Identification of a Spliced Gene from Kaposi’s Sarcoma-Associated Herpesvirus Encoding a Protein with Similarities to Latent Membrane Proteins 1 and 2A of Epstein-Barr Virus

Abstract

Kaposi’s sarcoma-associated herpesvirus (KSHV) or human herpesvirus 8 (HHV-8) is a novel herpesvirus implicated as the causative agent of Kaposi’s sarcoma (KS), primary effusion lymphoma, and some cases of multicentric Castleman’s disease. KSHV persists in the majority of KS spindle (endothelial tumor) cells and lymphoid cells in a latent form, with only a limited set of viral genes expressed in a tissue-specific manner. Here, we report the identification of a family of alternatively-spliced transcripts of approximately 7.5 kb expressed in latently infected body cavity-based lymphoma (BCBL) cell lines which are predicted to encode membrane proteins with similarities to the LMP2A and LMP1 proteins of Epstein-Barr virus. In two highly divergent sequence variants of the right end of the KSHV genome, alternative splicing of eight exons located between KSHV ORF 75 and the terminal repeats yields transcripts appropriate for proteins with up to 12 transmembrane domains, followed by a hydrophilic C-terminal, presumably cytoplasmic, domain. This C-terminal domain contains several YxxI/L motifs reminiscent of LMP2A and a putative TRAF binding site as in LMP1. In latently (persistently) infected BCBL cells the predominant transcript utilizes all eight exons, whereas in phorbol-ester-induced cells, a shorter transcript, lacking exons 4 and 5, is also abundant. We also found evidence for an alternative use of exon 1. Transfection of an epitope-tagged cDNA construct containing all exons indicates that the encoded protein is localized on cell surface and intracellular membranes, and glutathione S-transferase pull-down experiments indicate that its cytoplasmic domain, like that of LMP1, interacts with TRAF1, -2, and -3. Two of 20 KS patients had antibodies to the hydrophilic C-terminal domain, suggesting that the protein is expressed in vivo.

Human herpesvirus 8 (HHV-8), also called Kaposi’s sarcoma (KS)-associated herpesvirus (KSHV), is a gamma-2 herpesvirus which has been detected in all forms of KS, in AIDS-associated body cavity-based lymphomas (BCBL) or primary effusion lymphoma (PEL), and in a number of cases of multicentric Castleman’s disease (12, 13, 65; for a review, see reference 60). Classic KS, which presents mainly in elderly men of Mediterranean origin, usually consists of indolent skin tumors, but KS associated with human immunodeficiency virus (HIV) infection or iatrogenic immunosuppression following organ transplantation and endemic KS of HIV-negative individuals in central regions of Africa are typically more aggressive and can affect systemic organs (3). Viral DNA can be detected in all forms of KS, regardless of epidemiological origin (6, 8, 14, 47). Serological studies suggest that KSHV is not ubiquitous in Northern Europe, North America, and some parts of Asia but is relatively common in those regions of Southern Europe (Italy, Greece), with a higher incidence of classic KS, and is also widespread in Africa (10, 26, 31, 41, 61, 72).

KSHV has been detected in endothelial and spindle cells of KS lesions, in circulating endothelial cells, B cells, CD8⁺ cells, T cells, macrophages, and prostate gland epithelial cells (2, 7, 17, 28, 36, 42, 54, 62, 66–68). In KS spindle (endothelial tumor) cells and PEL cells, KSHV persists in a latent form, as demonstrated by the presence of circular genomes of KSHV in PEL cells and KS tissue (20, 56). During this state, only a few genes are expressed, some in a tissue-specific manner. In KS tissue these include a putative apoptosis inhibitor, v-FLIP (ORF 71/K13), a D-type cyclin homologue, v-cyclin (ORF 72), the latency-associated nuclear antigen, LNA/LANA, (ORF 73), and the T0.7/K12 transcript (18, 54, 66, 73). The same genes are also expressed in PEL cell lines, together with those for the interferon regulatory factor homologue vIRF-1(K9), the interleukin-6 (IL-6) homologue vIL-6, and the chemokine homologue vMIP-II (27, 48, 58).

The lack of inducibility of v-FLIP, v-cyc, and LANA by treatment of PEL cell lines with phorbol esters and/or sodium butyrate led Sarid et al. (58) to designate them class I genes. Several other genes are expressed at a basal level in uninduced PEL cell lines and may be further upregulated by chemical treatment; these are designated as class II genes (58). Class II transcripts include a nontranslated nuclear RNA T1.1, K12/kaposin, vIRF-1, MIP-I/MIP-1A, MIP-II/MIP-1B, and vIL-6 (48, 58, 73). Genes for structural proteins are not expressed in unstimulated PEL cells but can be induced chemically. These are classed as group III genes (58). The small number of cells expressing these class II and III genes in KS tissue are thought to be productively infected (66–68). Lytic KSHV gene expression has also been reported in monocytes/macrophages (4), and the presence of linear KSHV genomes in peripheral blood mononuclear cells suggests productive infection in at least some cells (20).

Infection of primary B lymphocytes in vitro with Epstein-Barr virus (EBV), a gamma-1 herpesvirus, leads to restricted expression of several EBV genes and cell immortalization. Of those genes which are expressed, five nuclear proteins (EBNAs 1, 2, 3A, 3C, and -LP) and an integral membrane protein (LMP1) have been found to be critical to cell immortalization (15, 30, 40, 70). Other genes that are expressed in latently infected B cells but which are not required for immortalization are EBNA3B, LMP2A, and the EBV-encoded small RNAs (37, 38, 39, 69). Of these, LMP2A is known to be expressed in nasopharyngeal carcinoma cells as well as in latently infected B cells of asymptomatic individuals. Its role may be related to maintaining EBV latency in B cells in vivo (24, 25, 38, 39, 43, 44).

