ABSTRACT
Epstein-Barr virus (EBV) nuclear protein 2 (EBNA-2) is a transcriptional transactivator of cellular and viral gene expression and is essential for the transformation of resting human B lymphocytes into long-term lymphoblastoid cell lines (LCLs). Previous molecular genetic analyses identified three domains that are critical for transformation and showed that the rest of EBNA-2 is not critical. We now find that codons 231 to 280 that were part of one of the critical domains (J. I. Cohen, F. Wang, and E. Kieff, J. Virol. 65:2545–2554, 1991) can be deleted with only a small effect on the ability of EBNA-2 to transactivate gene expression. In transient transfection assays, EBNA-2 deleted for codons 231 to 280 accumulated to higher levels and was similar to wild-type EBNA-2 in activation of the BamC promoter and in association with RBPJk, a cellular transcription factor that is important for EBNA-2 interaction with promoter regulatory elements. However, EBNA-2 d231–280 activated the viral latent membrane protein 1 (LMP1) promoter with only 60% of wild-type efficiency. Recombinant EBVs specifically deleted for EBNA-2 codons 231 to 280 were efficient in initiating the transformation of resting primary human B lymphocytes into LCLs. However, these LCLs grew less well than wild-type EBV-transformed LCLs, and 4- to 10-fold more cells were required for outgrowth following limit dilution. EBNA-2 d231–280 accumulated to unusually high levels in the recombinant transformed LCLs, and this was associated with somewhat higher EBNA-1 and lower LMP1 expression, consistent with the near-wild-type activation of the BamC EBNA promoter and the abnormally low activation of the LMP1 promoter in transient transfection assays. Thus, EBNA-2 d231–280 modestly perturbed the regulation of viral gene expression and resulted in less LMP1, while having surprisingly subtle effects on LCL outgrowth. Deletion of EBNA-2 codons 292 to 310, which are closer to the site that specifies interaction with RBPJk, was more disruptive of RBPJk association and of the ability to transform B lymphocytes.
Epstein-Barr virus (EBV) can efficiently transform resting human B lymphocytes to long-term lymphoblastoid cell lines (LCLs). LCL outgrowth is associated with the expression of at least six virus-encoded nuclear proteins (EBNAs), two virus-encoded integral membrane proteins (LMPs), and several RNAs of uncertain function (reviewed in references 20 and34). Five EBNAs and one LMP are critical for efficient resting B-lymphocyte proliferation (6, 11, 19, 29,40). EBNA-2 and EBNA-LP are particularly important, since they are the first two proteins expressed from the viral genome after lymphocyte infection, and they up-regulate EBNA, LMP, and cellular gene expression. EBNA-2 is a direct transactivator of cell and viral gene expression (4, 38, 45, 47, 51), while EBNA-LP is a coactivator and is dependent on EBNA-2 for its effects (12,32).
