The human cytomegalovirus US3 immediate-early protein ... - NCBI

\./ 1993 Oxford University Press

Nucleic Acids Research, 1993, Vol. 21, No. 12 2931 -2937

The human cytomegalovirus US3 immediate-early protein lacking the putative transmembrane domain regulates gene expression Daniel J.Tenney+, Linda D.Santomenna, Karyn B.Goudie§ and Anamaris M.Colberg-Poley* The Du Pont Merck Pharmaceutical Co., PO Box 80328, Wilmington, DE 19880-0328, USA Received July 22, 1992; Revised and Accepted May 6, 1993

ABSTRACT Through alternative transcript splicing, the human cytomegalovirus (HCMV) US3 immediate-early (IE) locus encodes multiple products including potential membrane-bound glycoproteins. To characterize the US3 products and determine which encode regulatory activity, individual cDNAs were cloned and expressed. Three transcript species were confirmed through the isolation of cDNAs; an unspliced transcript, a transcript spliced once from exon 3 to exon 5 and a transcript spliced both at exon 1 to exon 3 and at exon 3 to exon 5. The predicted signal sequences and N-linked glycosylation sites in the US3 products were confirmed using expression in reticulocyte lysates containing microsomal membranes. Regulatory activity of the individual US3 products was demonstrated using transient transfection assays. The unspliced cDNA and the cDNA containing the exon 3 to exon 5 splice, encoded products which increased expression of the human heat shock protein 70 (hsp70) promoter, while the product of the doubly-spliced US3 cDNA did not. Transactivation was synergistically increased by coexpression with the HCMV UL37 protein. We conclude that the first 132 amino acids common to the unspliced and the singly-spliced US3 gene products are sufficient for hsp70 transactivation; while the aminoterminal 28 amino acids, encoded by the doubly-spliced US3 cDNA, are not. These results demonstrate that a US3 IE protein lacking the putative transmembrane domain has regulatory activity. INTRODUCTION The HCMV genome US3 region (EMBL accession # M18921; Ref. 1) is transcribed during infection in the absence of de novo viral protein synthesis (1-4), thus experimentally establishing it as an immediate-early (IE) gene. The expression of the US3 region is controlled by a multicomponent, complex enhancer (1) with homology to the other identified HCMV enhancer, that of

the major immediate early region (MIE; 1; reviewed in 5, 6, 7, 8). Two other regions of LE gene expression witiin the HCMV genome have been characterized, the UL36-38 region (encoding UL36; UL37; UL37x1; 2, 9, 10) and the TRSl/IRS1 region (11), but neither contain discernible transcriptional enhancers. Products of the HCMV MIE genes have long been recognized to regulate expression of cellular and viral promoters (reviewed in 5, 6, 7, 8). However, only recently have gene products from the the US3 (12), UL36-38 (12) and TRS1 (11) genes been shown to transactivate gene expression either alone or in concert with the MIE gene products. The deduced amino acid sequence of the US3 (1) and UL37 (9) gene products suggests that these genes encode membrane-bound glycoproteins with hydrophobic amino-terminal signal sequences, hydrophobic carboxy-terminal membrane spanning regions, and potential N-linked glycosylation sites. The possible participation of HCMV membrane-bound glycoproteins in the regulation of nuclear gene expression represents an unusual and interesting pathway in herpesvirus gene regulation. Alternative processing of US3 transcripts gives rise to several US3 gene products (1). To define precisely the processing of US3 transcripts and to isolate reagents suitable to determine the function of the individual products of US3, we isolated their cDNAs. In this work, cDNAs corresponding to the individual US3 IE RNAs have been isolated and examined. We have defined unambiguously the splice sites of the mRNAs and their coding capability, as well as corrected a proposed splice site and the US3 polyadenylation site (1). The isolation of these complete cDNAs has enabled us also to express individual US3 IE products in vitro in reticulocyte lysates and assay their ability to undergo primary processing events. We have also expressed the cDNAs in transient transfection assays and demonstrated transactivation of the human heat shock protein 70 (hsp7O) promoter (13) by individual US3 gene products alone and in concert with that of the HCMV UL37 gene. We found that the unspliced and the singly-spliced (exon 3 to 5) US3 cDNAs encode products which regulate human hsp70 promoter expression while the doublyspliced (exon 1 to 3 and exon 3 to 5) cDNA did not.

