Latent Membrane Protein 1 of Epstein-Barr Virus Inhibits as Well as Stimulates Gene Expression

Abstract

The latent membrane protein 1 (LMP-1) of Epstein-Barr virus (EBV) functionally resembles a constitutively active, CD40-like receptor and contributes to the maintenance of proliferation of EBV-infected primary human B lymphocytes. LMP-1 is targeted to the plasma membrane, where it binds TRAF, TRADD, and JAK molecules to activate NF-κB-, AP-1-, and STAT-dependent pathways as does CD40. Yet LMP-1 appears to lack a ligand to regulate its signaling. We have found that LMP-1, when expressed at physiologic levels, inhibits gene expression detectably. Higher levels of LMP-1 expression eventually inhibit both the steady-state level of RNA produced from a BamHI C promoter reporter and general cellular protein synthesis. These findings indicate that LMP-1 can limit its signaling and that this control is manifest at two levels. The domain of LMP-1 that binds TRAF, TRADD, and JAK/STAT molecules is not required for this regulation. A derivative of LMP-1 that contains only its amino-terminal and membrane-spanning domains is sufficient to inhibit reporter activity when the reporter genes are expressed from the BamHI C and LMP-1 promoters. This same derivative of LMP-1 in parallel assays is sufficient to inhibit wild-type LMP-1's stimulation of NF-κB-dependent gene expression. We suggest that LMP-1 encodes stimulatory and inhibitory activities; the latter could limit signaling in the apparent absence of ligand-dependent down-regulation.

Ligand-dependent surface receptors can limit their signaling by requiring ligand for this function and, in the presence of ligand, by being internalized and removed from their signaling compartments. LMP-1 may have evolved from ligand-dependent receptors of the tumor necrosis factor family (16), but it fails to conform with this paradigm for receptor regulation because it signals apparently in the absence of a ligand (29). LMP-1 contributes to the proliferation of cells infected by Epstein-Barr virus (EBV) (24, 27) and can affect the growth properties of some cell lines, identifying it as an oncoprotein (3, 8, 31, 34, 38). However, LMP-1 has been shown to limit the proliferation of both epithelial and lymphoid cell lines, a phenotype which indicates that LMP-1's expression or activities must be regulated for efficient survival of the infected cell (12, 18, 25).

Several characteristics of LMP-1 are consistent with its functioning as a ligand-independent, constitutively active growth factor receptor. It is an integral membrane protein with an intracellular amino terminus of 25 amino acids, six hydrophobic membrane-spanning domains, and an intracellular carboxy terminus that is known to bind cellular proteins and activate signal transduction pathways as does CD40 (4, 5, 7, 9–11, 13, 15, 20, 22, 28, 32, 35). LMP-1 stimulates NF-κB-, AP-1-, and STAT-mediated transcription in cells (10, 15, 17, 21, 26, 30). LMP-1 stimulates these activities by aggregating, which is dependent on its amino-terminal and membrane-spanning domains, apparently independently of a ligand (16, 29). Ligand-dependent stimulation of cellular receptors yields NF-κB-, AP-1-, and STAT-mediated transcription which is down-regulated in normal cells following their stimulation. Such down-regulation is important because, for example, both AP-1 and NF-κB family members are proto-oncogenes whose aberrant activation contributes to tumor formation (2, 23). We have searched for mechanisms by which LMP-1, in spite of its functioning like a constitutively active receptor-like molecule, could limit its activation of these critical signaling pathways consistent with survival of both the host cell and the infected human being.

