Latent Membrane Protein 1 of EBV Activates Phosphatidylinositol 3-Kinase to Induce Production of IL-10¹

Epstein-Barr virus is a ubiquitous human γ-herpesvirus that infects B lymphocytes and persists as a chronic infection in the majority of the adult population with no ill effects. Under normal circumstances, EBV resides in a subset of resting memory B cells. Periodic reactivation of the virus is controlled effectively by the CTL arm of the immune response. However, in the absence of an effective cellular immune response, such as during drug-induced immunosuppression, EBV infection can result in the development of malignant B cell lymphomas as in posttransplant lymphoproliferative disorder (PTLD)³ (1). EBV has also been linked to other B cell malignancies, including Hodgkin’s disease and Burkitt’s lymphoma, and to tumors of epithelial origin such as nasopharyngeal carcinoma.

Latent membrane protein 1 (LMP1), a key viral protein expressed during EBV infection, is a proven oncogene in rodent fibroblasts, and, along with the nuclear proteins EBNA1, EBNA2, and EBNA3a and c, is considered essential for EBV transformation of human B cells in vitro (2). LMP1 is required for oncogenesis of the B cell tumors that arise in immunocompromised SCID mice injected with human B lymphocytes from EBV⁺ donors (3, 4). Moreover, transgenic mice expressing LMP1 under control of the Ig H chain promoter have a significantly increased incidence of lymphoma, demonstrating that LMP1 can be oncogenic in vivo (5). Together, these findings indicate that LMP1 is critical to B lymphocyte transformation by EBV.

LMP1 itself is an integral membrane protein containing six transmembrane-spanning domains and a long C-terminal tail (6). The transmembrane domains function to induce oligomerization of LMP1 complexes in the membrane. This clustering of LMP1 proteins brings individual C-terminal tails into proximity, creating suitable docking sites for a variety of signaling adaptor proteins, and allowing LMP1 to act similarly to a constitutively active CD40 (6). Chimeric proteins that replace the LMP1 extracellular and transmembrane domains with a ligand-responsive protein subunit permit controlled oligomerization of LMP1, and use of these systems has shown that LMP1 signal transduction is mediated through the intracellular tail of LMP1 (7, 8). Two regions of the intracellular tail are most important for propagation of downstream signaling, as follows: the membrane-proximal C-terminal activating region (CTAR) 1, which interacts with cellular TNFR-associated factor (TRAF) adaptor proteins, and the membrane distal CTAR2, which interacts with the cellular TNFR-associated death domain protein. Interactions between LMP1 and adaptor molecules lead to signal transduction through the MAPK p38, ERK, and JNK, as well as NF-κB, PI3K, and other cellular pathways (9, 10, 11, 12, 13). Activation of these cellular signaling pathways by LMP1 results in a variety of phenotypic changes associated with malignancy, including the up-regulation of anti-apoptotic genes and the induction of cellular IL-10 (14, 15).

Cellular IL-10 is a multifunctional cytokine that has been implicated in the pathogenesis of EBV-related B cell malignancies. Associations between IL-10 and EBV⁺ PTLD in transplant patients are prevalent, and IL-10 transcripts have been identified in PTLD tumors (16). We have previously shown that IL-10 is an autocrine growth factor for EBV-infected B cell lines derived from patients with PTLD, and that neutralization of IL-10 can significantly inhibit the proliferation of these cells in vitro (17). Similarly, Masood et al. (18) showed that IL-10 acts as an autocrine growth factor in EBV⁺ B cell lymphomas in AIDS patients. Tumor-derived IL-10 may also act to inhibit cellular immune responses, because it has been observed that B cell lymphoproliferative disorders often coexist with functional T cell anergy (19). Thus, it is important to understand the mechanisms by which LMP1-induced signaling pathways promote IL-10 production.

Although p38 has been reported to participate in LMP1-induced IL-10 production (20), PI3K has not previously been implicated. The PI3K family of enzymes, which provides important survival signals in human cancer, generates lipid second messengers that activate downstream signal transduction proteins such as the serine/threonine protein kinase Akt. Class I PI3K are receptor regulated and consist of a catalytic p110 subunit complexed to a p85 or p55 regulatory subunit, which activates PI3K upon recruitment to the cell membrane (21). Akt activation as a result of PI3K-dependent membrane translocation leads to the inactivation of Akt target proteins that negatively regulate survival and cell cycle progression, including the tuberous sclerosis complex and glycogen synthase kinase (GSK)3β (21). Repression of these enzymes by PI3K activation lifts the negative regulation of the serine/threonine protein kinase mammalian target of rapamycin (mTOR) by active tuberous sclerosis complex and blocks substrate degradation targeting by active GSK3β.