The experiments reported here investigate one of the three essential components of EBNA-2 for resting B-lymphocyte growth transformation. Previously, EBNA-2 codons 2 to 88, 97 to 122, 112 to 141, 143 to 231, 337 to 354, 359 to 383, 385 to 430, and 462 to 482 have been deleted, with at least 10% residual transformation efficiency (Fig. 1), indicating that these sites are not essential for EBNA-2-transforming activity (5, 44, 48). Three deletions have been persistently negative for transformation. These deletions likely identify codons that specify key functional domains of EBNA-2 for transformation. One type of deletion leaves fewer than three codons for proline (the polyproline domain corresponds to residues 59 to 95). The second deletion eliminates codons 230 to 336, while the third eliminates codons 426 to 462 (5, 48). Residues 230 to 336 include the GPPW319W320PP sequence, which interacts with RBPJk, a cellular sequence-specific DNA binding protein (10, 13,21, 24, 25, 37, 49). Mutation of W319W320to SS abrogates the ability of the specifically mutated EBNA-2 to participate in EBV-mediated LCL outgrowth (49). EBNA-2 is highly associated with RBPJk, and RBPJk mediates much of the EBNA-2 promoter specificity (10, 13, 17, 18, 24, 25, 48, 49). However, the same region of EBNA-2 can deplete PU.1 from nuclear extracts (17), and PU.1 is also important for EBNA-2-mediated activation of the LMP1 and LMP2a promoters (17,21, 30, 36). EBNA-2 codons 426 to 462 encode an acidic domain that can recruit TAF40, TFIIB, TFIIH, TBP, and a p100 nuclear protein to promoters, thereby facilitating transcription (4,41-43). Thus, the essential parts of EBNA-2 aside from the proline requirement mediate either interaction with specific promoters (residues 230 to 336) or recruitment of transcription factors (residues 426 to 462). We now further evaluate the importance of the residues between residues 230 and 336 for resting B-lymphocyte growth transformation. Our analysis focuses on residues 230 to 310, since linker insertion and point mutations of W319W320 to S319S320have already established the importance of this site for RBPJk interaction and for transformation.
MATERIALS AND METHODS
Cell culture.BJAB is an EBV-negative B-lymphoma cell line (27). P3HR-1 clone 16 cells are infected with the EBV strain P3HR-1, which is replication competent but unable to growth transform human B lymphocytes (31). IB4 is an EBV-transformed lymphoblastoid cell line. Cells were cultured in RPMI 1640 medium (Gibco) supplemented with 10% heat-inactivated fetal calf serum (FCS; HyClone) and gentamicin. LCLs and in vitro EBV-infected human peripheral B lymphocytes were cultured in RPMI 1640 medium supplemented with 15% heat-inactivated FCS, gentamicin, and amphotericin B (Boehringer).
Cosmids and plasmids.TheHindIII-CpoI fragment of the W91 strainEcoRI A cosmid (corresponding to residues 48039 to 52589 of the EBV B95-8 sequence (2), was subcloned into a plasmid vector. The EBV W91 strain EBNA-2 used in most of the genetic analyses has 483 amino acids and differs from the B95 prototype in having five fewer prolines in the polyproline repeat and a leucine between B95 amino acids 211 and 212 (5). The sequences of interest were deleted either by PCR mutagenesis (14) or restriction enzyme digestion followed by the insertion of oligomers. To delete codons 231 to 280, inside primers of CTACTCACGGTACTACAAAGGATTCCACCTATGCCATTACCC and GGGTAATGGCATAGGTGGAATCCTTTGTAGTACCGTGAGTAG were used for PCR mutagenesis. To delete codons 231 to 310, inside primers CTACTCACGGTACTACAAAGGCATAATCTACCCTCGGGGCCA and TGGCCCCGAGGGTAGATTATGCCTTTGTAGTACCGTGAGTAG were used for PCR mutagenesis. To delete codons 292 to 310, two synthetic oligomers, AATTGCATAATCTACCCTCGGGGCCACCATG and GTGGCCCCGAGGGTAGATTATGC, were annealed to be cloned into the gap between the MunI and BstXI sites of the EBNA-2 coding sequence. The mutatedHindIII-CpoI fragments were cloned back into the EcoA cosmid, or the mutatedBstUI-DraI fragments were cloned into the eukaryotic expression vector pSG5, whose expression was under the control of the simian virus 40 early promoter (Stratagene). Mutated clones were verified by dideoxynucleotide sequencing.
Transformation assay.P3HR1 cells (1.5 × 107) were electroporated at 220 V in the presence of 10 μg of cosmid DNA containing mutant or wild-type EBNA-2 and 40 μg of the BZLF1 expression plasmid pSVnaeIZ. The transfected cells were cultured in 15 ml of medium for 5 days (48). Culture supernatant containing virus was filtered through a 0.45-μm-pore-size filter and used to infect freshly prepared human peripheral blood B lymphocytes. The infected cells were plated in 96-well plates at 5 × 104 cells per well in 150 μl of RPMI medium supplemented with 15% FCS. Medium was changed 2 weeks after plating, and then the cells were fed once a week with fresh 15% FCS-RPMI medium. LCLs were macroscopically visible 4 to 6 weeks after plating. Cultures were maintained for at least 3 months.
