Molecular studies in insect olfaction

Insects, at various stages of their development, use a range of environmental cues to locate their mates, food sources and oviposition and hibernation sites, and to avoid dangerous situations or unsuitable habitats and hosts. Many of these interactions are mediated by chemicals that act as signals without having direct physiological effects, and these signalling compounds, or semiochemicals, are classified according to the relationship between the organisms involved. Pheromones are secreted and released by an organism and elicit a specific response in a receiving organism of the same species, whereas allelochemicals comprise a range of semiochemicals produced by one species and causing a response in another. Molecular biological techniques are now being used to study aspects of insect olfaction. Thus, insect genes encoding the proteins involved in the biosynthesis of pheromones, the perception of semiochemicals and the processing of the signals are being cloned and characterized. More general techniques, for detecting changes in gene expression, mapping genetic loci and studying mutants, are also being used to identify genes involved in behavioural responses elicited by semiochemicals. The pathways involved in insect pheromone biosynthesis appear to have evolved by the addition of one, or a few, tissue-specific enzymes that convert the products of normal primary metabolism into compounds of high stereochemical specificity, which then act as the pheromones (for an overview see Tillman et al., 1999). Thus, pheromones arise from the pathways for isoprenoid biosynthesis or by the transformation of amino acids, fatty acids and some other cellular precursors. A number of the genes encoding the enzymes involved in these conversions are now being studied. For example, components of the sex pheromones of lepidopterans are usually derived from fatty acids by a species-specific combination of desaturation, chain shortening and reductive modification at the carbonyl carbon. cDNAs encoding the enzymes involved in the desaturation step have been cloned from Trichoplusia ni (Knipple et al., 1998; Liu et al., 1999). This was facilitated by the similarity between insect desaturase sequences and those of vertebrates and fungi which made it possible to use previously cloned genes to design degenerate primers for reverse transcription polymerase chain reaction (RT-PCR) on messenger RNA (mRNA) from insect tissues (Knipple et al., 1998). Most pest aphids, which are mainly from the subfamily Aphidinae, use the sesquiterpene (E)-β-farnesene as an alarm pheromone to signal when a colony is under attack (Pickett & Griffiths, 1980; Hardie et al., 1999). This compound is produced from farnesyl diphosphate (FPP) by both plants and animals, and a cDNA encoding the synthase responsible for its production in the mint plant (Mentha×piperata, L.) has been cloned and sequenced (Crock et al., 1997). Indeed, many plant genes involved in isoprenoid biosynthesis have been cloned (for reviews see Scolnik & Bartley, 1996;Bohlman et al., 1998), but whether the sequences will be of use in strategies to clone homologous genes from insects remains to be seen. It may be that the pathways in insects and plants have arisen by convergent evolution and there will be little, or no, similarity between genes to allow the use of degenerate PCR. In our laboratory we have attempted to use primers, designed from plant sesquiterpene synthases, to prime RT-PCR fragments of the (E)-β-farnesene synthase gene from aphids, but so far this has been unsuccessful (L. M. Field & R. J. Tregaskis, unpublished). However, a cDNA encoding the FPP synthase from the moth Agrostis ipsilon has been obtained by PCR, using primers designed to rat, human, yeast and plant sequences (Castillo-Gracia & Couillaud, 1999), suggesting that there is a common ancestral origin of the genes in the pathway up to FPP. The production of pheromones by insects is regulated by at least three hormonal messengers: juvenile hormone III, ecdysteroids and a 33 or 34 amino acid pheromone biosynthesis activating neuropeptide (PBAN). In the latter case, cDNAs encoding the neuropeptides have been cloned from a number of insect species, including Bombyx mori (Xu et al., 1996;Kawano et al., 1997), Mamestra brassicae (Jacquin-Joly et al., 1998) and Helicoverpa assulta (Choi et al., 1998). Molecular studies have shown that the PBAN of B. mori is produced, together with a diapause hormone, from a common polypeptide precursor (Xu et al., 1995). A cloned cDNA for the PBAN of Helicoverpa zea has also been used to study its tissue and temporal expression (Xu et al., 1995; Ma et al., 1998). Overall, there are many opportunities for studying genes involved in pheromone production which, in turn, will lead to a better understanding