Honey bees (Apis mellifera L.) were adult New World Carniolan foragers of unknown age obtained from colonies maintained at Arizona State University. The bees were collected at the entrance of the hive when they returned from the field with pollen, usually in the afternoon.

Tyramine antiserum (AB124; EMD Millipore) was raised in rabbits using p-tyramine conjugated to N-alpha-acetyl-L-lysine-N-methylamide using glutaraldehyde (Geffard et al., 1984b). This antiserum has been used in the locust and fruit fly to describe tyramine-like immunoreactivity in the brain (Nagaya et al., 2002; Kononenko et al., 2009; Homberg et al., 2013). Antiserum specificity of immunostaining was tested using tyramine/octopamine conjugated to bovine serum albumin (BSA). The conjugates were prepared as described elsewhere (Geffard et al., 1984a; Mons and Geffard, 1987). Brains were dissected out of the head in fixative containing 3% glutaraldehyde (Electron Microscopy Sciences (EMS), Hatfield, PA, USA) in 0.1 M cacodylate buffer (EMS, pH 7.0) with 1% sodium metabisulfite (SMB, Sigma-Aldrich, St.Louis, MO, USA). Each dissected brain was then transferred into 1 ml of fresh fixative and left overnight at 4°C. To saturate double bonds, after fixation the brains were treated with 0.5% sodium borohydride (NaBH4; Sigma-Aldrich) in 0.05 M Tris-HCl buffer containing 0.45% SMB (Tris-HCl-SMB; Sigma), pH 7.4 for 20 min. After washing with Tris-HCl-SMB buffer (4 × 10 min), the brains were embedded in 8% agarose (low melted point A0169, Sigma-Aldrich) in water. Brain sections (70 μm) were made with a Vibratome Leica 1000S (Leica Biosystem, Germany).

Brain sections were washed (6 × 20 min) in Tris-HCl-SMB buffer containing 0.5% Triton-X100 (TX) and were incubated with 1% normal donkey serum (Jackson ImmunoResearch Laboratories) in Tris-HCl-TX for 15 min. Then, tyramine antiserum was added to a final dilution of 1:500 to each brain and left for 48 h. After washing in 0.05 M Tris-HCl-TX, pH 7.5 (6 × 20 min), F(ab’)2 fragments of donkey anti-rabbit IgG conjugated to Cy3 (Jackson ImmunoResearch Laboratories, diluted 1:250 in Tris-HCl-TX) were applied as the secondary antibody overnight. All incubations were at room temperature. After final washing in 0.05 M Tris-HCl pH 7.4 (6 × 20 min) sections were mounted on slides in 80% glycerol in phosphate-buffered saline (PBS) mix.

To test the specificity of tyramine immunostaining, working dilutions of the anti-tyramine antibodies were preincubated overnight without and with tyramine conjugated to BSA, (10⁻⁴ M concentration of tyramine in the conjugate, Figures 1A,B). After preincubation of the primary antiserum with tyramine-G-BSA, anti-tyramine staining was absent (Figure 1B). However, after preincubation of the antibodies with octopamine-G-BSA (10⁻⁴ M concentration of octopamine; Figure 1C), immunolabeling in the cell body and fine processes was present. Therefore, the working dilution of the tyramine antiserum specifically recognizes tyramine in cell bodies and their processes in fixed honey bee brain sections. Twenty bee brains were processed and analyzed in our studies, eight brains that were fixed and processed on the same days were used to count cell bodies (Table 1).

Immunocytochemistry with mouse monoclonal anti-synapsin antibody (SYNORF1; clone 3C11, Developmental Studies Hybridoma Bank, The University of Iowa) was used to label the synaptic neuropil in brain whole-mounts according to the protocol of Brandt et al. (2005). Brains (n = 3) were fixed in 4% paraformaldehyde in PBS overnight, washed in PBS containing 1% Triton-X (PBS-TX), preincubated 1 h with normal donkey serum, and then the anti-synapsin antibody 1:1000 was applied for 3 days at room temperature under gentle shaking. After six washes of 1 h each with PBS-TX, the brains were incubated with the secondary antibody (F(ab’)2 fragments of donkey anti-mouse conjugated to Alexa 488, Jackson ImmunoResearch Laboratories) diluted 1:250 in PBS-TX for 2 days at room temperature. After final washing in PBS, the whole-mounts were dehydrated, cleared and embedded in methyl salicylate. The whole-mount brains were embedded in methyl salicylate for observation (n = 3). To illustrate the location of the anti-tyramine staining cell bodies, we used the consecutive protocol staining technique on sections. Three brains that were sectioned and labeled with anti-tyramine were post-fixed for 20 min with 4% paraformaldehyde, and then brain sections were processed for anti-synapsin immunostaining as described above to obtain tyramine immunoreactive cell bodies in the sections with the labeled synaptic structure of neuropil (n = 3).

The structural model of AmTyr1 (NCBI Reference Sequence: NP_001011594.1) was generated by I-TASSER (Zhang, 2008; Roy et al., 2010; Yang and Zhang, 2015; Yang et al., 2015) based on the top 10 proteins from the PDB that have the closest structural similarity. Besides the threading-based restraints, no additional external restraints were specified. Molecular graphics and localization of the antigenic peptides were performed with the UCSF Chimera package (Pettersen et al., 2004) developed by the Resource for Biocomputing, Visualization and Informatics at the University of California, San Francisco.

