The human histamine H₄ receptor (hH₄R), a prototypical member of the superfamily of G-protein-coupled receptors (GPCRs), has been identified recently through the use of bioinformatics by several groups simultaneously (Oda et al., 2000; Liu et al., 2001; Morse et al., 2001; Nguyen et al., 2001; Zhu et al., 2001). The H₄R couples to members of the G_i/o family of heterotrimeric G-proteins to mediate the inhibition of adenylyl cyclase. In addition, the receptor may activate phospholipase C and induce calcium mobilization (de Esch et al., 2005). The H₄R expression is almost exclusively restricted to hematopoietic cells and is suggested to mediate functions of the immune system. As such, the H₄R is a target for the development of anti-inflammatory drugs (Hofstra et al., 2003; Thurmond et al., 2004; de Esch et al., 2005).

The use of various biochemical and biophysical approaches has revealed recently that members of the GPCR family may exist as homo- and hetero-oligomers at the cell surface. When considering the heterotrimeric G protein, which is approximately twice the size of the GPCR (Lambright et al., 1996), it seems reasonable that GPCRs need to oligomerize to interact with the G protein, as suggested for the leukotriene B₄ receptor (Baneres and Parello, 2003). Hetero-oligomerization has been shown to be pivotal for the GABA_BR1, which needs to associate with GABA_BR2 receptors to be transported to the cell membrane (Jones et al., 1998), and for the T1R taste receptors, which require hetero-oligomerization to form receptors that can recognize sweets (Nelson et al., 2001) or amino acids (Nelson et al., 2002). In other cases, heterooligomerization may change the ligand binding characteristics, potentially giving rise to a new dimension in GPCR drug discovery (Devi, 2001; Terrillon and Bouvier, 2004; Waldhoer et al., 2005).

We have reported previously the detection of homo-oligomers of the human histamine H₁ receptor (H₁R) by applying biochemical and tr-FRET experiments and by the formation of H₁R radioligand binding sites upon the coexpression of two ligand binding-deficient mutant H₁Rs (Bakker et al., 2004). The H₁R is a well-known target for the treatment of seasonal allergies but has also been shown to mediate inflammatory responses in keratinocytes (Giustizieri et al., 2004; Matsubara et al., 2005). The H₁R is ubiquitously expressed and is coexpressed together with the H₄R in leukocytes, including monocytes and T lymphocytes (Cameron et al., 1986; Morse et al., 2001), suggesting that on these cells, histamine may modulate inflammatory actions through the action on both H₁Rs and H₄Rs. We therefore investigated the potential homo-oligomerization of the H₄R and the heterooligomerization of the H₁R with the H₄R using heterologous expression systems.

Herein, we report on the generation of specific antibodies raised against the H₄R, the detection of homo-oligomers of the H₄R, and the potential formation of H₁R-H₄R heterooligomers by using biochemical and BRET and tr-FRET approaches. Using these methodologies, we show the human H₄R to constitutively form homo-oligomers at the cell surface and that the oligomerization is independent of ligand stimulation of the receptors. Furthermore, N-glycosylation of the H₄R receptor is not a prerequisite for oligomer formation. Although we can detect H₄R homo-oligomers at physiologically relevant H₄R expression levels and in endogenously H₄R expressing PHA blast cells, the detection of H₁R-H₄R hetero-oligomers requires higher receptor expression levels.

Materials and Methods

Materials. Reagents for tr-FRET were from Cis Bio International (Bagnols-sur-Cèze Cedex, France). Coelenterazine was purchased from Chemicon International (Temecula, CA). Sheep anti-mouse IgG horseradish peroxidase was from GE Healthcare (Little Chalfont, Buckinghamshire, UK). Bovine serum albumin, chloroquine diphosphate, DEAE-dextran (chloride form), histamine (2-[4-imidazolyl-]ethylamine hydrochloride), mepyramine (pyrilamine maleate), monoclonal mouse anti-FLAG (DYKDDDDK), and polyethyleneimine were purchased from Sigma (St. Louis, MO). Calf serum (Integro BV, Dieren, The Netherlands). Cell culture media, penicillin, and streptomycin were obtained from Invitrogen (Merelbeke, Belgium). Cell culture plastics were from Greiner Bio-one (Wemmel, Belgium). Tris was from AppliChem (Darmstadt, Germany). [³H]Histamine (12.40 and 18.10 Ci/mmol) and [³H]mepyramine (23.00 Ci/mmol) were purchased from PerkinElmer Life Science (Boston, MA). Oligonucleotides were purchased from Isogen Biocience (Maarsen, The Netherlands). Pfu Turbo DNA polymerase was purchased from Stratagene (La Jolla, CA). Restriction enzymes were from MBI Fermentas (St. Leon-Rot, Germany). Thioperamide, iodophenpropit, clobenpropit, and JNJ 7777120 were synthesized at the Department of Medicinal Chemistry, Vrije Universiteit Amsterdam (Amsterdam, The Netherlands). Gifts of mouse anti-hemagglutinin (anti-HA) antibody (Dr. J. van Minnen), pcDNA3.1-eYFP vector (Dr. T. Schmidt), pRL-CMV vector (Dr. G. Milligan), pCR3.1-HA-H₁R and pcDEF₃-c-myc-H₁R (Dr. S Hill), expression vector pcDEF₃ (Dr. J. Langer) (Goldman et al., 1996), and mianserin (Organon NV, Oss, The Netherlands) are greatly acknowledged.

Wild-type human H₄R in pcDNA3.1 was purchased from Guthrie cDNA resource center (Sayre, PA). The vector was subcloned into the pcDEF₃ using BamHI/XbaI sites.

Construction of Epitope-Tagged Proteins for tr-FRET. An N-terminally FLAG (DYKDDDDK) epitope-tagged H₄R was created by PCR. The coding sequence of the hH₄ gene was amplified using the sense oligonucleotide primer 5′-GGGAAGCTTGCCACCATGGACTACAAGGACGACGATGACAAGGATCCAGATACTAATAGCAC-3′ and the antisense primer 5′-GGAAGG CACGGGGGAGGGC-3′. The amplified gene was first cloned into the pCRII-Topo vector by TOPO TA cloning (Invitrogen BV, Breda, The Netherlands) and subsequently subcloned into the pcDEF₃ expression vector using EcoRV/XbaI sites.

