Molecular, pharmacological, and signaling properties of octopamine receptors from honeybee (Apis mellifera) brain



G protein-coupled receptors are important regulators of cellular signaling processes. Within the large family of rhodopsin-like receptors, those binding to biogenic amines form a discrete subgroup. Activation of biogenic amine receptors leads to transient changes of intracellular Ca2+-([Ca2+]i) or 3′,5′-cyclic adenosine monophosphate ([cAMP]i) concentrations. Both second messengers modulate cellular signaling processes and thereby contribute to long-lasting behavioral effects in an organism. In vivo pharmacology has helped to reveal the functional effects of different biogenic amines in honeybees. The phenolamine octopamine is an important modulator of behavior. Binding of octopamine to its receptors causes elevation of [Ca2+]i or [cAMP]i. To date, only one honeybee octopamine receptor that induces Ca2+ signals has been molecularly and pharmacologically characterized. Here, we examined the pharmacological properties of four additional honeybee octopamine receptors. When heterologously expressed, all receptors induced cAMP production after binding to octopamine with EC50s in the nanomolar range. Receptor activity was most efficiently blocked by mianserin, a substance with antidepressant activity in vertebrates. The rank order of inhibitory potency for potential receptor antagonists was very similar on all four honeybee receptors with mianserin >> cyproheptadine > metoclopramide > chlorpromazine > phentolamine. The subroot of octopamine receptors activating adenylyl cyclases is the largest that has so far been characterized in arthropods, and it should now be possible to unravel the contribution of individual receptors to the physiology and behavior of honeybees.


The biogenic amine octopamine is an important modulator of behavior and physiology in arthropods. Binding of octopamine to specific G protein-coupled receptors causes elevation of [Ca2+]i or [cAMP]i. Only one honeybee octopamine receptor inducing Ca2+ signals has been experimentally characterized, yet. Here, we present the pharmacological properties of four additional members of the honeybee octopamine receptor family causing cAMP production.

Abbreviations used

G protein-coupled receptors


hemagglutinin A


phosphate-buffered saline



Neuronal activity strongly depends on the interplay of chemical and electrical signals. Beyond cellular all or none responses leading to action potentials, transient changes of cellular physiology are equally important for proper neuronal function. One group of neurotransmitters that induces transient changes of cellular signaling properties are biogenic amines. These substances can act as neurotransmitters, neuromodulators, and neurohormones as well. Biogenic amines exert their activity by binding to membrane proteins that belong to the superfamily of G protein-coupled receptors (GPCRs). Structural data for these proteins have been obtained more recently (see e.g., Palczewski et al. 2000; Cherezov et al. 2007; Rasmussen et al. 2007) and confirmed the predicted membrane organization with seven transmembrane (TM) segments (for review see Lefkowitz 2004). Agonist binding to GPCRs results in their activation and causes changes in the concentration of intracellular second messengers such as 3′,5′-cyclic adenosine monophosphate (cAMP) or 3′,5′-cyclic guanosine monophosphate (cGMP), inositol-1,4,5-trisphosphate (IP3) and Ca2+. In non-vertebrate species, the biogenic amines octopamine and tyramine are considered to complement the catecholamines epinephrine and norepinephrine present in vertebrates (for reviews see: Roeder 2005; Scheiner et al. 2006). However, both octopamine and tyramine may act in vertebrates as trace amines as well (Burchett and Hicks 2006).

Octopamine has been discovered in the salivary glands of the octopus (Erspamer and Boretti 1951). Rather high concentrations of the phenolamine then were found in neuronal and non-neuronal tissues of nematodes, annelids, arthropods, and molluscs (David and Coulon 1985). Octopamine's role as a physiologically relevant and nervous system modulating compound is experimentally well documented. It controls neuromuscular transmission (Malamud et al. 1988), lipid and carbohydrate metabolism (Orchard et al. 1993), and contributes to learning and behavior (for review see Scheiner et al. 2006; Schwärzel and Müller 2006; Barron et al. 2010). In the honeybee (Apis mellifera L) a number of studies have shown that octopamine can modulate the responsiveness of sensory receptors, interneurons, as well as motoneurons and thus affect complex behavioral responses originating from olfactory or visually perceived inputs (for review see Scheiner et al. 2006; but also Braun and Bicker 1992; Erber et al. 1993; Pribbenow and Erber 1996; Scheiner et al. 2002; Farooqui et al. 2003; Schröter et al. 2007).

At the cellular level octopamine induces either Ca2+ signals or production of cAMP by activation of specific GPCR subtypes (for reviews see Roeder et al. 2003; Scheiner et al. 2006; Verlinden et al. 2010). These receptors are classified as α-type or β-type octopamine receptors (Evans and Maqueira 2005) because of functional similarities to vertebrate adrenoceptors. Up to now only one octopamine receptor has been molecularly identified and functionally characterized from the honeybee. The AmOctαR1 (formerly called AmOA1; Grohmann et al. 2003) has been shown to induce Ca2+ oscillations after heterologous expression. This behavior is reminiscent to the signaling of an orthologous receptor from Drosophila (DmOctαR1B; Balfanz et al. 2005) where phosphorylation of a single threonine residue has been found to be necessary and sufficient to desensitize and thereby shut off receptor signals (Hoff et al. 2011). In Drosophila, three additional genes have been characterized that encode β-type octopamine receptors generating cAMP signals (Balfanz et al. 2005; Maqueira et al. 2005). Analyses of the completely sequenced honeybee genome uncovered four gene candidates coding for β-type octopamine receptors in this insect (Hauser et al. 2006). Here, we have combined bioinformatics with physical cloning and functional expression to characterize these receptors. All four receptors stimulated adenylyl cyclase activity in an octopamine-dependent fashion. Pharmacologically, the four receptors (AmOctβR1 to AmOctβR4) share the property that mianserin efficiently blocks receptor signals in the low nanomolar range. With a total of five members and the signaling and pharmacological properties of all members being identified, we expect that the contribution of these receptors to behavior and physiology of honeybees can now be addressed.