In contrast, LMP1 has transforming properties in rodent fibroblasts, is essential for B-cell immortalization (30, 71), and inhibits epithelial differentiation in vitro (19). LMP1 induces the expression of Bcl-2, activation markers, adhesion molecules, and autocrine growth factors by inducing downstream signaling through its interaction with TRAFs (21, 22, 29, 46, 49).

Using probes generated from the right end of the genome of an isolate of KSHV from a classic Greek KS biopsy, we have screened Northern blots from two PEL cell lines, BCP-1 and HBL-6, and a latent cDNA library of HBL-6. We have identified a 7.5-kb transcript that appears to be expressed in latently infected cells but which is upregulated following induction with either tetradecanoyl phorbol acetate (TPA) (BCP-1 cells) or sodium butyrate (HBL-6 cells) and would thus be considered a class II transcript. Analysis of the sequence of cDNA and reverse transcription (RT)-PCR clones and of the BCP-1 and HBL-6 genomic sequences showed the presence of a gene composed of multiply-spliced exons with the potential to encode a family of proteins with similarities to EBV LMP2A and LMP1, for which we suggest the name latency-associated membrane protein (LAMP).

MATERIALS AND METHODS

Cell culture.

The HBL-6 (BC-1) cell line was grown at 37°C in RPMI 1640 supplemented with 20% fetal calf serum (FCS), 50 IU of penicillin/ml, and 50 μg of streptomycin/ml in the presence of 5% CO₂. To induce lytic replication, cells were exposed to sodium butyrate (3 mM) and harvested after 48 h. The BCP-1 cell line was grown in the same manner. To induce lytic replication, phorbol myristate acetate (TPA/PMA) was added to 20 ng/ml, and the cells were harvested after 48 h.

Cloning of KSHV genome from a classic KS biopsy.

DNA was prepared from an excision biopsy of a case of classic KS (GK18) by proteinase K digestion and phenol-chloroform extraction by using conventional protocols. A cosmid library was constructed by partial digestion with Sau3A and screened with a probe derived from ORF 72/v-cyclin. A positive clone (cos83) was completely sequenced, and the sequence was deposited in GenBank.

RNA analysis.

Total RNA was extracted from both latent and induced cells by the RNAzol B method (Cinna/TEL-TEST). For each sample, 12 μg of total RNA was fractionated on a 1% agarose-2.2 M formaldehyde gel and blotted onto nylon membranes (Hybond N⁺; Amersham) by alkaline transfer. Prior to the addition of the radiolabeled probe, the blots were prehybridized at 42°C for 2 to 4 h in prehybridization buffer containing 100 μg of denatured sheared salmon sperm DNA/ml. The first probe consisted of a 4.3-kb EcoRI fragment generated from a KSHV cosmid clone obtained from a classic KS lesion (see text). To obtain a probe specific for the family of transcripts described here, the end of exon 1 and exons 2 to 5 were amplified by RT-PCR by using primers ex1c3′ (5′-CCCCTCCCATTGGACCATCTAT-3′; nucleotides 136549 to 136570 for HBL-6) and ex5crev (5′-CGTTATTGGTGTCTTGTTCGCGC-3′; nucleotides 135609 to 135631 for HBL-6) see Fig. 1. Hybridization of the probes to the filters was done at 42°C overnight. Blots were washed twice in 2× SSPE (1× SSPE is 0.18 M NaCl, 10 mM NaH₂PO₄, and 1 mM EDTA [pH 7.7]) for 15 min at room temperature, followed by washing in 2× SSPE-2% sodium dodecyl sulfate (SDS) at 42°C. Blots were exposed to Kodak XAR-5 film for up to 5 days with intensifying screens at −80°C.

FIG. 1.

Location of exons for the KSHV membrane protein in the two major KSHV sequence variants represented by HBL-6 and GK18/BCP-1. (Top) Schematic diagram of conserved genomic regions (open boxes) and genes found only in KSHV or closely related gamma-2 herpesviruses (black boxes). The orientation of the black boxes reflects their transcriptional orientation. The position and organization of a cosmid clone (Cos 83) obtained from a classic KS case (GK18; see text) is shown enlarged. (Bottom) Location of exons 1 to 8 for KSHV K15/LAMP in HBL-6 and GK18/BCP-1. The location of the 4.3-kb probe that was generated from GK18 and which was used to screen a latent HBL-6 cDNA library is indicated. Short horizontal lines mark the locations of primers used in this study. A bracket above the HBL-6 diagram indicates the position of the ORF K15 initially predicted by Russo et al. (56). An additional T residue within exon 1 of the HBL-6 sequence results in the entire exon 1 being in frame with exons 2 to 8 (see text).

RT-PCR.

Poly(A) RNA was isolated by using the Dynabead mRNA purification kit (Dynal), and 500 ng was reverse transcribed with Superscript II (Life Technologies) by using primer LAMParev (5′-TCACTCTCCAACCACTGCCCAGTGACG-3′; nucleotides 19143 to 19169 of GK18) for BCP-1 or LAMPcrev (5′-TCACTCTCCAACCACAGCCC-3′; nucleotides 134592 to 134614) for HBL-6. These primers are located inside ORF 75 (see Fig. 1), downstream of the eighth exon in the family of transcripts described here. PCR amplification was performed with Bio-X-Act polymerase (Bioline, London, United Kingdom) by using either LAMParev plus ex1afor (5′-TTGTAAGCCCTGTGGATACC-3′; nucleotides 21498 to 21517 of GK18) for BCP-1 or LAMPcrev plus initially LAMPcfor (5′-GTTTCAAGCTTCCCC-3′; nucleotides 136567 to 136582) and later by using ex1cfor (5′-CTTCATTTTTGGGCCTTGGGC-3′; nucleotides 136785 to 136805) for HBL-6. PCR products were cloned into pGEM-T (Promega Corp.) and sequenced by using an ABI 377 automated sequencer.

Southern blot analysis.