An N-terminally HA (YPYDVPDYA) epitope-tagged H₄R was created by PCR in two steps. The H₄R gene was amplified by PCR with a 5′ SacII site and without start codon using the sense primer 5′-ACCGCGGCCCCAGATACTA ATAGCACAATC-3′and the antisense primer 5′-GGAAGGCACGGGGAGGGC-3′. The fragment was directly cloned to the pCRII-Topo vector. The gene was subsequently subcloned using SacII/XbaI sites into the pcDNA3.1-HA-rH_3AR vector (Bakker et al., 2006). The HA-H₄ gene was finally subcloned using BamHI/XbaI sites to pcDEF₃. The HA-H₁R gene was subcloned from the pCR3.1-HA-H₁R into the pcDEF₃ using Bsp1407I/SpeI restriction sites.

Construction of Fusion Proteins for BRET. For the BRET assay, H₄Rs were C-terminally fused to either a Renilla reniformis luciferase (H₄R-Rluc) or a yellow fluorescent protein (H₄R-eYFP) in two steps. The coding sequence of the hH₄R gene was amplified without its stop codon using the sense primer 5′-TCGGATCCACCATGCCAGATACTAATAGC-3′ and the antisense primer 5′-CCGCGGC CGCACTAGTAGAAGATACTGACCGAC-3′, harboring unique BamHI and NotI restriction sites, respectively. The gene was cloned directly into the pCRII-Topo vector and subsequently subcloned to a pcDEF₃ vector using BamHI/NotI sites [pcDEF₃-H₄R (Del stop)].

The coding sequence for the Rluc gene was amplified from the pRl-CMV vector lacking a start codon and harboring a NotI restriction site using the sense primer 5′-AGCGGCCGCGACTTCGAAAGTTTATGATCC-3′ and the antisense primer 5′-TCTAGAATTATTGTTCATTTTTGAG-3′. The gene was directly cloned to the pCRII-Topo vector and subsequently subcloned in frame using NotI/XbaI sites into the pcDEF₃-H₄R (Del stop) vector.

The coding sequence for the eYFP gene was amplified from the pcDNA3.1-eYFP vector lacking a start codon and harboring a NotI restriction site using the sense primer 5′-CGCGGCCGCGGTGAGCAAGGGCGAGGAG-3′ and the antisense primer 5′-GTCTAGATTACTTGTACAGCTCGTCCATG-3′. The gene was directly cloned to the pCRII-Topo vector and subsequently subcloned in frame using NotI/XbaI sites into the pcDEF₃-H₄R (Del stop) vector.

An hH₁R-eYFP fusion was generated by PCR using the sense primer 5′-AAGAGAATTCTGCATATTCGCTCCATGGTGAGCAAGGGCG-3′ and the antisense primer 5′-TTCTCTAGATTACTTGTACAGCTCGTCC-3′, harboring unique EcoRI and XbaI restriction sites, using pcDNA3.1eYFP as template. The PCR fragment was digested using EcoRI and XbaI, and the purified fragment was subsequently ligated together with the fragment that was obtained by digestion of the pcDEF₃-hH₁R plasmid using EcoRI/XbaI sites.

An hH₁R-Rluc fusion was generated by PCR using the sense primer 5′-AAGAGAATTCTGCATATTCGCTCCATGACTTCGAAAGTTTATGATCC-3′ and the antisense primer 5′-CGCTCTAGAATTATTGTTCATTTTTGAGAACTCGC-3′, harboring unique EcoRI and XbaI restriction sites. The PCR fragment was digested using EcoRI and XbaI and the purified fragment was subsequently ligated together with the fragment that was obtained by digestion of the pcDE-F₃-hH₁R plasmid using EcoRI/XbaI sites. Each construct was fully sequenced before its expression and analysis.

Construction of His₁₀-Tagged Proteins for Immobilization. An N-terminally c-myc (EQKLISEEDL) and C-terminally His₁₀-epitope-tagged H₄R was created as follows. First, a c-myc epitope-tagged H₄R was created by PCR in two steps. The c-myc tag was amplified by PCR using a pcDEF₃-c-myc-H₁R vector as template with a 3′-NheI site using the sense primer 5′-GGGTGGAGAC TGAAGTTAGGCC-3′and the antisense primer 5′-GTGCTAGCAGGTCCTCCTCGGAG-3′. The fragment was directly cloned to the pCRII-Topo vector (pCRII-topo-myc). The H₄R gene was amplified without start codon and contained a 5′-NheI restriction site using the following sense 5′-CCGCTAGCCAGATACTAATAGCAC-3′ and the antisense primer 5′-TCTTTAAGAAGATACTGACC-3′. The gene was directly cloned to the pcDNA3.1/V5-His-Topo vector. The H₄R gene was subsequently subcloned in frame using NheI/NotI into the pCRII-topo-c-myc vector (pCRII-topo-c-myc-H₄R). The c-myc-H₄R gene was subsequently subcloned into the pcDEF₃ expression vector using the BamHI/XbaI sites.

Second, the gene of the wild-type H₄R was amplified by PCR without a start and stop codon with a 5′-BamHI site and a 3′-SpeI site using the sense primer 5′-CCGG ATCCCCAGATACTAATAGCACAATCAA-3′ and the antisense primer 5′-CCGCGGCCG CACTAGTAGAAGATACTGACCGAC-3′ and directly cloned into the pC-RII-Topo vector. The H₄R gene was then subcloned in frame from the pCRII-topo-vector using BamHI/SpeI sites in the pSFV2genB vector. An N-terminally tagged FLAG and C-terminally tagged H₄R-His₁₀ gene was subcloned from the pSFV2genB-FLAG-H₄R-His₁₀ behind the p10 promoter of the pFastbac_DUAL vector using NcoI/NheI restriction sites.

The H₄R-His₁₀ gene was amplified by PCR from the pFastbac_ DUAL-FLAG-H₄R-His₁₀ vector without start codon and a 3′-XbaI site using the sense primer 5′-CATCTAGATTAATTACCCACTGGGCCC-3′and the antisense primer 5′-GAGGATCCGCCAGATACTAATAGCACAATC-3′ and directly cloned into the pcDNA3.1/V5-His-Topo vector by TOPO TA cloning and subsequently subcloned into the pcDEF₃-c-myc-H₄R vector using BoxI/XbaI restriction sites. Each construct was fully sequenced before its expression and analysis.

