (DB) and activation domain (A)/IE2 fusion proteins and showed that IE2 could also dimerize or oligomerize through the C-terminus of the protein in yeast cells. Domains required for this interaction were all mapped to within the region between codons 388 and 542, which is coincident with the domain mapped previously for dimerization by co-translation and immunoprecipitation in vitro. Comparison of the domains of the IE2 protein required for CRS binding and dimerization in yeast suggests that these activities correlate precisely with requirements for the negative autoregulation function of the IE2 protein in mammalian cells. © 1998 Elsevier Science B.V. Keywords: HCMV; Autoregulation; Transactivation; Cis-repression signal; Domain mapping

1. Introduction The lytic cycle of human cytomegalovirus (HCMV ) progresses through a programmed series of events * Corresponding author. Tel: +1 410 9558684; Fax: +1 410 9558685; e-mail: [email protected] Abbreviations: CRS, cis-repression sequence; EMSA, electrophoretic mobility gel shift assay; GST, glutathion S transferase; HCMV, human cytomegalovirus; Staph Prot-A, Staphylococcus Protein-A; pm, point mutation; UAS , upstream activation sequence of the GAL1 promoter. G 0378-1119/98/$19.00 © 1998 Elsevier Science B.V. All rights reserved. PII S 03 7 8 -1 1 1 9 ( 9 8 ) 0 0 0 56 - 0

including a cascade control of viral gene expression. The viral genome is expressed in sequential fashion, and lytic gene expression is divided into three categories; immediate-early (IE ), early and late (Spaete and Stinski, 1985). IE gene transcription, which occurs during the first few hours after infection, even in the absence of de-novo protein synthesis, includes the major IE1 and IE2 transcripts ( Wathen and Stinski, 1982; Stenberg et al., 1985). The functional roles of the IE1 ( UL123) protein, a 72-kDa form encoded by exons 1, 2, 3, and 4 of the major IE (MIE ) gene and of the IE2 ( UL122) protein, an 86-kDa form encoded by exons 1, 2, 3, and 5 of

26

J.-H. Ahn et al. / Gene 210 (1998) 25–36

MIE are generally considered to be pleiotropic. Together, they initiate the process of preparing the cell to allow the virus to usurp control of the cellular macromolecular biosynthesis machinery and may also be involved in regulation of the switch between latent and lytic infection. The IE2 proteins of all beta herpes viruses show relatively high levels of amino acid conservation within the C-terminal half between the equivalents of positions 320–546 of the 579 amino acid HCMV protein (Chiou et al., 1993; Chang et al., 1995). The intact 86-kDa form of the HCMV IE2 protein displays the properties of a powerful non-specific transactivator of many viral and cellular promoters (Hermiston et al., 1987; Pizzorno et al., 1988, 1991), and is thought to play a major role in initiating and maintaining lytic cycle HCMV gene regulation pathways. Previous maping studies have revealed that IE2 contains two distinct acidic acivator domains (one at the N-terminus and one at the C-terminus) that function independently of one another in GAL4 fusion proteins, but are both required to be present for transactivation of most target reporter genes within the context of IE2 itself (Pizzorno et al., 1991). Also, IE2 has the ability to specifically down-regulate the MIE promoter through binding to the CRS motif near the cap site, both in transient cotransfection assays (Pizzorno et al., 1988; Pizzorno and Hayward, 1990; Cherrington et al., 1991; Liu et al., 1991; Chiou et al., 1993; Lang and Stamminger, 1993) and in a cell-free in vitro transcription assay system (Macias and Stinski, 1993). This autoregulation mechanism and the partially palindromic CRS DNA target site motif 5∞-CGTTTN4AACCG-3∞ are conserved between human and African green monkey CMVs (Pizzorno et al., 1988). Recently, HCMV IE2 has been shown to bind in vitro to several cellular proteins including TBP, RB, p53, CREB and EGR-1 (Hagemeier et al., 1992; Caswell et al., 1993; Jupp et al., 1993a,b; Hagemeier et al., 1994; Sommer et al., 1994; Speir et al., 1994; Lang et al., 1995; Yoo et al., 1996). Interestingly, IE1 and IE2 together are able to fully complement E1A-negative mutants of adenovirus (Spector and Tevethia, 1986) as well as to inhibit some apoptosis pathways (Zhu et al., 1995) and to co-localize with and disrupt PML-associated nuclear bodies in infected HF cells (Ahn and Hayward, 1997). DNA-binding, oligomerization and heterologous protein–protein interactions are all likely to contribute to the role of the IE2 protein in HCMV gene regulation pathways, but the mechanisms and domain structures involved are not yet well understood. The intact in vitro translated 86-kDa form of the IE2 protein forms dimers in solution (Chiou et al., 1993) and self-interacts with a bacterially expressed GST/IE2 fusion protein (Furnari et al., 1993). Both synthesized intact IE2 protein and Staph Prot-A/IE2 or GST/IE2 fusion proteins synthesized in E. coli bind specifically to the CRS site in

electrophoretic mobility gel shift assays ( EMSA) (Chiou et al., 1993; Lang and Stamminger, 1993). Furthermore, the DNA-binding and dimerization domains mapped in vitro occupy overlapping segments of the most highly conserved C-terminal portion of the protein (Chiou et al., 1993). However, as yet, there is no direct in-vivo evidence for DNA-binding or dimerization of the mammalian IE2 protein, and there is no detailed explanation available as to how these activities lead to negative autoregulation by IE2. In fact, although IE2 binding to the CRS site can be demonstrated with the E. coli fusion proteins, neither in vitro translated IE2 nor the IE2 protein recovered from extracts of HCMV infected or DNA-transfected mammalian cells was found to bind to CRS DNA by EMSA (Chiou et al., 1993). In this study, we employed one- and two-hybrid yeast genetic systems to study and map the DNA-binding and dimerization functions of intracellular forms of the IE2 protein and to reveal their relationship to autoregulation activity. We show evidence that the IE2 protein both binds to CRS DNA and dimerizes in vivo in yeast, and that these activities correlate precisely with requirements for the autoregulation function of the IE2 protein.