In this study we describe LMP-1's negative regulation both of gene expression from EBV's BamHI C (Bam C) and LMP-1 promoters and of its own stimulation of NF-κB-dependent gene expression. For the purpose of this study, we define gene expression as the sum of all cellular processes required to generate reporter activity. EBNA-1 is a positive regulator of both the Bam C and LMP-1 promoters (14, 33, 37). These two promoters drive the expression of all EBV genes known to be required for immortalization of B cells in culture. Here we show that increasing levels of LMP-1 inhibit gene expression from the Bam C or LMP-1 promoters to as much as 3% of uninhibited levels and that this inhibition is first detectable at levels of LMP-1 normally expressed in six different clones of EBV-infected B cells. This inhibition correlates with a decrease in the steady-state levels of RNA synthesized from the Bam C promoter and with a decrease in cell protein synthesis. A derivative of LMP-1 that contains only its amino-terminal and membrane-spanning domains, and thus cannot bind the cellular signaling molecules known to bind to LMP-1, is sufficient to mediate LMP-1's inhibitory activity. Finally, this derivative of LMP-1 cannot stimulate NF-κB-mediated transcription but can inhibit intact LMP-1's stimulation of NF-κB-mediated transcription. These combined observations support a model in which LMP-1 signals independently of a ligand but dependent on its own concentration; beginning at physiological and at higher concentrations, it inhibits gene expression in a dose-dependent manner. The levels of LMP-1 expressed in EBV-infected B cells likely represent a balance between its stimulation and inhibition consistent with survival of the infected host cell.

A more detailed description of protocols for methods used in this study can be found in the website http://mcardle2.oncology.wisc.edu/sugden-main.htm.

LMP-1 inhibits EBNA-1-mediated transactivation of the Bam C and LMP-1 promoters.

We tested if LMP-1 could regulate EBV's Bam C promoter, which can express transcripts for all of EBV's nuclear antigens, with a vector which contains all of the EBV DNA from oriP up to and including the Bam C promoter fused to luciferase (oriP-Bam Cp-luciferase) (Fig. 1). oriP-Bam Cp-luciferase was cotransfected with an expression vector for EBNA-1 (oriP-EBNA-1) into EBV-negative BJAB cells (Fig. 2A) and assayed as described previously (35). All transfections in this study used the same amount of DNA with empty vector as the filler. EBNA-1 stimulates expression of luciferase (measured as relative light units [RLU]) from oriP-Bam Cp-luciferase 35-fold; this level of stimulation was set to 100% activity. Cotransfection of increasing amounts of an expression vector for LMP-1 (SVLMP-1) with a constant amount of oriP-EBNA-1 and oriP-Bam Cp-luciferase yields a dose-dependent decrease in the detected luciferase activity. LMP-1 can therefore inhibit EBNA-1-mediated transactivation of the Bam C promoter. The level of EBNA-1 was measured to vary by less than twofold in these experiments (data not shown). Parallel experiments performed in GG68 cells transfected with an expression vector for EBNA-2 (to complement this cell's deletion) yielded comparable results (data not shown).

FIG. 1.

Diagrams of the B95-8 strain of EBV, oriP-Bam Cp-luciferase, and oriP-LMP-1p-luciferase reporters. Shown are the elements expressed and used by EBV in latent infection of resting B cells in cell culture. Not shown are the more than 80 genes used during the lytic phase of infection. Letters inside the circle indicate the fragments of the B95-8 strain of EBV resulting from digestion of the genome with BamHI. The promoters used in latent infection are indicated by arrowheads, with the primary RNA transcripts generated from these promoters indicated by dashed lines and the open boxes representing exons of these transcripts. The origin of DNA replication used in the latent life cycle (oriP), the origin of DNA replication used in the lytic life cycle (ori Lyt), and the site of DNA circularization after infection, the terminal repeats (TR), are indicated by black boxes. Above the genome are diagrams of the oriP-Bam Cp-luciferase and oriP-LMP-1p-luciferase reporters used. oriP consists of 20 EBNA-1 binding sites found in the family of repeats (FR) and four binding sites for EBNA-1 in the dyad symmetry element (DS). The luciferase open reading frame was inserted at the locations indicated. The base pair numbers indicate locations of the corresponding DNAs found in the B95-8 strain of EBV (1).

FIG. 2.