For all immunostaining on bee brain sections, we used the anti-AmTyr1 antibodies that were raised against the peptide-1, and the protocol for testing their immunostaining specificity on bee brain sections is described below.

Honey bee forager brains were removed from the head capsule under fixative containing 4% paraformaldehyde in PBS, then each brain was placed in one milliliter of the same fixative overnight at 4°C. The next morning, brains were washed in PBS and embedded in 8% agarose. Each agarose block was sectioned (70 μm) and processed for immunostaining. First, brain sections were preincubated with 1% normal donkey serum and then anti-AmTyr1 (affinity purified polyclonal antibodies raised against the peptide-1 in rabbit) at 1:500 dilution in the PBS-0.5%TX solution was added to each brain. To visualize the staining, F(ab)’2 fragments of donkey anti-rabbit IgG conjugated to Cy3 were used at a dilution of 1:500. After each step, there were at least six washes of 20 min each in PBS-TX (6 × 20 min). PBS only was used in the final wash before embedding brain sections on a glass slide. Twenty brains were processed with anti-AmTyr1 antibodies in these experiments.

To identify cell types that express AmTyr1 in the antennal lobe, we labeled ORN and PN terminals by injecting neurobiotin into the antenna and antennal lobe. For both procedures, each bee was cooled and placed into a plastic restraining harness, and the head was immobilized with low melting point wax. Then each antenna was gently immobilized with ecosaine, the isomeric hydrocarbons obtained from paraffin wax (Aldrich).

To label ORN endings in the glomeruli, neurobiotin was injected into the antenna, where the ORN axons take up the tracer, which then undergoes anterograde transport to the axon terminals in glomeruli. For neurobiotin injection into the antenna, a small hole was cut in the scape at the base of the antenna, and 50–70 nL of 2% neurobiotin (weight/volume in water) was injected using a picospritzer. To label ORN cell bodies in the antenna, neurobiotin was injected into the antennal lobe from where the tracer was transported retrogradely to the cell body and dendritic fibers in the sensory receptor pocket. 4′,6-diamidino-2-pheylindole (DAPI) was used as a fluorescent marker of cell nuclei in the antenna. Five bees were processed to trace the ORNs and co-labeled with anti-AmTyr1 and anti-synapsin.

To label the PN axon terminals we injected neurobiotin into the antennal lobe, where the PNs take up the tracer and anterogradely transport it to the mushroom body calyx, where axons form the synapses in the KC dendritic field. For this procedure, a small window was cut in the head capsule allowing access for neurobiotin injection into each antennal lobe. The bee then was detached from the holder and placed in a small wooden box with available food (1.5 M sucrose and pollen) and a humidified environment. The bees were sacrificed the next day 16–20 h after dye injections. Brains were dissected and fixed as described above for anti-AmTyr1 staining. Five bees were processed to trace the PNs and co-labeled with anti-AmTyr1 and anti-synapsin.

To label synaptic neuropil, the anti-synapsin antibody (dilution 1:1000) was added together with anti-AmTyr1 (1:500) overnight. Then sections were washed, and secondary antibodies F(ab’)2 fragments donkey anti-rabbit IgG conjugated to Cy3 (1:250) and F(ab’)2 fragments donkey anti-mouse IgG conjugated to 488 (1:250) were added to the sections together with streptavidin conjugated to Cy5 (Jackson ImmunoResearch Laboratories, 1:250) to reveal anti-AmTyr1, anti-synapsin and neurobiotin respectively in the brain sections. Preparations were then thoroughly washed in PBS and embedded in 80% glycerol. To control the specificity of the secondary antibodies, all secondaries were incubated with sections that had only one of the primary antibodies. The staining did not show any cross-reaction between the secondary antibodies and Streptavidin-Cy5. Streptavidin-Cy5 did not interact with any structure in the absence of the neurobiotin in the bee brain.

Data were collected on a Leica SP5 confocal laser scanning microscope (Leica, Bensheim, Germany) using a Leica HCX PLAPO CS 40_oil-immersion objective (numerical aperture: 1.25) with appropriate laser and filter combinations. Stacks of optical sections at 1 μm spacing were processed using Leica software (1024 × 1024 pixel resolution) either as a single slice or flattened confocal stacks (maximum intensity projections). Size, resolution, contrast, and brightness of final images were adjusted with Adobe Photoshop software. To generate the general view of a tyramine-containing cell in the whole brain we used whole-mount immunocytochemistry with anti-synapsin antibodies. Serial sections were made at 1 μm using a Leica ×10 objective and reconstructed to create 3D images of the brain using AMIRA software (FEI visualization science group). Then the agarose brain sections labeled with both anti-tyramine and anti-synapsin were compared with digital serial sections of the whole-mount brain, and cell bodies were manually added in the appropriate serial digital layer to create the image of tyramine cell body distribution in the whole-mount brain (Figure 3A).

