The Epstein-Barr Virus-encoded G Protein-coupled Receptor BILF1 Hetero-oligomerizes with Human CXCR4, Scavenges Gα_i Proteins, and Constitutively Impairs CXCR4 Functioning^*

Materials

Cell culture media, trypsin, and accutase were obtained from PAA Laboratories (Pasching, Austria). Forskolin, pertussis toxin, and α-bungarotoxin-tetramethylrhodamine (TMR-αBTX) were purchased from Sigma-Aldrich. Penicillin and streptomycin were obtained from Lonza (Verviers, Belgium). Fetal bovine serum was purchased from Integro BV (Dieren, The Netherlands). HEPES and d-luciferin were purchased from Duchefa Biochemie (Haarlem, The Netherlands). Bovine serum albumin fraction V (BSA) was obtained from Roche Applied Science. [¹²⁵I]α-Bungarotoxin (αBTX; 2200 Ci/mmol), [¹²⁵I]CXCL12 (2200 Ci/mmol), [³H]histamine (18.1 Ci/mmol), and [³⁵S]GTPγS (1250 Ci/mmol) were obtained from PerkinElmer Life Sciences. CXCL12 was purchased from PeproTech (Rocky Hill, NJ), and Vectashield Hardset mounting medium was obtained from Vector Laboratories.

Plasmids

DNA plasmids encoding the full-length and complementation protein fragments of mVenus and Rluc8 were kindly provided by Dr. J. A. Javitch (Columbia University, New York) (22). To facilitate subcloning, the 5′-end of the original 24-amino acid N-terminal linker sequence of these constructs was substituted with SpeI/NotI restriction endonuclease sites by PCR, whereas the XhoI/XbaI sites were introduced immediately downstream of the stop codon. FLAG-BILF1, BBS-BILF1, BBS-BILF1-Rluc, and CXCR4-eYFP constructs have been described previously (21). Fusion of the 13-amino acid αBTX-binding site (BBS) to the N terminus of BILF1 allows detection of receptor proteins by high affinity binding of TMR-αBTX or [¹²⁵I]αBTX. Rluc- and eYFP-encoding sequences were substituted with the DNA sequence of the full-length and complementation protein fragments of Rluc8 and mVenus using NotI and XbaI. The BILF1-K^3.50A construct was kindly provided by Dr. P. Beisser (Maastricht University Medical Center, The Netherlands). The K^3.50A mutation was subcloned into BBS-BILF1/pcDEF3 by using Psp5II and Van9Il. All constructs were verified by DNA sequencing. DNA encoding the GABA_B2-YFP fusion protein was a gift from Dr. J.-P. Pin (Institut de Génomique Fonctionnelle, Montpellier, France). Myc-tagged human histamine H₄ receptor (myc-H₄R) has been described previously (23). The cyclic AMP-responsive element-binding protein (CREB)-driven luciferase reporter gene plasmid pTLNC-21CRE was provided by Dr. W. Born (National Jewish Medical and Research Center, Denver, CO). The cDNA clone for human Gα_i1 was obtained from the Missouri S&T cDNA Resource Center.

Cell Culture and Transfection

HEK293T cells were cultured and transiently transfected with the indicated amounts of DNA using 25-kDa linear polyethylenimine (Polysciences) as described previously (21). Total DNA in gene dosing experiments was kept constant by the addition of “empty” pcDEF3 plasmid.

Protein Fragment Complementation and BRET

HEK293T cells were cultured and transfected in white-bottomed 96-well plates. BiFC and BiLC were monitored 48 h after transfection as fluorescence (excitation at 498 nm and emission at 535 nm; 1-s recording) and luminescence in the presence of 5 μm coelenterazine H (emission at 460 nm; 1-s recording at 5 min after the addition of coelenterazine H (Promega)), respectively, using a Victor³ 1420 multilabel plate reader (PerkinElmer Life Sciences). BRET was measured as described previously (21).