2. Materials and methods 2.1. Mammalian cell cultures and CAT expression assay Monolayers of Vero cells for transfection were grown in Dulbecco’s modified Eagle’s medium (DMEM ) supplemented with 10% fetal calf serum (FCS ). Transient DNA-transfection and CAT assay procedures were carried out by using the N,N-bis-(2-hydroxyethyl )2-aminoethanesulfonic acid-buffered saline version of the calcium phosphate procedure described previously (Pizzorno and Hayward, 1990). 2.2. Construction of plasmids for mammalian expression studies To generate IE2 cDNA expression vectors for use in transient expression for CAT assays in Vero cells, plasmid pJHA122 was generated by placing a 1.6-kb BglII fragment from pCJC186 (Chiou et al., 1993), which encodes the intact IE2(1–579) protein as a cDNA gene from HCMV ( Towne), into the BglII site of pSG5 (Clontech, California, USA), under the control of the SV40 early promoter and a 5∞-b globin intron. Because plasmid pJHA122 has two StuI sites (one in the SV40 promoter and the other in exon-5 of IE2), the former was eliminated by insertion of a 12-bp BglII linker into the StuI site in the SV40 promoter (pJHA124). All IE2 mutant expression vectors were subsequently generated in the pJHA124 background. The XhoI–StuI fragment containing codons 290–542 from pJH124 was replaced

J.-H. Ahn et al. / Gene 210 (1998) 25–36

by the equivalent XhoI–StuI fragments containing deletion or triple amino acid point mutations (Fig. 6) resulting in pJHA189(D290–313), pJHA190(D313–346), pJHA191(D346–358), pJHA129(D358–379), pJHA131 (D376–404), pJHA130(D403–419), pJHA132(D487–518), pJHA136(pm358–360), pJHA137(pm376–378), pJHA 139(pm403–405), pJHA138(pm417–419), pJHA133 (pm442–444), pJHA134(pm472–474) and pJHA135 (pm487–489), respectively. 2.3. Yeast strains, growth media and transforamtion The yeast strain yWAM2 (MATa gal80D URA3:: GAL1 lys2–801amber HIS3-D200 trp1--D63 leu2–3,−112 ade2–101ochre CYH2) was the host for transformation in a one-hybrid system to study IE2 binding to CRS DNA and was provided by Randall R. Reed (Johns Hopkins School of Medicine). The yeast strain Y190 (MATa gal4 D200 trp1-901 ade2–101 ura3–52 leu2–3,−112 URA3::GAL1-lacZ LYS2::GAL-HIS3 cyhR) was used in a two-hybrid system to study IE2 dimerization and was provided by Stephen J. Elledge (Baylor College of Medicine). The Y190 strain, a derivative of Y153 (Durfee et al., 1993), has two reporter genes (GAL-lacZ and GAL-HIS3). Complete ( YEPD) and synthetic (SC ) media for yeast growth and the method for yeast transformation were as described by Rose et al. (1990). 2.4. Construction of plasmids for yeast analyses All GAL4 DNA-binding (DB) domain (1–147) fusions for expression in yeast were generated in pAS1-CYH2, a derivative of pAS1 (Durfee et al., 1993), which was obtained from Stephen J. Elledge. Various portions of the IE2 coding region were placed in-frame behind the GAL4-DB domain. Plasmids pCJC420, pCJC442, and pJHA117 encoding fusion proteins containing IE2(290–579), IE2(290–542), and IE2(346– 542), respectively, were prepared by placing BamHI PCR fragments from the truncated mammalian IE2 expression plasmids pCJC190, pCJC62, and pCJC73 (Chiou et al., in preparation), respectively, into pAS1-CYH2. All GAL4 activation (A) domain (768–881) fusions for expression in yeast were constructed in pACT (Durfee et al., 1993) or pACTII, a derivative of pACT, provided by Stephen J. Elledge. Plasmid pCJC440 encoding a fusion protein containing IE2(290–542) was prepared by placing a BamHI–EcoRI PCR fragment from the truncated IE2 vector pCJC62 into pACTII. To construct pJHA140 containing the intact IE2(1–579) protein, a 2.0-kb BglII PCR fragment was prepared from an intact IE2 cDNA in pCJC186 (Chiou et al., 1993) and placed at the BamHI site of pACT. All IE2 mutant fusions were then generated in the pJHA140

27

background. The XhoI–StuI fragment containing codons 390–542 from pJHA140 was replaced by the equivalent XhoI–StuI fragment containing deletion or point mutations, and resulted in pJHA192(D290–313), pJHA193 (D313–346), pJHA197(D290–346), pJHA194(D346–358), pJHA141(D358–379), pJHA143(D376–404), pJHA142 (D403–419), pJHA144(D487–518), pJHA148(pm358– 360), pJHA149(pm376–378), pJHA151(pm403–406, pJHA150(pm417–419), pJHA145(pm442–444), pJHA 146(pm472–474), and pJHA147(pm487–489), respectively. Control plasmids pSE1112 containing GAL4-DB(1–147)/SNF1 and pSE1111 containing GAL4-A(768–881)/SNF4 (Durfee et al., 1993) were also provided by Stephen J. Elledge. To develop a yeast one-hybrid system to study IE2 binding to the CRS motif, we generated a target GAL1-HIS3 fusion gene containing either the wild-type or mutant CRS motif upstream of the GAL1 minimal promoter. Plasmid pJHA180, which has a TRP1 marker and GAL1-HIS3 fusion target gene, does not allow the expression of HIS3 because of lack of the activation of the GAL1 promoter. It was constructed by placing a 1.8-kb BamHI–SalI fragment containing GAL1-HIS3 fusion gene from pRS315HIS ( Wang and Reed, 1993) (provided by Randall R. Reed ) between the BamHI–SalI sites of pJHA178, a pAS1-CYH2 derivative, which has a 2.0-kb deletion of the EcoRI fragment containing the GAL1-GAL4(−DB) domain fusion. To generate pJHA181, which contains wild-type CRS sites upstream of the GAL1 promoter of pJHA180, two CRS oligonucleotides (30-mers), LGH546 and LGH547 (Chiou et al., 1993), were annealed and multimerized, and tandem unidirectional tetramers of the wild-type CRS motifs (wtCRS, as the 30-mer oligonucleotide pair LGH546/547) bounded by BamHI–BglII were selected for placement into the BamHI site upstream of the GAL1 promoter of pJHA180. To generate pJHA207 and pJHA209 containing mutant pmCRSI and pmCRSII sites, the LGH318/LGH319 (30-mer) and LGH821/LGH822 (23-mer) pairs were used instead of the wild-type oligonucleotides. The orientation and copy number of the CRS insertions were determined by restriction mapping and the sequences of the oligonucleotides used are described in Fig. 2B. 2.5. Growth assay in the yeast one-hybrid system The yeast strain yWAM2 was transformed with both target plasmids [GAL1-HIS3, (wtCRS) -GAL1-HIS3, 4 (pmCRSI ) -GAL1-HIS3 or (pmCRSII ) -GAL1-HIS3] 5 3 and GAL4-A fusion plasmids [GAL4-A alone or GAL4-A/IE2(1–579)]. Tranformants were selected on plates lacking Trp and Leu (SC-TrpLeu), and their growth abilities on plates lacking Trp, Leu and His (SC-TrpLeuHis+25 mM 3-AT ) were tested. The 25 mM of 3-Amino Triazol (3-AT ), an inibitor of the HIS3