The tyramine antiserum used in our studies specifically recognized tyramine, as shown through immunostaining pre-absorption tests (Figures 1A–C). Anti-tyramine immunostainings were repeatable across animals, revealing cell bodies, neurites and varicosities of neurons. The intensity of anti-tyramine staining in the cell bodies was variable, as illustrated in Figure 1A for ventral unpaired neurons: staining ranged from very bright to low intensities. After pre-incubation of working dilutions of anti-tyramine antibodies with tyramine-G-BSA, specific anti-tyramine labeling was abolished, as shown in Figure 1B for SEG frontal sections. When the same working dilution of anti-tyramine antibodies was preincubated with octopamine-G-BSA, the tyramine immunoreactive staining in the cell bodies and processes was still present (Figure 1C). In total, we treated 20 brains with anti-tyramine antibodies, and cell counts in Table 1 were based on eight of the brains.

We created a 3D model of the AmTyr1 receptor protein to show localization of the peptides used for immunization (Figure 2A). We used I-TASSER that employs composite approaches of threading, structural refinement, and ab initio modeling to generate a 3D model of the full-length AmTyr1 receptor. The best predicted model had a C (confidence)-score of −0.89, and the estimated TM (template modeling)-score and RMSD (root mean square deviation) of 0.60 ± 0.14 Å and 8.8 ± 4.6 Å, respectively. Generally, values of the C-score (typically, in the range of [−5, 2]) and the TM-score (range of [0, 1]) higher than −1.5 and 0.5, respectively, are indicative of correct global fold and main chain topology. RMSD is more sensitive to local errors, and it is not unusual to have big values for proteins of considerable length. In the light of the above, the generated model of the AmTyr1 receptor appears to be of reasonably high accuracy/quality.

The AmTyr1 receptor protein has seven transmembrane domains with a large intracellular loop between helixes 5 and 6 and a short C-terminus. The peptides used for immunizations are located extracellularly in the N-terminus and between helixes 4 and 5 (Figure 2A). After four immunizations, the AmTyr1 antisera were affinity purified and the anti-AmTyr1 antibodies were further characterized. First, we used western blotting to analyze anti-AmTyr1 antibodies for both peptides separately (Figure 2B). We observed one large band in the Western blot of brain homogenate proteins, and it corresponded to the predicted weight of the TYR protein 45 kD (Figure 2B) for both antibodies. Moreover, pre-incubation of anti-AmTyr1-peptide-1 and anti-AmTyr1 peptide-2 with corresponding peptide conjugated to KLH abolished the immunolabeling on Western blot (Figure 2B). In all our AmTyr1 receptor immunolabeling on bee brain sections, we used only the antibody raised against H₂N-TEDYDMTGCGPPEEET-amid (peptide-1, P1). Therefore, we used only this antibody in tests for specificity of immunolabeling.

Control immunolabeling on brain sections revealed that anti-AmTyr1 was strongly expressed in the cortex of glomeruli (Figure 2C), and staining was absent after pre-incubation of anti-AmTyr1 with peptide conjugated to carrier protein KLH (Figure 2D). Furthermore, knockdown experiments using dsiTyr1 RNA injections in an amount that reduced gene expression by approximately 50% (Figure 2E) significantly reduced labeling by anti-AmTyr1 antibodies (Figure 2F). In contrast, areas where the scrambled construct (dsiScr) was injected in the same amount as dsiTyr1 RNA failed to reduce labeling (Figure 2G). The average level of fluorescence intensity dropped to 42% in the area close to the dsiTyr1 injection site compared to the same area around the dsiSCR injection site (illustrated in Figure 2H). Figures 2F,G illustrate the frontal sections of the bee brains through the mushroom body vertical lobes embedded into the anterior protocerebrum. From all of these experiments, we conclude that the anti-AmTyr1 antibodies from rabbit specifically recognize the AmTyr1 receptor protein in Western blots and fixed brain sections.

The tyramine antiserum labeled clusters of cells in the brain (Figure 3A) and on the midline of the SEG (Figures 3A–D). Tyramine containing processes are illustrated in the antennal lobe (Figures 3E–G) and antennal nerve (Figure 3H), the lateral protocerebral lobe, the LH, the mushroom body calyces and lobes (Figures 3I–K) and in the anterior optic tubercule (AOTu; Figure 3K).

The major tyramine containing fibers in the neuropils of trito-, deuto- and proto-cerebral ganglia arose from the ventral neurons located in the midline of SEG (Figures 3A–D). These anti-tyramine labeled median neurons were classified based on the position of their primary neurites in tracts within the three SEG segments corresponding to the mandibular (Md, Figures 3B,C), maxillary (Mx, Figures 3B,D) and labial (Lb, Figure 3B) neuromeres, respectively. These neuromeres receive sensory projections from the nerves of the mouthparts and via the mandibular, maxillary and labial nerves supply muscles of the mouthparts involved in the proboscis-extension reflex (Rehder, 1988; Schröter et al., 2007). The SEG is also the relay station for information in descending and ascending neurons. We focused our study on median ventral neurons from the mandibular and maxillary neuromeres. In particular, we focused on the tyraminergic/octopaminergic ventral unpaired median (VUM) neurons (VUMmx and VUMmd) that send their symmetrical secondary neurites in each part of the brain to innervate the antennal lobe (Figures 3E–G), LH (Figure 3I) and mushroom body calyx (Figures 3I,J).