BiFC Microscopy

HEK23T cells were cultured and transfected in 10-cm dishes. Twenty-four hours after transfection cells were seeded on poly-l-lysine-coated coverslips in 6-well plates. The next day, coverslips were labeled with 0.6 μm TMR-αBTX in phosphate-buffered saline for 30 min at room temperature. Coverslips were then washed twice with phosphate-buffered saline and once with H₂O. To stain cell nuclei, 4′, 6-diamidino-2-phenylindole (1 μg/μl) was added during the last wash step. Coverslips were placed on a microscope slide with one drop of Vectashield Hardset mounting medium and stored at 4 °C until further analysis. Cells were visualized using an Olympus FSX100 fluorescence microscope.

Enzyme-linked Immunosorbent Assay (ELISA)

Forty-eight hours after transfection, protein expression was determined by ELISA as described previously (24), with some minor changes. Cells were incubated with the anti-CXCR4 monoclonal IgG_2a antibody, 12G5 (fusin) (1:800 dilution, 200 μg/μl, Santa Cruz Biotechnology). As secondary antibody, horseradish peroxidase-conjugated goat anti-mouse antibodies (1:2000 dilution, Bio-Rad) were used. Peroxidase activity was visualized using the 3,3′,5,5′-tetramethylbenzidine liquid substrate system (Sigma-Aldrich).

Radioligand Binding

Whole cell binding was performed 48 h after transfection using ∼0.25 nm [¹²⁵I]αBTX as described (21). Alternatively, cell membrane fractions from HEK293T cells expressing CXCR4, BBS-BILF1, and/or BBS-BILF1-K^3.50A were prepared as described previously (24). Cell membranes were incubated in 96-well plates in binding buffer (50 mm HEPES, 1 mm CaCl₂, 5 mm MgCl₂, and 100 mm NaCl, pH 7.4) containing 0.2% BSA with ∼0.25 nm [¹²⁵I]αBTX or 0.08 nm [¹²⁵I]CXCL12 for 2 h at room temperature. Aspecific radioligand binding was determined in the presence of 0.1 μm unlabeled αBTX or CXCL12, respectively. Incubations were terminated by filtration through Unifilter GF/C plates (PerkinElmer Life Sciences) presoaked in 0.3% polyethylenimine and washed three times with ice-cold binding buffer supplemented with 0.5 m NaCl. Radioactivity was measured using a MicroBeta Trilux (PerkinElmer Life Sciences).

[³⁵S]GTPγS Binding

Cell membranes were incubated in 96-well plates in assay buffer (50 mm HEPES, 10 mm MgCl₂, and 100 mm NaCl, pH 7.2), 5 μg of saponin, 3 μm GDP, and 0.5 nm [³⁵S]GTPγS, in the absence or presence of 100 nm CXCL12, in a total volume of 100 μl. Reactions were incubated at room temperature for 1 h and then terminated by filtration through Unifilter GF/B plates (PerkinElmer Life Sciences) and washed three times with ice-cold washing buffer (50 mm Tris-HCl and 5 mm MgCl₂, pH 7.4). Radioactivity was measured using a Microbeta Trilux.

CREB Reporter Gene Assay

HEK293T were cultured in white-bottomed 96-well plates and co-transfected with 500 ng of reporter gene plasmid, pTLNC-21CRE (consisting of a firefly luciferase gene controlled by a 21-cAMP-responsive element-containing promoter) and 10–50 ng of CXCR4, BBS-BILF1, and/or BBS-BILF1-K^3.50A DNA/10⁶ cells. Forty-eight hours after transfection, the culture medium was aspirated, and cells were stimulated with 100 nm CXCL12 in culture medium supplemented with 1 μm forskolin for 6 h at 37 °C. Luciferase activity was measured as described previously (25).

Data Analysis

All experiments were repeated using cells from independent transfections. Curve fitting and statistical analyses were performed using GraphPad Prism 4 software.

