Odorant-receptor genes encode proteins with a putative seven-transmembrane-domain structure and have been identified in vertebrates and invertebrates, including various insect species1. Generally, a given odorant-receptor gene is expressed in only a small fraction of olfactory receptor neurons (ORNs), and each ORN expresses a very small number of odorant-receptor genes². In insects, however, one highly conserved member of the odorant-receptor gene family is expressed in nearly all ORNs in addition to the conventional odorant receptor or receptors^{3, 4, 5}. The encoded proteins DOR83b in D. melanogaster, AgOr7 in Anopheles gambiae, HvirR2 in Heliothis virescens and AmelR2 in Apis mellifera have between 64% and 88% amino acid−sequence identity³, a level not observed for any other insect odorant-receptor gene. The function of this ubiquitously expressed protein is still unknown. Antennal neurons in D. melanogaster that express only DOR83b do not respond to any of a large panel of odors6, so it is unlikely that the receptor itself has ligand-binding properties. Alternatively, DOR83b could act as a subunit in heterodimeric receptor complexes, either modulating the binding specificity of the coexpressed receptor or regulating the assembly of functional receptor transduction complexes, both of which are known functions of G-protein-coupled receptors⁷. Heterodimerization might alternatively be necessary for the folding or membrane targeting of conventional odorant-receptor proteins^{3, 5}. Using a combination of electrophysiological and biochemical approaches, we investigated the role of DOR83b in the olfactory system of D. melanogaster.

Investigations of the function of DOR83b were done in a heterologous HEK293 expression system (Fig. 1)*. Bioluminescence resonance energy transfer (BRET)⁸ experiments showed that DOR83b interacts with other odorant receptors. Energy transfer occurred when a DOR83b-luciferase (Luc) fusion construct was expressed together with DOR43a−green fluorescent protein (GFP) or with DOR22a-GFP in HEK293 cells (Fig. 1a)*, indicating that dimers or oligomers were forming between DOR83b and the conventional D. melanogaster odorant receptors DOR43a or DOR22a (see Supplementary Methods online). The energy transfer was similar to the one that we measured with GFP and Luc-tagged beta2-adrenergic receptors, G-protein-coupled receptors for which homodimerization is already known to occur⁸. Similar results were obtained for coexpression of DOR83b-GFP and DOR83b-Luc and for coexpression of DOR43a-GFP and DOR43a-Luc, indicating that the receptors can also homodimerize. Receptor oligomerization was specific and not the result of receptor overexpression, as no BRET was detected when DOR83b-Luc was coexpressed with the GFP-tagged cyclic nucleotide−gated channel from D. melanogaster or with the GFP-tagged rat beta2-adrenergic receptor (Fig. 1a)*. Dimeric complexes persisted following denaturing gel electrophoresis and western blotting (Fig. 1g, arrows)*.

To test the functional consequences of coexpression of DOR83b with DOR43a, we performed calcium imaging measurements in the heterologous HEK293 system (Supplementary Methods online), which has frequently been used to assay odorant receptor function⁹. DOR43a can be activated by cyclohexanone^{10, 11}, but in HEK293 cells we obtained responses only when ligand concentrations were in the millimolar range, and with poor efficiency (below 1% of cells responding; Fig. 1d,h)*. In contrast, when DOR43a was cotransfected with DOR83b, the threshold dropped to the micromolar range and the number of cells responding significantly increased to 10−15% (Fig. 1d,i)*, even though expression of the receptors remained unchanged (30−40% of cells after transfection; Fig. 1b,c,e,f)*. The DOR43a-DOR83b heterodimer had a spectrum of active ligands similar to that reported for DOR43a-expressing neurons^{12, 13} and for recombinantly expressed DOR43a alone¹¹. Active substances were cyclohexanone, benzaldehyde, isoamyl acetate, cineole and cyclohexanol; inactive substances were ethylbutyrate, octanal and 2-octanal (c = 1 mM). To generalize these findings, we tested the responses of DOR22a transfected alone and with DOR83b (DOR22a/DOR83b) in HEK293 cells, and found similar effects. DOR22a/DOR83b cotransfected cells responded to lower (micromolar compared to millimolar) concentrations of ethylbutyrate, one of the best ligands for DOR22a-expressing neurons12, and the number of responding cells was significantly higher (<1% for DOR22a, 8−10% for DOR22a/DOR83b) despite similar transfection rates (both approx20%). To check the specificity of the effect of the DOR83b cotransfection, we transfected DOR43a together with DOR22a into HEK293 cells and, again, obtained responses only at ligand concentrations in the millimolar range and with poor efficiency (fewer than 1% of cells responding; data not shown). DOR83b did not respond to any of the odorants tested, even in the millimolar range.