Cell Culture and Transfection of COS-7 Cells. COS-7 African green monkey kidney cells were maintained at 37°C humidified in 5% CO₂/95% air atmosphere in Dulbecco's modified Eagle's medium supplemented with 5% (v/v) fetal calf serum, 50 IU/ml penicillin and 50 μg/ml streptomycin and grown in 100-mm dishes. Cells were transiently transfected using the DEAE-dextran method as described previously (Bakker et al., 2001). The total amount of DNA transfected was maintained constant by the addition of pcDEF₃.

Cell Culture and Transfection of HEK 293 Cells. HEK 293 cells were maintained at 37°C humidified in 5% CO₂/95% air atmosphere in Dulbecco's modified Eagle's medium/F-12 (Cambrex, Nottingham, UK) supplemented with 10% (v/v) fetal calf serum, 50 IU/ml penicillin, and 50 μg/ml streptomycin and grown in 100-mm dishes. HEK 293 cells were transfected with pcDEF₃-H₄R receptor essentially using the Lipofectamine Plus method described by Shenton et al. (2005). In brief, for each cDNA, two microtubes were prepared: tube 1 contained 2 μg of cDNA, 6 μl of Lipofectamine Plus reagent (Invitrogen), and 150 μl of Opti-MEM I media (Invitrogen); tube 2 contained 5 μl of Lipofectamine reagent (Invitrogen) and 150 μl of Opti-MEM I media. The mixtures were incubated at room temperature for 15 min, after which the contents of tube 2 were added to tube 1, followed by a further 15-min incubation. In the meantime, the HEK 293 cells at 50 to 80% confluence in 2-ml Petri dishes were washed three times with Opti-MEM I media. At the end of the second incubation period, the contents of tube 1 were increased to 1.5 ml with Optimem-I media and added to the washed HEK 293 cells. The cells were incubated at 37°C for 6 h. The transfection mixture was then removed and replaced with growth media. The cells were harvested 48 h after transfection, and cell homogenates were prepared for immunoblotting.

[³H]Histamine Binding Studies. Cells used for radioligand binding studies were harvested 48 h after transfection and homogenized in ice-cold H₄R binding buffer (50 mM Tris, pH 7.4). For saturation isotherms, cell membrane homogenates were incubated at 37°C for 60 min with 0 to 125 nM [³H]histamine in a total assay volume of 200 μl. Nonspecific binding was determined by incubation in the presence of 10 μM JNJ 7777120. For competition binding assays, the cell homogenates were incubated at 37°C for 60 min with 0.1 to 10,000 nM ligand in the presence of ∼15 nM [³H]histamine in a total volume of 200 μl. The incubations were stopped by rapid dilution with ice-cold H₄R binding buffer. The bound radioactivity was separated by filtration through GF/C filter plates (Whatman, Maidstone, UK) that had been treated with 0.3% polyethyleneimine. Filters were washed four times with H₄R binding buffer, and radioactivity retained on the filters was measured by liquid scintillation counting.

[³H]Mepyramine Binding Studies. Cells used for radioligand binding studies were harvested 48 h after transfection and homogenized in ice-cold H₁R binding buffer (50 mM Na²⁺/potassium phosphate buffer, pH 7.4). For saturation isotherms, cell membrane homogenates were incubated at room temperature for 30 min with 0 to 25 nM [³H]mepyramine in a total assay volume of 200 μl. Nonspecific binding was determined by incubation in the presence of 1 μM mianserin. For competition binding assays, the cell homogenates were incubated at room temperature for 30 min with 0.1 to 10,000 nM in the presence of ∼1.5 nM [³H]mepyramine in a total volume of 200 μl. The incubations were stopped by rapid dilution with ice-cold H₁R binding buffer. The bound radioactivity was separated by filtration through Whatman GF/C filter plates that had been treated with 0.3% polyethyleneimine. Filters were washed four times with H₁R binding buffer, and radioactivity retained on the filters was measured by liquid scintillation counting.

Anti-H₄R Antibody Generation. The unique peptide corresponding to the amino acids CIKKQPLPSQHSRSVSS of the human H₄R subtype was conjugated to thyroglobulin by the cysteine-coupling method (Chazot et al., 1998). The resultant conjugate was used to generate polyclonal antibodies in rabbits. Antibody production and affinity purification was performed as described previously (Chazot et al., 2001).

Production of Human PHA Blasts. Human peripheral blood mononuclear cells (PBMCs) stimulated with phytohemagglutinin (PHA blasts) were generated essentially as described previously (Bradford et al., 2005). In brief, heparinized human whole blood was obtained from healthy volunteers (with local ethical approval), and PBMCs were separated using Lymphoprep (Axis-Shield Poc AS, Oslo, Norway) and centrifuged at 400g for 25 min. The PBMCs were isolated from the interfacial layer, washed twice in RPMI 1640 medium without l-glutamine (Invitrogen), and resuspended in RPMI 1640 medium complemented with 10% (v/v) fetal calf serum, 1% (v/v) penicillin and streptomycin, and 1% (v/v) l-glutamine. Cell density was adjusted accordingly to 1 × 10⁶ cells/ml with RPMI 1640 medium. Next, 100 μl of PHA (Lectin; Sigma, Poole, Dorset, UK) was added to the cells to make PHA blasts. These were grown in culture for 24 h, harvested, and a cell homogenate was prepared in the presence of protease inhibitors (Protease Inhibitor Cocktail III; Calbiochem, Beeston, Nottingham, UK).

Immunoblotting. SDS-polyacrylamide gel electrophoresis was carried out using 6 or 7.5% polyacrylamide slab gels under reducing conditions. Samples of HEK 293 cells, COS-7 cells, and PHA blasts (20-50 μg of protein) were prepared using a chloroform/methanol method of protein precipitation, and immunoblotting was performed as described previously (Chazot et al., 2001; Bakker et al., 2006). Immunoblots were probed with anti-H₄ 374-390 antibody at a concentration of 0.5 μg/ml.

Blots containing FLAG or c-myc-tagged receptors were probed with primary antibodies, mouse anti-FLAG (1.5 μg/ml), or mouse anti-c-myc (1 μg/ml), respectively. Horseradish peroxidase-conjugated goat anti-mouse antibodies (1:2000-5000) were used as secondary antibodies.