Materials and methods

Identification and cloning of honeybee octopamine receptors

We screened a previous release of the genomic honeybee database (Amel_2.0) with octopamine receptor sequences from Drosophila melanogaster coding for α- and β-type receptor isoforms (DmoctαR1A/R1B, CG3856; DmoctβR1 – R3, CG6919, CG33976, and CG42244) as well as the honeybee (Apis mellifera) AmoctαR1 sequence (Grohmann et al. 2003; Q70TB0). This approach confirmed those four genes that have been assigned as β-type octopamine receptors in a bioinformatics approach to phylogenetically classify neurohormone and biogenic amine receptors from Drosophila and the honeybee (Hauser et al. 2006). In order to obtain functional data on the honeybee octopamine receptors, PCR cloning strategies were applied to amplify the corresponding cDNA sequences. As a template, single stranded cDNA was synthesized on adult honeybee brain mRNA. Receptor-specific primer sets were purchased from eurofins mwg/operon (Ebersberg, Germany). To facilitate expression in eukaryotic cells, the primer binding to the 5′-end of the coding sequence harbored a Kozak consensus motif (CCACC; Kozak 1984) preceding the ATG translational start codon. In addition, this primer introduced a unique restriction site enabling subsequent cloning procedures. For immunological detection of the expressed proteins, a hemagglutinin A (HA)-Tag sequence was engineered to the 3′-end of each receptor-encoding cDNA. For cloning purposes another restriction site was inserted 3′ to the HA-Tag sequence. Except for AmoctβR2 which was assembled from two overlapping PCR fragments, all other fragments were obtained from single PCR reactions. The primer pairs used to amplify AmoctβR1 to AmoctβR4 are listed below and given in 5′ to 3′ direction. AmoctβR1: 5′primer GAGGAATTCCACCATGGAAGCGAACGAAACGAC and 3′primer AGATGCACGCCTCGATGTGC; AmoctβR2: 5′-fragment, 5′primer GGACGGATCCACCATGACGACGATCGTGACGAGCAG and 3′primer GTGTTCCGACGGTCCTGCAACCCGTAGT; 3′-fragment, 5′primer GTATTTCGCCGTGTTTAAAGAGGCGAACAGG and 3′primer GGGATTCCATGCTAGACCAGATAGGCATGCTGCAGGGTCTAATACGTAGATCTAAAA; AmoctβR3: 5′primer CGGAAGCTTCCACCATGGAAGTCAGTGAGCCATCG and 3′primer AGACGAATTGCGACGGTGTTTTC; AmoctβR4: 5′primer GATCGAATTCCCACCATGTTGAACGTAATGGCAGCG and 3′primer AGACGAATTGCGACGGTGTTTTC. The amplified fragments were cloned into pBluescript vector (Stratagene, Amsterdam, the Netherlands) by standard cloning techniques (Sambrook and Russell 2001) and sequenced on both strands.

Construction of pcAmoctβR expression vectors

For heterologous expression of receptor constructs, the sequence verified individual receptor fragments were subcloned into pcDNA3.1-vector (Invitrogen, Groningen, the Netherlands) using the following restriction fragments: AmoctβR1 EcoRI/XbaI, AmoctβR2 BamHI/DraI and DraI/XbaI, AmoctβR3 HindIII/XbaI, and AmoctβR4 EcoRI/XbaI. The longest open reading frame (excluding the HA-Tag) consists of 1287 bp for AmoctβR1; 1242 bp for AmoctβR2; 1242 bp for AmoctβR3, and 1206 bp for AmoctβR4. The nucleotide sequences have been submitted to the EMBL European Bioinformatics Institute Database (AmoctβR1, accession no. HF548209; AmoctβR2, HF548210; AmoctβR3, HF548211; AmoctβR4, HF548212).

Heterologous expression of AmoctβR constructs

Stable transfection of cell lines with honeybee AmoctβR1-HA to AmoctβR4-HA constructs followed a previously established protocol (Wachten et al. 2006). Briefly, ~ 10 μg of AmoctβR1-HA to AmoctβR4-HA expression vectors were introduced into exponentially growing ~ 4 × 105 human embryonic kidney (HEK)293 cells by a modified calcium-phosphate method (Chen and Okayama 1987). AmoctβR-HA-transfected cells were selected in the presence of the antibiotic G418 (1 mg/mL; Invitrogen). Isolated foci were propagated and analyzed immunocytochemically for expression of AmOctβR-HA proteins. Cell clones constitutively expressing AmOctβR1-HA to AmOctβR4-HA were called AmOctβR1 to AmOctβR4, respectively. All further experiments were performed with cell clones that expressed the highest amount of the respective receptor.

Western blot analysis and immunocytochemistry of stably transfected cell lines

Membrane proteins were prepared from non-transfected HEK293 cells and AmOctβR-expressing cells by standard methods. Briefly, cells were hypotonically lysed in a buffer consisting of 10 mM NaCl, 25 mM HEPES, pH 7.5, 2 mM EDTA, and protease inhibitors. After centrifugation, the pellet was solubilized in 100 mM NaCl, 25 mM HEPES, pH 7.5, 1% (w/v) 3-[(3-Cholamidopropyl)dimethylammonio]-1-propanesulfonate and protease inhibitors. AmOctβR-containing samples were either treated or not-treated with PNGase F to examine for potential glycosylation of receptors. Protein samples were separated on 10% sodium dodecyl sulfate–polyacrylamide gels and transferred onto Immobilon-P polyvinylidene difluoride membranes (Millipore, Schwalbach, Germany). Membranes were blocked in 5% (w/v) low-fat dry milk in phosphate-buffered saline (PBS) buffer (PBS = 130 mM NaCl, 7 mM Na2HPO4×2H2O, 3 mM NaH2PO4×H2O) and then incubated with anti-HA antibodies (rat monoclonal antibody 3F10, Roche Applied Science, Mannheim, Germany; dilution 1 : 1000 in PBT = PBS containing 0.02% (w/v) Tween 20). Membranes were washed with PBT and incubated with horseradish peroxidase-conjugated anti-rat secondary antibodies (Sigma, Taufkirchen, Germany; dilution 1 : 10 000 in PBT). After several successive rinses with PBT and PBS, immunoreactive bands were detected with an enhanced chemiluminescence-Kit (AppliChem, Darmstadt, Germany).

