Kaposi's Sarcoma-associated Herpesvirus K8 Exon 3 Contains Three 5′-Splice Sites and Harbors a K8.1 Transcription Start Site*

Kaposi's sarcoma-associated herpesvirus (KSHV) K8 and K8.1 open reading frames are juxtaposed and span from nucleotide (nt) 74850 to 76695 of the virus genome. A K8 pre-mRNA overlaps the entire K8.1 coding region, and alternative splicing of KSHV K8 and K8.1 pre-mRNAs each produces three isoforms (α, β, and γ) of the mRNAs. We have mapped the 5′ end of the K8.1 RNA in butyrate-induced KSHV-positive JSC-1 cells to nt 75901 in the KSHV genome and have shown that exon 3 of the K8 pre-mRNA in JSC-1 cells covers most part of the intron 3 defined previously and has three 5′-splice sites (ss), respectively, at nt 75838, 76155, and 76338. Selection of the nt 75838 5′-ss dictates the K8 mRNA production and overwhelms the RNA processing. Alternative selection of other two 5′-ss is feasible and leads to production of two additional bicistronic mRNAs, K8/K8.1α and -β. However, the novel bicistronic K8/K8.1 mRNAs translated a little K8 and no detectable K8.1 proteins in 293 cells. Data suggest that production of the K8/K8.1 mRNAs may be an essential way to control K8 mRNAs, especially K8α, to a threshold at RNA processing level.

Like other gamma herpesviruses, KSHV characteristically establishes latent infections with only a few genes being expressed in PEL-derived B cells. Five genes have been identified as latent: ORF K13 (ORF71, vFLIP), ORF72 (v-cyclin), ORF73 (LANA 1), ORF K10.5 (LANA 2), and ORF K12 (kaposin A) (8, 10 -15). Other genes are the inducible genes that can be expressed by induction of various chemicals such as tetradecanoyl phorbol acetate (TPA) and n-butyrate (16 -19). Induction by the chemicals initiates the viral lytic life cycle of KSHV in PEL-derived B cells, with the accumulation of progeny virus in culture medium (18,20). Although KSHV gene expression studies remain controversial, a number of studies, based on transcription kinetics, have classified KSHV genes into three categories during lytic infection induced by chemicals. These include immediate-early genes, early genes, and late genes (21,22). More recently, Jenner et al. (7) clustered the lytic genes as primary (0 -10 h), secondary (10 -24 h), and tertiary (48 -72 h) lytic genes according to its expression pattern by DNA arrays. A typical immediate-early gene is the KSHV ORF50 (Rta) that is expressed at 4 h of butyrate induction and is resistant to cycloheximide inhibition. An early gene is defined by its expression after 8 -20 h of chemical induction and its being sensitive to cycloheximide but resistant to antiviral drugs such as phosphonoacetic acid (PAA). KSHV genes encoding K3, K5, K8, vIL6, vMIPs, vGCR, vDHFR, and viral thymidylate synthase belong to this group. Late genes differ from early genes by their absence prior to 24 h of chemical induction and by their sensitivity to PAA. These include KSHV ORF65 and several viral envelope glycoproteins such as K8.1 (21). Although the molecular events in correlation with each category gene during KSHV lytic infection are just beginning to be understood, considerable evidence indicates that expression of each category gene is coordinately regulated and sequentially ordered in a cascade fashion in a viral lytic life cycle.
A KSHV locus encompassing ORF50 (Rta), K8, and K8.1 extends for about 5,400 nts at map position ϳ0.43-0.46 of the virus genome. Although the three genes are positioned in the virus genome side by side, they, respectively, belong to three different categories of genes involved in different stages of viral lytic cycle. KSHV Rta gene is an immediate-early gene and expresses a homolog to EBV Rta that mediates transcription of several early genes including the K8, ORF57, K2 (vIL6), and PAN RNA (23,24). A recent report (25) indicates that interaction of the N terminus of KSHV Rta with a 12-bp binding site, AACAATAATGTT, in both K8 and ORF57 promoters is necessary for transcriptional activation. Thus, expression of KSHV Rta is essential for lytic viral reactivation. The K8 gene is an early gene encoding a basic-leucine zipper (bZIP) protein called K-bZIP (22,26,27). The K-bZIP protein is also a homolog to EBV Zta, a transcriptional activator that plays a crucial role in the initiation of the EBV lytic cascade. Accumulated data show that KSHV K-bZIP is a nuclear protein probably related to viral DNA replication (27)(28)(29). Finding that KSHV K-bZIP forms a dimer in vivo (22) and undergoes extensive phosphorylation by cyclin-dependent kinase (30) as well as colocalizes with cellular promyelocytic leukemia protein (PML) in (PML oncogenic domains) along with KSHV core replication proteins (29,31) implies its roles in initiation of viral transcription and DNA replication. Evidence of the K-bZIP repressing transcription through interaction with p53 and CREB-binding protein further supports such a notion (31)(32)(33). KSHV K8.1 gene is a late gene encoding two related glycoproteins, K8.1A and K8.1B (34 -36). The K8.1A and -B have identical amino acid sequences in both N terminus and C terminus but differ at the central domain because of alternative RNA splicing. The K8.1 A (gp35-37) is a viral envelop protein interacting with cellular receptors heparan sulfate-like moieties (37). However, function of the K8.1B remains unknown.