A 373-bp fragment of the gene was amplified by PCR from BCP-1 DNA by using primers ex8afor (5′-CAGGGATCCTAAATAGTTACCGACAGAGAGGCGG-3′; nucleotides 19796 to 19821 of GK18) and ex8arev (5′-ATACCCGGGCTAGTTCCTGGGAAATAAAACCTCCTC-3′; nucleotides 19415 to 19441 of GK18). The product was then labeled with [³²P]dCTP and used as a probe for Southern blot hybridization. Five microliters of the amplified RT-PCR products described above was separated by electrophoresis on a 1% agarose gel, blotted onto nylon membrane (Hybond N⁺) by alkaline transfer, and fixed by exposure to UV radiation (UV cross-linker; Stratagene). Hybridization and washing were carried out essentially as described for the Northern blot analysis, except that the temperature was increased to 65°C.

Expression of the putative cytoplasmic domain of the membrane protein and reactivity with sera from patients.

The coding region for the hydrophilic domain of exon 8 (see Fig. 1) of the GK18/BCP-1 transcript was obtained by PCR amplification with primer pair MBP8afor (5′-ACGGGATCCGTAAATAGTTACCGACAGAGAAGG-3′; nucleotides 19799 to 19822 of GK18) and MBP8arev (5′-GCAAAGCTTCTAGTTCCTGGGAAATAAAACCTCCTC-3′; nucleo-tides 19415 to 19441 of GK18). The PCR product was cloned into the prokaryotic expression vector pMALc2 (New England Biolabs), resulting in the construct pMAL8A. This construct encodes a fusion protein of maltose binding protein (MBP) and the predicted cytoplasmic domain encoded by exon 8. Recombinant, bacterially expressed protein was affinity purified on amylose agarose. To investigate the reactivity of the recombinant protein with sera from patients by Western blotting, the recombinant fusion protein was cleaved with factor Xa, separated on SDS–15% polyacrylamide gels, and transferred onto Hybond nitrocellulose membrane by using an electrotransfer apparatus (Bio-Rad). After being blocked with phosphate-buffered saline (PBS) plus 0.1% Tween-20 (PBS-T) containing 5% nonfat dry milk (PBS-M) for 1 h, membranes were probed with sera from KS patients (1:300 in PBS-M) for 1 h at room temperature. The blots were washed three times (10 min each) in PBS-T at room temperature. Bound sera were detected by incubation for 1 h at room temperature with horseradish peroxidase-conjugated anti-human immunoglobulin G (IgG) (1:1,000 in PBS-M containing 1% goat serum), followed by enhanced chemiluminescence (ECL) as recommended by the manufacturer (Amersham).

Expression of KSHV membrane protein in mammalian cells and subcellular localization.

A genomic construct, containing all eight exons and intervening introns and corresponding to nucleotides 21530 to 19438 of GK18 (databank accession no. AF148805) and nucleotides 1578 to 3652 of a previously reported isolate (53; GenBank accession no. U85269) was generated by PCR from BCP-1 DNA by using primers ex1bfor (5′-CTAGGATCCCAACTCTATTGTAAGCCCTGTGGATACCTAGTC-3′; nucleotides 21494 to 21526 of GK18) and LAMPamyc (5′-GATCTCGAGTCACAGATCCTCTTCTGAGATGAGTTTTTGTTCGTTCCTGGGAAATAAAACCTCCTC-3′; nucleotides 19418 to 19441 of GK18) and cloned into pCDNA3.1 (Invitrogen). Similarly, an expression construct for the largest transcript, containing all eight exons, was generated by RT-PCR by using the same primers. The second of these primers introduced the c-myc epitope tag QEKLISQQDL (the nucleotide sequence in the primer is underlined), recognized by monoclonal antibody 9E10, at the C-terminal end of the cytoplasmic domain. Both constructs were transfected into 293 cells by using Lipofectamine (Life Technologies) as described by the manufacturer, and transfected cells were plated onto chamber slides. Forty-eight hours after transfection, the cells were fixed directly onto the slides by washing with PBS followed by immersion in 4% paraformaldehyde for 10 min, permeabilized by immersion in 0.2% Triton X-100 for 10 min, and blocked by immersion in 100 mM glycine for 10 min. Prior to immunofluoresence, the coverslips were immersed in 10% FCS for 10 min. Monoclonal antibody 9E10 (2 μg/ml in 2% FCS) was used to stain the fixed cells for 1 h. Cells were washed extensively in PBS and then reacted with fluorescein isothiocyanate (FITC)-conjugated antihuman IgG (1:100 in 2% FCS) for 1 h. Cells were washed extensively and viewed by confocal microscopy.

Generation of recombinant GST-fusion proteins and in vitro binding assay.

Recombinant glutathione S-transferase (GST)-fusion proteins were generated by PCR amplification of the hydrophilic region of the cytoplasmic domain of LAMP from BCP-1 by using primers GST8afor (5′-CAGGGATCCTAAATAGTTACCGACAGAGAAGGGG-3′) and GST8arev (5′-ATACCCGGGCTAGTTCCTGGGAAATAAAACCTCTC-3′), followed by cloning into pGEX-3X (Pharmacia). Recombinant proteins from 10 ml of cultures were adsorbed onto 50 μl of glutathione beads for 1 h at 4°C, after which the beads were washed twice in PBS, followed by washing once in 1% Nonidet P-40 (NP-40) lysis buffer (50 mM Tris-HCl [pH 8.0], 150 mM NaCl, 1 mM EDTA, 1% NP-40).