LMP-1's effect on expression from oriP-Bam Cp-luciferase and oriP-LMP-1p-luciferase. (A) Increasing amounts of SVLMP-1 (0, 1, 3, 9, and 27 μg) were transfected as indicated into BJAB cells along with 5 μg of oriP-EBNA-1 and 5 μg of oriP-Bam Cp-luciferase. In all transfections, the same total amount of DNA was used with an empty vector added as a filler. The average number of LMP-1 molecules per transfected cell was calculated as described below. Relative activity of oriP-Bam Cp-luciferase was determined 48 h posttransfection by setting the luciferase detected in the presence oriP-EBNA-1 and absence of SVLMP-1 to 100%. The average RLU detected at 100% of oriP-Bam Cp-luciferase activity is 1.5 × 10⁶. The data shown are averages of three independent transfections. Error bars indicate 1 standard deviation from the mean; where no error bar is indicated, the standard error was smaller than the size of the symbol. Wild-type LMP-1 as it resides in the plasma membrane is shown as an inset. (B) oriP-LMP-1p-luciferase (10 μg) was cotransfected into BJAB cells with 0, 1, 3, or 9 μg of SVLMP-1. The cells were assayed for luciferase activity 48 h later. The average number of LMP-1 molecules per transfected cell was calculated as described below. The data represent the averages of four independent experiments, with 100% oriP-LMP-1p-luciferase activity corresponding to the 5.5 × 10³ RLU detected in the absence of SVLMP-1. Error bars indicate 1 standard deviation from the mean. (C) Measuring LMP-1 expression levels in the EBV-immortalized B-cell line 721 and transfected BJAB cells. Luciferase data are taken directly from the experiments represented in panel A. The average number of LMP-1 molecules per cell was calculated by the following method. GST–LMP-1(181-386) was isolated from E. coli DH5α as described previously (35) and found to be 50% pure. GST–LMP-1(181-386) extracts of transfected cells were assayed for LMP-1 by quantitative Western blotting (35) with LMP-1 signals corrected for the efficiency of transfection, which was approximately 50%. All data represent the averages of three independent experiments.

We also tested if LMP-1 could regulate its own promoter. To determine if LMP-1 can also inhibit EBNA-1-mediated expression of oriP-LMP-1p-luciferase (Fig. 1), increasing amounts of SVLMP-1 were transfected into BJAB cells with oriP-EBNA-1 and oriP-LMP-1p-luciferase (Fig. 2B). EBNA-1 positively stimulates oriP-LMP-1p-luciferase fourfold in these cells; this level of stimulation was set to 100% activity. LMP-1 when expressed from 3 × 10⁴ to 9 × 10⁴ molecules per cell progressively inhibits this stimulation. We used BJAB cells in this experiment to be consistent with our other experiments, although the EBNA-1-mediated stimulation of oriP-LMP-1p-luciferase in BJAB cells is less than that in EBV-positive cells (14). The decrease in luciferase mediated by 3 × 10⁴ to 9 × 10⁴ molecules of LMP-1 per cell was statistically significant (P = 0.02). LMP-1 has the ability to inhibit expression from its own promoter in the natural context of LMP-1's regulatory domain. LMP-1 inhibited oriP-Bam Cp-luciferase in a dose-dependent manner in the absence of oriP-EBNA-1 (data not shown). Parallel experiments performed in GG68 cells transfected with an expression vector for EBNA-2 yielded comparable results (data not shown).

We determined whether LMP-1's ability to inhibit EBNA-1-mediated transactivation occurs at levels of LMP-1 expression found in EBV-immortalized lymphoblastoid cells. Quantitative Western blots were used to measure the amount of LMP-1 expressed in six clones of EBV-immortalized B cells (721, RPMI-1788, JO-L-B1, 11/17-1, 11/17-3, and 11/17-10) and in BJAB cells transfected with SVLMP-1. A glutathione S-transferase (GST)–LMP-1 derivative purified from Escherichia coli was used as a standard for these measurements. These measurements, performed as described previously (25), indicate that the average amount of LMP-1 required to inhibit EBNA-1-mediated expression of luciferase to 50% in BJAB cells is equivalent to the average amount of LMP-1 detected in the EBV-immortalized cell clone 721 (Fig. 2C). The level of LMP-1 expressed in 721 cells is the same (within twofold) as that in the five other EBV-immortalized cell clones studied (data not shown) (36). This finding supports a model in which LMP-1 is expressed in EBV-immortalized cells at a level balanced by its positive and negative activities that is consistent with their continued proliferation.