28

J.-H. Ahn et al. / Gene 210 (1998) 25–36

product, was used because the GAL1 minimal promoter in the GAL1-HIS3 fusion gene allows enough weak expression of HIS3 to permit the yeast cells containing this plasmid to grow on plates lacking His alone. 2.6. Assay of b-galactosidase activity in the yeast twohybrid assay For rapid in-situ assays of lacZ expression from yeast colonies, an XGal filter assay was used. The nitrocellulose filters were laid on to the plate and allowed to wet completely, then lifted off the plate carefully to avoid smearing the colonies and placed in liquid nitrogen to permeablize the cells. After 10 s, the filters were removed from the liquid nitrogen and placed cell side up in a Petri dish containing 3MM paper soaked with Z buffer (60 mM Na HPO , 40 mM Na H PO .2H O, 2 4 2 2 4 2 10 mM KCl, 1 mM MgSO .7H O, and 50 mM b4 2 Mercaptoethanol ) plus 1 mg/ml XGal. The filters were then incubated at 30°C for appropriate times for the development of a positive blue colour. For quantitation of the b-galactosidase activity in yeast, 2-ml cultures were grown in the appropriate SC medium to an OD of 2.0, and then 0.4 ml of the culture was 600 harvested, and the b-galactosidase activity within the cells was assayed by the standard method using ONPG after permeabilizing the cells with chloroform and sodium dodecyl sulfate (SDS) (Guarente and Ptashne, 1981).

3. Results 3.1. The HCMV IE2 protein binds to CRS DNA in a yeast one-hybrid assay A yeast functional assay was used for directly studying the in vivo requirements for binding of IE2 to the CRS motif near the cap site of the MIE promoter. In this system, GAL1 promoter driven yeast HIS3 fusion genes were used as target reporter genes for IE2 DNA-binding activity. Initially, two reporter genes were generated: one was the control parent GAL1-HIS3 in which the HIS3 coding region was fused to the GAL1 minimal promoter, and the other was (wtCRS) -GAL1-HIS3 in 4 which four tandem copies of the wild-type CRS site were placed upstream of the GAL1 promoter (Fig. 1A). Plasmids containing each of these two reporter genes had a TRP1 marker for selection of yeast transformants. For effector genes, a GAL4-A/IE2(1–579) fusion in which the cDNA for the intact 86-kDa IE2 protein was fused to the yeast GAL4 ativation domain (amino acids 768–881) was constructed, and GAL4-A alone was used as a negative control. These two plasmids containing the GAL4-A fusion also had a LEU2 selection marker. The yeast yWAM2 transformants ( Trp+Leu+) contain-

ing both the plasmids encoding reporter genes GAL1-HIS3 or (wtCRS) -GAL1-HIS3 and the effector 4 gene plasmids GAL4-A alone or GAL4-A/IE2(1–579) were selected on plates lacking Trp and Leu. Subsequently, the growth abilities of the recovered double transformants on His-negative plates were investigated (Fig. 1B). As expected, all yeast transformants ( Trp+Leu+) showed healthy growth in control TrpLeunegative plates. However, only cells containing both the (wtCRS) -GAL1-HIS3 reporter plasmid and the 4 GAL4-A/IE2(1–579) fusion plasmid were able to grow on His-negative plates. This result demonstrates that the wild-type IE2 protein in the context of the GAL4-A fusion protein binds efficiently to the CRS DNA in yeast cells. The specificity of the DNA-binding activity of the IE2 protein in the yeast one-hybrid system was further investigated using mutant target CRS motifs. To generate mutant CRS-GAL1-HIS3 reporter genes, two different mutant CRS sites referred to pmCRSI or pmCRSII, in which positions 10–14 of the consensus CRS sequence had been changed, were multimerized and placed upstream of the minimal GAL1 promoter ( Fig. 2B). The yeast yWAM2 transformants ( Trp+Leu+) containing both reporter plasmids and GAL4-A fusion plasmids were selected in TrpLeu-negative plates, and subsequently, their growth abilities on His-negative plates were investigated (Fig. 2A). As expected, no double transformants ( Trp+Leu+) containing the GAL4-A alone (Leu+) reporter plasmids could grow on His-negative plates. However, when the GAL4-A/IE2(1–579) fusion plasmid (Leu+) was introduced, only those transformants ( Trp+Leu+) containing (wtCRS) -GAL1-HIS3 reporter plasmids ( Trp+) 4 were able to grow on His-negative plates. This result with the mutant CRS motifs in the yeast one-hybrid assay showed that the IE2 protein produced in yeast binds to CRS DNA in the correct sequence-specific manner. 3.2. The C-terminal segment of the IE2 protein is required for binding to CRS DNA in yeast We next wished to map the domain of IE2 required for CRS DNA-binding in yeast. The previous study using bacterially expressed IE2 proteins in EMSA had shown that the C-terminal region of IE2 between codons 346 and 579 is both sufficient and necessary for DNAbinding in vitro. In this study, we generated eight internal in-frame deletions and seven triple amino acid substitution mutants in the C-terminal region of the IE2 protein within the GAL4-A/IE2(1–579) fusion background. The map locations of these mutations relative to the known functional domains of IE2 are shown in Fig. 5. Yeast transformants ( Trp+Leu+) containing both the reporter (wtCRS) -GAL1-HIS3 plasmid and 4

J.-H. Ahn et al. / Gene 210 (1998) 25–36

29

A

B

Fig. 1. Use of a yeast one-hybrid system to study IE2 binding to the CRS site. (A) Construction of reporter plasmids. The wild-type CRS containing target gene referred to as (wtCRS) -GAL1-HIS3 was prepared by placing four tandem repeats of the annealed oligonucleotide pair LGH546/547 4 upstream of the yeast GAL1 minimal promoter fused to the HIS3. The GAL1-HIS3 fusion without the CRS site upstream of the GAL1 promoter (GAL1-HIS3) was used as a negative control. When yeast cells were transformed by both the wild-type CRS reporter plasmid and a functionally active GAL4-A/IE2 fusion plasmid, binding through protein–DNA interations between IE2 and the CRS site resulted in transactivation of the GAL1 promoter to give high levels of HIS3 expression. (B) In-vivo evidence for IE2 binding to the CRS site. A yeast strain yWAM2 was transformed both with reporter plasmids containing either GAL1-HIS3 or (wtCRS) -GAL1-HIS3 and with GAL4-A fusion plasmids expressing either GAL4-A 4 or GAL4-A/IE2(1–579). Transformants were selected on plates lacking Trp and Leu (SC-TrpLeu) and grown on His-positive plates (SC-TrpLeu) or His-negative plates (SC-TrpLeuHis+25 mM 3-AT ).