HEK 293 cells (8 × 10⁵) were transfected with FLAG-tagged cDNA expression constructs of TRAF1, TRAF2, TRAF3 (55, 59), kindly provided by Mike Rothe (Tularik Inc.), or p45-FLAG, a control construct generated from a novel cytoplasmic protein (2a), by using Lipofectamine according to the manufacturer’s instructions. Forty-eight hours after transfection, cells were lysed in 1% NP-40 lysis buffer (1 ml/10⁷ cells) containing protease inhibitors (50 μM leupeptin, 1 μM pepstatin A, 200 μM benzamidine, 100 U of aprotinin/ml, 1 mM phenylmethylsulfonyl fluoride) for 15 min on ice. Cell lysates were centrifuged at 14,000 × g for 15 min at 4°C and precleared by incubating with 50 μl of glutathione beads for 1 h at 4°C. Cleared cell lysates were then incubated for 2 h at 4°C with either GST control protein or fusion protein immobilized on 20 μl of glutathione beads, prepared as described above. The beads were then washed three times in 1% NP-40 lysis buffer. Proteins adsorbed to beads were boiled in Laemmli buffer and analyzed by SDS-polyacrylamide gel electrophoresis (10% polyacrylamide) and transferred to nitrocellulose. The resulting blots were analyzed by ECL-Western blotting as described above by using mouse monoclonal antibody M2 (Sigma) which recognizes the FLAG epitope at a concentration of 3 μg/ml, followed by an antimouse antibody.

Nucleotide sequence accession numbers.

The sequence of the positive cosmid clone (cos83) of the KSHV genome from a classic KS tumor biopsy (GK18) was deposited in GenBank under accession AF148805.

RESULTS

Sequence analysis of KSHV from a case of classic Kaposi’s Sarcoma.

A cosmid clone (cos83) of the KSHV genome from a classic KS tumor biopsy (GK18) was sequenced. This clone contains an insert of 28,559 kb that extends from a Sau3A site in ORF 68 (position 115304 in the HBL-6 sequence [56]) to a Sau3A site in a copy of the terminal repeat (TR). A comparison of the right end of this clone to the corresponding regions of the prototypic HBL-6 sequence (56) and another previously reported sequence (53) demonstrated near identity to the latter but a high degree of sequence divergence to the HBL-6 sequence (Fig. 1). While a putative open reading frame, K15, had been identified in the HBL-6 sequence (56), no similar reading frame could be assigned in GK18 (Fig. 1). GK18 and HBL-6 are highly homologous up to the 5′ end of ORF 75, but the region between ORF 75 and the TR is highly divergent and also shorter than that of HBL-6. The same observation was recently reported (53) for the KSHV sequence BCBLR (accession no. U85269). Sequencing of several cDNA and genomic clones generated by PCR from another PEL cell line (BCP-1; see below) showed that the BCP-1 sequence is also nearly identical to GK18 and the sequence reported by Nicholas et al. (53).

Northern blot analysis reveals the presence of two transcripts expressed in unstimulated PEL cell lines.

To investigate the existence of transcripts that originate from the region between ORF 75 and TR, we used a 4.3-kb probe extending from an EcoRI site within ORF 75 (position 132227) to an EcoRI site close to the TR of GK18 (this site is not present in the HBL-6 genome; Fig. 1) to probe Northern blots of latent and chemically induced HBL-6 and BCP-1 RNA. We identified two large transcripts, of approximately 4.5 and 7.5 kb (Fig. 2). The 7.5-kb band was present in HBL-6 and BCP-1 RNA from unstimulated cells but appeared to be upregulated in chemically induced cells, therefore representing a class II transcript as defined by Sarid et al. (58). The 4.5-kb message was present in unstimulated cells and was not upregulated following chemical treatment; it therefore corresponds to a class I transcript.

FIG. 2.

(a) Northern blot analysis of two KSHV cell lines. A 4.3-kb genomic probe generated from GK18 (Fig. 1) was used to probe total cellular RNA as described in Materials and Methods. Total RNA isolated from the RAJI cell line was used as a negative control. This probe detects both the 7.5-kb class II K15/LAMP transcript and the class I 4.5-kb LT3 transcript (see text and reference 58). The equal intensity of the (noninducible) class I transcript serves as a loading control. (b) The same blot was analyzed with a probe generated from exons 2 to 5 of the HBL-6 version of the spliced transcript described here which detects only the 7.5-kb K15/LAMP transcript (see text). I, induced; U, uninduced.

Mapping of transcripts from the right end of the KSHV genome.

The same 4.3-kb probe was used to screen a cDNA library derived from latent (uninduced) HBL-6 mRNA (54) in order to identify transcripts from this region of the genome. A cDNA clone covering the 5′ end of ORF 75 and most of the region between ORF 75 and the TR was isolated and found to lack the region between nucleotides 136456 and 136542. The presence of canonical splice donor and acceptor sites on either side of this deletion suggested the existence of an intron (Fig. 1). However, its unusual polyadenylation site (nucleotide 134270, within ORF 75) suggested that it might represent an incompletely processed and aberrantly polyadenylated mRNA, and it was therefore not pursued further.

Following the indication that splicing may have been involved in generating the mRNA represented by this clone, we inspected the HBL-6 and GK18 genomic sequences for conserved potential splice donor and acceptor sites. We looked for consensus splice donor and acceptor sites that allowed expression of a sizeable open reading frame (initially on either strand) and were helped by the striking similarity that emerged between the predicted splicing patterns in both the GK18 and HBL-6 sequences in terms of exon number, size, and arrangement. Table 1 lists predicted splice donor and acceptor sites in the GK18 and HBL-6 sequences that were subsequently confirmed by RT-PCR (see below). Since the cDNA clone obtained from the HBL-6 library extended into ORF 75, we carried out RT-PCR by using an antisense primer (LAMPcrev) inside ORF 75 and a sense primer upstream of the splice donor found in this cDNA clone (LAMPcfor). Cloning and sequencing of the products confirmed all the predicted splice donors and acceptors (Table 1) as well as identifying an alternative splice acceptor in exon 3 (Table 1). Additional PCR primers (ex1cfor, HBL-6; and ex1afor, GK18) were then employed in RT-PCR in combination with LAMPcrev, HBL-6, and LAMParev GK18 (Fig. 1) to locate the beginning of this open reading frame. Sequencing of RT-PCR products indicated that the first in-frame start codon is at position 136771 for the HBL-6 sequence and at position 21493 for GK18. We also identified an additional T residue at position 136608 in the published HBL-6 sequence that placed the ATG at position 136771 in frame with exon 1 (Fig. 1). In-frame stop codons were found in RT-PCR products upstream of these ATG codons at position 21562 for GK18 and position 136850 for HBL-6, indicating that these ATG codons represent the beginning of this open reading frame.