We previously reported that rapamycin, an immunosuppressive drug that targets the mTOR pathway, significantly inhibits IL-10 production by EBV-infected lymphoblastoid cell lines derived from patients with PTLD (22). Thus, we reasoned that signaling pathways upstream of mTOR activation could be involved in the induction of IL-10 by EBV, and by extension, by LMP1. In this study, we provide the first direct evidence that LMP1 activates the PI3K/Akt/mTOR axis in B lymphocytes and demonstrate that this pathway is critical for induction of IL-10 by LMP1. Our findings indicate that one mechanism by which PI3K enhances IL-10 induction by LMP1 is through the regulation of GSK3β. This kinase is known to regulate CREB, a transcription factor that we show to be activated by LMP1 through p38 activation. We link p38 activation and PI3K activation to distinct regions of the LMP1 signaling tail and identify the underlying mechanisms that determine LMP1-induced IL-10 production downstream of p38 and PI3K activation.

Abs to β-actin and nerve growth factor receptor (NGFR) (clone 20.4) were obtained from Sigma-Aldrich. Biotinylated anti-NGFR was from Chromaprobe. Secondary Abs including PE-conjugated goat anti-mouse IgG, unconjugated goat anti-mouse F(ab′)₂, unconjugated goat anti-mouse IgG, HRP-conjugated polyclonal goat anti-rabbit, and HRP-conjugated polyclonal donkey anti-mouse Abs were purchased from Jackson ImmunoResearch Laboratories. Anti-LMP1 Ab CS.1-4 and isotype control Abs were obtained from DakoCytomation. Anti-TRAF1, anti-phospho-ERK (Tyr²⁰⁴), anti-ERK, and anti-p38 were from Santa Cruz Biotechnology. Anti-phospho-p38 (Thr¹⁸⁰/Tyr¹⁸²) was from New England Biolabs, whereas Abs to IκB, Akt, phospho-Akt (Ser⁴⁷³), p70, phospho-p70 (Thr³⁸⁹), MAPK-activated protein 2 (MAPKAP2), phospho-MAPKAP2 (Thr³³⁴), CREB, phospho-CREB (Ser¹³³), GSK3β, and phospho-GSK3β (Ser⁹) were from Cell Signaling Technology. Streptavidin-PE was obtained from BD Pharmingen, and streptavidin was obtained from Sigma-Aldrich.

AB5, JB7, and MF4 are EBV⁺ cell lines derived from the peripheral blood or lymph nodes of patients with EBV⁺ PTLD (22). The Burkitt’s lymphoma lines BL30 and BL41, the BJAB B lymphoma line, and their EBV-infected counterparts BL30_B95, BL41_B95, and BJAB_B95, were provided by E. Kieff (Harvard Medical School, Boston, MA). BL30.NGFR-LMP1 (clone 12), BL41.NGFR-LMP1 (clone 1), BJAB.NGFR-LMP1 (clone 18), BL41.NGFR-LMP1.CTAR1mut (clone 9), and BL41.NGFR-LMP1.CTAR2mut (clone 4) were generated, as described below. All B cell lines were maintained in RPMI 1640 supplemented with 10% heat-inactivated FCS (Mediatech), 2 mM l-glutamine, and 50 U/ml penicillin-streptomycin (Invitrogen Life Technologies). Growth medium for the NGFR-LMP1 transfectants was additionally supplemented with geneticin (Sigma-Aldrich). The 293 cells were obtained from American Type Culture Collection, maintained in FCS-supplemented DMEM, and transfected using Fugene6 (Roche) in serum-free Opti-mem, according to the manufacturer’s instructions, to confirm expression of plasmid vectors.