Virus passage.LCLs established by infection with recombinant EBV from transfected P3HR-1 cell supernatants were induced to enter lytic cycle by transfection with pSVnaeIZ and treatment with 20 ng of phorbol-12-myristate-13-acetate (Gibco) per ml. Five days after induction, virus was harvested and used to infect human peripheral B lymphocytes (5, 44). Established LCLs were designated second-generation LCLs.
Endpoint dilution cell outgrowth assays.Second-generation LCLs infected with EBV recombinants with wild-type or mutant d231–280 EBNA-2 were incubated at various dilutions in 96-well plates to measure the endpoint for LCL outgrowth. Four wild-type and five mutant LCLs were analyzed. Incubations were done with and without CRL 1634 feeder cells (diploid human fibroblasts; American Type Culture Collection). Cells were serially diluted twofold from 2 × 104 to 0.5 × 104 per well. Half the medium was changed weekly thereafter. Positive outgrowth was assessed 8 weeks after plating.
CAT assays.BJAB cells (1.5 × 107) in log-phase growth were transfected with 10 μg of reporter plasmid, 5 μg of β-galactosidase expression plasmid, and 30 μg of the wild-type or mutant EBNA-2 expression plasmid DNA or control pSG5 vector DNA. Reporter plasmids were p−234/+40LMP1CAT, which contained −234 to +40 sequences of LMP1 promoter region cloned into the promoterless chloramphenicol acetyltransferase (CAT) gene (45), or pCpTKCAT, which had −330 to −380 ofBamC promoter sequences upstream of the herpes simplex virus type 1 tk promoter-driven CAT gene (49). Transfected cells were harvested 48 h after electroporation as previously described (12). Each sample was divided into two portions and used for Western blotting analysis with anti-EBNA-2 antibodies or for CAT activity (9). CAT activity was measured with ImageQuant software and a PhosphoImager (Molecular Dynamics).
Immunoprecipitation and Western blotting.BJAB cells were transfected with 30 μg of pSG5-derived expression plasmids with wild-type or mutant EBNA-2. Twenty hours after electroporation, cells were lysed in 1 ml of 1% Nonidet P-40 buffer (50 mM Tris-Cl [pH 7.4], 1 mM EDTA, 150 mM NaCl, 3% glycerol, 1 mM phenylmethylsulfonyl fluoride, 5 μg of aprotinin per ml, 2.5 μg of leupeptin per ml, and 1% Nonidet P-40). After 4 h of incubation with normal rabbit serum and protein A-Sepharose beads (Pharmacia), half of the precleared lysates were incubated with rabbit antiserum against RBPJk-glutathioneS-transferase fusion protein (35), followed by additional incubation with protein A-Sepharose. The other half of the lysates were incubated with anti-EBNA-2 monoclonal antibody PE2 and protein G-Sepharose (Pharmacia). Antigen-antibody complexes were recovered, separated on sodium dodecyl sulfate (SDS)–8% polyacrylamide gel, and transferred to nitrocellulose membrane filters. Filters were probed with the anti-RBPJk antiserum or PE2 anti-EBNA-2 monoclonal antibody (50). Proteins were detected by horseradish peroxidase-conjugated secondary antibodies and enhanced chemiluminescence.
Western blots of LCL lysates.LCLs were lysed in SDS sample buffer at 108 cells/ml. Lysates were separated by SDS-polyacrylamide gel electrophoresis and transferred to filters, and proteins were detected by enhanced chemiluminescence. EBNA-2 and LMP1 were detected with monoclonal antibodies PE2 and S12 (28), respectively. Human anti-EBNA-1-positive serum was used for detecting EBNA-1.