TABLE 1.

Positions of splice donor and acceptor sites in the K15 region of HBL-6 and GK18^a

Cell line	Exon	Splice acceptor		Splice donor
		Sequence	Position	Sequence	Position
		YYYYYYYYYYYNYAGGb		MAGGTRAGTb
HBL-6	1			TTGGTAAGT	136536–136544
GK18	1			TTGGTAAGT	21268–21276
	1			CTGGTATGTc	21445–21437
HBL-6	2	TGTCTTTTGTTTTAGG	136456–136471	CTGGTAAGT	136358–136366
GK18	2	CATTTGTTTTTATAGC	21164–21179	CAGGTAGGT	20142–20150
HBL-6	3	GTATGTGTTTTTCAGC	136267–136282	GTGGTGGTA	136019–136027
	3	ATATCTTTGCAACAGAc	136072–136087	GTGGTGGTA	136019–136027
GK18	3	TTTTGTATTTTATAGC	20988–21003	CAGGTAAGT	20740–20748
HB-6	4	TATTTTTTACTACAGG	135939–135954	TTGGTATAT	135844–135852
GK18	4	TTATCTTTTTTATAGG	20657–20672	TGGGTACAG	20562–20570
HBL-6	5	ATAACTTTATTACAGG	135762–135777	AAGGTATGT	135601–135609
GK18	5	CGACATTTTTTGTAGG	20490–20505	AAGGTTTGT	20332–20340
HBL-6	6	ATTTTTTATTTACAGG	135512–135527	TGGGTAAGG	135399–135407
GK18	6	TTTATGTATAAACAGG	20255–20270	CAGGTAGGT	20142–20150
HBL-6	7	GGTTTATTTCCTTAGC	135326–135341	TAAGTGAGT	135219–135227
GK18	7	TCCTCTATTTTTTAGC	20065–20080	TCAGTAAGT	19958–19966
HBL-6	8	ATTTTTTCTTATTAGG	135144–135159
GK18	8	TATTTTAAAATTTAGG	19878–19893

^aThe sequences are listed in the orientation of the coding strand, i.e., in the opposite orientation to the genomic sequence. Intron/exon boundaries are in bold type. 

^bConsensus, sequence of splice acceptors and donors in the K15 gene. 

^cAlternative, splice acceptor/donor (see text). 

Of several primers located further upstream in the GK18 sequence, LRH6for (nucleotides 21663 to 21682) was the most upstream oligonucleotide with which this transcript could be amplified. Thus, this transcript may start within the 81 bp of sequence between the TR and position 21682 in GK18. Alternatively, a transcriptional start could be envisaged at the other end of the viral genome and involve splicing across the TR to a putative splice acceptor between the TR and the position of LRH6for (Fig. 1). However, we have not detected the 7.5-kb band corresponding to this spliced transcript (data not shown) on Northern blots of BCP-1 and HBL-6 by using probes from nucleotides 68 to 1344 and 1903 to 4914 (HBL-6 sequence [56]). Any additional upstream exons would be expected to be noncoding in view of the in-frame stop codons at positions 21562 for GK18 and 136850 for HBL-6 prior to the predicted start codons. Experiments to more accurately determine the 5′ end of this transcript (primer extension analysis and 5′ rapid amplification of cDNA ends) are in progress.

Cloning and sequencing of RT-PCR products also revealed that multiple alternatively-spliced transcripts are generated from this gene (Fig. 3A). In untreated BCP-1 and HBL-6, the most abundant transcript contained eight exons (Fig. 3A). Alternatively-spliced forms containing fewer exons were also found (Fig. 3A) but were less abundant in uninduced cells, as shown by Southern blot of RT-PCR products (Fig. 3A), than the eight-exon version. In addition to the examples shown in Fig. 3A, a number of other multiply-spliced transcripts were also identified by sequence analysis of cloned RT-PCR products (e.g., exons 1, 5, and 8 or 1 and 8 for HBL-6; exons 1, 2, 3, 6, 7, and 8 or 1, 6, 7, and 8 for BCP-1). In contrast, induction of lytic viral replication by TPA (BCP-1) and Na-butyrate (HBL-6) led to an increase in a transcript lacking both exons 4 and 5 (Fig. 3A and data not shown). As for uninduced cells, fewer more-extensively-spliced forms were found by cloning and sequencing of PCR products (e.g., exons 1, 4, 5, 6, 7, and 8 or 1, 2, 5, 6, 7, and 8 for HBL-6; 1, 2, 3, 5, 6, 7, and 8 for BCP-1). One of these less-abundant splice forms, found only in HBL-6 cells, used an alternative splice acceptor in exon 3 (Table 1).

FIG. 3.

Relative abundance of spliced mRNAs. (A) Southern blot of RT-PCR products of BCP-1. RNA was reverse transcribed and amplified by using primers ex8arev and ex1afor. Resulting products were analyzed by Southern blotting with a probe derived from exon 8 and were also cloned into pGEM-T and sequenced to obtain the splicing patterns. Lanes 1 and 2, TPA-treated BCP-1 cells, lanes 3 and 4, untreated BCP-1 cells; lane 5, genomic BCP-1 DNA; lane 6, water control. RT was added during first-strand synthesis in lanes 1 and 3 but not in lanes 2 and 4. The positions of the two most abundant transcripts containing all eight exons (i) or lacking exons 4 and 5 (ii) are indicated, as are two minor transcripts lacking exons 2, 3, and 5 (iii) or exons 2, 3, 4, and 5 (iv). The splicing patterns of these transcripts are shown in the lower panel. (B) Southern blot of RT-PCR of BCP-1 showing alternative splice within exon 1. cDNA was synthesized from uninduced BCP-1 RNA by using primer LAMParev, followed by PCR amplification of the region encompassing exons 1 and 2 with primers LRH4rev and ex1afor. Lane 1, BCP-1 DNA; lanes 2 and 3, BCP-1 RNA, respectively, with and without RT during first-strand synthesis. The conventional splice yields a product of 294 bp (i), while the alternative splice yields a product of 115 bp (ii), and unspliced RNA yields a product of 404 bp. The splicing patterns of the conventional splice (i) and the alternative splice (ii) are shown in the lower panel. Because of the locations of the primers used, we cannot infer which exons downstream of exon 2 are present in transcripts utilizing this alternative splice site. Hence, these are indicated by a dotted line in the lower panel.