LMP-1 inhibits steady-state RNA levels produced from the Bam C promoter.

S1 nuclease mapping was used to determine if LMP-1 exerts its inhibition of expression of a reporter gene by affecting steady-state levels of RNA encoded by the reporter. Chloramphenicol acetyltransferase (CAT) was used as the reporter and assayed as described elsewhere (37) because we found its RNA to accumulate to higher levels than that of luciferase. As with oriP-Bam Cp-luciferase, EBNA-1 positively stimulates oriP-Bam Cp-CAT and LMP-1 inhibits EBNA-1's transactivation of this reporter to 15% of uninhibited levels (Fig. 3A). Cytoplasmic RNA was therefore isolated from these cells and subjected to S1 nuclease mapping, as described previously (14), using the S1 CAT oligonucleotide as a probe for CAT RNA and the glyceraldehyde-3-phosphate dehydrogenase (GAPDH) oligonucleotide as a probe for GAPDH RNA. In the absence of cytoplasmic RNA (Fig. 3B, lane 3), no specific GAPDH product was detected. RNA from untransfected and three transfected BJAB cell populations yielded an S1 nuclease-protected product for the GAPDH probe that migrated at the expected size of 45 nucleotides (Fig. 3B, lanes 4 to 7). These signals were quantified and used to normalize the input RNA for the signals they yielded with the CAT probe. In the untransfected cell population, no CAT RNA was detected (Fig. 3B, lane 9); however, in the transfected BJAB cell populations, a protected CAT product of the expected size of 55 nucleotides was detected (Fig. 3B, lanes 10 to 12). The CAT RNA increased in the presence of EBNA-1 and decreased when LMP-1 was coexpressed (Fig. 3B, lanes 11 and 12). The level of GAPDH RNA was not found to change significantly when SVLMP-1 was transfected into the cells. This finding presumably reflects both the stability of GAPDH RNA and the fact that only half of the cells received SVLMP-1. The protected signals were verified to be dependent on RNA and not contaminating DNA by their sensitivity to digestion with RNase A prior to S1 mapping (data not shown). The protected RNAs were verified to increase linearly within the range of RNA studied (data not shown). LMP-1 decreases the steady-state levels of CAT RNA produced by oriP-Bam Cp-CAT. The LMP-1-mediated decreases in both CAT activity and steady-state CAT RNA levels as judged by the Wilcoxon rank sum test are significant (P = 0.01). However, LMP-1 inhibited CAT activity by 87% and steady-state CAT RNA levels by only 40%. The decrease in steady-state CAT RNA levels is, therefore, unlikely to account for the entire decrease in CAT activity. Although LMP-1 inhibits the accumulation of CAT RNA, LMP-1 is also likely to inhibit the accumulation of CAT activity by some other means.

FIG. 3.