RESULTS
Transformation marker rescue with wild-type or specifically mutated (EBNA-2 d231–280, d231–310, ord292–310) EBV DNA.The EBV P3HR-1 strain is replication competent but lacks the ability to transform resting B lymphocytes because of the deletion of a DNA segment that includes the last exon of EBNA-LP and the entire EBNA-2 open reading frame (6,11, 31). Transfection of P3HR-1-infected cells with wild-type EBV DNA fragments that span the 6.8-kbp deletion site enables homologous recombination between the transfected DNA and the endogenous EBV P3HR-1 DNA and marker rescue of the transforming phenotype in the resulting virus preparations. DNA fragments that have been specifically mutated in the EBNA-2 open reading frame can thereby be evaluated for their marker rescue efficiency relative to that of wild-type DNA fragments (5, 6, 11, 29, 44, 48) (Fig.1).
Schematic map of EBNA-2 indicating the deletions and their effects on EBV-mediated transformation. The polyproline sequence (PP), RBPJk binding site (Jκ), and acidic transactivating domain (ATD) are indicated. The EBNA-2 deletions studied here are indicated immediately below the schematic map. Beneath the dotted line are previously characterized deletions (5, 44, 48). Empty boxes indicate deletions that are compatible with primary B-lymphocyte transformation. Solid boxes indicate deletions that markedly affect the ability of EBV to transform B lymphocytes.
EBNA-2 d231–280, d231–310, d292-310, or otherwise isogenic wild-type DNA fragments were assayed for their ability to marker rescue a resting human B-lymphocyte transformation phenotype from EBV P3HR-1 strain-infected cells (Fig. 1). EBNA-2 d231–310 was first evaluated for transformation marker rescue relative to that of wild-type EBV DNA. Nine transfections with four independently derived EBNA-2 d231–310 DNA clones failed to rescue virus capable of transforming B lymphocytes into LCLs. In contrast, parallel control transfections with wild-type EBV DNA rescued transforming activity in every experiment (Table 1). These results indicate that deletion of codons 231 to 310 results in a mutated EBNA-2 gene whose ability to participate in B-lymphocyte transformation is severely impaired.
Efficiency of primary lymphocyte transformation marker rescuea
To evaluate the specific sequence requirements within codons 231 to 310, EBNA-2 d231–280 and d292–310 were then compared to wild-type EBV DNA regarding transformation marker rescue. Six transfections of P3HR-1 cells with four independent EBNA-2 d231–280 DNA fragments resulted in virus that was able to transform B lymphocytes in each assay (Table 1). The yield of transforming EBV from EBNA-2 d231–280 transfections was usually 2- to 10-fold less than that with wild-type EBV DNA transfections done in parallel (Table 1). In contrast, four independently derived EBNA-2 d292–310 clones failed to marker rescue transforming virus in eight attempts in three separate experiments in which wild-type DNA readily rescued transforming virus (Table 1). These results indicate that residues 292 to 310 are critical for transformation, while residues 231 to 280 are not.
Although residues 231 to 280 are not essential for transformation, the efficiency of marker rescue with EBNA-2 d231–280 was less than wild-type DNA. LCL outgrowth was also retarded relative to LCLs transformed with virus from wild-type DNA transfections. For example, in the experiment in which 95 wild-type recombinant virus-infected LCLs grew out of 96 wells by 3 weeks after plating, only 7 EBNA-2 d231–280 -infected LCLs were evident by 3 weeks, and 77 were evident at 8 weeks.