the GAL4-A/mutant IE2 fusion plasmid were selected, and subsequently, their growth abilities on His-negative plates were investigated. Only Trp+Leu+ transformants expressing GAL4-A/IE2(D290–313) were able to grow on His-negative plates, whereas all transformants grew on plates lacking Trp and Leu as a control (Fig. 2C ). The results of this experiment demonstrated that the C-terminus (at least between codons 313 and 518) is required for DNA binding activity of IE2 in yeast. Interestingly, a deletion from codon 313 to codon 346 could not bind to DNA, although the same region was not required for DNA-binding activity in the previous in vitro EMSA experiments (Chiou et al., 1993) Furthermore, all of the internal point mutants failed to retain DNA-binding activity. This suggests that the overall context within the entire C-terminal half of the IE2 protein is important to constitute DNA-binding activity in yeast. 3.3. The HCMV IE2 protein interacts with itself in a yeast two-hybrid assay In the previous in vitro study, we showed that two different-sized versions of the IE2 protein when co-translated formed cross-linked dimers in solution and

could be co-immunoprecipitated, but they did not do so when simply mixed together (Chiou et al., 1993). Here, the self-interaction of IE2 was investigated in vivo using the yeast two-hybrid assay. Because a contiguous C-terminal domain (from codon 388 to codon 542) of IE2 was sufficient for dimerization in the previous in vitro study, we generated three GAL4-DB/IE2 fusions containing codons 290–579, codons 290–542 and codons 346–542. For the GAL4-A/IE2 fusion, the intact IE2 cDNA encompassing codons 1–579 was used. The yeast Y190 cells containing both GAL4-DB/IE2 fusion plasmids ( Trp+) and GAL4-A/IE2 fusion plasmids (Leu+) were selected on plates lacking Trp and Leu, and then the production of b-galactosidase was measured by both an XGal filter assay and a quantitative assay using ONPG. The results of this two-hybrid assay are shown in Table 1. Because the IE2 protein contains an acidic activator domain (A2) at the C-terminal end (between codons 544 and 579) (see Fig. 5), we first tested whether the three GAL4-DB/IE2 fusion activated the UAS -GAL1-HIS3 reporter gene in the absence of the G GAL4-A/IE2 fusion. As expected, the transformants ( Trp+Leu+) containing both GAL4-DB/IE2(290–579) and the GAL4-A control did activate the reporter gene. However, when IE2(290–542) and IE2(346–542) con-

30

J.-H. Ahn et al. / Gene 210 (1998) 25–36

A

C

B

Fig. 2. Sequence-specific binding of IE2 to the CRS site and effects of mutations in the IE2 protein on binding to the CRS site. (A) Yeast yWAM2 cells were transformed with a control GAL1-HIS3 reporter plasmid without added CRS motifs (w/o CRS) or with versions containing four-tandem copies of wild-type CRS {(wtCRS ) ,LGH546/547}, five copies of mutant CRSI {(pmCRSI ) , LGH318/319} or three tandem copies of mutant 4 5 CRSII {(pmCRSII ) , LGH821/822}, together with a GAL4-A fusion plasmid expressing either GAL4-A or GAL4-A/IE2(1–579). Transformants 3 were selected on plates lacking Trp and Leu (SC-Trp-Leu) and subsequently grown on His-negative plates (SC-TrpLeuHis+25 mM 3-AT ). (B) Oligonucleotide sequences of the wild-type or mutant CRS sites used. Bold upper-case letters represent the partially palindromic (indicated as arrows) wild-type 14-bp consensus site of CRS. Mutated nucleotides are indicated as underlined italics (*). Sequences added for cloning purposes are indicated as lower case. (C ) All of the deleted and point mutant versions of IE2 described in Fig. 5 were expressed as GAL4-A fusion proteins. The yWAM2 cells were transformed both with a reporter plasmid containing (wtCRS) -GAL1-HIS3 and with one of the GAL4-A/mutant IE2 4 fusion plasmids. Transformants were selected on plates lacking Trp and Leu (SC-TrpLeu) and grown on His-positive plates (Sc-TrpLeu) or Hisnegative plates (Sc-TrpLeuHis+25 mM 3-AT ). The relative positions on the plate of transformants containing each GAL4-A/mutant IE2 fusion plasmids are indicated at the bottom of the diagram. Table 1 In vivo evidence for IE2 dimerization GAL4-DB Fusions ( TRP+)

GAL4-A fusions (LEU+)

X-Gal filter assay (colony color)

b-Galactosidase (units*)

GAL4-DB GAL4-DB/IE2(290–579) GAL4-DB/IE2(290–542) GAL4-DB/IE2(346–542) GAL4/DB GAL4-DB/IE2(346–542) GAL4-DB/SNF1 GAL4-DB/IE2(346–542)

GAL4-A GAL4-A GAL4-A GAL4-A GAL4-A/IE2(1–579) GAL4-A/IE2(1–579) GAL4-A/SNF4 GAL4-A/SNF4

White Blue White White White Blue Blue White

0.02±0.01 3.36±0.01 0.02±0.02 0.02±0.02 0.09±0.06 1.99±0.05 0.41±0.06 0.01±0.01

aThe unit of b-galactosidase was defined as 1000(A

−1.75 A )/(A xtxv); t, reaction time (min); v, reaction volume (ml ). 420 550 600

taining segments that lack the C-terminal A2 activation domain were fused to the GAL4-DB domain, they failed to activate the reporter gene in the presence of GAL4-A alone. This observation suggests that the A2 activation domain that was identified between codons 544 and 579 using a GAL4 swap assay in transient transfection CAT assays in Vero cells (Pizzorno et al., 1991) can function in yeast. On the other hand, the intact IE2(1–579)

protein did not stimulate the reporter gene expression on its own when fused to the GAL4-A domain in the reciprocal arrangement. On the other hand, when both plasmids containing the GAL4-DB/IE2(346–542) fusion and the GAL4-A/IE2(1–579) fusion were present together in yeast cells, they activated the expression of the UAS -GAL1-HIS3 reporter gene by a process that G presumably resulted from the reconstitution of a GAL4

J.-H. Ahn et al. / Gene 210 (1998) 25–36

31

activator complex by IE2 self-interaction. The strength of the presumed dimerization (or oligomerization) between IE2(346–542) and IE2(1–579) was three times higher than the interactions between two positive control yeast proteins that interact to form heterodimers (SNF1 and SNF4). This result shows that the HCMV IE2 protein forms either dimeric or higher oligomeric structures in yeast, and that the region between codons 346 and 542, which includes the domain (between codons 388 and 542) required for in vitro dimerization (Chiou et al., 1993), is sufficient for self-interaction in yeast.