Measuring LMP-1's effect on oriP-Bam Cp-CAT in BJAB cells by CAT assay and S1 nuclease mapping. (A) Extracts of BJAB cells cotransfected with the indicated DNAs were assayed for CAT activity as described previously (37); 5 μg of oriP-Bam Cp-CAT, 5 μg of oriP-EBNA-1, and 9 μg of SVLMP-1 were cotransfected where indicated. The percentage of chloramphenicol acetylated in each extract is indicated at the bottom. (B) S1 nuclease mapping of GAPDH and CAT RNAs in the cell extracts shown in panel A was performed as described previously (14). The expected sizes of signals of undigested probes and digested probes are indicated at the right. Lane 1 contains markers; their sizes are indicated in nucleotides at the left. Above each lane is indicated the probe used in the S1 reaction and the DNAs transfected in each set of BJAB cells. Lane 3 is the GAPDH probe digested in the absence of any cellular RNA. The fold induction of CAT activity was determined by averaging the CAT activity observed in four independent experiments, with the signal of oriP-Bam Cp-CAT set to 1. The difference between the CAT activity observed for each point was determined by the Wilcoxon rank sum test and has a P value of 0.01. The CAT RNA levels were determined by PhosphorImager quantitation of signals in the S1 nuclease mapping gels shown in panel B. The fold induction of CAT RNA is the average of four independent S1 nuclease mapping experiments after normalizing each for its detected GAPDH signals. The statistical significance of the levels of CAT RNA in each point was determined by the Wilcoxon rank sum test and has a P value of 0.01. (C) Metabolic labeling of cells was performed by incubating 10⁷ BJAB cells with 100 μCi of [³⁵S]Met-Cys for 60 min at 37°C 48 h after transfection. The cells were washed once with RPMI 1640 containing 10% fetal bovine serum and then lysed in 1 ml of 0.2 N NaOH for 5 min at 25°C; 10 ml of 10% trichloroacetic acid containing unlabeled methionine (30 μg/ml) was added, and the samples were bound to glass fiber filters. The filters were washed twice with 10 ml of 10% trichloroacetic acid–unlabeled methionine (30 μg/ml) and once with 10 ml of 100% ethanol. Filters were measured for bound radioactivity in a liquid scintillation counter. Normalized ³⁵S incorporation was determined by setting the radioactivity detected in labeled BJAB cells transfected with the oriP-Bam Cp-CAT reporter alone to 1. ³⁵S incorporation data are averages of three independent experiments; ± indicates 1 standard deviation from the mean.

LMP-1 inhibits cellular protein synthesis.

Metabolic labeling of transfected BJAB cells with [³⁵S]Met-Cys was used to determine if LMP-1 can inhibit cellular protein synthesis. BJAB cells were cotransfected with oriP-Bam Cp-CAT, oriP-EBNA-1, and SVLMP-1. The CAT activity in these transfected cells was measured along with their protein synthetic capacity. oriP-EBNA-1 did not have a detectable effect on protein synthesis; however, SVLMP-1 caused a decrease of protein synthesis (Fig. 3C). In these experiments, 40 to 50% of the transfected cells took up and expressed green fluorescent protein. In a more sensitive assay, up to 60% of similarly transfected BJAB cells take up and are killed by a Fas-expressing vector (data not shown). The observed 40% reduction in protein synthesis at 48 h is therefore distributed in up to 60% of the cells, indicating that in this population protein synthesis is inhibited up to 68%. In addition to inhibiting the steady-state levels of CAT RNA, LMP-1 inhibits the majority of cellular protein synthesis when it is expressed at levels twofold above that found in clones of EBV-immortalized B-cells.

The amino-terminal and membrane-spanning domains of LMP-1 are sufficient to inhibit gene expression.

To determine if any portion of the carboxy-terminal domain of LMP-1 is required for LMP-1's inhibitory activity, we analyzed a derivative of LMP-1 that contains only its amino-terminal and membrane-spanning domains (6MHALMP-1EE) expressed from a cytomegalovirus immediate-early promoter. The hemagglutinin (HA) epitope was placed between amino acids 2 and 3 of the amino terminus, and two copies of an EE epitope were placed at amino acid 194 to form the carboxy terminus of this truncated derivative. The EE epitopes maintain the proper charge distribution of the amino acids located adjacent to the last membrane-spanning domain found in LMP-1, which have been proposed to be important for proper protein insertion into the plasma membrane (40).

Unpermeabilized and permeabilized cells expressing 6MHALMP-1EE were stained with fluorescent antibodies to the two epitope tags to define the location within the cells of the amino and carboxy termini of this derivative (Table 1). The unpermeabilized cells were not stained specifically with antibodies directed to both the HA and EE epitopes, while the permeabilized cells were, indicating that the amino and carboxy termini of 6MHALMP-1EE are both inside the cell, as are those of wild-type LMP-1 (6). 6MHALMP-1EE was also verified to be at the plasma membrane by indirect immunofluorescence (data not shown).