Since recombinant virus from transfected P3HR-1 cells is mixed with a vast excess of parental P3HR-1 virus, and P3HR-1 can inhibit transformation and exaggerate differences in transforming activity (29, 31), we directly compared the resting B-lymphocyte transforming activity of recombinant virus from four EBNA-2 d231–280 -infected LCLs with that of recombinant virus from four wild-type-infected LCLs (Table2). While the transforming activity of virus preparations from the LCLs varied, EBNA-2 d231–280 -infected LCLs did not differ significantly from wild-type-infected LCLs overall. Wild-type virus stocks that gave 94 × 102 transformations had similar levels of viral DNA by endpoint dilution PCR to EBNA-2 d231–280 virus stocks that gave 93 × 102 transformations (data not shown).
LCL resulting from infection by EBNA-2 mutant d231–280a
EBNA-2 d231–280-infected LCLs differ in growth from wild-type recombinant virus-infected LCLs.In most experiments, LCLs infected with EBNA-2 d231–280 EBV recombinants in the absence of the EBV P3HR-1 cells differed only marginally from wild-type recombinant virus-infected LCLs in their time to outgrowth and in their continued growth in culture. Since there was substantial variability and overlap among wild-type and mutant LCL clones in these parameters, we searched for another assay that would distinguish the growth of mutant and wild-type LCLs. Assays of the endpoint dilution from which the two types of LCLs could regrow distinguished the growth phenotypes of the EBNA-2 d231–280 and wild-type recombinant-infected LCLs that had been derived and maintained in parallel. Five EBNA-2 d231–280 -infected LCLs differed from four wild-type LCLs in their regrowth following limiting dilution. While 104 cells per microwell were necessary for 100% growth of EBNA-2 d231–280 -infected LCLs, only 103 cells per microwell were necessary for 100% growth of wild-type virus-infected LCLs. Similarly, the 50 and 1% endpoint dilutions for growth of EBNA-2 d231–280 -infected LCLs required four-times-higher cell concentrations than wild-type-infected cells. A similar 4- to 10-fold difference in endpoint dilution growth of EBNA-2 d231–280 -infected LCLs relative to that of wild-type-infected LCLs was also evident when the infected cells were grown on fibroblast feeder layers (Fig.2). These results are compatible with a cell concentration-dependent, fibroblast feeder layer-independent growth defect in EBNA-2 d231–280 mutant-infected LCLs, consistent with altered autocrine growth factor dependence or secretion from the EBNA-2 d231–280 -infected LCLs.
Endpoint dilution growth of wild-type (WT) and mutant (mt) recombinant EBV-transformed LCLs. Second-generation LCLs transformed by either wild-type or an EBNA-2 d231–280 mutant EBV recombinant were analyzed for their ability to regrow after endpoint dilution without (A) or with (B) fibroblast feeder cells. The data shown are the average results among four wild-type and five mutant LCLs. Cells (2 × 104) were plated into the first well, and twofold serial dilutions were plated into wells 2 to 12. The y axis indicates the average percentage of wells that had growth at 8 weeks.
Interaction of EBNA-2 d231–280, d231–310, and d292–310 and wild-type EBNA-2 with RBPJk.Since residues 231 to 310 border on the RBPJk-interacting domain that is centered about GPPW319W320PP, we evaluated the extent to which the deletion mutations from codons 231 to 310 affect EBNA-2 association with RBPJk in BJAB, a non-EBV-infected B-lymphoma cell line. BJAB cells were transfected with a vector expressing wild-type EBNA-2, EBNA-2 d231–280, d292–310, or d231–310, and the extent of mutant or wild-type EBNA-2 coimmunoprecipitation with endogenous RBPJk was assayed (Fig.3). An EBNA-2 monoclonal antibody that interacts with the carboxyl-terminal acidic domain of EBNA-2 was used to immunoprecipitate EBNA-2 and to detect the immunoprecipitated EBNA-2 by Western blotting. Rabbit antibody to a glutathione S -transferase-RBPJk fusion protein was used to immunoprecipitate RBPJk and to detect RBPJk by Western blotting. EBNA-2 d292–310 accumulated in cells to slightly lower amounts than did wild-type EBNA-2, while EBNA-2 d231–280 was more abundant than EBNA-2 and EBNA-2 d231–310 was slightly more abundant than EBNA-2 d231–280. EBNA-2 d231–280 associated with nearly as much RBPJk as wild-type EBNA-2. However, EBNA-2 d231–310 associated substantially less well with RBPJk, and EBNA-2 d292–310 associated poorly with RBPJk. Thus, EBNA-2 residues 292 to 310 are important for EBNA-2 interaction with RBPJk.