Cells of stably transfected cell lines were grown on cover slips and fixed for 15 min in 4% paraformaldehyde. Fixed cells were incubated in phosphate buffer (PB = 81 mM NaH2PO4, 19 mM Na2HPO4, pH 7.4) containing 0.5% (w/v) Triton-X 100 and 5% (w/v) ChemiBLOCKER (CT; CHEMICON-Millipore, Schwalbach, Germany) for 30 min at 22°C, followed by incubation for 1 h with anti-HA antibodies (dilution: 1 : 100 in CT) and rabbit anti-giantin antibodies (dilution 1 : 3000 in CT; Convance, Princeton, NJ, USA) or with anti-HA antibodies and rabbit anti-early endosomal antigen antibodies (dilution 1 : 100 in CT; Sigma). Samples were rinsed with PBS and incubated with goat anti-rat secondary antibodies labeled with Alexa 488 (dilution 1 : 500 in CT; Dianova, Hamburg, Germany) and goat anti-rabbit antibodies labeled with Cy3 (anti-giantin; dilution 1 : 500 in CT; Dianova) or Alexa 568 (anti-early endosomal antigen; dilution 1 : 500 in CT; Life Technologies, Darmstadt, Germany) for 1 h at 22°C. After several rinses with PBS, the cover slips were mounted on microscope slides (Aqua Poly/Mount; Polysciences, Warrington, PA, USA). Immunoreactive cells were visualized by confocal microscopy (Leica TCS, Heidelberg, Germany).

Functional characterization of AmOctβ receptors

The ability of each AmOctβ receptor to stimulate adenylyl cyclase activity in the stably transfected cell lines was performed as described earlier (Balfanz et al. 2005). AmOctβR-expressing cells were grown in minimal essential medium (MEM + GlutaMAXI, Invitrogen) with 10% (v/v) fetal bovine serum (Invitrogen), 1% (v/v) non-essential amino acids (Invitrogen), and 1% (v/v) antibiotics (Invitrogen). Incubations with different ligands were performed at 37°C for 30 min in the presence of the phosphodiesterase inhibitor isobutylmethylxanthine (IBMX; final concentration 100 μM) in PBS. The effects of antagonists to inhibit cAMP production in receptor-expressing cell lines was examined with increasing antagonist concentrations in the presence of the following octopamine/tyramine concentrations: AmOctβ1 (100 nM octopamine/1 μM tyramine), AmOctβ2 (10 nM/100 nM), AmOctβ3 (10 nM/50 nM), and AmOctβ4 (5 nM/5 nM). Reactions were stopped by adding 0.5 mL of ice-cold ethanol. After 2 h at 4°C, the lysate was transferred to an Eppendorf cup and lyophilized. The amount of cAMP produced was determined with the TRK 432 cyclic AMP assay kit (GE Healthcare, Freiburg, Germany) or the cAMP-Screen System (Applied Biosystems, Darmstadt, Germany). Mean values of cAMP per mg total protein were determined in quadruplicate from two to three independent measurements. Maximal values of octopamine-induced cAMP amounts (pmol/mg protein) for receptor-expressing cell lines were: AmOctβ1 (~ 200); AmOctβ2 (~ 320); AmOctβ3 (~ 110); AmOctβ4 (~ 160). Data were analyzed and displayed by using PRISM 5.04 software (GraphPad, San Diego, CA, USA).

Phylogenetic analysis

For phylogenetic analyses only those dopamine-, serotonin-, tyramine-, and octopamine-receptor sequences were included which have been functionally and pharmacologically characterized. In addition to honeybee (Apis mellifera) sequences, receptors from Balanus amphirite, Bombyx mori, Drosophila melanogaster, and Periplaneta americana were examined. Sequences were obtained from NCBI databases. Multiple sequence alignments were performed with ClustalW using the MEGA 5.05 software package (Tamura et al. 2011). Full-length amino-acid sequences were used for pairwise and multiple alignments. The BLOcks Substitution Matrix was used for calculating values of identity and similarity. Phylogenetic trees were generated with MEGA 5.05. To calculate the genetic distances a program for neighbor-joining trees was used with 2000-fold bootstrap resampling and p-distance modeling. The FMRF-receptor sequence from Drosophila (CG2114) formed the outgroup.


Molecular and structural properties of the octopamine receptors from honeybee brain

We initiated a combined bioinformatics and physical cloning approach to gain a comprehensive understanding of the octopamine receptor system in the honeybee. In addition to the previously characterized AmoctαR1 gene (Grohmann et al. 2003), four related genes could be identified in different releases of the honeybee genome database (Amel_2.0 and 4.5; see also Hauser et al. 2006). Receptor-encoding cDNAs were amplified by PCR. As a template single stranded cDNA was synthesized on mRNA isolated from adult worker bee brains. The deduced amino-acid sequences consist of 428, 413, 413, and 401 residues for AmOctβR1 to AmOctβR4, respectively. An alignment of all five honeybee octopamine receptor (AmOctR) sequences is shown in Figure S1. Characteristic features of GPCRs (Lefkowitz 2004), for example, seven transmembrane regions, potential N-glycosylation sites, as well as consensus motives for phosphorylation by different kinases are present in all AmOctβRs (Table 1).

Table 1. Potential sites for post-translational modification of honeybee AmOctβRs
  N-glycosylationPKA phosphorylationPKC phosphorylation
  1. Consensus motives for glycosylation in extracellular loops, phosphorylation in intracellular loops by cAMP/cGMP-dependent protein kinase (PKA) or protein kinase C (PKC) were identified with the prosite package on Potentially modified residues are indicated by their position in the primary receptor sequence.





