A NGFR-LMP1 expression vector was provided by W. Hammerschmidt (GSF National Research Center for Environment and Health, Munich, Germany) (8). The NGFR-LMP1 coding sequence was first modified to replace several amino acids from the NGFR intracellular domain with the equivalent amino acids from the LMP1 intracellular domain, so that aa 272–276 of NGFR (FKRWN) was replaced with aa 185–189 of LMP1 (HGQRH). NGFR-LMP1.CTAR1mut was generated by replacing aa 206–210 of LMP1 (PQQAT) with AQAAT using overlapping PCR primers. NGFR-LMP1.CTAR2mut was generated by replacing aa 384–386 of LMP1 (YYD) with ID by PCR. The modified constructs were then cloned into pcDNA3 (Invitrogen Life Technologies) using PCR-appended HindIII and XbaI sites. Sequencing confirmed the identity of each construct, and construct functionality was confirmed by NF-κB luciferase reporter assays in transiently transfected 293 cells. Stable BL30, BL41, and BJAB clones expressing NGFR-LMP1 constructs were generated by electroporating 5 × 10⁶ parental cells with 15 μg of plasmid using a Bio-Rad GenePulser at 210 V and 960 μF. After 48 h of recovery, transfected cells were selected by limiting dilution plating in 96-well plates with geneticin (Invitrogen Life Technologies) at concentrations of 1 mg/ml (BL30), 700 μg/ml (BL41), or 3 mg/ml (BJAB). Expression of NGFR-LMP1 was verified in the selected clones by immunofluorescent staining and flow cytometry for membrane NGFR and by Western blotting of cell lysates for LMP1.

For cell surface staining, one million cells were washed with cold FACS buffer (PBS, 1% BSA, 0.02% sodium azide) and incubated on ice for 30 min with 1 μg of primary Ab, washed, then incubated 30 min on ice with 0.4 μg of the fluorophore-conjugated secondary Ab. Stained cells were then washed and analyzed on a BD Biosciences FACScan using CellQuest software.

To induce signaling by NGFR-LMP1 cross-linking, cells were first resuspended at 5–10 × 10⁶ cells/ml in RPMI 1640 with 4% FCS. Cells were then labeled with 0.5 μg of unconjugated mouse anti-NGFR per 10⁶ cells for 15 min at room temperature, diluted to 1 × 10⁶ cells/ml, and triggered to signal with 2 μg of goat anti-mouse IgG per 10⁶ cells for the indicated amount of time at 37°C. Results were comparable when NGFR-LMP1 cross-linking was initiated with mouse anti-NGFR and goat anti-mouse F(ab′)₂ or with biotinylated anti-NGFR and streptavidin at the same concentrations. Excess cross-linking Ab was neutralized by the addition of mouse IgG isotype control Abs when required.

To analyze the effect of cellular signaling pathways on phosphorylation events and on IL-10 production, cells were first preincubated for 1 h in RPMI 1640 medium containing 4% FCS and either the signaling inhibitor compound rapamycin (10 ng/ml), PD98059 (5 μM), SB203580 (5 μM), LY294002 (2.5 μM for BL30 lines, 5 μM for BL41 and patient-derived lines, and 10 μM for BJAB lines), or SB216763 (5 μM for BL30 lines and 10 μM for BL41 and BJAB lines). Cells were then washed and incubated with the indicated inhibitor for the duration of the experiment, up to 2 h for phosphorylation studies, or up to 24 h for IL-10 studies. All inhibitors were obtained from Calbiochem and prepared as stock solutions at a minimum of ×2000 in DMSO. Control cultures were exposed to an equivalent concentration of DMSO. Proper dosage of inhibitor for each cell line was determined in preliminary experiments by its ability to specifically block phosphorylation of the relevant signaling molecule without inducing cell death or toxicity (data not shown).

IL-10 measurements were made, as previously described, with some modifications (17). Briefly, 96-well immulon plates were coated overnight with anti-IL-10 mAb (clone 9D7; Pierce Biotechnology). Wells were blocked with 5% nonfat dry milk, supernatant samples were incubated 2 h, and following washes bound cytokine was detected with biotinylated anti-IL-10 (clone JES3-12G8) and streptavidin-HRP (both from BD Pharmingen). Data from substrate (tetramethylbenzidine; Sigma-Aldrich) activity were generated as directed by the manufacturer and quantified using a 96-well microplate reader (Molecular Devices). IL-10 produced per million cells was calculated based on the live cell counts, as determined by trypan blue exclusion, at the time that cytokine-containing supernatants were harvested.