EBNA-2 mutants interact with RBPJk. BJAB cells were transfected with pSG5 expression vector DNAs which encode wild-type (WTE2) or three EBNA-2 deletion mutants (d231–280, d292–310, and d231–310). Cell lysates were immunoprecipitated with anti-EBNA-2 monoclonal antibody (PE2) or anti-RBPJk rabbit antibody (anti-Jk). The precipitates (PE2 ip and α-Jk ip) were analyzed by Western blotting with PE2 and anti-Jk antibodies.
EBNA-2 d231–280 association with RBPJk correlates with EBV BamC but not LMP1 promoter responsiveness in transfected B-lymphoma cells.Previous mutational analyses of EBNA-2 were consistent with the model that EBNA-2 transactivation of the EBV BamC EBNA promoter is completely dependent on RBPJk association, while transactivation of the EBV LMP1 promoter is partially dependent on RBPJk association and fully dependent on interaction with PU.1 (17, 49). We therefore evaluated the ability of EBNA-2 d231–280, EBNA-2 d231–310, and EBNA-2 d292–310 to transactivate the BamC and LMP1 promoters. Transactivation of the BamC promoter was fully consistent with the ability of the EBNA-2 mutants to associate with RBPJk. EBNA-2 d231–280 was similar to wild-type EBNA-2, while EBNA-2 d231–310 was moderately impaired and EBNA-2 d292–310 was severely impaired (Fig.4B). Surprisingly, EBNA-2 d292–310 transactivated the LMP1 promoter as well as the wild type, and both EBNA-2 d231–310 and d231–280 were somewhat impaired relative to the wild type (Fig. 4A). These data are consistent with RBPJk association being essential for BamC promoter activation and less important for LMP1 promoter activation as has been previously noted (17,49). Most interestingly, residues 231 to 280 appear to be critical for EBNA-2 interaction with an LMP1 promoter-specific transcription factor. Since the LMP1 promoter is critically dependent on PU.1 for EBNA-2 up-regulation (17), residues 231 to 280 could be important for EBNA-2 interaction with PU.1 or with a PU.1-associated protein.
Mutant or wild-type (WT) EBNA-2 effects on CAT expression. BJAB cells were transiently transfected with pSG5-derived EBNA-2 expression plasmids together with an LMP1-promoter CAT plasmid (p−234/+40LMP1CAT) (A) or a BamC promoter CAT plasmid (pCpTKCAT) (B). Fold transactivation activity is relative to that of control plasmid without EBNA-2. The data shown in panels A and B are the averages and standard deviations (error bars) of 12 and 4 independent experiments, respectively.
EBNA-2 d231–280 is less active in up-regulating LMP1 expression in transformed LCLs.The level of EBNA and LMP1 expression in whole-cell lysates of eight second-generation LCLs transformed by EBNA-2 d231–280 recombinants was compared with EBNA and LMP1 expression in four wild-type recombinant-infected LCLs derived in parallel (Fig. 5). The most striking and consistent difference was that EBNA-2 d231–280 was more abundant than wild-type EBNA-2, the difference being even greater than that observed in transfected BJAB cells (Fig. 5A and 3). EBNA-1 levels were only slightly higher in the EBNA-2 d231–280 -infected LCLs, probably reflecting a higher level of activation of the EBNA promoter in response to increased stability of EBNA-2 d231–280. In contrast, LMP1 expression was similar or lower in EBNA-2 d231–280 -infected LCLs than in the wild-type LCLs (Fig. 5C). These results are most compatible with the differentially higher activity of EBNA-2 d231–280 on the BamC EBNA promoter versus the LMP1 promoter in the transient transfection assays.