of insect–insect signalling and the evolution of associated metabolic pathways. In insects, perception of volatile semiochemicals is mediated via specialized olfactory organs, the sensilla, located mainly on the antenna. Information on the perception of semiochemicals can be obtained using electrophysiological recording techniques, the simplest of which is the electroantennogram (EAG), which measures an overall polarization of the antenna and represents the summated response of many receptor neurons. More detailed information on the nature of the olfactory neurons can be obtained from extracellular recordings of individual sensilla, achieved by either implanting an electrode at the base of a sensillum or putting an electrode over a cut sensillum end. Thus, the response profiles of individual olfactory neurons have been investigated, leading to their classification as either specialists or generalists (Strausfeld & Hildebrand, 1999), with specialists being narrowly tuned to the detection of one particular semiochemical whilst generalists may show a response to a number of compounds. However, the latter categorization is proving incorrect as more detailed studies are made. The sex pheromones in the Lepidoptera have been studied extensively and their multicomponent nature, comprising mixtures of long-chain aldehydes, alcohols or acetates, has been established. Electrophysiological studies have demonstrated the presence of olfactory cells specialized for the detection of the individual pheromone components. This allows the male moth to distinguish the sex pheromone blend emitted by females of its own species from those of other species, even where the species share the same principal pheromone components, thereby minimizing mating attempts with the wrong females (Baker, 1998). Recent studies have shown that, in many cases, neurones tuned to the detection of agonists are co-located on the same olfactory sensillum as those tuned to antagonists. This allows males to discriminate between pheromone and antagonist sources separated by as little as 1 mm (Baker et al., 1998). Pheromone receptors in other insect orders can be equally specialized and chirality can play a key role in species isolation. Thus, a number of bark beetles have olfactory neurons tuned to the detection of individual chiral pheromone components (e.g. Wadhams et al., 1982). Some economically important pest aphids have specific olfactory cells for the perception of particular structural isomers of either the nepetalactone or nepetalactol components of their sex pheromone, and this discrimination is maintained even at high stimulus concentrations (Hardie et al., 1999). Again, compartmentalization of olfactory neurons within the same sensillum allows accurate discrimination of the conspecific ratio of pheromone components. Pheromones frequently act in conjunction with other semiochemicals, particularly plant volatiles. For example, the aggregation pheromone of the pea and bean weevil, Sitona lineatus, is synergized by so-called 'green leaf volatiles' from the host plant (Blight et al., 1991). Although these are ubiquitous plant volatiles, perception is mediated by specialized olfactory neurons. Indeed, electrophysiological studies, using physiologically/biologically relevant stimulus concentrations on a wide range of insects from several orders, have demonstrated the widespread occurrence of this phenomenon (Pickett et al., 1998). Even aphids, which possess relatively few olfactory receptors, have cells on the primary rhinaria which respond selectively to the green leaf volatile (E)-2-hexenal. In a situation analogous to that observed in the vertebrate system, the detection of semiochemicals by insects involves proteins which carry the compounds from the porous cuticular surface of the antennal sensilla through the sensillum lymph to the G-protein-coupled receptors and the olfactory neurons, which in turn activate signalling cascades (for reviews see Prestwich, 1996; Krieger & Breer, 1999). It has been suggested that such insect odourant binding proteins (OBPs) may participate in the recognition of odourant molecules (e.g. Steinbrecht, 1996) and clearly, although specificity would be expected to reside in the receptors, the diversity of the system could be extended if the carrier proteins also embody aspects of molecular recognition. Insect OBPs and the subgroup of pheromone binding proteins (PBPs) share some characteristics with the OBPs of vertebrates in that they are all small (15–20 kDa), soluble proteins with acidic isoelectric points, concentrated in the sensillum lymph (Pelosi & Maida, 1995). Genes and cDNAs encoding the OBPs of many insect species have now been cloned and this is allowing some of the questions on the evolution and role of OBPs to be addressed. During the late 1980s