Structural, functional, and expression analysis of individual viral genes have demonstrated that KSHV Rta, K8, and K8.1 genes share a common polyadenylation signal at nt 76714 downstream of the K8.1 coding region. KSHV Rta and K8 transcribe their RNAs from two separate promoters (22,24,30), whereas the K8.1 promoter and its transcription start site remain unknown. As a result, a pre-mRNA transcribed from the Rta promoter overlaps the K8 and K8.1 coding regions and thus is tricistronic with 5 exons and 4 introns. However, a pre-mRNA transcribed from the K8 promoter only overlaps the K8.1 coding region and thus is bicistronic with 4 exons and 3 introns. Chen et al. (38) recently showed that the Rta promoter p71560 is heavily methylated in latently infected cells, and demethylation of the Rta promoter by TPA treatment induces the KSHV lytic life cycle. KSHV K8 promoter p74845 identified by Lin et al. (22) has an Rta-binding site and can be directly activated by Rta (25). It has been documented that maturation of both Rta and K8 mRNAs undergoes extensive alternative splicing which leads to the generation of three spliced forms of the mRNAs (␣, ␤, and ␥) (22,27,30). In each case, a 3Ј-terminal intron is excluded from all isoforms of both Rta and K8. The KSHV K8.1 gene shares the 3Ј-terminal exon with Rta and K8 but utilizes a 3Ј-terminal intron of the Rta and K8 as its own coding region, which is also alternatively spliced to the same 3Ј-terminal exon of the Rta and K8 by using two alternative 5Ј-splice sites (ss) and subsequently produces the following two spliced mRNAs: K8.1␣ and K8.1␤ (34,35). RNA K8.1␥ is the one left unspliced.
Altogether, these findings suggest that alternative promoter usage and alternative splicing of the three primary transcripts are cascaded and play important roles in KSHV lytic reactivation. We are interested in regulation of KSHV RNA splicing in the virus life cycle. The structures of Rta, K8, and K8.1 pre-mRNAs and their sophisticated splicing provide an ideal model for understanding how a viral pre-mRNA recruits cellular splicing machinery to define its splice sites for removal of an unwanted intron. Our initial approach focuses on a 3Ј-terminal intron because it involves the processing of all three transcripts Rta, K8, and K8.1. Because of complexity of their RNA structures and splicing, however, we simplify our focus on description only on K8 and K8.1 RNAs for better understanding in this report. Here we provide evidence that 5Ј part of the K8 terminal intron is part of the K8 exon 3 (Rta exon 4) which extends to include two additional 5Ј-ss being used for splicing of the K8.1 pre-mRNA. Alternative selection of three 5Ј-ss in the K8 exon 3 results in several novel forms of bicistronic K8/K8.1 mRNAs. In addition, we also demonstrate that transcription of late K8.1 RNA is initiated at nt 75901, 14 nts upstream of the first AUG at nt 75915 in KSHV genome.

EXPERIMENTAL PROCEDURES
Cells-Human 293 cell was obtained from the American Type Culture Collection (ATCC, Manassas, VA) and was maintained in Dulbecco's modified Eagle's medium supplemented with 10% inactivated fetal bovine serum (HyClone), 2 mM L-glutamine, 100 units penicillin/ml, and 100 g streptomycin/ml at 37°C under a 5% CO 2 . The KSHV-positive B cell line JSC-1 cells (PEL-derived, coinfected with Epstein-Barr virus (EBV)) (20) was a generous gift from Dr. Richard Ambinder (The Johns Hopkins University), and the BCP-1 cell line (KSHV ϩ , EBV Ϫ ) (CRL-2294) was purchased from ATCC. Both cell lines were maintained at 37°C in RPMI 1640 medium (Invitrogen) containing 2 mM L-glutamine, 100 units of penicillin/ml, 100 g of streptomycin/ml, and 10 (JSC-1 cells) or 20% (CBP-1 cells) fetal bovine serum in the presence of 5% CO 2 . For chemical activation of cells, sodium butyrate (Aldrich) was added to a final concentration of 3 mM (27) and incubated for 24 h or at indicated time points as described in each figure. To inhibit butyrate-induced KSHV DNA replication, JSC-1 cells were simultaneously treated with PAA (0.4 mM) and butyrate (3 mM) for 48 h (21). For anti-IgM activation, JSC-1 cells at 1 ϫ 10 7 , after three washes with serum-free RPMI 1640, were resuspended in 1 ml of serum-free RPMI 1640, incubated with 10 g of anti-human IgM at 37°C for 30 min, and then fed with 9 ml of RPMI 1640 with 10% fetal bovine serum at 37°C in the presence of 5% CO 2 for another 30 min.
RNase Protection Assay (RPA)-The radioactive antisense RNA probes were prepared by in vitro transcription in the presence of [␣-32 P]GTP from XbaI-linearized plasmid pST22 DNA templates or from template DNAs prepared by PCR using an antisense chimeric T7/KSHV primer, Pr76136 (oST46, 5Ј-TAATACGACTCACTATAGGGA/ TAGTCTGGTCCCGTTTAGG-3Ј) combined with a sense primer Pr75679 or Pr75843 (oST44, 5Ј-GAAGGAATTGTTATCCGGCAGC-3Ј) for the experiment in Fig. 1, or using a sense primer Pr75605 combined with an antisense chimeric T7/KSHV primer Pr75935 (oST33, 5Ј-TA-ATACGACTCACTATAGG/GCGAATCTGTGTGGAACTCA-3Ј) for the experiments in Fig. 2, D and E. Human cyclophilin antisense [ 32 P]RNA probe prepared from a DNA template from Ambion was used as an internal control. Three nanograms of each probe (specific activity 35,000 cpm/ng) was hybridized overnight to 15 g of total JSC-1 RNA in hybridization buffer and then followed by RNase digestion with RPA III TM kit (Ambion) according to the manufacturer's instructions. Protected RNA fragments were separated by using a 6% polyacrylamide gel containing 8 M urea. Autoradiographic data captured with a Molecular Dynamic PhosphorImager Storm 860 were analyzed with ImageQuant software. A sequence reaction of pUC19 was also performed with a forward primer (5Ј-GTTTTCCCAGTCACGACGTTGTA-3Ј) and used as a sequencing ladder for sizing.