On the basis of the alignment, we calculated the overall identity and similarity, that is, identical and conservative substituted positions, between the five honeybee receptors (Table 2). The closest relationship exists between AmOctβR3 and AmOctβR4 with 67.8 % identity and 76.8 % similarity, followed by AmOctβR1 and AmOctβR2, with 53.5 % identity and 68.8 % similarity. In contrast to these pairs of receptors, the homology between AmOctαR1 and AmOctβRs is less pronounced (Table 2). Another feature of AmOctβR3 and AmOctβR4 is worth mentioning. In both receptors, a stretch of 107 residues in the C-terminal third of the proteins is completely conserved (Figure S1). This domain starts within the third intracellular loop, covers TM region six and seven, and the C-terminal loop of the proteins. In a previous study (Balfanz et al. 2005), we had uncovered that Drosophila DmOctαR1A and -R1B (CG3856) receptors share identity in their N-terminal halves. In this case, the proteins are derived from a single gene by alternative splicing of two exons that encode the respective C-terminal halves of the receptors (Balfanz et al. 2005). In order to unravel whether alternative splicing might also lead to synthesis of AmOctβR3 and AmOctβR4, we examined the genomic organization of all five AmoctR genes. Comparison of cloned cDNAs with the honeybee genome database revealed that the AmoctαR1 gene (Grohmann et al. 2003) is located on chromosome LG15, whereas the AmoctβR genes are located on chromosome LG7. All receptors are encoded by several exons. While AmOctβR1 and -βR2 are encoded by individual structural genes harbored on contig NW_0033774867 and NW_003377885, respectively, AmOctβR3 and -βR4 most likely originate by alternative splicing of a gene located on contig NM_003377871. The N-terminal two-thirds of both receptors are encoded by physically separated exons, whereas the identical C-terminal parts of the proteins are encoded by two exons located further downstream on the contig. The alternatively spliced exons were assigned to positions 60000–59383 and 53675–53375 for AmOctβR3 and 130295–129651 and 127399–127162 for AmOctβR4. In Fig. 1 the exons are depicted with respect to the encoded part of the receptor proteins. The nucleotide sequences comprising the splice donor and acceptor sites of each AmoctR gene are summarized in Table S1. Notably, the most 3′ exon of all AmoctβRs starts at a conserved position which is located at the end of TM6 (Fig. 1).

Table 2. Sequence identity and similarity between honeybee AmOctRs
  1. The percentage of identical as well as identical and conservatively substituted amino acids (in parenthesis) between all five honeybee octopamine receptors (AmOctα1R; Grohmann et al. 2003) was calculated from pairwise alignments of the complete amino acid sequences.

AmOctα1R100%25.4 (40.9)24.2 (37.1)26.1 (38.3)23.5 (34.4)
AmOctβ1R 100%53.5 (68.8)49.9 (62.2)47.4 (60)
AmOctβ2R  100%50.8 (66.3)49.4 (62.5)
AmOctβ3R   100%67.8 (76.8)
AmOctβ4R    100%
Figure 1.

Organization of AmoctβR-encoding genes. A pictogram of a G protein-coupled receptor (GPCR) with its seven transmembrane segments (yellow bars) is displayed. For each AmoctβR gene, the positions of splice donor and acceptor sites were identified by comparing cloned cDNA and genomic DNA deposited in the honeybee genomic database ( Black arrows point to the relative positions of these sites in the primary structures of the receptors. Conserved splice sites are indicated by red arrows. For each exon, the last amino acid residue is indicated and numbered according to its position in the primary sequence.

Binding of GPCRs to their cognate ligand is achieved by amino-acid side chains buried in a binding cleft formed by the membrane spanning segments of the protein. For biogenic amine receptors an aspartic acid residue in TM3, serine residues in TM5, and aromatic residues in TM6 and/or TM7 are considered as the main interaction partners. Residues fulfilling these criteria are present at conserved positions in each of the AmOctβRs (see Figure S1). Although the basic characteristics of GPCRs and especially biogenic amine receptors are maintained in all AmOctβRs, the downstream effects resulting from receptor activation are difficult to predict ab initio. We therefore performed a multiple alignment using a total of 42 biogenic amine receptor sequences from Drosophila, the honeybee (Apis mellifera), the American cockroach (Periplaneta americana), Bombyx mori, as well as barnacle (Balanus improvisus). The selected non-vertebrate sequences cover especially those receptors whose pharmacological and cellular signaling properties have been precisely investigated. In Fig. 2, the phylogram of this analysis is shown. In this tree, dopamine-, octopamine-, serotonin-, and tyramine receptors form distinct, ligand-specific clades. All AmOctβRs assemble in a group that contains β-like octopamine receptors from Drosophila (Balfanz et al. 2005; Maqueira et al. 2005), Bombyx (Chen et al. 2010), and Balanus (Lind et al. 2010). This suggests that AmOctβR1 to AmOctβR4 are likely to stimulate adenylyl cyclase activity once activated by octopamine. To examine this hypothesis experimentally, cell lines were generated that constitutively expressed the individual receptor genes.

Figure 2.