* To whom correspondence should be addressed at: Center I Research, R-213, Children's National Medical Center, 111 Michigan Avenue, NW, Washington, DC 20010-2970, USA Present addresses: +Department of Virology, Bristol-Myers Squibb Pharmaceutical Research Institute, Princeton, NJ 085434000 and §DUMC, Box 3020, Department of Microbiology and Immunology, Durham, NC 27710, USA

2932 Nucleic Acids Research, 1993, Vol. 21, No. 12

MATERIALS AND METHODS RNA isolation and Northern blot analysis Total infected cell RNA was isolated at 4 and 8 hr after permissive HCMV (strain AD 169) infection of human foreskin fibroblast cells (multiplicity of infection of 3). Total RNA enriched for IE transcripts was isolated from cells infected for 8 hours in the presence of 100,M anisomycin (2, 10). RNA was isolated by the guanidinium-isothiocyanate/CsCl centrifugation procedure (14), electrophoresed in a 0.9% agarose/2M formaldehyde gel (5Ag/lane), and transferred to GeneScreen Plus membranes (DuPont NEN Research Products). Membranes were hybridized to a US3-specific antisense 32P-riboprobe [289 bp PstI fragment, HCMV genome nucleotides (nt) 194452 to 194741; HCMV sequence: EMBL accession #X17403, ref. 15], as described (10).

cDNA synthesis Initial cDNA synthesis used 5 ,tg of polyA+ that was selected by oligo-dT chromatography. RNA from 4 and 8 hours after permissive infection was primed with oligo-dT and used to generate cDNAs by the AMV reverse-transcriptase/S 1 nuclease method (16). These cDNAs were cloned into lambda ZapIl (Stratagene, La Jolla, CA) using EcoRI linkers, as described (17). Approximately 1 x 106 unamplified phage plaques were screened by hybridization to the US3 PstI fragment labelled with 32p, and several recombinants were isolated. Nucleotide sequencing of the isolated US3 cDNAs was determined with CsCl-banded doublestranded DNA using a Sequenase kit (US Biochemical). To increase the abundance of cDNAs for the spliced US3 IE transcripts, cells were infected for 8 hr in the presence of anisomycin (100 jiM). cDNAs were synthesized from polyA+ RNA using commercial kits (Invitrogen, San Diego, CA). Three different reverse transcription reactions were performed using (i) a specific US3 3' primer (see PCR reactions, below), (ii) an oligo dT12>18 primer, or (iii) random hexamer primers. The cDNA reaction was terminated by addition of 2 4d of 0.5 M EDTA, phenol/chloroform extraction and filtration through a G-25 column and, finally, dilution with H20 to 200 ,1l before PCR amplification.

PCR amplification Ten /l of single-stranded cDNA from each of the three individual reverse transcription reactions were amplified by PCR using a Perkin-Elmer Cetus GeneAmp kit, except that Vent DNA polymerase (New England Biolabs), reportedly having a 3'-5' exonuclease proofreading function, was used instead of Taq DNA polymerase. All three single-stranded cDNA preparations used specific 5' and 3' US3 primers for PCR amplification. The 5' US3 amplification primer was 5'-CCgaattcACGCTACTTCTCAGCGAAG-3 ', where the small case letters represent the inserted EcoRI site for cloning and the underlined sequences represent HCMV nt 194767-194748. The 3' US3 amplification primer (also used for reverse transcription) was 5'-CCggatccTAGAGTTACGAATAAACCGG-3 ', where the small case letters represent the inserted BamHI site for cloning and the underlined sequences represent HCMV nt 193889-193908. PCR amplifications were carried out for 30 cycles using 99°C/1 min denaturation, 56°C/30 seconds annealing, and 75 °C/2 min extension. The amplified US3 cDNAs were CHCl3-extracted, digested with EcoRI and BamHI, ligated to a similarly cleaved pBSIISK+ vector (Stratagene), and transformed into E. coli (XL-1 blue).