Mapping of KSHV K8.1 Transcription Start Site by Primer Extension Assay-Total cell RNA (10 g) extracted from JSC-1 cells treated with butyrate was used to map the K8.1 transcription start site by primer extension assay (Promega). The 32 P-labeled primer Pr75986 (oST64, 5Ј-GCACGCCACCAGACAAAGG-3Ј) was used for both primer extension and KSHV DNA sequencing to determine where the extended products were terminated. The primer extension products were analyzed along sequencing reactions on an 8% denaturing polyacrylamide gel.
Plasmid Construction and Transfection-To construct an expression vector containing both K8 and K8.1 genes (pST1), a 1.8-kb DNA fragment was amplified from total JSC-1 DNA by PCR using a primer Pr74850 (oST41, 5Ј-CACC/ATGCCCAGAATGAAGGAC-3Ј) combined with an antisense primer Pr76687 (oST42, 5Ј-TCC/AGGGTTTCTTAC-GCCG-3Ј). The amplified DNA fragment was inserted into an expression vector pcDNA3.1D/V5-His-TOPO (Invitrogen) at a polylinker region downstream of a CMV promoter. Plasmid pST22 containing an inverted KSHV fragment (nt 76136 to 75679) for the RPA assay in Fig.  1C was constructed by inserting into the same expression vector a fragment amplified by PCR using a primer pair of Pr75679 and Pr76136b (oST80, 5Ј-CACC/TAGTCTGGTCCCGTTTAGG-3Ј). To construct the plasmid (pST2) with a U to G mutation at 75838 5Ј-ss, a high fidelity PCR Taq enzyme (Roche Molecular Biochemicals) was used for overlapping PCR (39). Two primer sets, Pr74850 and Pr75857 (oST49, 5Ј-GATAACAATTCCTTCTTCCCACACAAAGTCTGGC-3Ј) as well as Pr75824 (oST48, 5Ј-GCCAGACTTTGTGTGGGAAGAAGGAATTGTTA-TC-3Ј) and Pr76687, were utilized to amplify PCR fragments, respectively. The amplified PCR products from each set of the primers were mixed and re-amplified using a primer set of Pr74850 and Pr76687. Following gel purification, the PCR fragment with a U to G mutation at 75838 5Ј-ss was inserted into the expression vector pcDNA-3.1D/V5-His-TOPO as described above. The same strategy was used to create plasmid pST5 derived from plasmid pST2 by overlapping PCR using two other primer sets, Pr74850 and Pr76180 (oST57, 5Ј-GATAC-GGAGACATTTGGATTGTCATTCGACGGAGAT-3Ј) and Pr76145 (oST-56, 5Ј-ACTCCGTCGAATGACAATCCAAATGTCTCCGTATC-3Ј) and Pr76687. The individual PCR products were then mixed, re-amplified by using a primer set of Pr74850 and Pr76687, and cloned into the same expression vector. The resulting plasmid pST5 had mutations in two 5Ј-ss, respectively, at nt 75838 (GU to GG) and at nt 76155 (GUGAGU to AUGACA). All mutations described in plasmids pST2 and pST5 were verified by sequencing. To construct a K8␤ cDNA plasmid (pST3), the RT-PCR fragment corresponding to K8␤ amplified from the JSC-1 total RNA using the primer set Pr74850 and Pr76687 was cloned. Plasmid pST6 was derived from plasmid pST1 by deletion of the CMV promoter with BamHI and SpeI. LipofectAMINE TM 2000 (Invitrogen) was used to transfect 2 g of plasmid DNA into 293 cells according to the manufacturer's instructions. Total cellular RNA was extracted from the transfected cells after 24 h.
Pre-mRNA Substrates and in Vitro Splicing-DNA templates used for in vitro transcription were obtained from plasmids pST1 and pST2 by PCR using a primer set of chimeric SP6/Pr75680 (oST47, 5Ј-ATTTAGGTGACACTATAGA/AGCGGCGAGATTGGAGGC-3Ј) combined with Pr76576 or Pr76576/U1 (oST50, 5Ј-GTACTCACCCC/TG-TAGTGCGCGTCTCTTCCTC-3Ј). In vitro transcription using SP6 RNA polymerase was performed in the presence of the cap analog (m7GpppG) and [␣-32 P]GTP. In vitro splicing of the transcribed pre-mRNAs and analysis of the spliced products were carried out as described in our previous publication (41).