Dendrogram of AmOctβRs and protostomian biogenic amine receptors. Phylogenetic comparison of representative biogenic amine receptors from arthropod and molluscan species that have been pharmacologically characterized. The amino acid sequences used for the analysis were obtained from NCBI databases. The accession or identification numbers (#) are listed in the order of receptor appearance in the tree: octopamine α-type receptors from Drosophila melanogaster (DmOctαR1A/R1B, #CG3856), Periplaneta americana (PeaOctαR1, #AAP93817), Bombyx mori (BomOctαR1, #BAF33393), Apis mellifera (AmOctαR1, #Q70TB0), Balanus (BalOctαR1, #Q93126), dopamine receptors from Bombyx (BomDop2, #BAF98648), Apis (AmDop2, #NP1011567), Drosophila (DAMB, #CG18741), tyramine receptors from Drosophila (DmTyr2, #CG16766; DmTyr3, #CG7431; DmTyr1, #CG7485), Apis (AmTyr2, #NP1032395; AmTyr1, #NP1011594), Bombyx (BomTyr, #BAD11157), Periplaneta (PeaTyr1, #CAQ48240), Balanus (BalTyr1, #Q93127), serotonin receptors from Apis (Am5HT7, #NP1071289; Am5HT1, #NP1164579), Drosophila (Dm5HT7, #CG12073; Dm5HT1A, #CG16720; Dm5HT1B, #CG15113), Bombyx (Bom5HT1, #Q17239), Periplaneta (Pea5HT1, #CAX85666), dopamine receptors from Apis (AmDop3, #NP1014983; AmDop1, #NP1011595), Drosophila (DmDop2, #CG33517; DmDop1, #CG9652), Bombyx (BomDop1, #BAG14260), octopamine β-type receptors from Balanus (BalOctβR1, #ADI56272; BalOctβR2, #ADI56273; BalOctβR3, #ADI56274; BalOctβR4, #ADI56275), Drosophila (DmOctβR2, #CG33976; DmOctβR1, #CG6919; DmOctβR3, #CG42244), Bombyx (BomOctβR, #BAJ06526), and Apis (this study). The Drosophila FMRF receptor (DmFMRF, #CG2114) was chosen for outgroup. Alignments and graphical display of the tree were performed with MEGA 5.05 software package (Tamura et al. 2011). The numbers at the nodes of the branches represent the percent bootstrap support for each branch. The scale bars allow conversion of branch lengths in the dendrogram to genetic distance between clades (0.1 = 10% genetic distance).

Expression of AmOctβR1 to AmOctβR4 in HEK293 cells

To investigate the pharmacological properties and the downstream effectors of each β-type honeybee octopamine receptor, the HA-tagged cDNAs were stably transfected into HEK293 cells. Expression and distribution of receptor proteins was examined by co-immunostaining the transfected cells with α-HA antibodies for HA-tagged receptors and with α-giantin antibodies for localizing the Golgi complex as well as antibodies detecting endosomes (anti-early endosomal antigen). In Fig. 3 the staining pattern of each AmOctβR expressing cell line is depicted as revealed by confocal microscopy. In Fig. 3A1–D1 and A3–D3 the distribution of the receptors is shown. Fig. 3A2–D2 represents merged images of receptor (green) and giantin (red) staining. Fig. 3A4–D4 represents merged images of receptor (green) and anti-early endosomal antigen (red) staining. The receptors are present in the plasma membrane (arrow heads) as well as in intracellular compartments like the Golgi complex and endosomes (yellow color in A2/A4–D2/D4). Western blotting revealed that the apparent molecular weights for the non-glycosylated forms of heterologously expressed AmOctβR1-HA to AmOctβR4-HA agreed well with the values calculated from the deduced amino-acid sequences with 49.9 kDa, 48.1 kDa, 47.3 kDa, and 45.7 kDa including the HA-tag, respectively (Fig. 4).

Figure 3.

Immunofluorescent staining of AmOctβR-expressing cells. Confocal images of cells stably transfected with hemagglutinin A (HA)-tagged AmoctβR1 (A1–A4), AmoctβR2 (B1–B4), AmoctβR3 (C1–C4), and AmoctβR4 (D1–D4) were obtained after co-immunostaining with α-HA, α-giantin (A2, B2, C2, D2), and anti-early endosomal antigen primary antibodies (A4, B4, C4, D4). The α-HA antibodies were detected with Alexa 488 labeled secondary antibodies (green). The α-giantin antibodies were detected with Cy3 labeled secondary antibodies (red) for identifying the Golgi complex. The anti-early endosomal antigen primary antibodies were detected with Alexa 568 labeled secondary antibodies (red). In A1/A3 to D1/D3 the receptor distribution is depicted. Arrow heads indicate receptor localization on the plasma membrane. Merged images from an overlay of α-HA and α-giantin staining are shown in A2–D2. Asterisks indicate co-localization of receptors and giantin in the Golgi complex. Merged images from an overlay of α-HA and anti-early endosomal antigen staining are shown in A4–D4. Arrows indicate co-localization of receptors and endosomal marker protein. Scale bars 20μm.

Figure 4.

Western blot analysis of AmOctβRs. Membrane proteins from stably transfected AmoctβR-hemagglutinin A (HA) cell lines and non-transfected HEK293 cells were prepared by hypotonic lysis. Membrane proteins from receptor-expressing cell lines were either treated (+) or non-treated (−) with PNGase F to test for receptor glycosylation. Samples were separated by sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS–PAGE) and blotted to polyvinylidene difluoride (PVDF) membranes. Blots were probed with rat anti-HA-antibodies. The sizes of marker proteins in kDa are given on the left margin.

Pharmacological profiles of AmOctβRs

We examined the ability of different biogenic amines, that is, dopamine, histamine, octopamine, serotonin, and tyramine to evoke AmOctβR-dependent cAMP production in the cell lines. No stimulation of adenylyl cyclase activity was recorded after incubation with 1 μM dopamine, histamine, and serotonin. In contrast, all cell lines showed increasing [cAMP]i after application of octopamine or tyramine. Since these measurements were performed at rather high ligand concentrations, that is, 1 μM, series of octopamine and tyramine concentrations were applied to the cell lines. The amount of cAMP produced was determined and used to calculate concentration response curves for each receptor. The result for AmOctβR1 is displayed in Fig. 5a. In this graph the cAMP amount obtained with 3 μM octopamine (~ 200 pmol/mg protein) was set to 100%. The cAMP amounts obtained with all other ligand concentrations were calculated on that value. The EC50 values for octopamine and tyramine were 43.9 × 10−9 M and 217.7 × 10−9 M, respectively. Similarly, the EC50s for both amines were determined from normalized concentration response curves (Fig. 5b, c) for the remaining AmOctβRs and are summarized in Table 3. Half-maximal activation of the receptors with octopamine was in the low nanomolar range (~ 2–~ 44 nM) and octopamine was at least one order of magnitude more efficient than tyramine for each receptor (Table 3). In addition to cAMP signaling, all cell lines were tested for Ca2+ signals after stimulation with octopamine concentrations from 10−9 to 10−6 M. None of the four receptor-expressing cell lines showed a Ca2+ response.