BILF1 Forms Higher Order Hetero-oligomers with CXCR4

The optimized BRET sensors Rluc8 (26) and mVenus (27) were used to show close proximity between BILF1 and CXCR4. Bioluminescence and fluorescence emitted by GPCR-BRET sensor constructs increased linearly with increasing receptor numbers (data not shown), confirming previous observations (28, 29). BBS-BILF1 expression was determined by [¹²⁵I]αBTX binding, whereas expression of CXCR4 constructs was quantified by ELISA. Both BBS-BILF1-Rluc8 and CXCR4-Rluc8 emitted significantly (p < 0.0001) more light per receptor number than the previously described BBS-BILF1-Rluc and CXCR4-Rluc, respectively (Fig. 1A) (21). On the other hand, the mVenus variants of BILF1 and CXCR4 were less or equally fluorescent per receptor, respectively, as compared with their previously used eYFP counterparts (Fig. 1B) (21). BRET experiments were performed using these optimized BRET sensor constructs by transfecting 1 ng of receptor-Rluc8 (donor) DNA in combination with 0–1000 ng of receptor-mVenus (acceptor) DNA/10⁶ cells. Saturable BRET signals indicate specific interactions between BILF1-BILF1, BILF1-CXCR4, and CXCR4-CXCR4 BRET pairs, which is consistent with our previous work (21) (Fig. 1, C and D). Typically, BRET signals were higher for both BILF1 and CXCR4 homodimers as compared with BILF1-CXCR4 heterodimers. The latter was independent of the BRET sensor orientation, as swapping the sensors between BILF1 and CXCR4 yielded a similar BRET signal for the heterodimeric interaction. Nonspecific receptor interactions as a consequence of random collisions were observed as quasi-linear BRET signals in cells co-expressing BBS-BILF1-Rluc8 or CXCR4-Rluc8 with GABA_B2-YFP (Fig. 1, C and D).

To study higher order protein-protein interactions in living cells, we used BiLC and BiFC approaches in combination with BRET. N- and C-terminal complementation fragments of the optimized BRET couple Rluc8 and mVenus have been developed by the Javitch laboratory (22). Rluc8 was split into L1 (amino acids 1–229) and L2 (amino acids 230–311), whereas mVenus was split in V1 (amino acids 1–155) and V2 (amino acids 156–240). Expression of the individual receptor-L1, -L2, -V1, and -V2 fusion constructs in HEK293T cells did not significantly increase luminescence or fluorescence above background (data not shown), confirming previous observations (22). Co-expression of BBS-BILF1-L1 with CXCR4-L2 or CXCR4-L1 with BBS-BILF1-L2 resulted in BiLC (Fig. 2, A and B). Functional reconstitution of the Rluc8 protein occurred only when the L1 and L2 fragments were brought into close proximity by the proteins to which they were fused, confirming that BILF1 and CXCR4 form heterodimers. Co-expression of BBS-BILF1-L1 and BBS-BILF1-L2 in a 1:1 ratio at a total receptor surface level equal to that of BBS-BILF1-Rluc8 (Fig. 2C) yielded a 3.3-fold lower BiLC signal compared with BBS-BILF1-Rluc8 luminescence (Fig. 2D). Considering that three populations of BILF1 homodimers exist in cells co-expressing BBS-BILF1-L1 and BBS-BILF1-L2 (i.e. L1/L1, L1/L2, and L2/L2), the observed BiLC signal suggests that all BILF1 receptors are involved in dimers. Reconstitution of mVenus by BBS-BILF1-V1 and CXCR4-V2 and vice versa, in combination with the binding of TMR-αBTX to surface-expressed BILF1 (Fig. 2, E–H), revealed that BILF1-CXCR4 heterodimers are localized at the cell surface (Fig. 2H) confirming our previous findings (21). The punctuated green areas are newly formed intracellular BILF1/CXCR4 heterodimers, as indicated by the absence of co-localized TMR-αBTX fluorescence (Fig. 2, F–H). On the other hand, internalized receptor complexes are indicated by intracellularly localized TMR-αBTX-labeled BILF1/CXCR4 heterodimers. Higher order BILF1-CXCR4 protein complexes were detected by measuring BRET between reconstituted L1/L2 and V1/V2 heterodimers of BILF1 and CXCR4. To this end, cells were co-transfected with constant amounts of the split Rluc8 constructs (i.e. 10 ng of both L1 and L2 construct DNA/10⁶ cells) with increasing amounts of the split mVenus constructs (i.e. 0–950 ng of both V1 and V2 construct DNA/10⁶ cells). A saturable BRET signal was observed between all combinations of complemented Rluc8 and mVenus, indicating that BILF1 and CXCR4 form oligomeric complexes consisting of at least four protomers (Fig. 3).