Mapping of KSHV K8.1 RNA Transcription Start
Site-Several laboratories (22,26,42) have reported that the size of KSHV K8.1 mRNA isolated from chemically stimulated PELderived B cells is ϳ0.9 kb by Northern blot analysis. However, the exact 5Ј end of the transcript remains unknown. If the K8.1 does have its own promoter, the nascent transcript should have a 5Ј end distinguishable from that of the K8 and Rta transcripts. Otherwise, the K8.1 proteins must be translated from tricistronic Rta RNA or bicistronic K8 RNA by one of the following mechanisms: translation re-initiation, leaky scanning, or internal ribosome entry site. This seems unlikely because the published Northern blot results on the K8.1 mRNA do not support the latter scenario. Thus, locating the 5Ј end of the nascent K8.1 transcript will be very useful for our understanding on the K8.1 gene expression. To address this question, three K8.1 antisense RNA probes with different lengths of their 3Ј ends, but starting with the same 5Ј end at nt 76136 downstream of the K8.1 initiation codon AUG at nt 75915, were designed to cover K8.1 exon 1 (probe A in Fig. 1A) or both K8 exon 3 and K8.1 exon 1 as well as a putative junction region (probes B and C in Fig. 1A). RNase protection assays by using these probes were then performed to map the K8.1 transcription start site with total cell RNA isolated at different time points from JSC-1 cells induced by sodium butyrate. Results in Fig. 1, B and C, show that an RNA fragment with ϳ240 nts was protected from RNase digestion by all three probes. The protected RNA product appeared at 24 h of butyrate induction (lanes 3 and 9 in Fig. 1B and lane 2 in Fig. 1C) and reached a maximum at 48 h (lanes 4 and 10 in Fig. 1B and lane 3 in Fig.  1C), the latest time point of butyrate induction in this study. More importantly, the appearance of the protected 240-nt product induced by butyrate was sensitive to PAA inhibition (Fig.  1C, lane 4), suggesting that it was a transcript from a late gene and should represent a K8.1 transcript. Transcription of this product should be initiated within intron 3 of the K8 but upstream of the first K8.1 initiation codon AUG at nt 75915. By comparison with a sequencing ladder generated from pUC19, the protected RNA product was determined with a size of ϳ234 nts (data not shown). To map precisely the 5Ј end of the K8.1, primer extension analysis was further performed on those JSC-1 RNAs used for RPA with the 32 P-labeled antisense primer Pr75986. The same primer was also used in dideoxy DNA sequencing reactions on pST1 DNA, a KSHV K8 and K8.1 expression vector. Thus, the K8.1 transcription start site could be mapped by comparing the extended products with the DNA sequencing ladder. The data in Fig. 1D show that a product from butyrate-induced JSC-1 RNA (lanes 1 and 6) was extended and stopped at an A residue on the antisense strand at nt 75901. These results indicate that the K8.1 transcription is initiated at nt 75901, 14 nts upstream of the first AUG at nt 75915.
It was notable that expression of the K8 exon 3 (160 nts) protected by probes B and C also increased with butyrate induction (Fig. 1B, lanes 7-10, and Fig. 1C, lanes 2-4). However, its time course (Fig. 1B, lanes 7-10) was obviously different from that of the K8.1 mRNA. Importantly, expression of the K8 exon 3 was not sensitive to PAA inhibition (compare lane 4 to lane 3 in Fig. 1C), further confirming that the K8 is an early gene and the K8.1 is a late gene. In addition, we also found there were other protected fragments in the sizes of the fulllength probes A and B, suggesting that these RNAs retained intron 3 of the K8 (Fig. 1B, lanes 3 and 4 and 8 -10). The presence of this RNA species with the same sizes of probes A and B was further confirmed by using probe C (Fig. 1C, lanes 2  and 3). This was surprising because previous reports from several laboratories (22,24,27) have indicated that intron 3 of K8 is spliced out in all isoforms of Rta and K8 mRNAs. The expression profile of these RNAs with intron 3 retention in butyrate-induced JSC-1 cells (Fig. 1B, lanes 8 -10) was also different from that of K8.1 transcripts, but similar to the pattern of K8 exon 3; therefore, they might be a part of the K8 transcription products.
K8 mRNAs with Intron 3 Retention-Taking into account the juxtaposition and overlapping of the K8 and K8.1 in the KSHV genome, we assumed that the protected fragments with the size of a full-length probe in Fig. 1 might represent an mRNA containing both K8 and K8.1 open reading frames which has not been reported before. As a result, an antisense primer Pr75864 solely in intron 3 was paired with a sense primer Pr75183 to detect those potential RNAs from JSC-1 cell poly(A) ϩ RNAs by RT-PCR. As predicted, a major form of the mRNA with intron 3 retention was a ␤ form also having an intron 2 (Fig. 2B, lane 5). This amplification profile differed from detection of the K8 mRNAs from which K8␣ was overwhelming (Fig. 2B, lane 4). Interestingly, sodium butyrate treatment of JSC-1 cells not only increased K8␤ expression (Fig. 2C, lane 3) but also induced expression of the ␤ form RNA with retention of both intron 2 and 3 (Fig. 2C, lane 5), which was designated as a putative K8/K8.1 mRNA. Similar results were also observed in BCP-1 cells. The results were further verified by sequencing of the RT-PCR products amplified by using a primer pair spanning from nt 75605 to 75935 of the virus pre-mRNAs (data not shown) and by RPA using a probe spanning over part of K8 intron 2, the entire exon 3, and part of the intron 3 (Fig. 2, A and D).