We also found evidence for an alternative splice donor within exon 1 of the BCP-1 sequence (position 21443 of GK18). As shown by a Southern blot of LAMParev-primed cDNA products amplified with primers flanking exons 1 and 2 (ex1afor, LRH4rev), this splice event (corresponding to the 115-bp band in Fig. 3B) was less abundant than the one involving the donor at the end of exon 1 (position 21276; GK18) (294-bp band in Fig. 3B), was present in uninduced as well as induced BCP-1 cells (data not shown), and would be predicted to join the first part of exon 1 out of frame to the remainder of the transcript, thus presumably leading to the use of an alternative start codon at position 21522. This alternatively-spliced first exon was found in RT-PCR-generated cDNA clones primed with oligonucleotide LAMParev located within ORF 75 (Fig. 1 and 3B) and is thus present in transcripts containing at least exons 1, 2, and 8. As discussed below, this observation is somewhat reminiscent but different from the alternative use of a first exon in EBV LMP2A and 2B.

To define which of the 4.5- and 7.5-kb bands identified by Northern blot (Fig. 2a) corresponds to this family of transcripts, a probe was generated from exons 2 to 5 of HBL-6 by PCR and used to reprobe the blot. The probe was shown to specifically bind to the 7.5-kb transcript (Fig. 2b) in HBL-6. This probe did not cross-hybridize to BCP-1 RNA, as would be expected in view of the high degree of sequence divergence that exists at this end of the genome. This experiment confirmed our initial observation that the transcript is present in unstimulated cells but is upregulated following chemical induction and thus corresponds to a class II transcript. Additional Northern blots with probes derived from the 3′ end of ORF 75 (data not shown) indicate that the 7.5-kb mRNA extends through ORF 75. It may use a polyadenylation site immediately downstream of ORF 75 at position 130510 (56), as we isolated several cDNA clones extending into ORF 75 from our cDNA library which used this polyadenylation site (data not shown).

Predicted protein sequence encoded by this group of transcripts.

Inspection of the predicted protein sequences (Fig. 4) indicated that they contain 12 transmembrane domains followed by a hydrophilic, presumably cytoplasmic, C-terminal domain. The predicted protein sequences of the BCP-1, GK18, and the HBL-6 viruses were found to be very divergent, although the overall structure as well as some characteristic sequence motifs, involving YxxL/V/I and proline-rich regions within the hydrophilic domain of the protein, are conserved (Fig. 4).

FIG. 4.

Sequence alignment of LAMP proteins. Protein sequence alignment of the 8-exon form of LAMP from the two KSHV sequence variants represented by BCP-1/GK18 and HBL-6. The position of predicted membrane-spanning domains is indicated by dotted lines above and below the sequence, and that of individual exons is marked above the sequence by arrows. Conserved sequence motifs reminiscent of YxxI/L motifs in LMP-2A are double underlined, and the putative TRAF binding site reminiscent of CTAR-1 of LMP-1 is shown in bold (see text).

The hydrophilic C-terminal (putative cytoplasmic) domain is encoded by exon 8 and is common to all alternatively spliced transcripts (Fig. 3A). The more-abundant alternatively-spliced form, which lacks exons 4 and 5, would be predicted to encode a protein with eight transmembrane domains (Fig. 4) and more-extensively-spliced (but less-abundant) forms would have correspondingly fewer transmembrane regions.

Some of the sequence motifs conserved in both the GK18 and HBL-6 versions of this membrane protein are reminiscent of related sequences in the cytoplasmic domain of EBV-LMP2A. Thus, the KSHV motif YASIL resembles the EBV LMP2A motifs YPSA, YQPL, and YLGL, which all have homology with the group III tyrosine motifs of the general pattern Y-hydrophobic-X-hydrophobic (63, 64). The conserved YEEVL motif of KSHV is similar to the src phosphotyrosine kinase SH2 binding motif YEEI (YEEA in LMP2A). In addition, the presence of 12 transmembrane-spanning domains in the longest protein and its location at the right end of the viral genome (in the orientation generally used for gamma-2 herpesviruses) is also reminiscent of EBV LMP2A.

In contrast, the fact that the hydrophilic (presumably cytoplasmic) domain is located at the carboxy-terminal end in the case of the KSHV membrane protein and that some splice variants would encode a protein with only eight transmembrane domains is more reminiscent of EBV LMP1. In addition, the HBL-6 sequence contains a PFQPADE motif in the cytoplasmic domain of this protein that is reminiscent of the TRAF binding domain within CTAR-1 of LMP1 (PQQATDD; identical sequences are underlined [29]). The equivalent motif in the GK18/BCP-1 sequence is ATQPTDD (residues identical with LMP1 CTAR-1 are underlined).

Reactivity of the recombinant cytoplasmic domain with sera from patients.

A recombinant protein was generated consisting of MBP fused to the C-terminal hydrophilic cytoplasmic domain (i.e., amino acid sequences VNSY to the stop codon; nucleotides 19821 to 19415 of GK18) (Fig. 4). Sera from patients with classic and AIDS KS were screened on Western blots for the presence of antibodies to this fusion protein (Fig. 5). To be certain of the specificity of this reaction, the cytoplasmic domain was cleaved from MBP by factor Xa digestion (Fig. 5). We found that some KS patients (2 of 20 tested) reacted with the 57-kDa MBP-exon 8 fusion and the cleaved 15-kDa protein (Fig. 5) corresponding to the last 150 amino acids. This result suggests expression of this gene in at least some individuals. A similar result was obtained by using a recombinant HBL-6 exon 8 fusion protein in that only approximately 10% of sera samples tested reacted positively to the fusion protein (data not shown).