Constitutively Active BILF1 Inhibits CXCL12 Binding to CXCR4

To investigate whether hetero-oligomerization of BILF1 and CXCR4 affects chemokine (i.e. CXCL12) binding to the latter, BBS-BILF1 (2.5 or 25 ng DNA/10⁶ cells) and CXCR4 (50 ng DNA/10⁶ cells) were co-expressed in HEK293T cells. Radioligand binding assays were performed on membrane fractions expressing individual or both receptors. CXCR4 expression was evaluated by ELISA on whole cells of the same transfection. BBS-BILF1 significantly impaired binding of CXCL12 to CXCR4 in an expression level-dependent manner, without having a major effect on CXCR4 protein levels (Fig. 4). The slight reduction in CXCR4 protein expression at 25 ng of BILF1 DNA transfection was not the cause of the loss of CXCL12 binding (Fig. 4, B and C), as individually expressed CXCR4 at comparable levels still bound CXCL12 (supplemental Fig. I). To determine whether the constitutive activity of BILF1 hampers CXCL12 binding to CXCR4 by, for example, inducing a cross-conformational change in the assembled CXCR4 protomer, similar binding experiments were performed on membranes co-expressing the constitutively inactive mutant BBS-BILF1-K^3.50A (30). BILF1-K^3.50A has a propensity similar to BILF1 to heterodimerize with CXCR4 (supplemental Fig. II). BBS-BILF1-K^3.50A was co-expressed with CXCR4 at levels comparable to that of BILF1 (Fig. 4A). Similar to BILF1, the transfection of 25 ng of BBS-BILF1-K^3.50A DNA/10⁶ cells resulted in a slight reduction of CXCR4 expression as measured with the anti-CXCR4 antibody 12G5 (Fig. 4B). However, BBS-BILF1-K^3.50A had no significant effect on CXCL12 binding to membranes co-expressing CXCR4, as compared with membranes expressing only CXCR4 (Fig. 4C).

Constitutively Active BBS-BILF1 Inhibits CXCL12-induced CXCR4 Signaling through Gα_i/o Proteins

To evaluate the effect of BBS-BILF1 co-expression on CXCR4-mediated G protein activation, [³⁵S]GTPγS binding to membranes expressing both receptors was measured and compared with membranes that were derived from HEK293T cells that express either BBS-BILF1 or CXCR4 and were mixed in an 1:1 ratio (Fig. 5A). Constitutive binding of [³⁵S]GTPγS was observed in both co-expressing and mixed membranes as a result of constitutive signaling of BBS-BILF1 (Fig. 5A and supplemental Fig. III). Subsequent stimulation of membranes co-expressing BBS-BILF1 and CXCR4 with 0.1 μm CXCL12 did not increase basal [³⁵S]GTPγS binding (Fig. 5, A and B). In contrast, CXCL12 significantly increased [³⁵S]GTPγS binding to mixed membranes, which can be fully blocked by the CXCR4 antagonist AMD3100 (10 μm) (Fig. 5A). CXCL12 induced a significant 1.64-fold increase in [³⁵S]GTPγS binding to CXCR4-expressing membranes as compared with vehicle treatment (Fig. 5B). In contrast, CXCL12 did not change [³⁵S]GTPγS binding to mock, BBS-BILF1-, and BBS-BILF1-K^3.50A-expressing membranes differently from vehicle treatment (Fig. 5B and supplemental Fig. III). Interestingly, co-expression of the constitutively inactive mutant BBS-BILF1-K^3.50A with CXCR4 did not affect CXCL12-induced G protein activation, resulting in a similar increase in [³⁵S]GTPγS binding as observed in membranes that express CXCR4 alone (Fig. 5B and supplemental Fig. III).