EBNA-2, EBNA-1, and LMP1 expression in d231–280 recombinant virus-infected LCLs. Western blots of total cell lysates were incubated with antibodies or antiserum specific for EBNA-2 (A), EBNA-1 (B), and LMP1 (C), respectively. The experiment included four wild-type (WT) and eight d231–280 deletion mutant (mt) LCLs.
Since LMP1 and EBNA-2 independently or synergistically induce expression of B-lymphocyte adhesion and activation molecules, we assayed the effect of the EBNA-2 d231–280 mutation and the consequent lower LMP1 levels on surface adhesion and activation marker expression in the recombinant virus-infected LCLs. Four wild-type LCLs were compared with four mutant LCLs that had been derived in parallel. By fluorescence-activated cell sorter analysis, LFA-1, LFA-3, ICAM-1, CD40, CD21, and CD10 levels were similar on the surfaces of mutant and wild-type LCLs. These data indicate that EBNA and LMP expression in these cells are sufficient for wild-type expression of these surface adhesion, differentiation, and activation molecules.
DISCUSSION
The information about the ability of EBNA-2 d231–280 and the inability of EBNA-2 d292–310 to marker rescue transforming phenotypes in the background of the EBNA-2-negative P3HR-1 EBV genome completes a phase of the genetic dissection of the EBNA-2 open reading frame. In segments of various sizes, 80% of the EBNA-2 open reading frame has now been deleted, with substantial residual transforming activity. Only two to seven prolines of the polyproline domain (residues 58 to 95), residues 281 to 327, and residues 424 to 464 remain essential for EBNA-2 marker rescue of primary B-lymphocyte transforming activity (5, 6, 44, 48,49 and data therein). While the precise role of the prolines is uncertain, residues 281 to 327 mediate interactions with promoters that have nearby RBPJk and PU.1 sites, and residues 424 to 464 recruit basal and activated transcription factors to the promoters (4, 5,10, 13, 17, 18, 20, 41-43, 46).
The modest negative effect on transformation of the deletion of codons 231 to 280 and the critical importance of residues 292 to 310 in transformation correlate overall with their abilities to interact with RBPJk and activate the BamC promoter and with their sequence conservation among the two EBV types and the baboon lymphocryptovirus (26). Residues 292 to 305 are proline rich and somewhat hydrophobic and may have a role in effecting the intermolecular presentation of the nearby GPPW319W320PP RBPJk binding site. The RBPJk binding site is critically important in getting EBNA-2 to promoter sites for transcriptional activation (10, 13,45, 49). Mutation of GPPW319W320PP to GPPSSPP aborts RBPJk interaction, BamC promoter transactivation, and B-lymphocyte growth transformation (49). GPPW319W320PP is separated from the 292 to 310 sequence by several amino acids that have little obvious similarity between the two EBV types and between EBV and the otherwise closely related lymphocryptovirus of baboons. The divergence in these intervening sequences is compatible with the notion that the GPPWWPP sequence is the primary mediator of RBPJk interaction. Consistent with this notion, a yeast two-hybrid screen with RBPJk as bait yielded multiple antisense cDNAs that encode short in-frame oligopeptides with WWP motifs (50a). Also, the EBNA-2 WWP site is remarkably similar to the WFP site through which the cellular protein, Notch 1, primarily engages RBPJk (1, 39).
Notch 1 and EBNA-2 interact with the same part of RBPJk, and both exert activating effects through RBPJk (1, 3, 15, 23, 39). In mimicking Notch 1, EBNA-2 is likely to be activating cellular genes that are important in lymphocyte growth. Activated forms of Notch 1 have been associated with human T-cell leukemia and can cause leukemia when they are expressed in mouse bone marrow (7, 8, 33). Our finding that LCLs transformed by an EBV recombinant with the EBNA-2 d231–280 mutation are deficient in growth at limit dilution, despite normal surface adhesion and activation marker expression, is consistent with the likelihood that EBNA-2 has an important role in differentially effecting the expression of cellular genes that are critical for the growth of EBV-transformed cells.