In vitro transcription and translation 5 '-capped RNA was synthesized by T7-promoted run-off transcription from a pBSIISK+ vector, as suggested by the manufacturer (Stratagene). These transcripts were translated in rabbit reticulocyte lysates in the presence of 35S-methionine, using commercial kits (Stratagene). To study potential cotranslational processing, canine pancreatic microsomes were included in the translation reactions, according to the manufacturer's instructions (Promega Biotech). One Al of the 25 Al translation reactions was electrophoresed in precast 10-20% gradient polyacrylamide - SDS gels (Integrated Separation Systems, Natick, MA, USA) with Tris-Glycine SDS buffer (Fig. 4A) or Tris-Tricine SDS buffer (Fig. 4B).

Transient transfections and transactivation assays For expression of individual US3 gene products, cDNAs were subcloned into vector p394 (12) downstream of the HCMV MIE promoter and upstream of the SV40 early polyadenylation signal. HeLa cells were transfected in duplicate with 10 Ag human hsp70-chloramphenicol acetyl transferase reporter DNA (p221, ref. 13), with 5 Ag of US3 effector DNA and with S yg UL37 DNA (to assess activity in the presence of UL37) or with 5 ,tg UL37 nonsense mutant DNA (to assess the effect of US3 effector DNA alone). Effector DNAs included genomic US3 (p339, ref. 12), nonsense mutant US3 (p424, ref. 12), US3 unspliced cDNA (p516), US3 singly-spliced, exon 3 to exon 5, cDNA (p517), US3 doubly-spliced, exon 1 to 3 and exon 3 to 5, cDNA (p518), pUC 18, or HCMV glycoprotein B (gB; p370, ref. 12). UL37 DNAs included nonsense mutant UL37 (p428, ref. 12) and wild type UL37 cDNA (p414, ref. 12). Transfected cells were harvested 48 hours after transfection. Ten /kg of protein from duplicate samples were analyzed for conversion of chloramphenicol to its acetylated forms. The percentage conversion of each group was standardized to the value obtained for HeLa cells transfected with hsp70-CAT DNA, pUC 18, and the nonsense mutant UL37 DNA.

RESULTS US3 cDNAs generated from permissively-infected cells Northern blot analysis of total cellular RNA isolated from human foreskin fibroblast (HFF) cells at very early times after permissive HCMV infection revealed that the unspliced US3 transcript was primarily produced during early times of infection. The multiply spliced US3 transcripts were barely detectable during permissive HCMV infection, but were expressed in abundance during infection in presence of the protein synthesis inhibitor, anisomycin (Fig. 1). Because the US3 region is expressed at relatively low levels during permissive infection, there is often variation in the amount of US3 RNA detected on Northern blots, however, a consistent level of RNA is detected when protein synthesis is blocked using anisomycin (Fig. 1: two 4 hr lanes and two 8 hr anisomycin lanes are from two separate experiments). These results using total infected cell RNA complement previous reports of the levels and multiple US3 transcripts in the cytoplasmic fraction from infected cells (1, 2). Initial cDNA synthesis used polyA + RNA isolated from HFF cells after 4 and 8 hours of permissive HCMV infection. Several lambda phage containing US3 cDNA clones were identified by hybridization to a US3-specific probe and isolated for further analysis. Comparison of the nucleotide sequence of portions of these cDNAs to the published genomic sequence (4) revealed

Nucleic Acids Research, 1993, Vol. 21, No. 12 2933 A. 4

8 8A 4 8A Un kb -2.4 -2.0

-1.4

II -0.2

B. US 3 RNA mRNA

y -

2 ....

y

*> O > O¢... .O.B E. ..................