Quantitative analysis of the protected products by RPA showed a molar ratio of K8␣, -␤, and the putative K8/K8.1 to be 10:2:1 in uninduced JSC-1 cells (Fig. 2D, lane 1). Butyrate treatment of JSC-1 cells significantly increased expression of all three forms of the RNAs (16-fold for the ␣) with a relative constant molar ratio changed to approximately 5:1:1. More-  1-4) or B (lanes 7-10) each mixed with human cyclophilin antisense RNA probe (165 nts). The latter generated a protected product with a size of 103 nts and was used as an internal control for normalization of sampling. The cyclophilin probe was also used as a reference for efficiency of RNase digestion. Yeast RNA used as a negative control was also hybridized with probe A (lanes 5 and 6) or B (lanes 11 and 12), but RNase digestion was omitted in lanes 6 and 12. C, total cell RNA (ϳ15 g) from JSC-1 cells treated with butyrate (3 mM) for 0, 24, and 48 h (lanes 1-3) or together with PAA (0.4 mM) for 48 h (lane 4) was evaluated by RPA as described in B by using probe C. The probe C resembled probe B but had, respectively, 59 and 49 nts of vector sequences on each end. Yeast RNA used for hybridization with the probes is shown, respectively, in lanes 5 and 6 with omission of RNase digestion in lane 6. D, mapping of K8.1 transcription start site by primer extension analysis. Total cell RNA extracted from JSC-1 cells induced by butyrate was used for the assay, using an antisense 32 P-labeled primer Pr75986 (lanes 1 and 6). The same primer was also used for DNA sequencing on pST1 DNA. Sequencing reactions of pST1 DNA (lanes 2-5) were run in parallel with primer extension products (lanes 1 and 6, lane 1 having one-fifth of loading volume of the lane 6) on an 8% denaturing polyacrylamide gel. Arrows to the gel indicate the extended products. The sequence of the antisense strand is shown at the right along with nucleotide positions in KSHV genome. over, a 20-fold increase of the K8␤ was also accompanied by about a 32-fold induction of the putative K8/K8.1 mRNA (Fig.  2, D and E, lanes 2). The expression profile of all three RNAs in JSC-1 cells following anti-IgM (a B cell activator) stimulation was similar to the results obtained from butyrate treatment. Compared with butyrate, anti-IgM was a much weaker inducer (Fig. 2D, lane 3) that led to an increased expression of all three species of the RNAs by close to 5-fold with a molar ratio of 10:2:1.
Nevertheless, our data demonstrate that a K8 transcript containing both introns 2 and 3 could be a novel K8/K8.1 mRNA that was not reported before. Interestingly, a predicted 291-nt band corresponding to a protected fragment with intron 3 retention but no intron 2 was out of a detectable level in those RPA experiments (Fig. 2, D and E), indicating that an RNA with intron 3 retention most likely had an intron 2. This was consistent with our RT-PCR results (lane 5 in Fig. 2B and lanes 4 and 5 in Fig. 2C). Moreover, K8␥ RNA with both introns 1 and 2 retention but with no intron 3 previously described in other cell lines (22,27)) was hard to detect by RT-PCR in JSC-1 cells (Fig. 2B, lane 4; Fig. 2C, lanes 2 and 3), indicating that K8 intron 1 removal was very efficient during RNA splicing. To confirm this assumption, several sets of the primers from exon 1 or intron 1 and intron 2 were designed to amplify this region by RT-PCR. Results from those studies showed that an RNA with intron 1 retention always had an intron 2, but an RNA containing an intron 2 generally had no intron 1 (data not shown). The data conclude that removal of intron 1 preceded that of intron 2.

. Identification of KSHV K8 mRNAs with intron 3 retention in JSC-1 cells.
A, diagram of K8 pre-mRNA structure (22,27,30). An antisense RNA probe (heavy solid line) spanning over exon 3 with its position of hybridization to a sense transcript is shown at the top of the diagram. The sizes (nts in parentheses) of the probe and its protected products depicted above the probe are also shown. Primers used for the RT-PCR assays shown in B and C are indicated by horizontal arrows below the pre-mRNA and are named by the locations of their 5Ј ends. Tailed primers Pr75183 and Pr75838 had, respectively, 21 and 11 nts (nonspecific) attached to each 5Ј end as indicated by a dashed line. Shown below the primers are the RT-PCR products predicted from two sets of primers used for amplification in B and C. See other descriptions in Fig. 1. B, RT-PCR analysis of poly(A) ϩ mRNAs. Total cell RNA isolated from JSC-1 cells was subjected to RNase-free DNase I digestion before poly(A) selection. The poly(A) ϩ mRNAs with or without reverse transcription were then amplified by PCR using a 5Ј primer combined with a 3Ј primer as indicated. C, RT-PCR of total cell RNA isolated from JSC-1 cells with or without butyrate treatment. DNase I digestion of the RNA and amplification strategy were the same as in B. D and E, RNase protection assays of total JSC-1 cell RNA. Total cell RNA isolated from JSC-1 cells with or without (control) butyrate or anti-IgM treatment was hybridized with 1 ϫ 10 5 CPM of 32 P-labeled antisense viral probe mixed with a human cyclophilin antisense RNA probe (165 nts). The latter generated a protected product with size of 103 nts and was used as an internal control for normalization of sampling. The cyclophilin probe was also used as a reference for efficiency of RNase digestion. Yeast RNA was used as a negative control. The protected fragments from RNase A and T1 digestion were resolved on 6% polyacrylamide/8 M urea gel. The identities of corresponding products protected are shown between D and E. The numbers to the left of the D and to the right of the E are the size markers of a 100-bp ladder. lanes 5 and 7). The primary substrate RNAs used to generate ␣ and ␤ forms of the K8/K8.1 RNAs must be the K8 pre-mRNAs because the K8.1 transcription initiates at the 5Ј end of intron 3, and the sense primers upstream of the intron were used in those experiments. However, we cannot role out a tricistronic Rta pre-mRNA in butyrate-induced JSC-1 cells also being used as a substrate for the splicing because a tricistronic Rta pre-mRNA transverses both K8 and K8.1 coding regions.