TABLE 1.

Determining the topology of 6MHALMP-1 EE^a

Vector	% of cells staining positive for:
	Unpermeabilized			Permeabilized
	GAMb	Anti-HA + GAMc	Anti-EE + GAMd	GAMb	Anti-HA + GAMc	Anti-EE + GAMd
Empty	1.7 ± 1.0	7.7 ± 1.0	8.0 ± 1.0	2.7 ± 2.8	14 ± 3	9.0 ± 1.0
6MHALMP- 1EE	1.7 ± 0.6	9.0 ± 1.0	5.7 ± 1.0	2.0 ± 1.7	52 ± 2	28 ± 11

^a293EBNA-1 cells were transfected with an empty vector or one encoding 6MHALMP-1EE, and the percentage of cells staining positive for each antibody or antibody pair was determined. The data represent the average of three experiments ± 1 standard deviation. 

^bGAM, fluorescein isothiocyanate-conjugated goat anti-mouse antibody. 

^cAnti-HA. 

^dAnti-EE. 

6MHALMP-1EE was cotransfected into BJAB and GG68 cells with oriP-Bam Cp-luciferase and oriP-EBNA-1 (BJAB cells) or SVEBNA-2 (GG68 cells) (Fig. 4A). 6MHALMP-1EE inhibited luciferase activity from oriP-Bam Cp-luciferase, indicating that the amino-terminal and membrane-spanning domains of LMP-1 are sufficient for its inhibition of gene expression from the Bam C promoter. The diffuse migration of 6MHALMP-1EE in Western blots precluded our measuring the average number of 6MHALMP-1EE molecules expressed per transfected cell. Cotransfection of increasing amounts of 6MHALMP-1EE and oriP-Bam Cp-luciferase into the EBV-immortalized, EBNA-1-expressing cell line 721 yielded a dose-dependent inhibition of luciferase activity (data not shown). The inhibition of oriP-Bam Cp-luciferase in 721 cells demonstrates that the amino-terminal and membrane-spanning domains of LMP-1 can inhibit signaling in an EBV-immortalized cell line.

FIG. 4.

6MHALMP-1EE inhibits gene expression. Shown in the inset is 6MHALMP-1EE and its predicted secondary structure in the plasma membrane. (A) The oriP-Bam Cp-luciferase reporter was not or was cotransfected with 100 ng, 300 ng, 1 μg, 3 μg, or 9 μg of an expression vector for 6MHALMP-1EE into BJAB (squares) and GG68 cells (circles). After 48 h, the cells were assayed for luciferase activity; 100% reporter activity corresponds to the 1.3 × 10⁶ RLU in BJAB cells and 3 × 10⁴ RLU in GG68 cells detected in the absence of 6MHALMP-1EE. oriP-EBNA-1 was not included in the experiments performed in GG68 cells because these cells express EBNA-1 constitutively. The data represent the averages of three independent experiments. Error bars indicate 1 standard deviation from the mean. (B) BJAB cells were cotransfected with a constant amount of a NF-κB-responsive luciferase reporter, the indicated amounts of 6MHALMP-1EE, and either 1 μg of SVLMP-1 (circles) or 30 ng of an expression vector for a NF-κB p50/p65 fusion protein (squares); 100% reporter activity is calculated from the luciferase activity detected in the absence of 6MHALMP-1EE and in the presence of 1 μg of transfected SVLMP-1 (average of 2 × 10⁵ RLU) or 30 ng of transfected p50/65 expression vector (average of 1.5 × 10⁵ RLU). Results shown are the averages of three independent transfection experiments. Error bars indicate 1 standard deviation from the mean. (C) A clone of BJAB cells which expresses 6MHALMP-1EEGFP conditionally was tested for inhibition of protein synthesis by this truncated derivative of LMP-1. ³⁵S-incorporation was measured 48 h after addition of the indicated amounts of tetracycline. Cells were labeled as described in the legend to Fig. 3, with ³⁵S-incorporation in the uninduced cells set to 1. The percent ³⁵S incorporated is the average of four independent experiments, with ± indicating 1 standard deviation from the mean. The average number of 6MHALMP-1EEGFP molecules per cell was determined by measuring the amount of 6MHALMP-1EEGFP expressed in one of the four experiments used to determine the percent ³⁵S incorporated.