Immunoprecipitation. HEK 293 cells were transfected with HA-H₄ receptor and solubilized with 1% Triton X-100/0.15 M NaCl for 30 min at 4°C. Immunoprecipitation was performed essentially as described previously (Chazot et al., 1994). Solubilized HEK 293 cell extracts were incubated with 5 μg of rat anti-HA antibody (Roche Diagnostics, Mannheim, Germany) or rat nonimmune Ig (ADI, San Antonio, TX) at 4°C for 1 h. Prechilled, washed Protein G agarose slurry (50 μl; Sigma) was added and incubated for 1 h at 4°C on a rocking platform. Precipitation pellets were collected by centrifugation at 10,000g for 30 s at 4°C, washed with 3× PBS, resuspended in sample buffer, vortex-mixed, and heated to 90 to 100°C for 3 min. The sample was then recentrifuged, and the supernatant was subjected to immunoblotting. Control experiments were performed using untransfected HEK 293 cells.

Cross-Linking Experiments. The cross-linking method used was essentially as described by Shenton et al. (2005; Bakker et al., 2006). In brief, aliquots of COS-7 cells expressing c-myc-H₄Rs were pelleted, and the suspension buffer was removed and replaced with 150 μl of cross-linking buffer (150 mM NaCl, 100 mM sodium HEPES, 5 mM EDTA, pH 7.5, and 5 mM dithiothreitol) to give a final protein concentration of approximately 0.5 mg/ml. The cross-linker [bis(sulfosuccinimidyl) suberate sodium salt] was dissolved in 20 mM HCl to give a 100 mM stock solution. The tubes were incubated at room temperature with continual mixing for 12 min with 0.25, 0.5, 1.0, and 2 mM cross-linker, centrifuged at 10,000 rpm for 5 min, the cross-linking mixture was removed, and the resultant pellet was prepared for immunoblotting.

Tunicamycin Experiments. HEK 293 cells expressing H₄Rs were incubated with 2, 4, 6, and 8 μg/ml tunicamycin (stock dissolved in dimethyl sulfoxide at 2 mg/ml) immediately after transfection and were harvested 48 h after transfection, homogenized, and subjected to immunoblotting (Chazot et al., 1995). Cells grown in the absence of tunicamycin were incubated with the respective volume of dimethyl sulfoxide.

Deglycosylation of Native H₄ Receptor. Human PHA blast cell suspensions were resuspended in deglycosylation buffer (50 μM sodium phosphate, pH 6.0, containing 0.1% SDS, 0.1% β-mercaptoethanol, and 20 mM EDTA) and incubated with either water (control) or PNGase F enzyme (Sigma) at a final enzyme concentration of 400 IU/ml (test) for 16 h at 37°C. The samples were then subjected to immunoblotting and probed with anti-H₄ 374-390 antibody at a concentration of 2 μg/ml. The NMDAR1 transfected into HEK 293 cells was used as a positive control essentially as described by Chazot et al. (1992).

Receptor Immobilization. Membranes of COS-7 cells transiently expressing c-myc-H₄R-His and FLAG-H₄R or HA-H₁R-His and FLAG-H₄R receptors were homogenized, solubilized, and subsequently immobilized on Ni²⁺-NTA columns (Invitrogen) as described previously (Bakker et al., 2006). Immobilized receptors were eluted using 250 mM imidazole. Samples were prepared for immunoblotting and were subjected to chloroform/methanol extraction loaded on a 7.5% SDS page gel and subsequently blotted on nitrocellulose paper (GE Healthcare).

BRET Assay. Forty-eight hours after transfection, cells were detached with trypsin and washed twice with PBS. Approximately 50,000 cells per well were distributed in white-bottomed 96-well microplates (Corning BV, Schiphol-Rijk, The Netherlands). Coelenterazine was added to a final concentration of 5 μM, and readings were collected immediately after this addition using a Victor₂ allowing signal detection at 460 and 530 nm.

tr-FRET Assay.tr-FRET assays were performed using Europium (Eu³⁺)-labeled and allophyocyanin anti-FLAG and anti-HA antibodies as described by Bakker et al. (2006). In brief, tr-FRET was assessed in 1 × 10⁶ whole COS-7 cells transiently expressing the appropriate HA- and FLAG-tagged receptors. Cells were incubated in PBS containing 50% fetal calf serum (v/v), 0.8 nM Eu³⁺-labeled antibody, and 8 nM allophycocyanin-labeled antibody for 2 h at room temperature on a rotating wheel, after which the membranes were washed twice with PBS. The final pellet was resuspended in 50 μl of PBS and transferred to a 384-microtiter plate. Energy transfer was measured by exciting the Eu³⁺ at 320 nm and monitoring the XL-665 allophycocyanin emission for 500 μs at 665 nm using a Novostar (BMG LabTechnologies, Offenburg, Germany) configured for time-resolved fluorescence after a 100-μs delay.

Analytical Methods. Binding data were evaluated by a nonlinear least-squares curve-fitting program using Prism software (GraphPad Software Inc., San Diego, CA). Protein concentrations were determined according to Bradford's method (1976) using bovine serum albumin as standard. All data are represented as mean ± S.E.M. from at least three independent experiments performed in triplicate. Statistical significance was determined by a Student's unpaired t test (p < 0.05 was considered statistically significant).