3.4. Evaluation of the domains of the IE2 protein required for dimerization/oligomerization in yeast As shown above, the region of IE2 between codons 346 and 542 was sufficient for self-interaction in yeast. A similar stable in vitro dimerization domain with outer boundaries mapping between codons 388 and 542 was identified in the previous study by immunoprecipitation of [35S]-labelled co-translated version of the protein (Chiou et al., 1993). Both to characterize the internal region between codons 346 and 542 in the yeast in vivo assay and to compare the domains required for DNAbinding and dimerization/oligomerization in yeast, we employed the same set of mutant IE2 proteins used above in the one-hybrid assay in the two-hybrid assay. All mutant IE2 proteins described in Fig. 5 were expressed as GAL4-A fusion proteins, and their abilities to interact with the GAL4-DB/IE2(346–542) fusion protein, which encompasses the minimal region for dimerization in vitro (codons 388–542) were investigated (Fig. 3). As expected, deletions in the GAL4-A fusion in front of codon 346 (D290–313, D313–346 and D290–346) did not affect interactions with GAL4-DB/IE2(346–542). In addition, all deletions or point mutations between codons 346 and 379 (D346–358, D358–379, pm358–360 and pm376–378) were still able to dimerize with GAL4-DB/IE2(346–542). These results show that the N-terminal boundary of the domain or dimerization/oligomerization both in vitro and in vivo lies close to codon 388. Interestingly, the deletion mutant IE2(D346–358) showed about threefold higher dimerization activity compared to the wild-type IE2 protein, suggesting that the region between codons 346 and 358 acts as a repression domain for self-interactions in yeast. However, all other mutant forms of IE2 between codons 388 and 542 in the GAL4-A/IE2(1–579) background failed to interact with GAL4-DB/IE2(346–542), with the exception of the point mutant IE2(pm487–489). Overall, our results show that self-interaction of IE2 in yeast requires nearly all of the region (codons 388–542) identified as being necessary for dimerization/oligomerization by the in vitro study.

Fig. 3. Effects of mutant IE2 proteins on self-interaction in yeast. All of the deletion and point mutant IE2 versions described in Fig. 5, as well as wild-type IE2, were fused to the GAL4-A domain in the pACTII background. In this experiment, self-interactions with the GAL4-DB/IE2(346–542) fusion, which contains the minimum domain for dimerization in vitro was tested. Plasmids encoding GAL4-DB and GAL4-A fusion proteins were introduced together into Y190 cells. Transformants were selected on plates lacking Trp and Leu (SC-TrpLeu) and b-galactosidase activities of the transformants were measured as described in Materials and methods. Resulting b-galactosidase units are represented as vertical bars.

3.5. The region required for DNA binding in yeast is necessary for autoregulation activity in Vero cells To investigate the relationship between the DNAbinding property of the IE2 protein in yeast and the specific autoregulation activity in Vero cells, all of the mutant IE2 cDNA forms used in the yeast one and twohybrid assay were also expressed in a mammalian vector under the control of the SV40 promoter/enhancer from mRNA species containing added 5∞-b globin splice signals. Vero cells were transfected in a transient assay with mutant or wild-type IE2 cDNA expression plasmids together with target plasmids containing the IE68(−760/+9)-CAT gene, and the resulting levels of CAT activity were measured in cell extracts. Both wildtype IE2 and the mutant IE2(D290–313) gave approximately 17-fold repression of basal activity from the MIE enhancer/promoter region (Fig. 4, lanes 2 and 3). However, none of the other mutant IE2 proteins gave any autorepression activity. Our previous studies revealed that this activity is dependent upon an intact CRS motif in the promoter of the MIE reporter gene

32

J.-H. Ahn et al. / Gene 210 (1998) 25–36

Fig. 4. Effects of mutant IE2 proteins on CRS mediated autoregulation activity in transient DNA-cotransfection assays in Vero cells. The autoradiograph shows the results of CAT assays carried out with the resulting extracts. The target DNA was 0.5 mg of the IE68 (−760/+9)-CAT reporter gene (in plasmid pCATwt760) (Pizzorno et al., 1988), and the effector DNAs included 3 mg of pJHA124(wild-type IE2) and its derivatives pJHA189(D290–313), pJHA190(D313–346), pJHA191(D346–358), pJHA129(D358–379), pJHA131(D376–404), pJHA130(D403–419), pJHA132(D487–518), pJHA136(pm358–360), pJHA137(pm376–378), pJHA139(pm403–405), pJHA138(pm417–419), pJHA133(pm442–444), pJHA134(pm472–474), or pJHA135(pm487–489). The positions of the deletions or amino acid substitutions in each mutant IE2 were described in Fig. 5. All IE2 proteins (wild-type as well as mutant) in this experiment were cDNA versions expressed under the control of the SV40 promoter/enhancer with 5∞-globin splice signals.

(Pizzorno and Hayward, 1990). Therefore, these results showed that the lack of autoregulation activities of the mutant IE2 proteins in this transient CAT assay in Vero cells correlates precisely with the loss of DNA-binding activity in yeast cells, and provided in vivo evidence that the DNA-binding activity of the IE2 protein is required for the negative autoregulation function in Vero cells. Our current understanding of the domain requirements in IE2 for DNA-binding activity and dimerization in yeast one- and two-hybrid assays, and for both autoregulation and transactivation functions in mammalian cells is summarized in Fig. 5. The IE2 proteins distribute as a nuclear punctate pattern in virus-infected hyman diploid fibroblast (HF ) cells and give a nuclear punctate pattern with nuclear diffuse background in transiently DNA-transfected Vero cells (Ahn and Hayward, 1997). The expression levels of all mutant IE2 proteins used in this study were confirmed by indirect immunofluorescence assay (IFA) using mouse MAb 12E2 which recognizes a region between codons 131 and 274 of the exon-5 of IE2 (Ahn, J.-H., Brignole, E.J., Hayward, G.S., submitted for publication), and showed similar levels of expression compared to that of the wild-type protein, although some mutant proteins distributed in an abnormal nuclear aggregate pattern (not shown). Stable expression of the expected sized 75- to 80-kDa forms of the inactive IE2(D313–346) and IE2(D290–346) effector proteins was also validated by Western blotting with an IE2-specific MAb and both proved to produce two- to threefold more protein than the IE2(D290–313) version, which remained functionally active (not shown).