Both CXCR4 and BBS-BILF1 signal through Gα_i/o proteins (25, 31, 32). To examine downstream signaling, HEK293T cells were co-transfected with a CREB-driven reporter gene plasmid and CXCR4 and/or BBS-BILF1. Receptor surface expression of CXCR4 was determined by ELISA. BBS-BILF1 slightly reduced CXCR4 expression levels (Fig. 6A). To detect Gα_i/o-mediated inhibitory signaling to CREB, intact cells were incubated with 1 μm forskolin (FSK) leading to a direct activation of adenylyl cyclase. Expression of BBS-BILF1 resulted in a ligand-independent decrease of this FSK-induced CREB activity (supplemental Fig. IVA) as shown previously (25, 32). Stimulation of CXCR4-expressing cells with 0.1 μm CXCL12 resulted in a significant decrease of FSK-induced CREB activity as compared with vehicle treatment (Fig. 6B and supplemental Fig. IVB). However, co-expression of the constitutively active BBS-BILF1 with CXCR4 significantly attenuated this CXCL12-induced inhibition of CREB activity (Fig. 6B). In contrast, the constitutively inactive mutant BBS-BILF1-K^3.50A inhibited CXCL12-induced CXCR4 signaling to a much lesser extent than BBS-BILF1. Overnight pretreatment of the cells with pertussis toxin (100 ng/μl) abolished all BBS-BILF1- and/or CXCR4-mediated inhibition of FSK-induced CREB activity (supplemental Fig. V), confirming signaling through Gα_i/o proteins (25, 31, 32).

Constitutively Active BBS-BILF1 Competes with Co-expressed GPCR for Gα_i/o Proteins

The difference in efficacy between the constitutively active BBS-BILF1 and constitutively inactive mutant BBS-BILF1-K^3.50A in regard to CXCR4 signaling might be related to the stabilization of distinct CXCR4 conformations in their respective hetero-oligomeric complexes, resulting in changed CXCL12 binding and, consequently, signaling. Alternatively, the constitutively active BBS-BILF1 might scavenge Gα_i/o proteins to such an extent that only a very limited amount of G proteins are available for coupling to CXCR4. To evaluate the role of G protein coupling on CXCL12 binding to CXCR4, [¹²⁵I]CXCL12 binding experiments were performed in the presence of 3 μm GTPγS. Binding of CXCL12 to CXCR4 was fully inhibited by preventing G protein coupling to the receptor (Fig. 7A). On the other hand, co-expression of additional Gα_i1 protein with CXCR4 and BBS-BILF1 restored CXCR4-mediated inhibition of CREB activity in response to CXCL12 (Fig. 7B) without significantly affecting receptor surface expression (Fig. 7C). To further explore whether constitutive BBS-BILF1 signaling affects the functioning of other GPCRs that couple to Gα_i/o proteins, HEK293T cells were transfected with BBS-BILF1 or BBS-BILF1-K^3.50A and/or human H₄R. BBS-BILF1 and H₄R receptor were expressed at the cell surface and formed heterodimeric complexes as revealed by BiFC, although internalized TMR-αBTX-labeled BBS-BILF1/H₄R heterodimers were observed as well (Fig. 8, A–D). Histamine (10 μm) significantly decreased FSK-induced CREB activity in HEK293T cells expressing H₄R, as compared with vehicle treatment, without affecting BBS-BILF1 signaling to CREB (Fig. 8E). Co-expression of BBS-BILF1 with H₄R abolished the histamine responsiveness of H₄R, whereas BBS-BILF1-K^3.50A did not disturb histamine-induced signaling to CREB (Fig. 8E). In contrast to [¹²⁵I]CXCL12 binding to CXCR4, [³H]histamine binding to hH₄R-expressing membranes was not affected by BBS-BILF1 (Fig. 8F), which confirms that histamine binding to H₄R is not affected by G protein uncoupling with GTPγS (Fig. 8G) (33).