Cells were stimulated for the indicated amount of time with anti-NGFR Abs plus cross-linkers, with or without incubation with signal transduction inhibitors, as indicated. Lysates were generated, as previously described (14). Briefly, following signal quenching with ice-cold PBS containing phosphatase inhibitors, cells were lysed in ice-cold Nonidet P-40/deoxycholine phospholysis buffer with pulse sonication. Following sedimentation of insoluble materials and quantification using the Dc Protein Assay (Bio-Rad), samples (30–40 μ/lane) were separated by SDS-PAGE and transferred to a 0.45-μm nitrocellulose membrane. Membranes were blocked with either nonfat dry milk or BSA in TBST, as recommended by the Ab manufacturer, then probed with Abs against phosphorylated signaling molecules at the recommended dilutions. Ab binding was detected with either donkey anti-mouse IgG-HRP or goat anti-rabbit IgG-HRP, as required. Blots were developed with ECL, according to the manufacturer’s instructions (Amersham Pharmacia).

Cells were pretreated for 1–2 h with inhibitors or DMSO controls and induced to signal through LMP1. At 18 h, total RNA (from 5 × 10⁶ treated cells) was isolated with TRIzol reagent (Invitrogen Life Technologies), according to the manufacturer’s instructions. One microgram of mRNA was used for synthesis of first-strand cDNA with an oligo(dT) primer (Invitrogen Life Technologies). Quantitative real-time RT-PCR was done on a Stratagene Mx3000P machine, according to the manufacturer’s instructions, using SYBR Green PCR master mix (Applied Biosystems), including ROX internal controls. Specific primers used were as follows: 5′-ATGCTTCGAGATCTCCGAGA-3′ and 5′-AAATCGATGACAGCGCCGTA-3′ for human IL-10, and 5′-ACCACAGTCCATGCCATCAC-3′ and 5′-TCCACCACCCTGTTGCTGTA-3′ for human GAPDH. Results for IL-10 quantitation were normalized to GAPDH for each sample.

Cells were transfected using an Amaxa Nucleofector II, according to the manufacturer’s instructions. For plasmid transfections, cells were transfected with either 2 μg of a pcDNA.Myr-Akt construct (a gift from D. Stokoe, University of California, San Francisco, CA) or 2 μg of control pcDNA, and selected using 50 μg/ml hygromycin. For knockdown experiments, cells were transfected with GSK3β validated Stealth RNAi oligos (Invitrogen Life Technologies) or a control medium GC oligo at 200 mM. FITC oligo transfection controls indicated >95% transfection efficiency. Transfected cells were incubated for 96 h, then induced to signal through NGFR-LMP1 for 24 h, and supernatants were analyzed for IL-10 production per million cells, as described.

IL-10 induction by EBV has been reported to be partially dependent upon activation of the p38 MAPK pathway (20). Whether other EBV-driven signal transduction pathways contribute to IL-10 expression in virally infected cells is unknown. Our laboratory has shown that rapamycin, an immunosuppressive drug that targets mTOR, can inhibit constitutive IL-10 production by EBV-infected lymphoblastoid B cell lines from patients with PTLD (22). Using PTLD-derived cell lines (AB5, JB7, and MF4) as well as B lymphoma cell lines infected in vitro with the B95.8 strain of EBV (BL30_B95, BL41_B95, and BJAB_B95), we tested whether inhibition of known EBV-activated cellular signaling pathways could affect IL-10 production. As expected, the p38 inhibitor, SB203580, inhibited IL-10 production by all six EBV⁺ cell lines by 25–55% compared with untreated cells (Fig. 1). As previously observed, rapamycin inhibited IL-10 production by the patient-derived EBV-infected cell lines by 40–65%. Rapamycin also inhibited (40–60% decrease) IL-10 production by the in vitro generated EBV⁺ B cell lines BL30_B95, BL41_B95, and BJAB_B95 (Fig. 1). Because mTOR can be activated through the PI3K pathway, we tested whether the PI3K inhibitor, LY294002, altered IL-10 production in EBV-infected B cells. PI3K inhibition markedly diminished (25–85% reduction) IL-10 production by all six EBV-infected B cell lines. In fact, in five of the six EBV⁺ B cell lines tested, PI3K inhibition was more effective than p38 inhibition in suppressing IL-10 production. These signaling effects were specific, because neither the vehicle control nor PD98059, an ERK inhibitor, had an effect on IL-10 production by any of the EBV⁺ B cell lines (Fig. 1). Together, these data indicate that p38, PI3K, and mTOR pathways participate in EBV-induced IL-10 production.