4. Discussion The intact IE2 regulatory protein of HCMV synthesized in transient expression assays behaves as both a non-specific transactivator of viral and cellular promoters and as a specific repressor of its own promoter/enhancer region in target CAT reporter genes ( Hermiston et al., 1987; Pizzorno et al., 1988, 1991; Pizzorno and Hayward, 1990). In our previous in vitro studies (Chiou et al., 1993), the IE2 protein was shown to bind to CRS DNA and to form dimers (or oligomers) through overlapping C-terminal domains mapping between codons 346 and 579 and 388 and 542, respectively. Although IE2 obviously targets specifically to the CRS DNA sequence for negative autoregulation in transient expression assays, neither the DNA-binding nor dimerization/oligomerization properties have yet been demonstrated directly with IE2 extracted from either virus-infected or DNA-transfected mammalian cells. In fact, both the in vitro translated forms and mammalian cell nuclear extracts containing IE2 failed to bind to CRS DNA probes in EMSA experiments (Chiou et al., 1993). Therefore, we turned to the use of one- and two-hybrid genetic assays to demonstrate that both of these properties could be manifested in functional assays within yeast cells. In addition, the earlier studies only used truncated fragments of IE2 to identify the N-terminal and C-terminal boundaries of the DNAbinding and dimerization domains, whereas here we used intact versions of the protein with a set of 15 internal in-frame deletions or triple amino acid point mutations to investigate the need for internal segments

J.-H. Ahn et al. / Gene 210 (1998) 25–36

33

Fig. 5. Schematic representation of mutant IE2 proteins used in this work and summary of domain requirements for DNA-binding and dimerization and for both autoregulation and transactivation in CAT assays. The domains mapped in this study as well as the previously mapped segments of the IE2 protein that are essential for transactivation (codons 1–98 and 195–579) and autoregulation (codons 313–579) using transient DNAtransfection assays in Vero cells (Pizzorno and Hayward, 1990; Pizzorno et al., 1991) are shown at the top of the diagram. The domains required for DNA binding in vitro using an electrophoretic mobility shift assay ( EMSA) with E. coli synthesized Staph Prot-A and GST fusion proteins (Chiou et al., 1993) and for dimerization in vitro by immunoprecipitation of co-translated proteins (codons 388–542) (Chiou et al., 1993), as well as for binding to cellular proteins including TBP, p53, pRB and EGR-2 by in-vitro far Western and GST affinity binding assays (codons 290–388) (Caswell et al., 1993; Hagemeier et al., 1994; Sommer et al., 1994; Yoo et al., 1996; Chiou et al., in preparation) are also indicated. In addition, two acidic activator domains [codons 25–85 (A1) and codons 544–579 (A2), dark bars] and two independent repressor domains (codons 290–388 and codons 493–542, hatched bars) identified in GAL4 domain swap assays (Pizzorno et al., 1991; Waheed et al., submitted ) are indicated within the IE2 ORF (open bar). Two distinct nuclear localization signals (NLS) between codons 145 and 151 and between codons 321 and 328 (Pizzorno et al., 1991) are also indicated (closed circles). The numbers indicated are amino acid positions at each boundary. The generation of the IE2 mutants used in this work is described elsewhere (Chiou et al., in preparation). The locations of eight deletions (D) and seven three-amino acid substitutions (pm) are indicated as solid bars at the bottom of the figure. These mutant IE2 proteins as well as wild-type IE2 were expressed from cDNA genes in both yeast and mammalian expression vectors (see Materials and methods). The ability to bind to CRS DNA and dimerize in yeast, and to both autoregulate and transactive appropriate target promoter functions in Vero cells are indicated as ‘+’ for positive or ‘−’ for negative (*data from Chiou et al., in preparation).

of these rather large domains (233 amino acids for DNA binding and 154 amino acids for dimerization). The results confirmed that both direct CRS specific DNA binding and self-interaction by IE2 occurred in yeast cells and revealed that, with one small exception, the integrity of the entire domain was required in both cases. Interestingly, the self-interaction results in yeast matched precisely to those for in vitro dimerization, although the DNA-binding activity in yeast required an extra 33 amino acid region accompanying the WLS-2 signal between codons 313 to 346 compared to the in vitro binding assay results with GST fusion proteins. Considering that GST/IE2(346–579) is capable of efficient and specific CRS binding in EMSA experiments, the contribution of amino acids 313–346 must be something other than specific DNA recognition. We have shown elsewhere (Chiou, in preparation) that this region represents the core of a protein–protein interaction domain in which this segment of IE2 binds to proteins such as p53, pRB, and itself in both GST-affinity

columns and by far western blotting assays. Others have obtained similar results in vitro with TBP, pRB, EGR-1 and CREB (Caswell et al., 1993; Furnari et al., 1993; Jupp et al., 1993a,b; Hagemeier et al., 1994; Sommer et al., 1994; Lang et al., 1995; Yoo et al., 1996). The IE2 dimerization domain between codons 388 and 542 also interacts with itself independently of, and more efficiently than, the codon 290–388 protein–protein interaction domain in such assays. Therefore, we speculate that the 313–346 domain may contribute to tetramerization or higher order oligomerization of the protein, which may be a requirement for DNA binding in vivo. Alternatively, because the N-terminal segment of IE2 exhibits a strong masking effect on the protein–protein interaction domain in vitro, the difference between the two types of DNA-binding assay may be more to do with conformational effects in the presence of the N-terminus in the yeast system. A curious conformational effect of the N-terminus of IE2 on the position of DNA-bound in vitro translated OCT/IE2 dimeric