To determine if the amino-terminal and membrane-spanning domains of LMP-1 can inhibit NF-κB-mediated transcription, we tested the ability of 6MHALMP-1EE to inhibit the NF-κB reporter, 4×NF-κB-luciferase, when stimulated by LMP-1 or by NF-κB itself. BJAB cells were cotransfected with 4×NF-κB-luciferase and SVLMP-1 or an expression vector for a protein consisting of fused NF-κB members, p50/p65. Levels of the expression vectors were chosen to yield similar levels of luciferase activity, and this luciferase activity was set equal to 100%. Increasing levels of 6MHALMP-1EE were introduced with either an SVLMP-1 or p50/65 expression vector and 4×NF-κB-luciferase (Fig. 4B). 6MHALMP-1EE inhibits LMP-1's stimulation both of NF-κB-dependent gene expression and of exogenously derived NF-κB. This inhibition of NF-κB-dependent gene expression is consistent with LMP-1's inhibition of RNA accumulation and protein synthesis.

We tested directly if the amino-terminal and membrane-spanning domains of LMP-1 are sufficient to inhibit protein synthesis by expressing 6MHALMP-1EEGFP (green fluorescent protein at the carboxy terminus of 6MHALMP-1EE) conditionally. A clone of BJAB cells selected to express 6MHALMP-1EEGFP from a modified cytomegalovirus immediate-early promoter which is inhibited on binding a fusion of the tetracycline repressor fused to KRAB was induced to express the LMP-1 derivative by adding increasing levels of tetracycline (25). The levels of protein synthesis were measured for 1 h 48 h after addition of tetracycline. The derivative of LMP-1 lacking all of its carboxy-terminal signaling domain inhibited protein synthesis as efficiently as did intact LMP-1 when it is expressed efficiently (Fig. 4C).

In summary, we have found that at levels of expression of LMP-1 found in clones of EBV-immortalized B cells, LMP-1 detectably inhibits the activity of reporters from three promoters. Expression of LMP-1 at levels greater than twofold that found in EBV-immortalized cells can inhibit expression of reporters to 10% of their uninhibited levels. This inhibition is reflected by both a decrease in the steady-state RNA levels and an inhibition of protein synthesis. Levels of LMP-1 that inhibit the accumulation of RNA by 40% inhibit protein synthesis similarly. These findings indicate that at the level of expression of LMP-1 measured in EBV-infected B-cells and above, LMP-1 limits gene expression. We propose that LMP-1's ability to limit gene expression represents a solution to the problem inherent in its being constitutively active while usually supporting benign survival of the infected cell. The region of LMP-1 that encodes these functions may have a parallel in fibroblast growth factor receptor 3, for which the extracellular and membrane-spanning domain limits signaling of its cytoplasmic domain (39).

The inhibition of gene expression mediated by LMP-1 can be robust when LMP-1 accumulates to more than 10⁵ molecules per cell. This inhibition is likely to underlie LMP-1's inhibition of proliferation of cells (12, 19, 25). LMP-1 expressed in a variety of cells at levels of 2 × 10⁵ to 4 × 10⁵ molecules per cell inhibits their proliferation; when expression is reduced, the cells resume proliferation (25). A derivative of LMP-1 which lacks the entire carboxy-terminal domain, and is positioned in the plasma membrane as is wild-type LMP-1, inhibits both gene expression (Fig. 4) and cell proliferation (25). LMP-1's inhibition of gene expression and cell proliferation may therefore be mediated by the same activity.