If, as the BRET and calcium imaging data suggest, DOR83b functions as a co-receptor, a reduction in DOR83b expression in vivo should lead to a reduction in odor responsiveness of the ORNs. To investigate the influence of DOR83b expression in the fly, we used RNA interference¹⁴ (RNAi) to reduce the level of DOR83b and monitored the odor-evoked response of the antennae using electroantennogram (EAG) recordings¹⁰ (Fig. 2, Supplementary Methods online). In flies hatched from embryos injected with DOR83b double-stranded RNA (RNAi flies), the EAG amplitude was reduced compared to control (w1118) flies (Fig. 2a). Six different, commonly used odors evoked significantly smaller EAG responses in RNAi flies compared to control flies (P = 0.027 − 0.006, Fig. 2c), with average reductions ranging from 56% for 1-butanol (P = 0.0087) to 30% for ethanol (P = 0.0075). The residual response is likely to reflect incomplete knockdown of DOR83b mRNA, as suggested by the absence of detectable responses in DOR83b mutant flies¹⁵. Moreover, in situ hybridization on the antennae of the flies (RNAi versus non-RNAi) on which EAG recordings had been done (Supplementary Methods) showed that DOR83b mRNA levels were much lower, but not completely abolished, in flies with lowered EAG responses (Fig. 2b,d).

Here we show for the first time that insect odorant receptors form dimers and that heterodimerization improves the functionality of the receptors. Our results suggest that DOR83b functions as a co-receptor of conventional D. melanogaster odorant receptors. Further investigation will be needed to determine whether dimerization is a general property of odorant receptors in vertebrates as well.

REFERENCES

1. Mombaerts, P. Science 286, 707−711 (1999).

2. Mombaerts, P. Curr. Opin. Neurobiol. 14, 31−36 (2004).

3. Krieger, J., Klink, O., Mohl, C., Raming, K. & Breer, H. J. Comp. Physiol. A Neuroethol. Sens. Neural. Behav. Physiol. 189, 519−526 (2003).

4. Pitts, R.J., Fox, A.N. & Zwiebel, L.J. Proc. Natl. Acad. Sci. USA 101, 5058−5063 (2004).

5. Vosshall, L.B., Amrein, H., Morozov, P.S., Rzhetsky, A. & Axel, R. Cell 96, 725−736 (1999).

6. Elmore, T., Ignell, R., Carlson, J.R. & Smith, D.P. J. Neurosci. 23, 9906−9912 (2003).

7. Angers, S., Salahpour, A. & Bouvier, M. Annu. Rev. Pharmacol. Toxicol. 42, 409−435 (2002).

8. Angers, S. et al. Proc. Natl. Acad. Sci. USA 97, 3684−3689 (2000).

9. Spehr, M. et al. Science 299, 2054−2058 (2003). | Article | PubMed | ISI | ChemPort |

10. Stortkuhl, K.F. & Kettler, R. Proc. Natl. Acad. Sci. USA 98, 9381−9385 (2001).

11. Wetzel, C.H. et al. Proc. Natl. Acad. Sci. USA 98, 9377−9380 (2001).

12. Hallem, E.A., Ho, M.G. & Carlson, J.R. Cell 117, 965−979 (2004).

13. Wang, J.W., Wong, A.M., Flores, J., Vosshall, L.B. & Axel, R. Cell 112, 271−282 (2003).

14. Kennerdell, J.R. & Carthew, R.W. Nat. Biotechnol. 18, 896−898 (2000).

15. Larsson, M.C. et al. Neuron 43, 703−714 (2004).

Figure 1. Function of odorant-receptor homo- and heterodimers.
(a) BRET assays were done with HEK293 cells transfected with cytosolic GFP + luciferase, DOR83b-Luc alone, DOR83b-Luc + CNG-GFP, DOR83b-Luc + beta2-AR-GFP (all negative controls) and beta2-AR-Luc + beta2-AR-GFP (positive control), and with different combinations of D. melanogaster odorant receptors. Error bars, s.d. (b−i) HEK293 cells expressing DOR43a-GFP (b,e) or DOR43a-GFP + OR83b (c,f). Representative traces (integrated fluorescence ratio f340/f380 as function of time) showing the increase in intracellular calcium evoked by cyclohexanone in cells expressing DOR43a alone or together with DOR83b (approx1,000 cells were examined in each experiment). (d) [AU: Please specify units for vertical scale bar next to calcium recordings]. Cyclohexanone-induced increase in intracellular Ca2+ in HEK293 cells transfected with DOR43a alone (h) and with DOR43a + DOR83b (i). Western blot showing equal amounts of membranes from control (mock-transfected), DOR43a-GFP− and DOR43a-GFP + DOR83b−transfected HEK293 cells stained with anti-GFP antibodies (g). Arrowheads indicate bands at approx70 kDa and at approx140 kDa, corresponding to the sizes of receptor monomers and dimers.

Figure 2. Reduction of DOR83b mRNA abundance in D. melanogaster antennae impairs odor detection.
(a) EAG recordings of responses to ethyl acetate from a control (upper curve) and one RNAi fly (lower curve); black bar indicates the odorant pulse (1 s). The mutant fly curve represents one of the lowest responses measured. (b−d) In situ hybridization showed (incomplete) knockdown of DOR83b mRNA by double-stranded RNA in sections through the third antennal segment of control (b) and RNAi (d) flies. Scale bar, 5 mum. Bar graph shows average EAG response of control flies (black bars, n = 5) and RNAi flies (gray bars, n = 17) to several different odorants (c). Error bars, s.e.m.; *, P < 0.05; **, P < 0.01.