The group of anti-tyramine immunostained ventral neurons in the labial neuromere (Figure 3B) contains at least two dorsal unpaired neurons (not shown in this illustration). The secondary neurites from some of these neurons branch in the SEG, and there are at least two tyramine containing axons in the labial nerves. The DUM/VUM neurons from the labial neuromere innervate the labial nerves, the labial neuropil of the SEG and the corpora cardiaca (Figures 3B,C, Table 1). There were also tyramine containing axons in the connectives between the SEG and prothoracic ganglion (Figure 3B).

The mandibular and maxillary neuromeres each contain eight anti-tyramine immunoreactive neurons that have a laterally symmetrical morphology in the SEG and branch in the deuto- and proto-cerebrum, the antennal nerve and the lateral nerves (Figures 3E–H). At least two anti-tyramine staining branches are present in each (mandibular and maxillary) lateral nerve on each side of the SEG. The anti-tyramine labeled secondary neurites from some VUMmx and VUMmd neurons travel alongside the esophageal foramen (circle in Figures 3C,E) and run through the dorsal lobe and enter to the antennal tracts T5-T6. There were at least five tyramine containing axons in the ventral antennal nerve (Figure 3F and insert in Figure 3H). The antennal nerve also has large anti-tyramine positive varicose fibers on the surface (Figure 3H).

In our study we were particularly interested in VUMmd1 and VUMmx1 (Figures 3C,D), because of their importance for octopamine-driven behavioral conditioning (Hammer, 1993; Farooqui et al., 2003). They give rise to primary neurites in the corresponding MdT and MxT tracts and then to secondary neurites in the deutocerebrum caudal to the antennal lobe, as illustrated in the frontal section of the SEG (secondary neurites are circled in Figure 3C) and in the sagittal section through the left antennal lobe (Figure 3E). These two VUM neurons send branches into the antennal lobe (Figures 3F,G). In both Figures 3F,G, the anti-tyramine immunolabeled sagittal sections through the antennal lobe illustrate two tyraminergic branches from VUMx1 and VUMd1 (arrows). In Figure 3G, the sections made through the antennal T1 tracts show that each branch from VUMmx1 and VUMmd1 gives rise to additional branches that innervate glomeruli on the dorsal and ventral side of the antennal lobe (Figure 3G). The anti-tyramine labeled processes in each glomerulus have varicosities that could be release points for tyramine not only in glomeruli but also in the aglomerular neuropil.

The secondary neurites from VUMmx1 and VUMmd1 run through the lateral antenno-protocerebral tract (l-APT) (Figure 3I) and give rise to very fine branches in the LH and the calyx of the mushroom body with highest distribution in the lip and basal ring of mushroom body calyx (Figure 3J). These neurons arise from VUMmd1 and VUMmx1 and are identical to octopaminergic neurons in the same neuromere described in earlier studies (Schröter et al., 2007). That conclusion is based on our study with anti-tyramine and anti-octopamine labeling on the sections. No additional axons were labeled in the antennal lobe during these procedures [these data are not illustrated here].

The pedunculus and lobe of the mushroom body have little labeling with anti-tyramine staining, except for a few fine branches that were distinguishable in the gamma and vertical lobes (Figures 3I–K). In contrast, all protocerebral neuropils surrounding the mushroom bodies and lobes contain abundant tyramine containing varicosities (Figures 3I–K). Tyramine immunostaining in the mushroom body calyx has its origin from the VUMmx and VUMmd neurons. However, the mushroom body lobes are scarcely labeled with tyramine immunoreactivity, and staining in the gamma lobe originates from the lateral paired ventral cells (VPM, Figure 3D) as well from unidentified neurons in group G2. The tyraminergic neurons from group G3 also innervate the posterior protocerebrum and optic lobes. The tyramine containing neurons from group G4 innervate the protocerbral bridge and central complex (Figure 1 for cell bodies group location and Table 1). The terminology used for tyramine-containing cell groups is the same that is described in Sinakevitch et al. (2005).

We used anti-AmTyr1 antibodies to characterize the distribution of AmTyr1 in the antennal lobe and mushroom bodies (Figure 4). First, we performed single immunofluorescence staining in unknown age forager bee brains to identify the areas in the antennal lobe and mushroom bodies that labeled with anti-AmTyr1 antibodies (Figure 4). In the antennal lobe, anti-AmTyr1 staining is unevenly distributed within each glomerulus as well in the aglomerular neuropil (Figure 4A). High intensity anti-AmTyr1 staining is localized in the cortex area in each glomerulus, with low intensity staining in the core of the glomerulus and in the aglomerular neuropil (Figure 4A). Anti-AmTyr1 staining is absent in the antennal nerve and tracts in the antennal lobe (only tract T1 is shown in Figure 4A). Cell bodies surrounding the antennal lobe also labeled with different levels of intensity: high intensity labeling is in a subset of the medial group of cell bodies (asterisk) in Figure 4A. In contrast, the lateral group of cell bodies is not positive for anti-AmTyr1. Figure 4B illustrates a section through the mushroom body and central complex. Anti-AmTyr1 immunostaining is present in all mushroom body neuropils: calyx, peduncle, and lobes (Figure 4B). The peduncle and lobes have strong anti-AmTyr1 staining compared with KC bodies and calyx (Figure 4B). The anti-AmTyr1 labeling in the calyx and lobes is particularly distinct in the lip, collar and basal ring areas (Figure 4C). The gamma lobe of the mushroom bodies (staining is shown by the arrow in Figure 4C) had the most variable anti-AmTyr1 immunostaining distribution. Note that the ellipsoid body and ocelli also exhibited a high level of anti-AmTyr1 immunolabeling (Figure 4B).