Mutation of the nt 75838 5Ј-ss in K8 Exon 3 Leads to Production of a K8/K8.1␤ mRNA in Vitro-To simplify the data interpretation complicated by the presence of tricistronic Rta pre-mRNAs in JSC-1 cells, we constructed several K8 minigene expression vectors with different 5Ј-ss mutation to verify if the three 5Ј-ss within exon 3 of a K8 pre-mRNA are alternatively selected. A single nucleotide mutation (U to G) was introduced into the nt 75838 5Ј-ss of a K8 minigene expression vector to convert the wt 5Ј-ss GU (plasmid pST1) to a mt 5Ј-ss GG (plasmid pST2). In vitro splicing of the pre-mRNAs transcribed from those templates showed that the RNAs containing a wt nt 75838 5Ј-ss efficiently spliced and generated a K8 mRNA (fully spliced at nt 75838 to 76433) with a size of 301 (pre-mRNA 1 in Fig. 4) or 312 nts (pre-mRNA 2 in Fig. 4) regardless of whether or not a U1-binding site was attached to their terminal exons (Fig. 4, compare pre-mRNAs 1 and 2). A U1-binding site attached to a pre-mRNA terminal exon was thought to function as a splicing enhancer (39,43). The pre-mRNAs with a mt nt 75838 5Ј-ss spliced poorly, and no K8 mRNAs were detectable from the splicing, indicating that the nt 75838 5Ј-ss was the most favorable 5Ј-ss being used for the splicing. However, another alternative 5Ј-ss, obviously a nt 76338 5Ј-ss downstream of the nt 75838 5Ј-ss in the mt pre-mRNAs, was selected for the splicing, leading to production of more K8/K8.1␤ mRNAs (fully spliced at nt 76338 to 76433) with a size of 801 (pre-mRNA 3 in Fig. 4) or 812 nts (pre-mRNA 4 in Fig. 4). Thus, our data demonstrated that a K8 pre-mRNA was indeed a substrate RNA responsible for generation of the novel K8/K8.1 mRNA.

Novel K8/K8.1␣ and -␤ mRNAs Are Spliced Products of K8 Pre-mRNAs by Alternative Selection of Three 5Ј-ss in K8 Exon 3 in Transient Transfected 293 Cells-Transfection of 293 cells
with plasmid pST1 (wt K8 minigene) (Fig. 5A) showed that splicing patterns of the K8 pre-mRNAs were exactly the same as seen in JSC-1 cells. Selection of the nt 75838 5Ј-ss in K8 exon 3 also overwhelmed other two alternative 5Ј-ss downstream (Fig. 5B, compare lanes 2 with 5). However, the RNAs with a mt nt 75838 5Ј-ss obtained from plasmid pST2 transfection, when compared with their wt counterparts, were unable to use this 5Ј-ss (Fig. 5B, compare lanes 4 with 8). Instead, these RNAs switched to select the nt 76155 and 76338 5Ј-ss for the splicing (Fig. 5B, compare lanes 4 and 5 with 8 and 9). Transfection of 293 cells with plasmid pST5 that had mutations in both nt 75838 5Ј-ss and 76155 5Ј-ss showed a similar splicing profile as seen in pST2-transfected 293 cells (compare lanes 11 and 12 with lanes 6 and 7 in Fig. 5B). Because plasmid pST5 had only one 5Ј-ss (nt 76338 5Ј-ss) left after double 5Ј-ss mutations, only two RT-PCR products, one from unspliced and another from spliced (lanes 13 and 14 in Fig. 5B), could be amplified from pST5-transfected 293 cell RNA. In contrast, there were three RT-PCR products (lanes 8 and 9 in Fig. 5B) detectable from pST2-transfected cell RNA due to a single 5Ј-ss mutation made at nt 75838. These data provided strong evidence that a bicistronic K8 pre-mRNA could be spliced through alternative selection of three 5Ј-ss within its exon 3. Among the three 5Ј-ss, selection of the nt 75838 5Ј-ss overwhelms the other two, leading to production of a K8 mRNA. Selection of the nt 76155 5Ј-ss is favored if a blockade of the nt 75838 5Ј-ss occurs, resulting in production of a K8/K8.1␣ mRNA. The nt 76338 5Ј-ss is the least favorable one and its selection generates a K8/K8.1␤ mRNA.
Furthermore, our results also showed that blockade of the nt 75838 5Ј-ss or together with the nt 76155 5Ј-ss not only promoted the K8 pre-mRNAs to use other 5Ј-ss downstream but also affected efficiency of intron 2 removal (Fig. 5B, compare  lanes 2 with 6 and 11), consequently leading to accumulation of the K8 RNA retaining intron 2. Data suggest that selection of the nt 75838 5Ј-ss might play an important role in removal of the K8 intron 2.