K)

0.88 0.73

0.71 0.56 0 .40

P ROB E kbp

Figure 1. RNA expressed from the HCMV US3 IE region during infection. A. Northern blot analysis: Total infected cell RNA was isolated at the hours after infection indicated above each lane, from uninfected cells (Un), or from cells infected 8 hr in the presence of anisomycin (8A). The migration of RNA molecular size markers, run in parallel, is indicated. B. Map of the transcripts expressed from the HCMV US3 region at IE times of infection, proposed by Weston (1). Solid thick lines and stippled lines represent exons and introns, respectively. Polyadenylation sites and the direction of transcription are indicated by the arrowheads at the end of the transcripts. The common ATG initiation codon and different stop codons within the protein coding sequences are shown by a circle and boxes above the transcripts, respectively. The location of the IE transcriptional promoter is shown by the bent arrow below the map. Approximate predicted transcript sizes from Weston (1988; without polyadenylation) are shown. The location of the PstI fragment used for hybridization is indicated beneath the transcripts.

that the isolated US3 cDNAs corresponded to the unspliced transcript. The longest US3 LE cDNA extended from nt 194752, 14 nt downstream of the RNA cap site mapped previously (1) and terminated at nt 193889, 20 nt downstream of the consensus polyadenylation signal spanning nt 193914-193909. This RNA cleavage/polyadenylation site at 193889 is near the site mapped by Weston (1) who proposed two alternative RNA termini at nt 193893 and 193907. Another cDNA startpoint was located upstream of the US3 locus (nt 195568) and most likely corresponds to a US6 or US7 transcript (18, 19), that continues through the US3 locus and terminates at the US3 cleavage/polyadenylation site (nt 193889). We determined the complete nucleotide sequence of two different US3 cDNAs from unspliced RNAs isolated during permissive infection and confirmed the genomic sequence published by Weston (1, 4; data not shown).

PCR amplification of spliced cDNAs from anisomycin-treated cells To enable the cloning of cDNAs for the other, multiply spliced US3 transcripts, cDNA synthesis was performed using RNA isolated from cells infected in the presence of anisomycin. Three different preparations of single-stranded cDNA were first generated by reverse transcription using different 3' primers

Figure 2. PCR amplification of cDNA synthesized from HCMV IE RNA. Products of PCR amplification of cDNAs synthesized using RNA from cells infected for 8 hr in the presence of anisomycin. Three different preparations of single-stranded cDNA were made using three different 3' primers and each were amplified using PCR with 5' and 3' US3-specific primers. One fifth (20 Al) of the amplification reactions were electrophoresed in an 1.25% agarose gel and visualized by ethidium bromide staining. US3-specific PCR amplification products from single-stranded cDNA made with the specific US3 3' primer, A; oligo dT12_18 primers, B; and random hexamer primers, C were analyzed by agarose gel electrophoresis. Lane D shows a control amplification of a previously cloned, frill-length US3 fragment using the specific US3 primers. In lane M, a 123 bp DNA ladder (Bethesda Research Laboratories) is shown.

(oligo-dT, random, or the specific 3' US3 primer; Fig. 2). The US3 cDNAs were selectively amplified from each of the three preparations by the PCR using specific 5' and 3' US3 primers corresponding to the mapped mRNA start site (1) and the 3' terminus (determined on the unspliced cDNAs, above) common to all the US3 transcripts (Fig. 2). The primers contained EcoRI and BamHI restriction endonuclease sites for subsequent cloning. Upon PCR amplification of these pools using US3-specific primers, identical cDNA species were obtained. Thus, only the US3 cDNAs were amplified by PCR with the specific 5' and 3' US3 primers, even though libraries of HCMV-infected cell cDNAs were initially synthesized using reverse transcription with the three different reverse (3') primers. Moreover, the US3 cDNA products amplified by PCR reflected the US3 RNA species expressed during HCMV infection (compare lanes A-C, Fig. 2 to lanes 8A, Fig. 1). No products greater than 50-100 bp resulted from control amplification without template cDNA added (data not shown).