Results

Pharmacological Characterization of the hH₄R and hH₁R Expressed in COS-7 Cells. We have used COS-7 cells previously successfully for the heterologous expression of the hH₁R and for the identification of H₁R oligomers (Bakker et al., 2004). To investigate the potential oligomerization of the hH₄R, we therefore expressed the hH₄R heterologously in COS-7 cells. Transfection of these cells with cDNA coding for the hH₄R resulted in the expression of a high-affinity [³H]histamine binding site (Table 1). Subsequent displacement studies using [³H]histamine as a radioligand revealed these binding sites to display a characteristic H₄R pharmacological profile (Table 1). The H₁R and H₄R constructs used in the tr-FRET, BRET, and immobilization assays were also characterized by radioligand binding (saturation and displacement) assays. COS-7 cells were transiently transfected with cDNA encoding the HA-H₄R, FLAG-H₄R, H₄R-Rluc, H₄R-eYFP, c-myc-H₄R-His, HA-H₁R, H₁R-Rluc, or the H₁R-eYFP. [³H]Histamine bound to the H₄Rs according to a one-site saturable model with Hill slopes of approximately 1 and dissociation constants (K_d) similar to those of the wild-type hH₄R, although the B_max value is affected by fusion of the R. reniformis luciferase enzyme (Table 1). Bound [³H]histamine could be displaced from all of the N- and/or C-terminally tagged H₄Rs by the agonist histamine and the inverse agonist thioperamide with affinity values (pK_i) comparable with the wild-type (Table 1). Likewise, the H₁R radioligand [³H]mepyramine bound the various hH₁R constructs according to a one-site saturable binding model with K_d values similar to those of the wild-type hH₁R (data not shown). The agonist histamine and H₁R inverse agonist mepyramine were able to displace the radioligand with affinities equal to those of the wild-type H₁R (data not shown). The aforementioned H₄R constructs were functionally characterized using a cAMP response element-luciferase-luciferase reporter gene assay. In these assays, histamine behaved as a full H₄R agonist and thioperamide as a full inverse H₄R agonist for each H₄R construct, with pEC₅₀ values comparable with those obtained for the wild-type H₄R (Table 1).

Generation of hH₄R-Specific Antibodies. To enable our biochemical approaches and to study H₄R function in native tissue, we raised a rabbit polyclonal anti-hH₄ (374-390) receptor antibody, which represents the first published selective immunological probe for the hH₄R. The antibody was generated against the last 17 amino acids of the C-terminal tail of the H₄R (Fig. 1A). The selectivity of the anti-hH₄R antibody was confirmed by blockade with the C-terminal peptide of the H₄R (Fig. 1B, lane 3) and a lack of cross reactivity with the human H₃R, the most related GPCR (de Esch et al., 2005) (Fig. 1B, lane 1). In transfected HEK 293 cells, the antibody detects two major reactive species at 34 to 36 and 65 to 72 kDa (Fig. 1B, lane 2). The lower bands most likely represent monomeric H₄Rs. An additional band (approximately 45 kDa) was occasionally detected; such bands are likely to represent a proteolytic fragment. We suspect the 34-kDa species to be the unglycosylated product of the species at 36 kDa. The higher molecular mass species could either represent a heavily glycosylated form of the H₄R or dimeric H₄Rs.

Evidence that Native H₄R Are Robust Dimers. The H₄R clearly plays a role as an immune modulator, with mRNA expression shown in human mast cells, neutrophils, eosinophils, and T lymphocytes (Nakamura et al., 2000; Oda et al., 2000; Morse et al., 2001; Zhu et al., 2001; Gantner et al., 2002; Hofstra et al., 2003). A single major diffuse immunoreactive species (approximately 77 kDa) coincident with the putative recombinant dimeric hH₄R species expressed in COS-7 cells was detected in human PHA blasts (Fig. 1C, lane 1). This species was abolished by preincubation with the 374-390 peptide, again demonstrating the peptide selectivity of the antibody. Little or no protein monomers were detected in the native preparation, consistent with our previous data with the H₃R (Chazot et al., 2001; Bakker et al., 2006). It is noteworthy that these experiments were performed under reducing conditions, indicating the robust nature of the dimeric species in native tissue. An identical labeling pattern was detected with the anti-hH₄ 374-390 antibody probing human spleen lysates (data not shown). The putative dimeric recombinant hH₄R species expressed in HEK 293 cells was consistently smaller (approximately 72 kDa), which may reflect differential glycosylation in the two cell lines (Fig. 3). Coincident protein species were detected by the anti-hH₄ 374-390 and the anti-epitope-tagged antibodies in the respective cell lines, further confirming that the hH₄ receptor is being labeled by the anti-hH4 374-390 antibody (data not shown). No signal was detected in either COS-7 or HEK 293 cell lines, further supporting the selectivity of the anti-hH₄ 374-390 antibody. These data identify for the first time the H₄R protein in human T lymphocytes.

Immunoprecipitation of Recombinantly Expressed HA-H₄Rs from HEK 293 Cells. To further characterize the selectivity of the H₄R antibody, an immunoprecipitation assay was performed. HA-H₄Rs expressed transiently in HEK 293 cells (Fig. 2, lanes 3-5) were immunoprecipitated using anti-HA antibodies (Fig. 2, lane 4) or a nonimmune Ig (Fig. 2, lane 3). As negative controls, nontransfected HEK 293 cells (Fig. 2, lane 1) and nontransfected HEK 293 cells immunoprecipitated with anti-HA antibodies (Fig. 2, lane 2) were used. As positive control, solubilized HEK 293 cells expressing HA-H₄Rs (Fig. 2, lane 5) was used. All samples were subjected to immunoblotting using the anti-H₄R antibodies. Immunoreactive species were only detected for the HEK 293 cells expressing the HA-H₄R, which had been anti-HA-immunoprecipitated (Fig. 2, lane 4). The immunoreactive species represent the putative monomeric and dimeric H₄R and are identical with the reactive species in the positive control (Fig. 2, lane 5).

Cross-Linking of H₄Rs. To further investigate the homooligomeric structure of the H₄R, a cross-linking study was performed using N-terminally c-myc-tagged H₄R expressed in COS-7 cells. Upon application of increasing concentrations of the cell-impermeable cross-linker BS₃, a progressive reduction in the monomeric doublet species (34 and 36 kDa) was observed (Fig. 3A). Concomitant the appearance of, initially, a diffuse species of 77 kDa (putative glycosylated and unglycosylated dimers) and then higher molecular mass species (>175 kDa) at 0.25 mM and 2 mM BS₃, respectively, was noticed (Fig. 3A). These data are highly consistent with hH₄Rs expressed in HEK 293 cells (data not shown).