FIG. 5.

Western blot of recombinant cytoplasmic terminal domain of LAMP. A recombinant MBP-exon 8 fusion of BCP-1 was expressed in Escherichia coli, affinity purified, and reacted with sera from a KS patient as described in Materials and Methods. Lane 1, reactivity of a serum from a patient with classic KS with the uncleaved fusion protein, the expected size of which is 57 kDa. Lane 2, reactivity of the same serum with factor Xa-digested protein. The expected size of the free cytoplasmic domain of LAMP is 15 kDa. The same serum did not react with control MBP protein (lane 3) or with control protein plus factor Xa (lane 4). A serum sample from a healthy subject was used as a control on uncleaved fusion protein (lane 5), fusion protein plus factor Xa (lane 6), control MBP (lane 7), or MBP plus factor Xa (lane 8). Molecular size markers (in kilodaltons) are shown on the left, and the position of the uncleaved MBP-fusion protein and the cleaved C-terminal domain of K15/LAMP are indicated by arrowheads.

Expression of the KSHV membrane protein in 293 cells and subcellular localization.

Expression vectors containing the genomic sequence (i.e., all eight exons together with intervening introns) and a cDNA sequence corresponding to all eight exons were constructed with a c-myc tag at the C-terminal end of the cytoplasmic domain. After transfection into 293 cells, expression of the tagged protein was examined by immunofluorescence by using monoclonal antibody 9E10 that recognizes the c-myc epitope. This experiment indicated a membrane protein that formed medium to large patches on both external and cytoplasmic membranes (Fig. 6). The same subcellular localization was observed with both the cDNA and genomic constructs (Fig. 6).

FIG. 6.

Transfection of K15/LAMP expression constructs into 293 cells. Immunofluorescence assay of 293 cells transfected with mammalian expression constructs for K15/LAMP. 293 cells were transfected with either a genomic (A) or a cDNA (B) myc-tagged construct of the ORF for the membrane protein, and expression was analyzed by immunofluorescence with an anti-myc antibody on permeabilized cells.

In vitro binding of TRAF1, TRAF2, and TRAF3 to the carboxy-terminal domain of K15/LAMP.

Sequence analysis of the K15/LAMP protein suggested the presence of a TRAF binding domain similar to those found in EBV LMP1 (see above). To explore the potential for K15/LAMP to bind to members of the TRAF family of proteins, HEK 293 cells were transiently transfected with tagged cDNA expression constructs for TRAF1, TRAF2, and TRAF3, and detergent (NP-40) extracts were used for pull-down experiments by using a GST fusion protein with the C-terminal cytoplasmic domain of both GK18/BCP-1 and HBL-6, identical to the one used in the MBP-fusion protein (Fig. 5; see Materials and Methods). TRAF1, TRAF2, and TRAF3 were found to bind to K15/LAMP fusion proteins (Fig. 7 and data not shown). These results indicate that the carboxy-terminal domain of this KSHV membrane protein, like that of EBV LMP1, interacts with TRAF1, TRAF2, and TRAF3. Similar results were obtained for both the BCP-1 and the HBL-6 K15/LAMP carboxy-terminal domain proteins.

FIG. 7.

In vitro binding of TRAF1, TRAF2, and TRAF3 to the carboxy-terminal domain of K15/LAMP. FLAG-tagged TRAF1, TRAF2, or TRAF3 expression vectors were transiently transfected into HEK 293 cells, and the interaction between the carboxy-terminal domain of K15/LAMP of GK18/BCP-1 and the TRAF proteins was examined by pull-down assays as described in Materials and Methods. Lanes 1 to 3, TRAF1 transfected cells; lanes 4 to 6, TRAF2-transfected cells; lanes 7 to 9, TRAF-3-transfected cells. Lanes 1, 4, and 7, proteins bound to GST-LAMP cytoplasmic domain (GK18/BCP-1); Lanes 2, 5, and 7, proteins bound to GST; lanes 3, 6, and 9, total cell lysates. Similar results were obtained with GST-C-terminal fusion protein of the HBL-6 sequence (data not shown).

DISCUSSION

The results presented here demonstrate the presence between ORF 75 and the TR of a family of alternatively spliced KSHV transcripts of approximately 7.5 kb that encode products with features reminiscent of both EBV LMP2A and LMP1. The proteins encoded by this family of spliced transcripts contain up to 12 transmembrane domains, followed by a hydrophilic C terminus that is presumably cytoplasmic. This hydrophilic region contains several tyrosine-rich motifs and a proline-rich motif, which are conserved between two otherwise highly divergent KSHV sequences, represented here by GK18/BCP-1 and HBL-6. Some of its features, in particular the presence of 12 transmembrane domains in the largest protein and several sequence motifs in the cytoplasmic domain (see Results) as well as its location in the viral genome, are reminiscent of EBV LMP2A. However, other features, such as the position of the cytoplasmic domain at the C-terminal end, the fact that some spliced variants are predicted to encode proteins with fewer transmembrane segments, the presence of sequence motifs with homology to known TRAF binding sites, and the interaction of the cytoplasmic domain with members of the TRAF family in GST pull-down assays are reminiscent of the LMP1 protein of EBV (21–23, 49). It appears that this family of KSHV membrane proteins may combine features of both LMP2A and LMP1.

Most of the coding regions for LMP1 and LMP2A/B are located on opposite ends of the EBV genome (32). Both are transcribed from the same bidirectional promoter located at the left end of the viral genome in the orientation generally adopted for gamma-2 herpesviruses (1, 56), and expression of LMP2A involves a splicing event across the terminal repeats. We have so far not found any evidence (from Northern blots) for the 7.5-kb mRNA encoding this protein family being initiated at the left end of the viral genome (data not shown). Detailed mapping of the 5′ end of this 7.5-kb mRNA is currently in progress. The 7.5-kb mRNA extends through ORF 75 and is likely to use a polyadenylation site at position 130510 (56).