To identify the primary cell types that express AmTyr1 in the antennal lobe we used triple immunofluorescence staining. The images in Figures 5A1–A5, show triple immunofluorescence staining in glomeruli with anti-AmTyr1 (magenta, Figure 5A1), neurobiotin backfills of antennal axons (green, Figure 5A2) and anti-synapsin (blue, Figure 5A4). Close-ups of ORN endings in the box in Figures 5A1–A5 is illustrated in Figures 5B1–B5, respectively. The distribution of anti-AmTyr1 is in the cortex of the glomerulus (magenta, Figures 5A1,B1), where the ORN terminals are located (green, Figures 5A2,B2). The endings of ORNs represent varicosities that synapse on the neuronal processes located in the cortex of the glomerulus. In the merged image (magenta anti-AmTyr1 and green ORNs, Figures 5A3,B3), the white color demonstrates co-labeling of anti-AmTyr1 in the ORN axon endings (Figures 5A3,B3). Therefore, ORNs containing neurobiotin also have anti-AmTyr1 staining (Figures 5A3,B3). Anti-synapsin also co-labeled the neurobiotin containing ORNs (Figures 5A4,B5). An example of staining with anti-AmTyr1 and anti-synapsin in ORNs is shown by the white arrows in Figures 5B1–B5. However, anti-AmTyr1 labeled processes other than ORN axons. The yellow arrows in Figure 5B1 demonstrate co-staining of anti-AmTyr1 with anti-synapsin but not with neurobiotin. These data suggest that other cell types could express AmTyr1 in the cortex of glomeruli, or possibly not all ORNs were labeled with neurobiotin in our preparations. Also, due to the limitation of light microscopy we cannot exclude that AmTyr1 is expressed in processes of the glomerular cortex that are not co-labeled with anti-synapsin. More work needs to be done to identify all possible antennal lobe neurons and/or glial cells expressing AmTyr1. To reveal ORN cell bodies and dendrites in sensilla, we injected neurobiotin in the antennal lobe (Figure 6). Then the antenna was opened and processed for double immunofluorescence staining to reveal neurobiotin (green) and AmTyr1 (magenta). The neurobiotin labeled ORN cell bodies and processes in sensilla (Figure 6). Figures 6A,B illustrates a frontal view of antennal subsegments 5 and 6 of a flagellum. This antenna was not cut during immuno-procedures, and the images were taken in a confocal mode with overexposure of all laser channels for illustration of the surface area of an antenna with different types of sensilla. The three olfactory sensilla are identified as sensilla placodea (p in Figures 6A,B), sensilla basiconica (b, Figure 6B), sensilla trichodea type A (tA, Figure 6B) and type B1 (tB1, Figure 6B), sensilla trichodea type C (Figure 6A, tC). In the honey bee, sensilla placodea (or poreplate) house between 7 and 30 ORNs (Esslen and Kaissling, 1976). To reveal anti-AmTyr1 and neurobiotin simultaneously in the antennal processes, antennae were cut with a razor into two halves and each half processed for staining with anti-AmTyr1, neurobiotin and DAPI; the latter is the marker for cell nuclei (Figure 6C). Neurobiotin processes in the sensilla placodea are not co-labeled with anti-AmTyr1. However, the neurobiotin filled ORN cell bodies (green Figures 6C,D) co-labeled with anti AmTyr1 (magenta in Figures 6C,D). The example of a cell with co-localization with AmTyr1 is indicated by the arrow in Figure 6D. The anti-AmTyr1 labeled processes in the sensillum lymph area (Figure 6E1), but they are not co-localized with neurobiotin labeled dendrites in the sensilla placodea (Figure 6E2) as demonstrated by the absence of white color in the merged image (Figure 6E3).

The calyx of the mushroom bodies consists of intrinsic neuron (KC) dendrites and afferent neuronal terminals of uniglomerular PNs (uPNs) from the antennal lobe (lip and basal ring), gustatory inputs (area between lip and collar) and visual inputs (collar). Interestingly, the gustatory inputs to mushroom body calyx are from subesophageal neurons, the same brain region where the modulatory octopaminergic/tyraminergic VUM neurons originate. In addition, other modulatory neurons and GABAergic inputs are present in the calycal neuropil. The organization of the calyx is microglomerular, which reflects interactions of uPN axons (input) with dendrites of KCs (output) together with the inputs of inhibitory (GABA) and modulatory afferent neurons.