Biochemical Evidence that the hH₄Rs Is an N-Glycosylated Homodimer. In the N terminus of the hH₄R, Asn⁵ and Asn⁹ are potential sites for N-glycosylation (Fig. 3B). To study whether the higher molecular mass species are the N-glycosylated forms of the hH₄R, we expressed H₄Rs in the presence of the N-glycosylation inhibitor tunicamycin. In the absence of tunicamycin, two major putative monomeric species, 34 and 36 kDa, and a diffuse 65- to 72-kDa species were detected as in Fig. 1 (also see Fig. 3C, lane 1). In the presence of 2 μg/ml tunicamycin, a complete loss of the 36-kDa species and concomitant increase in intensity of the 34-kDa species and an additional species at 32 kDa were observed (Fig. 3C, lane 2). Furthermore, the diffuse 65- to 72-kDa species, detected in the absence of tunicamycin, was reduced to a single 65-kDa species. It is noteworthy that an increase in tunicamycin concentration had no further effect on either the 34- or 65-kDa species (Fig. 3C, lanes 3-5). The 32-kDa species observed upon tunicamycin treatment is probably a breakdown product of the glycosylated 36-kDa species in untreated cells. These data strongly suggest that the recombinant hH₄Ris N-glycosylated and forms dimers. This last process is independent of post-translational N-glycosylation. It is noteworthy that upon enzymic N-deglycosylation of PHA blasts, the 77-kDa species was greatly reduced in intensity, and a new 34-kDa species was detected, consistent with the monomeric hH₄R (Fig. 3D, lane 2).

HA-H₄Rs Associate with c-myc-H₄R-His₁₀. To further investigate whether the H₄Rs can associate with each other to form homo-oligomers, membranes of COS-7 cells coexpressing N-terminally c-myc and C-terminally His₁₀-tagged hH₄Rs (c-myc-H₄R-His₁₀) and N-terminally HA-tagged hH₄Rs (HA-H₄R) were solubilized and loaded on an Ni²⁺-resin column. The HA-H₄Rs, when coexpressed with the c-myc-H₄R-His₁₀, were retained on the Ni²⁺-column and could be eluted with 250 mM imidazole, as detected with anti-HA antibodies (Fig. 4, lane 3). When cells individually expressing c-myc-H₄R-His₁₀ and HA-H₄Rs were mixed before solubilization and subsequently loaded on the column, no HA-H₄Rs were found to interact with the Ni²⁺ resin (data not shown).

BRET Shows Constitutive Ligand-Independent Homo-Oligomerization of hH₄Rs. The use of biophysical techniques has been of great value to the study of GPCR oligomerization. We have used BRET to study in further detail the homo-oligomerization of the H₄R. BRET was performed on COS-7 cells expressing either the H₄R-Rluc or coexpressing the H₄R-Rluc with the H₄R-eYFP. After the addition of coelenterazine, a robust BRET signal could be observed in the cells coexpressing the two H₄Rs (Fig. 5B). As a negative control, cells individually expressing either of the H₄R constructs were mixed before adding coelenterazine (Fig. 5B). Previous studies have reported the ability of H₁Rs to oligomerize (Carrillo et al., 2003; Bakker et al., 2004) Therefore, cells in which the H₁R-Rluc and the H₁R-eYFP were coexpressed were taken as a positive control. In these cells, a BRET signal was detected that was approximately 2-fold lower than that observed for the H₄Rs (Fig. 5B).

To investigate the effect of ligands on H₄R oligomerization, cells coexpressing the H₄R-Rluc with the H₄R-eYFP were incubated with a 10 μM concentration of the agonist histamine, the neutral antagonist iodophenpropit (Lim et al., 2005) or the inverse agonist thioperamide for 15 min before the actual BRET measurement. No significant change was observed in BRET signal between stimulated and nonstimulated cells (Fig. 5C).

Agonist-induced increase in oligomerization of somatostatin receptors occurs at physiological expression levels (160 fmol/mg of protein) but not after overexpression (Patel et al., 2002). We therefore also tested the effect of histamine stimulation on H₄R oligomerization at different expression levels. While maintaining H₄R-Rluc expression level constant (approximately 0.2 pmol/mg of protein), we reduced the amount of expressed H₄-eYFP. The concomitant reduction of the donor/acceptor ratio resulted in an expected decrease in BRET signal. At total H₄R expression levels of 1.0, 0.6, or 0.3 pmol/mg, a significant BRET signal was observed. However, histamine also did not effect the H₄R oligomerization at lower H₄R expression levels (Fig. 5D).

tr-FRET Shows the Presence of hH₄R Oligomers at Cell Surface. To study whether the H₄Rs oligomers are actually present at the cell surface, we performed tr-FRET assays on COS-7 cells coexpressing N-terminally FLAG-tagged H₄Rs (FLAG-H₄R) and HA-H₄Rs. These cells were incubated with Eu³⁺-labeled anti-FLAG antibodies or a combination of the Eu³⁺-labeled anti-FLAG and allophycocyanin (APC)-labeled anti-HA antibodies. As control, cells individually expressing the FLAG-H₄Rs and the HA-H₄Rs were mixed and exposed to the two antibodies. Only from the cells coexpressing the FLAG-H₄Rs and HA-H₄Rs was a pronounced signal observed (Fig. 6B). This FRET signal can only be explained as a result of the resonance energy transfer from Eu³⁺ anti-FLAG antibodies bound to FLAG-H₄Rs to APC anti-HA antibodies bound to HA-H₄Rs. Because this resonance energy transfer can only take place within 100 Å, the data indicate the formation of H₄R oligomers at the cell surface of living cells. Stimulation of the COS-7 cells with 10 μM histamine or 10 μM thioperamide preceding tr-FRET measurement did not result in a significant change in signal (Fig. 6B). The used antibodies did not have an influence on the ligand binding to the H₄Rs because no significant difference was found in [³H]histamine binding in the absence or presence of the antibodies (data not shown).

Lack of Hetero-Oligomerization between H₄R and H₁Rs. We have subsequently used tr-FRET to investigate whether hetero-oligomerization occurs between H₄Rs and H₁Rs. tr-FRET was performed on COS-7 cells coexpressing the FLAG-H₄Rs and N-terminally HA-tagged histamine H₁Rs (HA-H₁Rs). As a control, cells individually expressing the FLAG-H₄Rs and the HA-H₁Rs were mixed and exposed to the two antibodies. No significantly increased tr-FRET signal could be observed compared with the signal obtained from cells individually expressing the two receptors that were mixed before incubation with the antibodies (Fig. 6C). The ratio and total amount of antibodies was maintained equal between experiments with H₁R-H₄Rs and H₄R-H₄Rs to ensure proper comparison.

Comparable results were obtained using Eu³⁺ anti-HA antibodies and APC anti-FLAG antibodies (data not shown). Stimulation of cotransfected cells with 10 μM histamine did not lead to a change in FRET signal (data not shown).