Analysis of amplified cDNAs Recombinant clones resulting from the ligation of vector and cDNA were screened by blue-white color selection on plates containing X-gal and IPTG. Ninety white colonies were selected; of these, forty one contained EcoRI to BamHI plasmid inserts of 1 kb or less and were chosen for further analysis because the largest predicted fragment generated from the PCR of genomic DNA is 878 nt. The two PstI sites located in the US3 region within the first and third exons (as shown in the map in Fig. 3) were used, along with the BamHI site in the 3' PCR primer, to categorize the cDNAs according to the transcript species. Figure 3A shows the number of cDNAs representative of each transcript of the 41 analyzed. The majority of cDNAs, 23 of 41, corresponded to the exon 3 to exon 5 singly-spliced species. The unspliced RNA was represented by 6 clones and the doublyspliced RNA (removing exons 2 and 4) was represented by 7 clones. The precise locations of splice junctions were determined by nucleotide sequencing and are presented in Fig. 3B. One clone of each of the three major species was sequenced in its entirety,

2934 Nucleic Acids Research, 1993, Vol. 21, No. 12

A. Number of cDNAs Isolated

Exons 5

4

RNA from: Uninhibited Anisomycin infection -treated

1

2

3

-90.. y

y

m

y y 7

m

y

Pst 1

mRNA {* 0.88 0.73 0.71 0.56 0.40

2

6 0 23

0

0 0

7

0

0

PsI

0 kbp

US3 RNA: 194766- 193889

B.

EXON

1 /

EXON 3

Splice Donor

Splice Acceptor

Splice Junction

GTG / GTATCG

ATCTrACATGGACAG / A

GTG-GTG-ACT-GCA 194607

EXON 3/ EXON 5

OGATTTATTTAACAG / G

CAG / GTGAGG

1

CCT-CAG-GTA-CGA 194295J

Consensus

A C 43 A GI G T 2AGT A 64 73 100100 G 688463

40

29

3' Terminus of US3 RNA

tL194454 t194126 G

Py Py PY Py Py Py Py Py Py Py Py N Py A G /52 96 100100 A 70 69 81 74 84 79 82 73 81 93 86

22

MTAAATAMCCGG1TTATTCGTAACTCT A N 1 939 18J

193889

3. US3 IE transripts and cDNAs. A. Map of the transcripts expressed from the HCMV US3 region at IE times of infection as proposed by Weston (1). The physical locations of the PstI sites, within the US3 locus, that were used for cDNA analysis are indicated. The numbers of each cDNA species isolated from 4 and 8 hr after infection without anisomycin (Uninhibited infection) and 8 hr with anisomycin (Anisomycin-treated), that correspond to the transcripts proposed by Weston (1), are indicated. B. Nucleotide sequence determination of the splice junctions and 3' termini of the US3 IE cDNAs. Shown are the splice junctions (donors and acceptors) and mapped polyadenylation site. Open letters indicate sequences removed from the spliced US3 transcripts and polyA sequences. Numbering refers to the genomic sequence (15) at the splice junction or the last nucleotide in the US3 cDNAs matching the genomic sequence (polyadenylation site). Also shown are the consensus splice donor and acceptor signals of eukaryotic mRNAs and the approximate percentage of each base (20). Figure

the regions involved in splicing were sequenced in two clones of each species, and the splice junction of exons 1 to 3 were sequenced in all seven clones of the doubly-spliced cDNA species. The exon 3 to exon 5 splice agrees with previous results obtained by nuclease protection assays while the exon 1 to exon 3 splice occurs at the reported splice acceptor, but three nucleotides downstream of the splice donor previously mapped (1) thus altering the coding sequence of this US3 RNA by one codon. In addition to the species listed (Fig. 3A), two clones were isolated that contained the identical exon 3 to exon 5 splice, but had an unusual exon 1 to exon 3 splice; that is, the same exon 3 splice acceptor site was used, but was joined to nt 194546 (61 bp downstream of the abundant exon 1 to exon 3 splice), which lacks a consensus splice donor signal. Because the correct splice acceptor was used, we interpret the results to indicate an aberrant splicing event. This unusual clone may be represented by the

minor band resulting from PCR amplification of cDNA (Fig. 2). Three other clones analyzed were small (