Novel K8/K8.1 mRNAs Are Poorly Translatable-Demonstration of the K8/K8.1 mRNAs derived from K8 pre-mRNAs through alternative selection of two 5Ј-ss previously identified for K8.1 mRNA processing also implies that the novel K8/K8.1 mRNAs should have coding potential for K8 and K8.1 and are bicistronic. It is predicted that a K8.1 protein will be expected to be translated in addition to K8 proteins in our transient transfection assay. Thus, Western blot analysis was further performed to look for both K8 and K8.1 proteins translated from the novel K8/K8.1 bicistronic mRNAs in several K8 minigene-transfected 293 cells by using anti-K8 or anti-K8.1 antibodies. Data in Fig. 5C showed that K8␣ (K-bZIP) but not K8␤ protein was efficiently translated in pST1-transfected 293 cells (lane 2). The K8␣ protein expression in 293 cells transfected with pST2 containing a mt nt 75838 5Ј-ss or with pST5 containing a mt nt 75838 5Ј-ss plus a mt nt 76155 5Ј-ss was greatly reduced along with appearance of a very small amount of K8␤ protein (lanes 3 and 5), even though a greater number of the bicistronic K8/K8.1 mRNAs existed in the cells (Fig. 5B, lanes  7-9 and 12-14). Moreover, several attempts were tried to detect K8.1 proteins and failed with negative results when the protein samples described above were probed with an anti-K8.1 antibody (Fig. 5D, lanes 2-6). Data suggest that the novel K8/K8.1 mRNAs are poorly translatable. This is very different from a bicistronic v-cyclin/vFLIP mRNA in which vFLIP is translated by using an internal ribosome entry site residing in the v-cyclin coding region (44 -46). DISCUSSION In this report, we mapped transcription start site for initiation of K8.1 transcription and found that K8.1 transcripts start at nt 75901, 14 nts upstream of the first AUG at nt 75915 in KSHV genome. By scanning the region 5Ј to the K8.1 transcription start site, we found no consensus TATA box but instead the presence of several potential transcription factor binding sites, including a TGACTCA-like site (AP-1) (47) at nt 75602, a CANNTG site , upstream stimulatory factor (48) at nt 75607, a CAAT (CP1, CP2, C/EBP, and ACF) site (49,50) at nt 75617, and a GC-like box (SP) (51) at nt 75766. This indicates that a potential K8.1 promoter is TATA-independent and lies in the body of K8 ORF. Nevertheless, implication of an existing K8.1 promoter upstream of the K8.1 transcription site mapped in this report suggests that expression of Rta, K8, and K8.1 genes is through alternative promoter usage in three stages (immediate-early, early, and late) of a lytic viral cycle. There are a number of reports suggesting that viral late promoters are generally TATA-independent promoters including KSHV TK promoter (52) and assembly protein promoter (53) as well as EBV late promoters (54), and their activities are associated mainly with viral DNA replication. In KSHV-positive JSC-1 cells, K8.1 mRNAs was not detected until 24 h of butyrate stimulation (Fig. 1B) and was inhibited by PAA, a viral DNA polymerase inhibitor. Moreover, the transcriptional activity of the potential K8.1 promoter was examined in 293 cells with or This report also provides some evidence that anti-IgM stimulates KSHV K8 and K8.1 gene expression in JSC-1 cells. Although expression induced by anti-IgM of KSHV K8 and K8.1 was less efficient compared with butyrate, this was unexpected because many PEL-derived B cells lack a B cell receptor, an IgM-type antigen on the B cell surface. Lack of B cell receptors may be due to Ig gene rearrangement or mutation (55)(56)(57) or probably due to blockade by KSHV K1 of BCR transport to cell surface (58). Immunoglobulin heavy chain rearrangements in JSC-1 cells have been demonstrated by Southern blot analyses (20). Attempts to detect such B cell receptors on JSC-1 cells were unsuccessful by Western blot (data not shown). However, in our study we cannot rule out the presence of any trace amount of B cell receptors on JSC-1 cells, which may be enough for the unexpected stimulation in our experiment.