Homo-Oligomerization of H₄Rs versus Hetero-Oligomerization between H₄Rs and H₁Rs. To further investigate hetero-oligomerization between H₄Rs and H₁Rs, BRET saturation curves were produced for both the H₄R homo-oligomer and the H₁R-H₄R hetero-oligomer. Experiments were performed in which COS-7 cells were transfected with a fixed amount of H₄-Rluc and increasing amounts of either H₄R-eYFP or H₁R-eYFP cDNA. Expression levels were determined by radioligand binding. Expression of the H₄R-Rluc was maintained at approximately 0.2 pmol/mg of protein. Expression levels of H₄R-Rluc were correlated with luminescence and expression levels for the eYFP fused H₁R and H₄R were correlated with fluorescence. A linear correlation was obtained for all three constructs. Expression for the H₄R-eYFP ranged from 0.3 to 2.5 pmol/mg of protein, whereas expression levels of the H₁R-eYFP ranged from 0.5 to 16 pmol/mg of protein. For the H₄R homo-oligomers, a steep increase in BRET signal is observed, showing detectable BRET when total H₄R expression is 0.3 pmol/mg of protein. For the H₁R-H₄R hetero-oligomers, a more gradual increase in BRET signal is observed upon increased expression of the H₁R-eYFP. A BRET signal is observed for the first time at expression levels of 1 pmol/mg of protein of the H₁R-eYFP. The H₄R-H₄R homo-oligomer showed a 2-fold lower BRET₅₀ value (0.77 versus 1.6) and a 2.5-fold higher B_max (0.1 versus 0.04) than the H₁R-H₄R hetero-oligomer as determined from the BRET saturation curve (Fig. 7).

Discussion

GPCR oligomerization has become a generally accepted phenomenon and has been reported to occur in all GPCR classes (George et al., 2002). Data obtained from atomic force microscopy (Fotiadis et al., 2003), electron microscopy, and Western blot analysis (Suda et al., 2004) have provided compelling evidence that the light-sensitive rhodopsin is predominantly present as a dimer in the retinal disc membrane. For histamine receptors, oligomerization has been shown convincingly for the hH₁Rs (Carrillo et al., 2003; Bakker et al., 2004), the hH₂Rs (Fukushima et al., 1997), and H₃Rs (Shenton et al., 2005; Bakker et al., 2006). In view of the emerging role of GPCR oligomerization in GPCR function and our interest in the H₄R as a new target for inflammatory conditions (de Esch et al., 2005), we investigated oligomerization of the human H₄R by various means. Combining biophysical measurements like tr-FRET and BRET (Angers et al., 2002; Boute et al., 2002) with biochemical approaches, like Western blot analysis and histidine-tag-based affinity chromatography, we provide compelling evidence for homo- and heterooligomer formation of hH₄Rs.

To enable our biochemical approaches and to study H₄R function in native tissues, we report in this study on the first polyclonal H₄R antibody that can successfully be used for Western blot analysis. This new molecular tool is directed against the C-terminal tail of the H₄R and detected monomeric and potential dimeric H₄R species after Western blot analysis of membranes from HEK 293 and COS-7-transfected cells. The selectivity of the new H₄R antibody was confirmed by blockade with the C-terminal peptide used to raise the antibody and the lack of cross-reactivity toward the highly related human hH₃R. Furthermore, through an immunoprecipitation study, the H₄R antibody was shown to detect the same HA-H₄Rs as detected by a commercially available anti-HA antibody.

Western blot analysis of the H₄R expressed in tunicamycin-treated cells indicates that the H₄R normally is N-glycosylated, most likely at Asn⁵ and/or Asn⁹ of the extracellular N terminus of the H₄R (Nguyen et al., 2001). However, inhibition of N-glycosolyation did not affect the presence of putative dimeric H₄R species on the Western blot. To show that H₄R proteins are in close proximity of each other, a requirement for oligomerization, a cross-linking experiment, using BS₃ was performed. With increasing concentrations of the cross-linker BS₃, the bands representing the monomeric H₄R disappeared. At the same time, bands representing oligomeric H₄Rs became more apparent. These data indicate that the H₄Rs are in close enough proximity for cross-linking by BS₃ and suggests that the 65- to 72-kDa species might represent dimerized H₄R proteins. Finally, the polyclonal H₄R antibody allowed us to study the presence of H₄R proteins in human PHA blasts. High-level H₄R mRNA expression has been shown in various white blood cells, including T-lymphocytes (Nakamura et al., 2000; Oda et al., 2000; Morse et al., 2001; Zhu et al., 2001; Hofstra et al., 2003), but H₄R protein expression so far has not been shown. Western blot analysis of membranes of PHA blasts with our polyclonal anti-H₄R antibody indeed revealed the presence of H₄R protein in PHA blasts. It is interesting to note that the endogenously expressed H₄R was only detected as a high molecular mass species. Enzymatic deglycosylation of the native H₄R protein resulted in a partial reduction of the high molecular mass species (77 kDa) to monomeric H₄Rs (34 kDa). These data in themselves do not directly exclude a heavily glycosylated (approximately 33 kDa) H₄R protein. However, the hH₄Rs in human HEK 293 cells is only moderately glycosylated (approximately 2 kDa), and the high molecular mass species coincide with the putative dimeric H₄Rs when recombinantly expressed in COS-7 (77 kDa) and HEK 293 cells (72 kDa). These data can be explained by assuming that in human PHA blasts, the hH₄R functions predominantly as a dimer. We hypothesize that N-glycosylation is not a prerequisite for dimerization, but it helps to stabilize the H₄R dimers. A similar stabilizing effect of glycosylation on receptor dimers has recently been shown for the human bradykinin B₂ receptors (Michineau et al., 2006). The putative H₄R dimerization in native tissue clearly warrants further investigation.

As an alternative biochemical method, we used an immobilized metal (Ni²⁺) affinity chromatography approach with a histidine-tagged H₄R protein. To this end, we coexpressed c-myc-H₄R-His₁₀ and HA-H₄Rs receptors to study oligomer formation via affinity column chromatography. In contrast to c-myc-H₄R-His₁₀ receptors, HA-H₄Rs are not robustly retained onto an Ni²⁺-NTA resin when expressed alone. Yet c-myc-H₄R-His₁₀ receptors immobilized onto Ni²⁺-NTA resin were shown also to retain coexpressed HA-H₄Rs on the Ni²⁺-NTA column, as determined by HA-immunoreactivity detected after elution of histidine-tagged proteins with high imidazole concentrations. These findings indicate that c-myc-H₄R-His₁₀ receptors physically interact with the coexpressed HA-H₄Rs to form oligomers that can be retained on the Ni²⁺-NTA resin through the C-terminal His₁₀ tag.