EBV-infected B cells express multiple latent viral genes, in addition to LMP1, that can affect B cell phenotype and function. We hypothesized that the suppression of IL-10 production by signal transduction pathway inhibitors in EBV-infected B cells was due to disruption of LMP1 signaling pathways because LMP1 is the major EBV gene product linked to IL-10 production. To directly test the role of LMP1 signaling in the production of cellular IL-10, we stably transfected a NGFR-LMP1 chimeric construct into the parental (EBV-negative) BL30, BL41, and BJAB cells and selected clones with similar expression levels. This chimeric NGFR-LMP1 system has the advantage of allowing controlled, inducible LMP1 signaling as opposed to the constitutive LMP1 signaling characterizing the native molecule. Transfected cells express NGFR-LMP1, which can be detected at the cell membrane using anti-NGFR Abs and flow cytometry, as well as in whole cell lysates by Western blot using anti-LMP1 Abs (Fig. 2, A and B). Because TRAF1 expression is known to be up-regulated upon LMP1 signaling (23), TRAF1 up-regulation is a good means of confirming the signaling functionality of the NGFR-LMP1 chimeric protein upon cross-linking of the surface NGFR moiety with anti-NGFR and secondary Abs. Following 24 h of NGFR-LMP1 signaling, TRAF1 up-regulation can be easily detected in cell lysates indicating NGFR-LMP1 is functional (Fig. 2 B).

We next established that LMP1 signaling induced IL-10 expression in B cell lines expressing the NGFR-LMP1 chimeric protein. Both BL30.NGFR-LMP1 cells and BJAB.NGFR-LMP1 cells secrete IL-10 in the absence of NGFR-LMP1 cross-linking, equivalent to that of the untransfected parental cell lines. However, induction of LMP1 signaling by cross-linking with anti-NGFR and secondary Abs increased IL-10 production nearly 5-fold in BL30.NGFR-LMP1 and BJAB.NGFR-LMP1 (Fig. 2,C, left and right panels). Whereas the BL41.NGFR-LMP1 cells produce minimal IL-10 in the absence of LMP1 signaling, cross-linking of NGFR-LMP1 induces significant IL-10 (180 pg/ml) (Fig. 2 C, middle panel). Identical Ab treatment of the untransfected parental cells did not affect IL-10 levels (data not shown). Together these data indicate that cross-linking of NGFR-LMP1 in BL30, BL41, and BJAB cells expressing the chimeric molecule triggers LMP1 signaling and induces IL-10 production.

To further investigate early intracellular signaling processes activated by LMP1, we used the BL41.NGFR-LMP1 stable transfectants, which lacked the background IL-10 production of the other two cell lines. Induction of LMP1 signaling by NGFR-LMP1 cross-linking led to time-dependent degradation of IκB, an early requirement for NF-κB activation, as well as rapid, time-dependent phosphorylation of p38, ERK, and Akt (Fig. 3). Additionally, the mTOR substrate p70-S6K was observed to undergo time-dependent phosphorylation in response to LMP1 signaling (Fig. 3). Together, these findings provide the first evidence that LMP1 signaling can activate PI3K in B cells and can lead to PI3K/mTOR-dependent phosphorylation of p70-S6K.

In EBV-infected B cells, disruption of p38, PI3K, or mTOR, but not ERK, signaling could inhibit IL-10 production, and each of these pathways could be activated by NGFR-LMP1 cross-linking. Therefore, we next directly investigated whether LMP1-dependent IL-10 production required p38, PI3K, and mTOR signaling. In NGFR-LMP1-transfected cell lines, inhibition of p38 reduced IL-10 induction by ∼30–50% in the three cell lines tested, whereas inhibition of PI3K reduced IL-10 induction by 45–95% (Fig. 4,A). mTOR inhibition diminished IL-10 induction in BL30.NGFR-LMP1 (by 45%) and in BL41.NGFR-LMP1 (by 35%), but had no effect on BJAB.NGFR-LMP1 cells (Fig. 4 A). In contrast, ERK inhibition had no effect on IL-10 production in any of the three NGFR-LMP1-transfected cell lines, consistent with its inability to alter IL-10 production in EBV-infected B cell lines. Thus, maximal induction of IL-10 by LMP-1 in B cells requires p38, PI3K, and mTOR signal transduction.