Anti-AmTyr1 antibodies labeled microglomeruli in the calyx of the mushroom bodies (Figure 7A1). The higher magnification in image (Figure 7B1) represents the equivalent of an area shown in the square box in Figure 7A1 and illustrates a few of the microglomeruli. The anti-AmTyr1 is in the cortex of the microglomeruli, but the core is free from immunostaining.

To mark the neuronal structure of uPN terminals in the calyx microglomeruli, neurobiotin was injected in the antennal lobe to reveal uPN boutons in the basal ring and lip areas of the mushroom body calyx (arrows in Figures 7A2,B2). The dye revealed uPN boutons (Figure 7B2), which are mainly presynaptic to KC dendrites. As demonstrated in Figures 7A3,B3, anti-AmTyr1 co-labeled with uPN boutons and was localized in the periphery of the bouton (white color on merged images of Figures 7A1,A2,B1,B2). Anti-synapsin labeled the presynaptic sites of the microglomeruli (Figures 7A4,B4). When the images (anti-synapsin, anti-AmTyr1 and uPNs terminals) are merged in Figures 7A5,B5, the white color demonstrates co-expression of anti-AmTyr1 and synapsin in the periphery of microglomeruli within the same boutons as uPNs. Thus, AmTyr1 is localized in presynaptic sites of uPNs in microglomeruli of mushroom body calyx.

The mushroom body KC bodies express a low level of AmTyr1 staining compared to staining in the calyx (Figure 7C1). The anti-AmTyr1 marks KC bodies in the basal ring, lip and collar and scattered cell bodies of the gamma lobe with very low intensity, although it is possible to distinguish a different level of staining in subsets of cells with higher expression in the basal ring KCs (Figure 7C1). The anti-AmTyr1 staining is strong in the vertical mushroom body lobe, especially in the areas that receive axons from KCs in the basal ring, the collar and the lip areas. In comparison, staining is much lower in the gamma lobe (Figure 7D1).

KCs are the intrinsic neurons that make up the mushroom body. They have dendrites in the calyx and axons in the lobe, and different types of afferent (input/output) neurons make their connection to the KCs in the calyx and lobe. To identify localization of the AmTyr1 receptor in KC dendrites and axons, we labeled KCs with neurobiotin before immunostaining procedures with anti-AmTyr1 (Figures 7C2,C3,D2,D3) and anti-synapsin (Figures 7C4,C5,D4,D5) antibodies. For neurobiotin injected in the calyx of the cell body layer (Figure 7C), the area of injection was marked by an ellipsoid in the peduncle of the mushroom body. The dye was taken up by subsets of KC bodies and their dendrites in the calyx as well in the axons in the peduncle (Figure 7C2) and lobes (Figure 7D2).

The neurobiotin-injected KC dendrites in the calyx were not labeled with anti-AmTyr1. The insert in Figures 7C1,C5 shows typical AmTyr1 staining in the microglomeruli. In the insert in Figure 7C2, fine green fibers of KCs are in close proximity to the AmTyr1 stained PNs, but there is no co-labeling as demonstrated in Figure 7C3 and in the insert, where the white color expected for co-labeling in the dendritic area of KCs is absent. The AmTyr1 receptor is co-expressed in the bundles of axons of neurobiotin labeled KC in the pedunculus (white arrow in Figures 7C2,C3). The subsets of neurobiotin labeled KCs (green bundles in Figures 7C2,C3,C5) could be traced to the mushroom body basal ring, the collar, the lip and gamma lobe areas (green in Figures 7D2,D3,D5). KC axons co-express AmTyr1 in the basal ring, collar and lip area of the mushroom body (Figure 7D3). The insert of Figures 7D1,D2 show the area of the lip KC axons in the vertical lobe, where only a few fibers express receptors as indicated by the yellow arrows in inserts. The Figures 7C4,D4,C5,D5 show anti-synapsin staining in the calyx and lobes and anti-synapsin co-labeled with KCs axons in the pedunculus and lobe, with subsets of KCs that co-express AmTyr1 in the same axons.

A total of approximately 160 tyramine immunoreactive neurons are organized in different clusters in the brain. They are the source of tyraminergic fibers with small varicosities in the optic lobes, antennal lobes, lateral protocerebrum, mushroom body calyces and gamma lobes, tritocerebrum and SEG. Also, tyramine-like immunoreactive fibers are present in the antennal nerve and in the nerve innervating the corpora cardiaca (NCC1). Since tyramine is a precursor of octopamine, it is not surprising that our studies revealed that tyramine immunostaining is largely in the same clusters of cells as octopamine, as previously described in the honey bee brain (Kreissl et al., 1994; Sinakevitch et al., 2005). We summarized the cell numbers in Table 1 and used the same nomenclature of cell groups as Sinakevitch et al. (2005). According to our present report and comparison with previous studies on octopamine immunoreactivity in the honey bee (Kreissl et al., 1994; Sinakevitch et al., 2005), tyramine containing neurons described here are largely the same as those that contain octopamine due to the position of their clusters.