We continued our investigation of the oligomerization of the H₄Rs in living cells using BRET and tr-FRET assays. Using the BRET assay, a clear signal could be detected when coexpressing H₄R-Rluc with H₄R-eYFP in nonstimulated cells, suggesting constitutive homo-oligomerization of H₄Rs. Because the oligomers detected in the immobilization and BRET assays do not necessarily have to be present at the cell surface, we also studied H₄R oligomerization on the cell membrane of living cells by tr-FRET. The tr-FRET approach uses antibodies that do not permeate the cell membrane and detects cell surface H₄R oligomers present at the cell surface. Similar to the BRET assay, we detected a robust signal, indicating the constitutive presence of H₄R homo-oligomers at the cell surface.

A number of studies have investigated the effect of agonists on receptor oligomerization. However, at present, the effects of ligand stimulation on GPCR oligomerization are not consistent. It has been found that agonists can promote or reduce GPCR oligomerization or are without effect on GPCR oligomerization (Angers et al., 2002; George et al., 2002; Pfleger and Eidne, 2005). In the case of the H₄R, we did not detect any significant difference in BRET signal if cells were treated with either the H₄R agonist histamine, the neutral H₄R antagonist iodophenpropit (Lim et al., 2005), or the inverse H₄R agonist thioperamide (Morse et al., 2001; Lim et al., 2005), suggesting that H₄R ligands do not modulate H₄R homo-oligomerization. Likewise, no agonist- or inverse agonist-induced modulation of H₄R oligomerization was detected in the tr-FRET assay. Patel et al. (2002) reported that agonist-induced oligomerization of somatostatin receptors was only detected at physiological expression levels but not after overexpression. We therefore performed BRET experiments at various H₄R expression levels. First, it is noteworthy that already at an H₄R expression level of approximately 300 fmol/mg of protein, significant BRET signals can be observed. These data indicate that at physiological expression levels, the H₄R can indeed homo-oligomerize and corroborate our findings of H₄R dimers, detected on PHA blasts with our anti-H₄R antibody. Second, from the BRET experiments, we also conclude that also at low expression levels of the H₄Rs, homo-oligomerization is not affected by agonist stimulation. Nevertheless, one should be aware that results concerning ligand effects on dimerization obtained with these biophysical assays can be difficult to interpret, because agonist-induced changes in H₄R conformation could potentially influence the energy transfer between the energy acceptor and donor (Angers et al., 2000).

The H₄R has been linked to play a role in inflammation based on its expression pattern and recent findings, showing that the H₄R induces chemotaxis of eosinophils (O'Reilly et al., 2002; Buckland et al., 2003) and mast cells (Hofstra et al., 2003) and stimulates the release of interleukin-16 from CD8⁺ T cells (Gantner et al., 2002) and the release of leukotriene B₄ in zymosan-challenged mice (Takeshita et al., 2003). The H₁R is colocalized with the H₄R in several white blood cells (Cameron et al., 1986; Morse et al., 2001) and plays a prominent role in inflammatory conditions (Giustizieri et al., 2004; Matsubara et al., 2005). Because both the H₁Rs (Carrillo et al., 2003; Bakker et al., 2004) and the H₄R (this study) are able to form homo-oligomers, we were prompted to study whether the H₄Rs can form hetero-oligomers with the H₁Rs. In fact, previous work with hetero-oligomeric opioid receptors has revealed that GPCR hetero-oligomerization brings an additional layer of complexity to the class of GPCR proteins (Bouvier, 2001; Devi, 2001; Franco et al., 2003; Waldhoer et al., 2005) but also offers opportunities to develop heterooligomeric-selective ligands (Waldhoer et al., 2005). Yet using tr-FRET assays, we were unable to detect H₁R-H₄R hetero-oligomers, suggesting that such GPCR hetero-oligomers are not present at the cell surface. In contrast to the tr-FRET experiments, we were able to detect an expression level-dependent formation of H₁R-H₄R heterooligomers using BRET. Distinct from the detection of H₄R homo-oligomers, H₁R-H₄R hetero-oligomers were only detected at high-expression levels, and we failed to detect hetero-oligomers at physiologically relevant conditions. Results from BRET saturation studies demonstrate a higher propensity for the formation of H₄R-H₄R homooligomers over H₁R-H₄R hetero-oligomers. We presume that the signal observed with BRET at high expression levels possibly originates from intracellular H₁R-H₄R hetero-oligomers. Whereas some receptors, such as the 5-hydroxytryptamine-1A receptor seem to readily form heteromeric receptors (Salim et al., 2002), our present H₄R data corroborate the idea that GPCR hetero-oligomerization is highly selective, as reported for the adrenergic receptors (Stanasila et al., 2003; Uberti et al., 2005) and the thyrotropin-releasing hormone receptors (Kroeger et al., 2001). Although hetero-oligomerization has been shown to occur even between receptors from different classes (Ferre et al., 2002), the relatively low homology (23%) between H₁R and H₄R (Oda et al., 2000) is apparently too low to readily form hetero-oligomers.

In conclusion, we have developed specific antibodies against the C terminus of the H₄R which allowed the detection of endogenously expressed H₄R proteins. This anti-hH₄R antibody is an important new molecular tool for studying the localization and function of the H₄R. Moreover, we determined by various methods that the H₄R constitutively forms cell surface homo-oligomers. Homodimeric H₄Rs are not only found using heterologous expression systems but are also present in PHA blasts and spleen lysates endogenously expressing H₄Rs. The formation of H₄R oligomers is not dependent on N-glycosylation or affected by ligand stimulation but is possibly destabilized by deglycosylation. Although H₁R-H₄R heterooligomers could be detected using BRET upon receptor overexpression, these hetero-oligomers are probably not present at the cell surface. Moreover, H₁R-H₄R heterooligomers were not found at physiologically relevant expression levels. Future studies will have to reveal whether the H₄R can form hetero-oligomers with other GPCR family members or if it preferentially exists as homo-oligomer.