34

J.-H. Ahn et al. / Gene 210 (1998) 25–36

fusion proteins was noted in our previous report (Chiou et al., 1993). However, no studies have yet directly addressed whether IE2 binds to the CRS DNA probe in a dimeric or oligomeric form. Presumably, the yeast self-interaction assay measures dimerization between two subunits, although higher order oligomers cannot be excluded. However, there was no evidence that inclusion of the in vitro protein–protein interaction domain enhanced self-interaction in yeast. On the contrary, it appeared to inhibit self-interaction slightly, because the affinity rose 1.5-fold when amino acids 313–346 were removed and increased up to threefold when amino acids 346–358 were removed. Therefore, the presence of this region simultaneously inhibited dimerization interactions, but was essential for DNA-binding in vivo (although not in vitro), which could conceivably be related to an oligomerization function, conformational effects or other protein–protein interactions. We showed previously that an antipeptide antibody directed against IE2 amino acids 367–382 disrupted DNA-binding by GST/IE2(346–479) in vitro, whereas antibody against amino acids 550–564, which are also required to DNAbinding but lie outside of the dimerization domain, instead produced supershifted complexes in EMSA experiments (Chiou et al., 1993). IE2 is unusual in that it binds to the CRS target DNA in the minor groove (Lang and Stamminger, 1994); however, we do not know at present which specific segment(s) of the IE2 DNA-binding domain are involved in DNA contact and recognition. With the generation of the internal deletions and triple mutations we took the opportunity to examine their effects on negative autoregulation by IE2 in Vero cell co-transfection assays. In this case, they were inserted into an SV40 enhancer (pSG5) driven wild-type cDNA background, which included a b globin 5∞ splice to ensure high level expression. We had previously shown that the IE2 region from codons 290 to 479 was sufficient for autoregulation, although truncation at 542 abolished activity (Pizzorno et al., 1988). The results presented here revealed that six of seven internal deletions including IE2(D313–346), and all seven triple point mutations, were devoid of CRS mediated down-regulation acivity on the IE68-CAT target, although IE2(D290–313) was unaffected. Therefore, this result correlates precisely with the yeast in vivo DNA-binding domain. The importance of the region between positions 290 and 363 for transactivation was also reported by Sommer et al. (1994). In addition, the ability of the mutant IE2(D290–313) to retain both autoregulation and transactivation functions in transient transfection assays was consistent with the observation by Stenberg et al. (1990) in which linker insertion mutations at amino acid positions 301 and 310 did not affect either

autoregulation or transactivation function. Curiously, our IE2 (pm487–489) mutant failed to down-regulate the MIE promoter in co-transfection assays and failed to bind to CRS DNA in the yeast one-hybrid assay, despite giving normal affinity dimerization. Therefore, this mutant appears to identify a third region within the conserved C-terminal domain that plays some specific role in DNA binding rather than in dimerization per se ( Fig. 5). Although the levels of expression of the various mutant and deleted forms of the GAL4-A/IE2 fusion proteins were not measured directly in the yeast system, the fact that the DNA-binding negative forms D313–346, D290–346, D358–379, pm358–360, pm376–378 and pm487–489 were all positive in the dimerization/oligomerization assay served as a control confirming that those versions (at least) were expressed as stable functional proteins in the yeast system. Furthermore, all of the yeast GAL4 activation domain fusions and their parallel mammalian pSG5-IE2 expression versions used here were derived from the same panel of in vitro translated IE2 proteins in which the mutations were first generated and characterized (Chiou et al., unpublished data). The stable expression of each of the deleted and mutant IE2 effector proteins in transfected Vero cells was confirmed here by IFA using MAb directed against an IE2 exon-5 N terminal epitope, and expression of the key IE2(D313–346) deletion was also confirmed by Western blotting. An examination of the abnormal intranuclear distribution of some of these deleted forms of IE2 with regard to interaction and co-localization with PML and nuclear bodies will be presented elsewhere (Ahn et al., submitted ). Overall, whereas these results do not explain why the IE2 proteins obtained in vitro translation and after transient expression or virus-infection from mammalian cells have not been found to bind to the CRS DNA probe in EMSA assays, they do show that these properties that had been previously observed directly only in vitro or with bacterial expressed forms, are indeed intrinsic characteristics of IE2 expressed inside eukaryotic cells. It should be mentioned that at 72 h in virusinfected cells, IE2 can be found in an apparent complex with the UL84 protein that can be detected by immunoprecipitation and might be capable of blocking the DNA-binding activity (Spector and Tevethia, 1994). Furthermore, our more recent studies have revealed that the DNA-binding activity of IE2 can be modulated by phosphorylation ( Waheed et al., submitted ).

Acknowledgement These studies were funded by Public Service Research Grant RO1 AI 24576 to G.S.H. from the National Institute for Allergy and Infectious Disease. We thank

J.-H. Ahn et al. / Gene 210 (1998) 25–36

Stephen J. Elledge (Baylor College of Medicine) and Randall R. Reed (Johns Hopkins School of Medicine) for gifts of yeast strains and plasmids. We also thank Sarah Heaggans for assistance in preparation of the manuscript.

References Ahn, J.-H., Hayward, G.S., 1997. The major immediate-early proteins IE1 and IE2 of human cytomegalovirus co-localize with and disrupt PML-associated nuclear bodies at very early times in infected permissive cells. J. Virol. 71, 4599–4613. Caswell, R., Hagemeier, C., Chiou, C.-J., Haywrd, G.S., Kouzarides, T., Sinclair, J., The human cytomegalovirus 86K immediate early (IE ) 2 protein requires the basic region of the TATA-box binding protein ( TBP) for binding and interacts with TBP and trascription factor TFIIB via regions of IE2 required for transcriptional regulation. 1993. J. Gen. Virol. 74, 2691–2698. Chang, Y.-N., Jeang, J.-T., Lietman, T., Hayward, G.S., 1995. Structural organization of the spliced immediate-early gene complex that encodes the major acidic nuclear (IE1) and trans-activator (IE2) proteins of African green monkey cytomegalovirus. J. Biomed. Sci. 2, 105–130. Cherrington, J.M., Khoury, E.L., Mocarski, E.S., 1991. Human cytomegalovirus ‘ie’2 negatively regulates alpha gene expression via a short target sequence near the transcription start site. J. Virol. 65, 887–896. Chiou, C.-J., Zong, J., Waheed, I., Hayward, G.S., 1993. Identification and mapping of dimerization and DNA-binding domains in the C-terminus of the IE2 regulatory protein of human cytomegalovirus. J. Virol. 67, 6201–6214. Chiou, C.-J., Ahn, J.-H., Hayward, G.S., in preparation. The central modulator domain of the human cytomegalovirus IE2 protein interacts with the C-terminus of p53 and the B/C pocket of Rb in vitro and is required for both the transactivator and repressor functions of the protein. Durfee, T., Becherer, K., Chen, P.L., Yeh, S.H., Yang, U., Kilburn, A., Lee, W.E., Elledge, S.J., 1993. The retinoblastoma protein associates with the protein phosphatase type 1 catalytic subunit. Genes Dev. 7, 555–569. Furnari, B., Poma, E., Kowalik, T.F., Huong, S.-M., Huang, E.-S., 1993. Human cytomegalovirus immediately-eary 2 protein interacts with itself and with several novel cellular proteins. J. Virol. 67, 4981–4991. Guarente, L., Ptashne, M., 1981. Fusion of Escherichia coli lacZ to the cytochrome c gene of Saccharomyces cerevisiae. Proc. Natl. Acad. Sci. USA 78, 2199–2203. Hagemeier, C., Walker, S., Caswell, R., Kouzarides, T., Sinclair, J., 1992. The human cytomegalovirus 80-kilodalton but not the 72-kilodalton immediate-early protein transactivates heterologous promoters in a TATA box-dependent mechanim and interacts directly with TFIID. J. Virol. 66, 4452–4456. Hagemeier, C., Caswell, R., Hayhurst, G., Sinclair, J., Kouzarides, T., 1994. Functional interaction between the HCMV IE2 transactivator and the retinoblastoma protein. EMBO J. 13, 2897–2903. Hermiston, T.W., Malone, C.L., Witte, P.R., Stinski, M.F., 1987. Identification and characterization of the human cytomegalovirus immediate-early region 2 gene that stimulates gene expression from an inducible promoter. J. Virol. 61, 3214–3221. Jupp, R., Hoffmann, S., Depto, A., Stenberg, R.M., Ghazal, P., Nelson, J.A., 1993a. Direct interaction of the human cytomegalovirus IE86 protein with the cis repression signal does not preclude TBP from binding to the TATA box. J. Virol. 67, 5595–5604. Jupp, R., Hoffmann, S., Stenberg, R.M., Nelson, J.A., Ghazal, P.,