Interaction through RBPJk is also one of the important effects of EBNA-2 on viral promoters. EBV has incorporated RBPJk binding sites into the regulatory domains of the BamC EBNA, LMP1, LMP2b, and LMP2a promoters and activates transcription in part through these sites (16, 17, 22, 25, 30, 37, 46, 49). The BamC EBNA promoter is not highly EBNA-2 responsive in BJAB cells, and much of the responsiveness is dependent on the RBPJk site and RBPJk interaction (44, 49). Consistent with this dependence, the deletion of residues 291 to 310 that markedly impairs EBNA-2 association with RBPJk also impairs BamC promoter up-regulation, while EBNA-2 d231–280 associates well with RBPJk and activates the BamC promoter as well as wild-type EBNA-2.
At least one other sequence-specific DNA binding protein appears to be critical for EBNA-2 transactivation of each EBV latency promoter (16, 17, 21, 36). The LMP1 promoter has substantial residual EBNA-2 responsiveness after mutation of its RBPJk sites, and this responsiveness is mediated by the Ets family protein, PU.1. Although EBNA-2 does not appear to stably associate with PU.1, EBNA-2 residues 310 to 376 can bind in vitro-translated PU.1 and can deplete PU.1 from a nuclear extract (17). Consistent with the importance of PU.1 in LMP1 promoter responsiveness, EBNA-2 d292–310 activated the LMP1 promoter as well as wild-type EBNA-2, despite its very poor association with RBPJk, while EBNA-2 d231–280 associated with RBPJk nearly as well as wild-type EBNA-2 and was impaired in LMP1 promoter up-regulation. These data are compatible with the hypothesis that the deletion of residues 231 to 280 affects the folding of EBNA-2 so that it does not interact as well with PU.1.
The EBNA-2 d231–280 recombinant EBV-infected LCLs are unusual in their high-level EBNA-2 expression, somewhat-increased EBNA-1 expression, and somewhat-lower-level LMP1 expression. The higher level of EBNA-2 and EBNA-1 expression appears to be due an increased stability of EBNA-2 d231–280 and its ability to associate with RBPJk and thereby maintain increased activation of the EBNA promoter. The higher-level accumulation also partially compensates for the lower efficiency of EBNA-2 d231–280 in activating the LMP1 promoter, probably accounting for the finding that LMP1 levels are only slightly lower in EBNA-2 d231–280 mutant recombinant-infected LCLs than in wild-type LCLs. Nevertheless, given the extent of abnormality in the expression of these important effector molecules, we are surprised by the efficiency with which EBNA-2 d231–280 recombinant EBV can transform resting human B lymphocytes into LCLs with wild-type activation and adhesion molecule expression and nearly wild-type cell growth under nonlimiting dilution conditions.
ACKNOWLEDGMENTS
This research was supported by grant CA4006 from the National Cancer Institute of the USPHS.
We thank Ellen Cahir Macfarland for help in use of the fluorescence-activated cell sorter and Eric Robertson and Steven Grossman for RBPJK antibody.
FOOTNOTES
- Received 9 May 1998.
- Accepted 8 September 1998.
↵* Corresponding author. Mailing address: Departments of Medicine and Microbiology and Molecular Genetics, Harvard Medical School and Brigham and Women’s Hospital, Channing Laboratory, 181 Longwood Ave., Boston, MA 02115. Phone: (617) 525-4250. Fax: (617) 525-4257. E-mail: ekieff{at}rics.bwh.harvard.edu.
REFERENCES
- Copyright © 1998 American Society for Microbiology