We also found differences from earlier studies (Sinakevitch et al., 2005), as reflected in Table 1. For example, we did not find tyramine immunoreactive cells in octopamine-positive clusters G0b, G1 and G5a. It could be that those cells convert tyramine to octopamine rapidly, such that the tyramine titers are below detection levels. We also found that there is a higher total number of cells containing tyramine immunoreactivity in comparison to octopamine immunoreactive cells reported by Kreissl et al. (1994) and Sinakevitch et al. (2005). Previous work on biogenic amine immunostaining reported that handling procedures could significantly alter octopamine/tyramine levels, and possibly make one or both undetectable by immunostaining (Sinakevitch et al., 2005; Kononenko et al., 2009; Homberg et al., 2013). However, the differences in numbers of octopaminergic and tyraminergic neurons in the honey bee brain might also be due to the presence of neurons that are only tyraminergic. Therefore, due to the higher number of tyramine-containing cells, and to a large amount of tyramine-containing profiles in neuropils, our data suggest that some cells could release only tyramine, some cells either release only octopamine or co-release both biogenic amines.

We also found that the antennal nerve contains tyramine-like immunoreactivity originating from processes in SEG neurons that enter the antennal nerve via T5-6. However, it is not clear what neurons in the SEG could be the source of these fibers: e.g., the neurons from cluster six or unidentified median neurons from mandibular and maxillary neuromeres. Tyramine in the antenna could be in the axons of neurons innervating the head and antennal muscles, and it could also act as a hormonal release in the antenna to modulate antennal sensory neurons. Furthermore, tyramine in the antennal lobe could also be involved in the regulation and modulation of sucrose responsiveness (Scheiner et al., 2002). It seems clear, however, that tyramine does not modulate ORN cell bodies via AmTyr1, because we did not observe AmTyr1 on the cell bodies of ORNs.

The primary sources of tyramine in the antennal lobe and calyx of the mushroom body are from at least two median neurons with cell bodies in the SEG: VUMmd and VUMmx neurons. In the antennal lobe, these neurons innervate the cortex of each glomerulus, where ORN axons terminate. They also branch to the LH and calyx of the mushroom body via the lateral antenno-protocerebral tract. The entire protocerebral neuropil, except for the mushroom body lobes, is penetrated by very fine varicosities of tyramine-containing fibers, and their sources are the VUM neurons from the SEG as well as the cells from clusters G2 and G3. The mushroom body lobe has very few tyramine containing fibers, only a few branches in the gamma lobe are clearly visible, and the source could be the laterally paired neurons from the SEG (Schröter et al., 2007).

Also, the presence of tyramine in NCC1 could indicate that the corpora cardiaca could receive tyramine-containing branches from some VUM neurons located in the SEG. The morphology of VUM neurons that give rise to axons in the NCC1 was previously described in the labial and maxillary neuromeres by Eichmüller et al. (1991). While the role of tyramine in NCC1 is unclear, it could be that tyramine is released from these neurons to act as a neurohormone, or/and it could control the release of other hormones from the corpora cardiaca.

Activation of AmTyr1 in heterologous expression systems leads to reduction of cAMP (Blenau et al., 2000; Reim et al., 2017). Both tyramine and octopamine reduce cAMP when bound to AmTyr1 (Blenau and Baumann, 2016), but tyramine is more potent than octopamine. Moreover, AmTyr1 binds yohimbine, which is an antagonist of the AmTyr1 receptor (Blenau and Baumann, 2001, 2016; Reim et al., 2017). In our 3D model, AmTyr1 has a long intracellular loop 3 and a short C-terminus. Those properties are similar to other GPCRs linked to inhibition of adenylyl cyclase activity; for example, the alpha 2-adrenergic receptors also have a long intracellular loop 3 and a short C-terminal tails (Kuhar et al., 1999; Rosenbaum et al., 2009). Based on the structural and pharmacological properties, AmTyr1 is similar to the vertebrate type alpha 2-adrenergic receptors, which when activated also reduce cAMP, and yohimbine has also high affinity of this vertebrate receptor type. The function of vertebrate-type alpha 2-adrenergic receptors is primarily for inhibitory presynaptic control of the release of norepinephrine, ATP and acetylcholine from the nerve (Rosenbaum et al., 2009).

The anti-AmTyr1 antibodies we used specifically recognize the AmTyr1 protein, and staining mapped with high intensity to honey bee brain neuropil areas. Some cell bodies immunolabeled with low intensity compared to the neuropil, which could reflect that the AmTyr1 receptor was translated in the cell body and transported to axonal terminals. The present localization studies on anti-AmTyr1 receptor protein distribution provide a further confirmation of the previous work of Mustard et al. (2005), where in situ hybridization of the AmTyr1 mRNA was reported to be expressed in the mushroom body and antennal lobe neurons.