We observed no off-target effects by the inhibitors in these studies. As demonstrated in the BL41.NGFR-LMP1 transfectants, the concentration of small molecule signal transduction inhibitors used resulted in target-specific inhibition of phosphorylation (Fig. 4,B). Predicted inhibitory effects on the phosphorylation of downstream signaling events could be detected, as seen with the p38 substrate MAPKAP2 in the presence of the p38 inhibitor, the PI3K substrate Akt in the presence of the PI3K inhibitor LY294002, and the mTOR substrate p70-S6K in the presence of either the mTOR inhibitor rapamycin or the PI3K inhibitor LY294002 (Fig. 4 B).

To further establish the importance of the PI3K/Akt pathway in LMP1-induced IL-10 production, we used constitutively active Akt constructs. Akt is normally activated by kinases following translocation to the membrane; addition of a myristoylation sequence targets Akt to the membrane leading to constitutive activation. Activated Myr-Akt can be detected as a higher m.w. phospho-Akt band on SDS-PAGE gel (24). Cells transfected with a Myr-Akt construct were found to express both endogenous Akt and constitutively active myristoylated Akt (Fig. 5,A). When BL41.NGFR-LMP1 cells expressing Myr-Akt were triggered to signal through LMP1, significantly more IL-10 protein (65% increase) was produced per cell in comparison with cells transfected with a control vector (Fig. 5,B). Expression of Myr-Akt in LMP1 signaling cells also enhanced IL-10 mRNA expression by more than 2-fold (Fig. 5 C). Together, these results confirm a key role for the PI3K/Akt pathway in the induction of IL-10 by LMP1.

LMP1 has two major intracellular signaling regions, designated CTAR1 and CTAR2 (25). We hypothesized that these regions might have differential contributions to the IL-10 induction capacity of LMP1. Mutational studies have identified the key site for TRAF binding within the CTAR1 region to be aa 206–210 (PQQAT), whereas the key site for TNFR-associated death domain protein binding within the CTAR2 region has been identified as the terminal 3 aa (YYD) (26, 27). We therefore transfected BL41 cells with NGFR-LMP1 chimeric proteins modified to contain either a mutation in the key CTAR1 region (PQQAT→AQAAT, CTAR1mut) or a mutation in the key CTAR2 region (YYD→ID, CTAR2mut) and selected clonal transfectants that expressed surface NGFR at approximately the same level as the BL41.NGFR-LMP1 transfectant expressing the unmodified EBV B95.8 LMP1 signaling domains (in this study referred to as wild type) (Fig. 6 A).

Using these transfected cells, we tested the contribution of the CTAR binding regions to Akt and p38 activation and to IL-10 induction by LMP1. We first investigated the time course and magnitude of phosphorylation of Akt and p38 signaling intermediates by CTAR mutant NGFR-LMP1. CTAR1 mutant LMP1 signaling generated reduced Akt activation compared with the wild-type and CTAR2 mutant transfectants (Fig. 6,B, top panel), in agreement with the recent report of CTAR1 dependence of PI3K activation by LMP1 in fibroblasts (28). Although both CTAR mutants had diminished p38 activation in relation to the wild type, the CTAR2 mutant was least effective in activation of p38 (Fig. 6,B, bottom panel). As demonstrated previously, wild-type LMP1 signaling results in significant IL-10 production by BL41.NGFR-LMP1 cells. Signaling by the Akt signaling-deficient CTAR1 mutant LMP1 resulted in a 90% reduction in IL-10 production, whereas signaling by the p38 signaling-deficient CTAR2 mutant LMP1 resulted in a 60% reduction in IL-10 production (Fig. 6 C). These data demonstrate that CTAR1 is required for maximal AKT activation, whereas CTAR2, and to a lesser extent CTAR1, are necessary for p38 activation by LMP1. Moreover, these results confirm the importance of both p38 and PI3K signal transduction in the induction of IL-10 by LMP1.

We next sought to determine which molecular mechanisms were responsible for the p38 and PI3K dependence of LMP1 induction of IL-10. Having established that inhibition of p38 or PI3K markedly diminished IL-10 protein levels in response to LMP1 activation, we tested whether IL-10 mRNA levels were similarly affected. Quantitative analysis of IL-10 mRNA levels by real-time RT-PCR following LMP1 signaling showed that inhibition of p38 reduced IL-10 transcription by ∼45%, whereas inhibition of PI3K reduced IL-10 transcription by over 90% (Fig. 7,A). These decreases in IL-10 mRNA expression closely reflect the decreases in IL-10 protein seen with p38 and PI3K inhibition (Fig. 3 A).