Alternative RNA splicing in eukaryotes and some animal viruses is an important mechanism for regulating diversity of gene expression. The mechanisms that control alternative RNA splicing remain understood. ORF50, K8, and K8.1 are three split genes positioned at the same locus side by side but represent three completely different categories of genes (immediateearly, early, and late) in the virus life cycle. Three genes share a common poly(A) signal downstream of the K8.1 coding region. With such an arrangement, expression of each gene takes place through regulation of alternative promoters and alternative splicing. Finding that the K8 exon 3 (Rta exon 4) has three 5Ј-ss and identification of bicistronic K8/K8.1 mRNAs in this study provide a strong evidence of an even more complex splicing regulation in expression of these genes. For better understanding, a model in Fig. 6 is proposed to lay out different splicing pathways involved in processing of K8 and K8.1 pre-mRNAs based upon what we have learned. In this model, alternative selection of three 5Ј-ss within exon 3 of a bicistronic K8 pre-mRNA at early stage of virus life cycle determines a divided coding function of the spliced products. Although the first step of the splicing is to remove intron 1 as we detected, selection of the nt 75838 5Ј-ss to remove an intron 3 appears to be the next crucial step, dictating production of the K8 mRNAs. Splicing of intron 3 in the K8 pre-mRNAs apparently activates the splicing of the intron 2 and results in formation of K8␣ mRNA, which is always a major product responsible for encod- ing K-bZIP protein (22,27,30). Removal of intron 2 is not as efficient as intron 1, leading to a minor product of K8␤. However, the nt 75838 5Ј-ss is not chosen sometimes, and consequently, alternative selection of other two 5Ј-ss downstream produces either a K8/K8.1␣ or a K8/K8.1␤ mRNA which often retains an intron 2. The nt 76155 5Ј-ss appeared preferentially selected over the nt 76338 5Ј-ss in our transient transfection assay (Fig. 5B). Selection of the nt 76338 5Ј-ss is problematic due to size of the exon. Because an internal exon is defined by cross-talking of a 3Ј-ss and a 5Ј-ss over the exon (59), an oversized exon (Ͼ500 nts) makes it more difficult to be defined by cellular splicing machinery (60). When the K8.1 promoter is activated at a late stage of the virus life cycle, the resultant K8.1 transcripts with only two exons and one intron undergo splicing through alternative selection of two 5Ј-ss in their exon 1 (34,35). A majority of K8.1␤ mRNAs (encoding for K8.1A protein) are produced by preferential selection of nt 76338. K8.1␣ mRNAs (encoding for K8.1B protein) are less prominent and come from selection of the nt 76155 5Ј-ss. This is consistent with our K8.1 protein data (Fig. 5D) and reports from other laboratories (34,35,40).
There are many examples of alternative selection of 5Ј-ss involved in expression of viral and eukaryotic genes including expression of SV40 large T and small t antigen and adenovirus E1A (61)(62)(63). Many studies (64 -68) show that the strength of the 5Ј-ss and the cis elements in the exon and intron and status and level of cellular splicing factors all devote to regulation of alternative 5Ј-ss selection. In this regard, we know nothing about how three 5Ј-ss in exon 3 of the K8 are alternatively selected. Considering that exon definition is limited by exon size (59) and recognition of a 5Ј-ss is scanned 5Ј to 3Ј by cellular splicing machinery during RNA splicing (69), three 5Ј-ss, respectively, at nt 75838, 76155, and 76338, should be selected in order by a simple scenario of the first come and first served. However, why the nt 76338 5Јss is preferentially selected over the nt 76155 5Ј-ss during KSHV K8.1 expression remains questionable. Looking for viral cis elements in this region and  (34 -36), and the 3Ј end of ORF50 are depicted at the top of KSHV genome (heavy line). Shown below the heavy line are the K8 and K8.1 pre-mRNA structures with exons (boxes) and introns (lines) as well as splicing directions (dashed lines) and splicing products (␣, ␤, and ␥ isoforms). The numbers above the ORFs and pre-mRNAs or below the heavy line are nt positions in KSHV (BC-1) genome (GenBank TM accession number U75698) (5). The K8 and K8.1 promoters are named based on the nt position where the transcription is initiated. The 5Ј-ss and 3Ј-ss are shown below each pre-mRNA. The K8 ORF terminates at nt 75791 for K-bZIP instead of nt 75569 (5) for K8␤ because the K-bZIP is a major form predominantly expressed in viral lytic infection. Downstream of the K8 termination codon are 3Ј-untranslational sequences including part of the exon 3 and the entire exon 4. The K8 shares with the K8.1 a single polyadenylation site (pA) at nt 76714 and a potential cleavage site at nt 76730 (22). Splicing of the K8 pre-mRNAs is first to remove intron 1. The resulting products are then further spliced by using a nt 75838 5Ј-ss to remove intron 3, leading to generation of the K8␤ mRNA. Finally, intron 2 is removed, and consequently the K8␣ mRNA is produced. Lack of intron 1 in all K8 mRNAs in this model is also consistent with the absence of a detectable protein containing a peptide sequence translated from the intron 1 in TPA-induced PEL cells (30). However, alternative splicing by using other 5Ј-ss in the K8 exon 3 is feasible, and the resulting K8/K8.1␣ and -␤ mRNAs are bicistronic due to their features of having both K8 and K8.1 coding regions. The K8.1 mRNAs are expressed only at late stage of virus infection because the K8.1 promoter P 75901 is a late promoter, and its activity requires viral DNA replication (Fig. 2C). Splicing by selection of two alternative 5Ј-ss (the same two 5Ј-ss in the K8 exon 3) in its exon 1 of the pre-mRNA produces either a K8.1␣ (splicing from the nt 76115 5Ј-ss to the nt 76433 3Ј-ss) or a K8.1␤ (splicing from the nt 76338 5Јss to the nt 76433 3Ј-ss) mRNA. The K8.1␥ RNA is the one left unspliced.
trans-acting factors involved in this regulation is currently under investigation.
It is also unclear why KSHV produces a bicistronic K8/K8.1␣ or K8/K8.1␤ mRNA even if it has little protein coding function. Because the bicistronic K8/K8.1␣ and -␤ mRNAs exist in a relative amount and are inducible in JSC-1 cells by butyrate, one of the possibilities is that these RNAs, in addition to T1.1 or PAN RNAs (70,71), may represent another class of regulatory RNAs. Through alternative selection of three 5Ј-ss in the K8 exon 3, the virus is able to keep the production level of the K8␣ mRNA in check. Thus, production of bicistronic K8/K8.1␣ and -␤ mRNAs or other derivatives may be an essential mechanism to control K8␣ mRNAs to a threshold during processing of bicistronic K8 pre-mRNAs.