Anti-AmTyr1 immunolocalization studies revealed that in the antennal lobe AmTyr1 is expressed in the presynaptic sites of ORN axons as they innervate the cortex of glomeruli. Similarly, in the mushroom body calyx, AmTyr1 is expressed in the presynaptic sites of uPN axons located primarily in the microglomeruli of the lip and basal ring calyx areas. AmTyr1 is expressed in areas innervated by VUM (md and mx) neurons. Therefore, release of tyramine from VUM (md and mx) neurons in the antennal lobe and mushroom body could target presynaptic sites of ORNs and uPNs (Figure 8). Because the AmTyr1 receptor is similar in structure and function to the vertebrate alpha-adrenergic receptor type 2 (Kuhar et al., 1999), we hypothesize that the release of tyramine from tyramine containing VUM neurons could inhibit excitatory neurotransmitter release in the presynaptic axons of ORNs and uPNs. The main excitatory neurotransmitter in ORNs and uPNs is acetylcholine, which plays important roles in olfaction and memory in bees (Gauthier, 2010). AmTyr1 in theses neurons might be involved in regulation of the release of acetylcholine. At the same time the release of octopamine from VUM neurons could coordinate excitation via AmOA1 in inhibitory neurons in the antennal lobe and mushroom bodies (Sinakevitch et al., 2013, Figure 8).

Only subsets of KCs express the AmTyr1 receptor in the axons and cell bodies, with a higher level of expression in the axons compared to cell bodies. We did not find any detectable staining in the dendrites of KCs located in calyx. In the KCs axons, AmTyr1 co-localized with synapsin, the presynaptic marker of the neurons. In the mushroom body lobe, the anti-AmTyr1 positive subsets of KCs axons are in the pedunculus, lip, collar and basal ring area of the lobe. The branching pattern of the KCs in the mushroom body lobes was described in detail by Strausfeld (2002).

Homolog of AmTyr1 have been studied in other insects: fruit fly (TyrR, Oct/TyrR, TYR, Tar1, CG7485; Saudou et al., 1990; El-Kholy et al., 2015); locust (Vanden Broeck et al., 1995); silkworm (Ohta et al., 2003); and cockroach (PeaTyr1, Rotte et al., 2009). Similar to AmTyr1, when investigated in expression systems, these Tyr1 receptors were also negatively coupled to adenylyl cyclase via Gi protein (Uzzan and Dudai, 1982; Aoyama et al., 2001; Blenau and Baumann, 2003). However, activation of locust and fruit fly Tyr1 receptors have also been shown to mobilize intracellular calcium (Ohta and Ozoe, 2014). The brain distribution of PeaTyr1 in the cockroach Periplaneta americana is similar to our results with AmTyr1, but that report did not identify specific neurons such as the presynaptic sites we show here (Rotte et al., 2009). Additionally, they reported expression in glial cells and peripherals organs. In our study, we also report the possible expression of AmTyr1 in glial cells and peripheral organs, but we did not illustrate it here. Expression of the moth (Agrotis ipsilon) OA/TYR (AipsOAR/TAR) was also reported in the antennae, the antennal lobes and the brain and in the antennal lobes was shown to be regulated with age (Duportets et al., 2010). In the fruit fly Drosophila melanogaster, the expression pattern of TyrRs was studied by utilizing the presumptive promotor regions of the TYR and the Gal4/UAS system (El-Kholy et al., 2015). The fruit fly TyrRs are expressed in the tracheal system, in salivary glands and in the mushroom body and ellipsoid body, glial cells, fat body and muscles. However, the fruit fly has three TYRs, one of which does not have an ortholog in the honey bee (Tyr3), which makes it difficult to compare our results on localization with the fruit fly (El-Kholy et al., 2015).

It was striking in our study that the mushroom body α- and γ-lobes do not have a high amount of tyraminergic (present studies) or octopaminergic fibers (Sinakevitch et al., 2005) in contrast to the calyx of the mushroom body and the protocerebral area that surrounds mushroom body lobes. However, the mushroom body lobes express both the AmTyr1 (present study) and AmOA1 receptors (Sinakevitch et al., 2011, 2013). We propose that the source of tyramine in the mushroom body lobe could be from the hemolymph or from the release of tyramine in the protocerebrum followed by diffusion into the lobes. The presence of tyramine in the hemolymph was reported in locust, and the level of tyramine is lower than octopamine in brain tissue as well in the hemolymph.

It may be important for behavioral studies to consider the ratio of tyramine and octopamine in the tissue rather than focusing exclusively on a single amine, since they have potentially antagonistic effects (Roeder, 2005). Fussnecker et al. (2006) showed that honey bee flight behavior was affected inversely by octopamine and tyramine treatment. Octopamine increased flight behavior, while tyramine treatment decreased it. Fussnecker et al. (2006) suggested that both biogenic amines affect central pattern generators or interact with sensory perception. Tyramine and octopamine have opposite effects on locomotion in the fruit fly (Saraswati et al., 2004), and have been shown (Brembs et al., 2007) to reduce fruit fly flight initiation. Tyramine also reduces the stimulatory effect of octopamine in the fly (Uzzan and Dudai, 1982). In the honey bee brain tyramine is involved in habituation of an appetitive reflex (Braun and Bicker, 1992) and in inhibition of the initiation of foraging behaviors (Schulz and Robinson, 2001). High brain tyramine in queenless honey bee workers might inhibit foraging behavior and encourage them to stay in the nest and become reproductive workers (Sasaki and Nagao, 2002).