Figure 5.

Concentration response curves of AmOctβRs. (a) Concentration-dependent production of cAMP was examined on AmOctβR1-expressing cells with octopamine (red) and tyramine (green). Non-transfected cells were incubated with octopamine (black) for control. Values were calculated based on the cAMP amount obtained with 3 × 10−6 M octopamine (= 100%). Mean values ± SD from quadruplicate determinations are displayed for a representative experiment. (b) Octopamine-dependent concentration response curves for AmOctβR1 (red), AmOctβR2 (black), AmOctβR3 (green), and AmOctβR4 (blue). For each receptor, values were normalized to the cAMP concentration obtained at the highest octopamine concentration (= 100%). Mean values ± SD from quadruplicate determinations are displayed from representative experiments. (c) Tyramine-dependent concentration response curves for AmOctβR1 (red), AmOctβR2 (black), AmOctβR3 (green), and AmOctβR4 (blue). For each receptor, values were normalized to the cAMP concentration obtained at the highest tyramine concentration (= 100%). Mean values ± SD from quadruplicate determinations are displayed from representative experiments. GraphPad Prism 5.04 software was used to calculate and display graphs.

Table 3. EC50 and log EC50 values for octopamine and tyramine on all four AmOctβRs
EC50 [M]log EC50EC50 [M]log EC50
  1. Concentration response curves were calculated with GraphPad Prism 5.04. Values for half maximal stimulation (EC50 [M] and log EC50) of receptor activity were obtained from non-linear fitting of the data.

AmOctβR143.93 × 10−9−7.357 ± 0.05421.77 × 10−8−6.662 ± 0.139
AmOctβR21.815 × 10−9−8.741 ± 0.1163.803 × 10−8−7.420 ± 0.088
AmOctβR33.302 × 10−9−8.481 ± 0.0992.436 × 10−8−7.613 ± 0.098
AmOctβR41.474 × 10−9−8.832 ± 0.1391.362 × 10−8−7.886 ± 0.146

We next examined the ability of six potential antagonists to impair octopamine- and tyramine-stimulated AmOctβR signaling. Non-saturating concentrations of both biogenic amines (see 'Materials and methods') were co-applied with increasing antagonist concentrations. The reduction of cellular cAMP production was quantified and normalized to the value obtained without adding antagonists (= 100%). These normalized data were used to construct inhibitor concentration response curves. The calculated IC50s for each antagonist displacing either octopamine or tyramine are summarized in Tables 4 and 5, respectively. The most efficacious antagonist on octopamine-stimulated AmOctβRs was mianserin with IC50s ranging from ~ 5 × 10−9 M (AmOctβR3) to ~ 23 × 10−9 M (AmOctβR2; Table 4). Similarly low IC50s for mianserin were determined on tyramine-activated AmOctβRs. Here, the values ranged from ~ 1 × 10−9 M (AmOctβR2) to ~ 36 × 10−9 M (AmOctβR1; Table 5). The activity of all four receptors was also inhibited by cyproheptadine, metoclopramide, chlorpromazine, and phentolamine. Yet, half-maximal inhibition required higher, that is, sub- to micromolar ligand concentrations (Table 4/5). Representative data for inhibition of octopamine-stimulated AmOctβR-expressing cell lines are shown in Fig. 6a–d. Here, the inhibitory effects of mianserin, cyproheptadine, and phentolamine are displayed. Cyproheptadine inhibited the activity all four receptors with IC50s ranging from ~ 2 × 10−7 M (AmOctβR3) to ~ 12 × 10−7 M (AmOctβR2) and thus was more efficient than chlorpromazine or metoclopramide (see Table 4). Phentolamine, an antagonist acting on different types of biogenic receptors (Roeder 2005), caused inhibition of AmOctβ-receptors only at high ligand concentrations (see Fig. 6). Except for AmOctβR1 whose activity was reduced to ~ 25% at the highest phentolamine concentration (10−5 M), the activity of AmOctβR2 - 4 was only reduced by 50% of the original value (Fig. 6b–d). Yohimbine which is considered a tyramine-receptor specific antagonist (Roeder 2005) did not impair octopamine-stimulated AmOctβR activity.

Figure 6.

Inhibitor concentration response curves of AmOctβRs. Inhibition of octopamine-dependent cAMP production was examined on AmOctβR1- (a), AmOctβR2- (b), AmOctβR3- (c), and AmOctβR4-expressing (d) cells with cyproheptadine (green), mianserin (black), and phentolamine (brown). To stimulate receptor activity octopamine concentrations were applied that, according to the concentration response curves, did not exceed 75 % of the maximal cAMP amount generated at saturating ligand concentrations. For each receptor, values were normalized to the cAMP concentration obtained with octopamine in the absence of antagonist (= 100%). Mean values ± SD from quadruplicate determinations are displayed for representative experiments. GraphPad Prism 5.04 software was used to calculate and display graphs.

Table 4. IC50 and log IC50 values for antagonists on octopamine-activated AmOctβRs
  1. Concentration response curves were calculated with GraphPad Prism 5.04. Values for half maximal inhibition (IC50 [M] and log IC50 ± SD) of receptor activity were obtained from non-linear fitting of the data.