The eight exons between ORF 75 and the TR are positioned such that any combination of exons found in our study would be in the same open reading frame. The only exception to this rule is the alternative splice donor in exon 1 (position 21443 of GK18), which would be predicted to join a part of exon 1 out of frame with the other exons, resulting in an alternative six amino acids at the N-terminal end of the resulting protein. This observation is somewhat reminiscent but different from the alternative use of a first exon by LMP2A and -2B; while the first exon of LMP2A encodes the N-terminal cytoplasmic domain of this protein, the first exon of LMP2B is noncoding (57).

The expression of LMP2A in EBV-negative B-lymphoma cell lines prevents normal calcium mobilization which follows surface Ig (sIg) cross-linking (45). More specifically, LMP2A interacts with the Src family of protein tyrosine kinases (PTKs) (9) and the Syk PTK (43). Important to the functioning of LMP2A is an immuno-receptor-like tyrosine activation motif. This motif consists of paired tyrosine and leucine residues and is important for binding and subsequent activation of SH2-containing proteins following sIg receptor stimulation (5, 11). It is thought that LMP2A may not be required for transformation of primary B lymphocytes by EBV (33, 37–39). EBV⁺ LMP2A⁺ lymphoblastoid cell lines (LCLs) are blocked in sIg-stimulated calcium mobilization, tyrosine phosphorylation, and lytic activation compared with EBV⁺ LMP2A⁻ LCLs (43, 44). It has therefore been suggested that LMP2A is essential for maintaining viral latency in EBV-infected LCLs. EBV LMP2A is, however, expressed in nasopharyngeal carcinoma tumor cells and is also one of the few EBV genes expressed in circulating B lymphocytes. In contrast, EBV LMP1 is essential for B-lymphocyte immobilization by EBV and considered one of the key EBV oncogenes (32, 71). The cytoplasmic domain of LMP1 contains two sites involved in downstream signaling, CTAR-1 and CTAR-2. CTAR-1 binds TRAF1, -2, and -3 and is involved in the activation of NF-κB-mediated gene expression, whereas CTAR-2 triggers NF-κB and C-Jun N-terminal kinase activation through interaction with TRADD and TRAF2 (21, 22, 29). KSHV has so far not been found to contain a homologue of LMP1. At the left end of the viral genome, where LMP1 is encoded, KSHV contains a type I transmembrane protein, K1, which has been shown to have transforming properties (35) but which appears to be expressed only after chemical induction of the lytic cycle and not during latency (34, 58). Another KSHV transcript (T0.7/kaposin/k12) has the potential to encode a membrane protein and is strongly expressed in latently infected PEL cell lines as well as in KS spindle cells (66, 73). It is, however, located in a different region of the viral genome (56), and although reported to have transforming properties (50), has no structural similarities to LMP1 (73).

The family of membrane proteins described here is expressed in unstimulated PEL cell lines (Fig. 2), including HBL-6, which is considered to be the cell line with the most-restrictive pattern of KSHV gene expression in the absence of chemical stimulation (58). Compared to other latent KSHV transcripts, such as T0.7/K12 (73), LT1 (LANA), and LT2 (v-cyc/v-FLIP) (58), the 7.5-kb mRNA encoding this family of membrane proteins shows only weak basal expression in PEL cell lines. This is illustrated in Fig. 2a by a comparison of the intensity of the 7.5-kb mRNA with the 4.5-kb mRNA, also detected by the 4.3-kb probe used at the beginning of this series of experiments, which most likely corresponds to the latent LT3 transcript described by Sarid et al. (58). The relatively weak expression of the 7.5-kb mRNA is also reminiscent of LMP2A.

In view of these similarities to the latent membrane proteins of EBV and its basal expression in unstimulated PEL cell lines, we suggest the name LAMPs for this family of proteins. We acknowledge, however, that basal expression of a KSHV gene in PEL cell lines does not necessarily imply its expression in tumor cells in vivo, as recently demonstrated for ORF K9 (vIRF-1) (27, 46a). Similar to EBV, KSHV may adopt different expression patterns in different tissues or tumors, and the expression of LAMP in vivo is currently under investigation. Our observation that at least some KS patients have antibodies to this protein suggests its expression in vivo. It is possible that more KSHV-infected individuals mount an antibody response to conformational determinants in the membrane-spanning domains (as is the case of LMP2A) than to linear epitopes in the cytoplasmic C-terminal domain investigated here.

In spite of the high degree of sequence divergence among the KSHV variants represented by HBL-6 and BCP-1/GK18 in the region between ORF 75 and the TR, a similar multiply-spliced gene is encoded in both, and several of the sequence motifs discussed above are conserved. In contrast, most of the KSHV genome, with the exception of ORF K1, is highly conserved among different isolates (51, 53, 56). While sequence divergence in K1 is under positive selection, as shown by the high ratio of nonsynonymous to synonymous nucleotide substitutions (16, 53), this is not the case for the K15/LAMP region. The fact that the GK18 sequence is virtually identical to BCP-1 and other previously reported sequences from this region (53) suggests that one of the two KSHV variants may have resulted from a recombination event with another closely related rhadinovirus. Others have come to a similar conclusion (53).

In summary, we describe a novel family of KSHV proteins with similarities to the latent membrane proteins of EBV. This characterization provides the basis for a functional comparison of LMP1, LMP2, and K15/LAMP and is likely to provide interesting new insights into the evolution of these virally encoded proteins with multiple spanning domains.

ADDENDUM IN PROOF

Similar independent observations are reported in this issue by Poole et al. (L. J. Poole, J.-C. Zong, D. M. Ciufo, D. J. Alcendor, J. S. Cannon, R. Ambinder, J. M. Orenstein, M. S. Reitz, and G. S. Hayward, J. Virol. 73:6646–6660, 1999) and have also been made by J. Jung and colleagues at Harvard University (personal communication).