The transcription factor CREB is known to participate in TLR- and CD40-dependent IL-10 production in macrophages (29, 30). We therefore examined whether CREB was activated in response to NGFR-LMP1 signaling and whether p38 or PI3K pathways contributed to CREB activation. We found LMP1 signaling led to a time-dependent phosphorylation of CREB that peaked at 30–60 min following cross-linking of NGFR-LMP1 (Fig. 7,B). CREB activation by LMP1 signaling was p38 dependent, because pretreatment with SB203580 could diminish CREB phosphorylation (Fig. 7,C). In contrast, inhibition of PI3K did not affect LMP1-mediated CREB phosphorylation (Fig. 7 C). These findings indicate that whereas LMP-1 regulates IL-10 transcription through both p38 and PI3K, only p38 activation leads to phosphorylation and activation of the CREB trancription factor.

Although LMP-1-mediated PI3K activation was found to not contribute to CREB phosphorylation, PI3K has been reported to repress the activity of GSK3β, a kinase with multiple functions, including the inhibition of CREB-dependent IL-10 production (31, 32). Inhibition of GSK3β is achieved via phosphorylation at Ser⁹ (33). We therefore investigated whether cross-linking of NGFR-LMP1 in BL41.NGFR-LMP1 transfectants resulted in the phosphorylation of GSK3β at Ser⁹ associated with GSK3β inhibition. LMP1 signaling induced a time-dependent increase in phospho-GSK3β (Fig. 8,A). This phosphorylation of GSK3β was decreased when PI3K activation was inhibited, and conversely, was increased when the specific GSK3β inhibitor SB216763 was present before NGFR-LMP1 signaling. These findings directly demonstrate a role for PI3K in GSK3β inhibition (Fig. 8 B).

Pharmacologic inhibition of GSK3β by SB216763 led to increased expression of IL-10 mRNA and protein following NGFR-LMP1 signaling in BL41.NGFR-LMP1 transfectants, demonstrating that GSK3β participates in regulation of IL-10 production (Fig. 8,C). Importantly, concomitant GSK3β inhibition could partially reverse the effects of PI3K inhibition on IL-10 protein and mRNA in BL41.NGFR-LMP1 cells (Fig. 8,C). To further establish the role of GSK3β in regulation of LMP1-induced IL-10 expression, we introduced GSK3β-targeted small interfering RNA into BL41.NGFR-LMP1 cells. Delivery of GSK3β small interfering RNA (siRNA) resulted in knockdown of GSK3β protein levels in BL41.NGFR-LMP1 cells by 80% (Fig. 8,D). LMP1 signaling in cells with knocked-down GSK3β resulted in enhanced IL-10 production compared with control-treated cells (Fig. 8 E). Taken together, these data indicate that LMP-1-induced activation of PI3K is critical for IL-10 production and that GSK3β plays a central role in the regulation of this pathway.

The EBV-encoded protein, LMP1, is a proven oncogene that usurps cellular signal transduction pathways to promote the survival and proliferation of infected host cells. In B cells, LMP1 is known to activate production of cellular IL-10, a multifunctional cytokine that can act both to suppress the generation of T cell immunity and as a potent autocrine growth factor for B cell lymphomas (17, 20, 34, 35). In this study, we show for the first time that activation of the PI3K/Akt pathway is essential for optimal induction of IL-10 by LMP1, adding to the prior knowledge that the p38 pathway is involved. We demonstrate that in B cells PI3K activation by LMP1 primarily requires the CTAR1 region of LMP1, whereas the CTAR2 region of LMP1 is most associated with p38 activation. We further show that enhanced Akt activation increases IL-10 induction. Although LMP1 can activate a variety of cellular signaling pathways, inhibition of only the p38 or the PI3K/mTOR pathway results in decreases in B cell-derived IL-10.

These findings suggest a mechanism by which drugs targeting the p38 or PI3K/Akt/mTOR pathways could have direct antitumor effects through inhibition of the autocrine growth factor IL-10. In fact, the potent immunosuppressant rapamycin, which targets the mTOR pathway, has been shown in animal models to have antitumor properties and to inhibit tumor-derived IL-10 (22). High serum IL-10 levels are associated with EBV-dependent PTLD, increasing before the clinical detection of PTLD and decreasing with drug-induced resolution (16, 36). In light of the observed connection between high IL-10 levels and tumor progression, treatment strategies targeted at decreasing tumor-signaling pathways for IL-10 production should be of great value in treating B cell lymphomas, especially those induced by EBV infection. Interestingly, one of the mechanisms by which the anti-CD20 Ab Rituximab has been reported to sensitize non-Hodgkins lymphoma cells to chemotherapy is by down-regulating IL-10 production through effects on the p38 pathway (37).