ChlorpromazineIC50 [M]8.840 × 10−56.979 × 10−72.299 × 10−79.025 × 10−6
log IC50−4.054 ± 1.301−6.156 ± 0.414−6.638 ± 0.129−5.045 ± 0.158
CyproheptadineIC50 [M]7.633 × 10−71.216 × 10−61.514 × 10−75.995 × 10−7
log IC50−6.117 ± 0.124−5.915 ± 0.143−6.820 ± 0.06−6.222 ± 0.084
MetoclopramideIC50 [M]2.121 × 10−62.554 × 10−71.991 × 10−74.588 × 10−6
log IC50−5.673 ± 0.497−6.593 ± 0.349−6.701 ± 0.092−5.338 ± 0.126
MianserinIC50 [M]1.575 × 10−82.25 × 10−85.232 × 10−99.151 × 10−9
log IC50−7.803 ± 0.090−7.648 ± 0.083−8.283 ± 0.042−8.039 ± 0.048
PhentolamineIC50 [M]8.613 × 10−56.005 × 10−71.272 × 10−61.245 × 10−6
log IC50−4.129 ± 1.550−6.221 ± 0.181−5.895 ± 0.165−5.905 ± 0.184
YohimbineIC50 [M]n.a.n.a.n.a.n.a.
log IC50n.a.n.a.n.a.n.a.
Table 5. IC50 [M] and log IC50 values for antagonists on tyramine-activated AmOctβRs
  1. Concentration response curves were calculated with GraphPad Prism 5.04. Values for half maximal inhibition (IC50 [M] and log IC50 ± SD) of receptor activity were obtained from non-linear fitting of the data.

ChlorpromazineIC50 [M]2.189 × 10−62.272 × 10−67.87 × 10−77.204 × 10−7
log IC50−5.66 ± 0.192−5.644 ± 0.574−6.104 ± 0.065−6.142 ± 0.15
CyproheptadineIC50 [M]1.617 × 10−68.307 × 10−87.678 × 10−83.126 × 10−7
log IC50−5.791 ± 0.219−7.081 ± 0.145−7.115 ± 0.094−6.505 ± 0.132
MetoclopramideIC50 [M]1.849 × 10−61.733 × 10−74.252 × 10−71.247 × 10−7
log IC50−5.733 ± 0.119−6.761 ± 0.313−6.371 ± 0.115−6.904 ± 0.299
MianserinIC50 [M]3.582 × 10−81.389 × 10−93.558 × 10−91.619 × 10−9
log IC50−7.446 ± 0.115−8.857 ± 0.115−8.449 ± 0.075−8.791 ± 0.114
PhentolamineIC50 [M]3.302 × 10−71.063 × 10−73.208 × 10−71.274 × 10−7
log IC50−6.481 ± 0.143−6.974 ± 0.292−6.494 ± 0.09−6.895 ± 0.559
YohimbineIC50 [M]n.a.n.a.n.a.n.a.
log IC50n.a.n.a.n.a.n.a.

Applying the different antagonists to tyramine-stimulated AmOctβRs resulted in almost the same rank order of potency as described for octopamine-stimulated receptors (Table 5). For the antagonists used, we did not observe any cAMP stimulatory effect in non-transfected cells. In summary, the receptors were efficiently blocked by mianserin, cyproheptadine, metoclopramide, and chlorpromazine, whereas yohimbine was non effective at all (Table 4/5).


The availability of completely sequenced genomes of model organisms greatly facilitates bioinformatics approaches to uncover hidden members of a gene family. Here, we took advantage of annotated genes potentially encoding honeybee (Apis mellifera L.) biogenic-amine receptors and successfully cloned four members of the octopamine receptor clade. Activation of all four receptors by nanomolar octopamine concentrations led to cAMP production. Pharmacological profiling of the AmOctβRs uncovered mianserin as the most efficient inhibitor blocking receptor activity in the low nanomolar range. With a total number of four receptors belonging to the β-class of phenolamine receptors, they form the largest subfamily in arthropods, so far.

Molecular and functional properties of the AmOctβ receptors

In recent years the number of molecularly identified octopamine receptors has considerably increased. Members of this receptor family have been reported from, for example, Bombyx mori (Ohtani et al. 2006; Chen et al. 2010), Drosophila melanogaster (Han et al. 1998; Balfanz et al. 2005; Maqueira et al. 2005), Periplaneta americana (Bischof and Enan 2004), Aplysia (Chang et al. 2000), or Balanus improvises (Lind et al. 2010). In addition to these fully characterized proteins, further receptors have been identified by sequence homology (see e.g., Hauser et al. 2006, 2008). With the four receptors presented here, the family of octopamine receptors in the honeybee now lists five, fully characterized members. The proteins share the typical topology of GPCRs. Cognate residues considered to participate in ligand binding are well conserved, for example, the Asp3.32 residue (labeling according to Ballesteros and Weinstein 1995) in TM3 or Ser5.42/5.43/5.46 residues in TM5 harbored in a SerSerXXSer motif of the AmOctβRs (see Figure S1). Experimental evidence that these residues participate in ligand binding has been obtained from a couple of site-directed mutagenesis studies performed on different members of the octopamine and tyramine receptor clade. Binding studies and intracellular signaling of mutated receptors uncovered Asp3.32 as a key residue interacting with the protonated amino group of octopamine or tyramine (Huang et al. 2007, 2008; Chen et al. 2011). Interaction of the p-OH group of both biogenic amines with one or even two serine residues in TM5 (Ser5.42/5.46) has also been shown (Chatwin et al. 2003; Ohta et al. 2004; Chen et al. 2011). In contrast to these amino acids, unequivocal experimental evidence for the binding partner of the β-OH group of octopamine is missing. Whether Tyr6.55 in TM6 accounts for the favored binding of a receptor to octopamine over tyramine is still a matter of debate (see Chatwin et al. 2003; Ohta et al. 2004). Notably, Tyr6.55 is fully conserved in tyramine receptors (Blenau et al. 2000). Tyramine is the Cβ-unsubstituted precursor of octopamine and thus cannot form hydrogen bonds with the hydroxyl group of Tyr6.55. Consequently, other residues in the binding pocket of octopamine receptors must serve as interaction partners of the β-OH group of octopamine (see also Chatwin et al. 2003) with one candidate being Asp3.32 in TM3 as suggested by Chen et al. (2011).