35

1993b. Human cytomegalovirus IE86 protein interacts with promoter-bound TATA-binding protein via a specific region distinct from the autorepression domain. J. Virol. 67, 7539–7546. Lang, D., Stamminger, T., 1993. The 86-kilodalton IE-2 protein of human cytomegalovirus is a sequence-specific DNA-binding protein that interacts directly with the negative autoregulatory response element located near the cap site of the IE-1,2 enhancer-promoter. J. Virol. 67, 323–331. Lang, D., Stamminger, T., 1994. Minor groove contacts are essential for an interaction of the human cytomegalovirus IE2 protein with its DNA target. Nucleic Acids Res. 22, 3331–3338. Lang, D., Gebert, S., Arlt, H., Stamminger, T., 1995. Functional interaction between the human cytomegalovirus 86-kilodalton IE2 protein and the cellular transcription factor CREB. J. Virol. 69, 6030–6037. Liu, B., Hermiston, T.W., Stinski, M.F., 1991. A ‘cis’-acting element in the major immediate-early (IE ) promoter of human cytomegalovirus is required for negative regulation by IE2. J. Virol. 65, 897–903. Macias, M.P., Stinski, M.F., 1993. An in vitro system for human cytomegalovirus immediate early 2 protein (IE2)-mediated site-dependent repression of transcription and direct binding of IE2 to the major immediate early promoter. Proc. Natl. Acad. Sci. USA 90, 707–711. Pizzorno, M.C., O’Hare, P., Sha, L., LaFemina, R.L., Hayward, G.S., 1988. Trans-activation and autoregulation of gene expression by the immediate-early region 2 gene products of human cytomegalovirus. J. Virol. 62, 1167–1179. Pizzorno, M.C., Hayward, G.S., 1990. The IE2 gene products of human cytomegalovirus specifically down-regulate expression from the major immediate-early promoter through a target sequence located near the cap site. J. Virol. 64, 6154–6165. Pizzorno, M.C., Mullen, M.-A., Chang, Y.-N., Haywrd, G.S., The functionally active IE2 immediately-early regulatory protein of human cytomegalovirus is an 80-kilodalton polypeptide that contains two distinct activator domains and a duplicated nuclear localization signal. 1991. J. Virol. 65, 3839–3852. Rose, M.D., Winston, F., Hieter, P., 1990. Methods in Yeast Genetics. A Laboratory Course Manual. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY. Sommer, M.H., Scully, A.L., Spector, D.H., 1994. Transactivation by human cytomegalovirus IE 86-kilodalton protein requires a domain that binds to both the TATA box-binding protein and the retinoblastoma protein. J. Virol. 68, 6223–6231. Spaete, R.R., Stinski, M.F., 1985. Regulation of cytomegalovirus gene expression; a and b promoters are trans-activated by viral functions in permissive human fibroblast. J. Virol. 56, 135–143. Spector, D.J., Tevethia, M.J., 1986. Identification of a human cytomegalovirus virus DNA segment that complements an adenovirus 5 immediate-early mutant. Virology 151, 329–338. Spector, D.J., Tevethhia, M.J., 1994. Protein–protein interactions between human cytomegalovirus IE2-580aa and pUL84 in lytically infected cells. J. Virol. 68, 7549–7553. Speir, E., Modali, R., Huang, E.-S., Leon, M.B., Shawl, F., Finkel, T., Epstein, E.S., 1994. Potential role of human cytomegalovirus and p53 interaction in coronary restenosis. Science 265, 391–394. Stenberg, R.M., Witte, P.R., Stinski, M.F., 1985. Multiple spliced and unspliced transcripts from human cytomegalovirus immediate-early region 2 and evidence for a common initiation site within immediateearly region 1. J. Virol. 56, 665–675. Stenberg, R.M., Fortney, J., Barlow, S.W., Magrane, B.P., Nelson, J.A., Ghazal, P., 1990. Promoter-specific trans activation and repression by human cytomegalovirus immediate-early proteins involves common and unique protein domains. J. Virol. 64, 1556–1565. Wang, M.M., Reed, R.R., 1993. Molecular cloning of the olfactory neuronal transcription factor Olf-1 by genetic selection in yeast. Nature 364, 121–126.

36

J.-H. Ahn et al. / Gene 210 (1998) 25–36

Wathen, M.W., Stinski, M.F., 1982. Temporal patterns of human cytomegalovirus transcription: mapping the viral RNAs synthesized at immediate early, early, and late times after infection. J. Virol. 41, 462–477. Yoo, Y.D., Chiou, C.-J., Choi, K.S., Yi, Y., Michelson, S., Kim, S.,

Haywrd, G.S., Kim, S.-J., 1996. The IE2 regulatory protein of human cytomegalovirus induces expression of the human TGF-b1 gene through an Egr-1 binding site. J. Virol. 70, 7062–7070. Zhu, H., Shen, Y., Shenk, T., 1995. Human cytomegalovirus IE1 and IE2 proteins block apoptosis. J. Virol. 69, 7960–7970.