PI3K has been recently implicated in regulation of IL-10 expression following ligation of cellular CD40 or TLR in monocytes and macrophages. CD40-dependent and TLR2-dependent induction of IL-10 in human peripheral monocytes was decreased by inhibitors of PI3K (29, 30). Similarly, overexpression of constitutively active Myr-Akt in a murine macrophage line increased LPS-induced IL-10 production, whereas knockdown of Akt by siRNA decreased LPS-induced IL-10 production (24). Although these studies have enhanced our understanding of the regulation of IL-10 in monocytes/macrophages, none addressed the regulation of IL-10 in B cells or the role of EBV. In this study, we show that LMP1-induced IL-10 production in both EBV-infected and NGFR-LMP1-transfected B cells is PI3K dependent as well as p38 dependent. Despite reports that ERK is essential to TLR-dependent monocyte production of IL-10 (38, 39), our findings indicate that ERK is dispensable for LMP1-induced IL-10 production in B cell lymphomas. Thus, the cellular context is likely to influence mechanisms of IL-10 production; along those lines, recent reports indicate that in macrophage ERK activation is key to enhancing accessibility of the IL-10 promoter locus (40), raising the possibility that the B cell lines used in this study may have a more accessible IL-10 promoter locus.

Our data demonstrate that both p38 and PI3K pathways strongly affect the expression of IL-10 mRNA. The PI3K/mTOR pathway through p70-S6K or other regulators was found to have an effect on IL-10 production as well, most likely through translational mechanisms. The mechanisms by which B cell IL-10 expression may be regulated are complex, because the IL-10 promoter contains over 74 potential transcription factor sites, including several cAMP response element sites (41). We investigated the effects of the p38 and PI3K pathways on CREB, a transcription factor known to bind to cAMP-responsive elements in the IL-10 promoter region and to be involved in IL-10 induction by LPS (42). Our results indicate that p38 activation by LMP1 leads to phosphorylation of CREB. However, other transcription factors may also play a role in the induction of IL-10 through LMP1-induced p38 and PI3K activation. C/EBP is another p38-regulated transcription factor that can bind to cAMP response element sites in the IL-10 promoter (43), whereas Sp1 is a p38-regulated transcription factor for which IL-10 promoter binding sites exist (44). In addition, PI3K activation may also affect Sp1 function, as has been demonstrated for the induction of vascular endothelial growth factor (45). Finally, STAT3 sites are also present in the IL-10 promoter, which allows for an autocrine loop by which IL-10R activation of STAT3 could further drive IL-10 production (41). In fact, this mechanism might account for the differential signaling pathway sensitivity of IL-10 production that we observed in high IL-10-producing BJAB cells. Future studies should continue to improve our understanding of which transcription factors contribute to the mechanism by which LMP1, and its endogenous homologue CD40, regulate IL-10 production in B cells.

We propose in this study that LMP1 signaling through PI3K plays an essential role in IL-10 production in B cells through secondary effects on CREB as mediated by GSK3β. Although the PI3K pathway does not have a direct effect on CREB phosphorylation, the PI3K-regulated kinase GSK3β has been found by others to inhibit CREB DNA-binding activity (31). GSK3β may negatively regulate CREB through inhibitory phosphorylation at Ser¹²⁹ (46, 47) to modulate its association with regulatory proteins such as CBP (48). We demonstrate that GSK3β is inactivated through PI3K-dependent LMP1 signaling in human B cells, and that GSK3β inactivation by exogenous inhibitors or by siRNA enhances LMP1-dependent IL-10 production. Thus, we provide a possible mechanistic framework by which the dual activation of p38 and PI3K/Akt pathways by LMP1 in B cell lymphomas could lead to production of the autocrine growth factor and immunosuppressive cytokine IL-10. Understanding how LMP1 might use such endogenous signaling pathways to induce IL-10 production should be of great use in determining new methods of targeting EBV-related diseases.

We thank Sheri M. Krams and Andrew L. Snow for helpful discussions, and Maria Vaysberg and Lauren Ehrlich for reviewing the manuscript.

The authors have no financial conflict of interest.