Binding of a ligand to a GPCR is the initiating event of cellular signaling cascades that finally result in altered activity profiles of downstream effectors. The efficacy of signal transduction can be strengthened by the assembly of signaling molecules in protein complexes, like transducisomes (Tsunoda et al. 1997; Montell 1998). Here, several signaling partners are physically tied together by protein–protein interaction domains. For several GPCRs it has been shown that C-terminal sequence motives interact with PDZ-containing scaffolding proteins (Chimura et al. 2011; Magalhaes et al. 2012). The primary sequences of all AmOctβRs, however, do not contain such motives. Although in silico analyses cannot exclude potential interaction, it seems unlikely though that AmOctβRs incorporate into larger signaling complexes by canonical protein-protein interaction domains.

Another property of the AmOctβRs is worth mentioning. All four receptors are activated by octopamine with EC50 values between ~ 2 nM (AmOctβR4) and ~48 nM (AmOctβR1; see Table 3). Thus, activation of these receptors in vivo is possible over a wide dynamic range of ligand concentrations. The orthologous receptors from Drosophila (Maqueira et al. 2005) displayed EC50s of ~ 5 nM (DmOctβ1R), 15 nM (DmOctβ2R), and 14 nM (DmOctβ3R). However, it is noteworthy that the Drosophila receptors were examined in CHO rather than in HEK293 cells used in this study, which may account for different pharmacological properties. Two other β-type octopamine receptors characterized from Bombyx mori (Chen et al. 2010) and the rice stem borer, Chilo suppressalis (Wu et al. 2012) were half-maximally activated at ~ 1.7 nM and ~ 2.2 nM octopamine, respectively. In addition to octopamine, also tyramine activated all four AmOctβRs. The EC50s, however, were shifted at least by one order of magnitude to higher concentrations (Table 3). Similar observations have been reported for the Drosophila (Maqueira et al. 2005), Bombyx mori (Chen et al. 2010), C. suppressalis (Wu et al. 2012), and Balanus β-type receptors (Lind et al. 2010). In summary, these data show that honeybees can generate octopamine-dependent signals over a wide, ~ 20-fold range of ligand concentrations, whereas the activation profile of Drosophila octopamine receptors is rather narrow (3-fold). Whether this pharmacological fingerprint physiologically corresponds to either neurotransmitter- or neurohormone-like actions of octopamine awaits a detailed neuroanatomical investigation of the receptor distributions and correlation to known octopamine release sites in honeybee tissue. To achieve this goal, receptor subtype-specific antibodies have to be established. Unfortunately, our attempts so far have not resulted in antisera of sufficient specificity to detect the proteins either on Western blots or in tissue sections.

Pharmacological characteristics of the AmOct β-type receptors

In vivo pharmacological studies in honeybees (e.g., Braun and Bicker 1992; Pribbenow and Erber 1996; Menzel et al. 1999; Scheiner et al. 2002; Farooqui et al. 2003) and other arthropods (e.g., Zornik et al. 1999; Kaczer and Maldonado 2009; Mizunami et al. 2009; Rigby and Merritt 2011; Rillich et al. 2011) were performed to uncover the functional role of biogenic amines in these animals. These approaches, however, suffered from some caveats because the number of proteins constituting the receptor families as well as the pharmacological properties and intracellular signaling mechanisms of individual receptors were not precisely known. Furthermore, off-target effects especially ligand cross-reactiviy to other aminergic receptors could hardly be excluded. Therefore, establishing pharmacological profiles of molecularly identified and heterologously expressed receptors is a worthwhile challenge. The data can be used to reevaluate in vivo pharmacological experiments and may, in retrospect, allow correlating physiological reaction(s) with receptor-subtype activities. In addition, uncovering substances acting through individual receptors may foster designing novel in vivo pharmacological experiments that precisely address a defined physiological trait.

Here, we have studied the interaction of the four AmOctβRs with six potential antagonists. From these substances, chlorpromazine, cyproheptadine, metoclopramide, and mianserin consistently impaired receptor activity. The most potent and efficacious inhibitor on all four receptors was mianserin. A similar observation has been made for Drosophila β-type receptors (Maqueira et al. 2005). In mammals, mianserin acts as an antidepressant. The sedative effect is mediated by blocking the activity of serotonin (5-HT2) and α2-adrenergic receptors and ultimately leads to an enhancement of serotonin and norepinephrine neurotransmission (Landgrebe et al. 2002). Physiological and behavioral effects of mianserin have also been investigated in honeybees, either after direct injection or external application of the substance to the animal. These treatments resulted in reduced gustatory and olfactory learning abilities (Farooqui et al. 2003; Vergoz et al. 2007) as well as reduced cardiac activity (Papaefthimiou and Theophilidis 2011). In these studies, high micromolar to millimolar concentrations of mianserin was used. Since we have observed very efficient inhibition of AmOctβRs at rather low mianserin concentrations (IC50s: ~ 5–23 nM; Table 4), the ligand concentrations administered in vivo most likely blocked all AmOctβR-subtypes simultaneously. Therefore, it might be worth considering future in vivo pharmacological experiments at reduced antagonist concentrations or to apply combinations of antagonists to block AmOctβR activity. However, some compounds used in this study have also been shown to act on other biogenic amine receptors as well (c.f. chlorpromazine on AmDop1 (Blenau et al. 1998); mianserin on AmDop2 and AmOctαR1 (Beggs et al. 2011) emphasizing that interpretation of in vivo pharmacological results might be rather complex. Nevertheless, incorporating pharmacological knowledge from the increasing number of well characterized receptors into future in vivo experiments may still be advantageous over RNAi-based techniques to knock down receptor gene-expression which awaits further improvement to be routinely applicable in the honeybee CNS. The detailed pharmacological characterization of five honeybee octopamine receptors (Grohmann et al. 2003; this study) thus marks an important step in the ongoing endeavor to precisely understand octopamines' role to higher brain function in this eusocial organism.


We gratefully acknowledge technical assistance of J. Schmitz. For helpful suggestions on the manuscript we thank A. Meisenberg. This study was supported by grants Ba 1441/6-1 (to A. B.) from the Deutsche Forschungsgemeinschaft and conducted in compliance with the ARRIVE guidelines. The authors declare no competing financial interests.