and early 1990s, cDNAs encoding OBPs were cloned from a range of lepidopterous insects, including Manduca sexta (Györgyi et al., 1988), Antheraea polyphemus (Raming et al., 1989), Antheraea pernyi (Breer et al., 1990) and Heliothis virescens (Krieger et al., 1993). Comparisons of the predicted amino acids sequences revealed that PBPs from different species were very similar to each other and to the so-called general odourant binding proteins (GOBPs), which can themselves be divided into two groups (Vogt et al., 1991). Since then, more lepidopterous PBP cDNAs have been identified, e.g. from B. mori (Krieger et al., 1996) and M. brassicae (Maïbèche-Coisnéet al., 1998a), and a general picture has emerged. All encode proteins with a signal peptide, a major hydrophobic domain (between residues 40 and 60), a conserved Arg 46 and six conserved cysteines. Beyond these there is also considerable similarity between amino acid sequences (Maïbèche-Coisnéet al., 1998b). Following the cloning of lepidopterous cDNAs encoding OBPs, cDNAs were cloned from Drosophila melanogaster (McKenna et al., 1994 and Pikielny et al., 1994), showing that there is very little similarity between the amino acid sequences of lepidopterous OBPs and those of D. melanogaster. However, there was still conservation of the six cysteine residues at similar positions to those in the moth sequences. Many other sequences of cDNAs encoding 'putative' OBPs and PBPs from a range of insect species have now been reported, including those of Carausius morosus (Tuccini et al., 1996), Ceratitiscapitata (Thymianou et al., 1998), Apis mellifera (Danty et al., 1997) and Phyllopertha diversa (Wojtasek et al., 1999). This has confirmed that, although there are only low levels of amino acid similarity between the OBPs of different species, they do all have a signal peptide and are of a similar size, and there is conservation of the cysteine pattern. Thus, most insect OBPs have X18-C-X30-C-X3-C-X42-C-X8–10-C-X8-C-X24–26, although with some variation in the spacing. A subgroup of proteins from C. capitata and Tenebrio molitor share only four of the six cysteine residues: X19–20-C-X30-C-X37–39-C-X19-C-X17 (Christophides et al., 2000). What has gene cloning and sequencing told us about the evolution and role of OBPs? The finding that the binding proteins of unrelated species have only low levels of amino acid similarity suggests that they have arisen from a common ancestral gene which has diverged considerably. The conservation of the cysteines shows that they are important for protein function. Alignments of members of the PBP/OBP family across species can be used to generate phylogenetic trees. One such tree indicated that moth proteins belong to one branch (which can be subdivided into two groups, the PBPs and GOBPs), with neither D. melanogaster, T. molitor nor C. capitata proteins being closely related (Thymianou et al., 1998). However, three regions, near cysteines 1, 3 and 6, are more conserved and may be structurally important (Thymianou et al., 1998). There is good evidence from molecular studies for the presence of multigene families, gene duplication and divergence events (e.g. Merritt et al., 1998;Christophides et al., 2000). However, with the lepidopterans there is no evidence that where species have converged on the same pheromone component they have also converged in PBP sequence (Willett, 2000). As discussed earlier, there have been suggestions that OBPs may be involved in conferring some of the specificity of the odourant molecular recognition system. So, have molecular genetic studies contributed to this debate? The demonstration of the presence of two different genes encoding two PBPs in A. pernyi in 1991 provided preliminary evidence for a specific function for the proteins (Krieger et al., 1991). This work was important because it looked at the genes and not just cDNAs, and therefore the production of multiple PBPs by alternative splicing of a primary transcript could be ruled out. However, subsequent studies detected a single PBP in Anomala osakana (Wojtasek et al., 1998) and a single PBP gene in two Ostrinia species (Willett & Harrison, 1999), suggesting that the PBPs in these species are not involved in discriminating between pheromone isomers. Recent advances in molecular techniques have shown that an expressed sequence tag (EST) library from M. sexta has seven OBPs which, together with evidence from the D. melanogaster genome project for seventeen members of the OBP family, is very strong support for a model of olfaction involving OBPs selectively transporting and presenting odours to the olfactory receptors (Robertson et al., 1999). The other advantage of cloning OBP cDNAs is that it allows expression of recombinant proteins in sufficient quantities (multigram) for structural analyses (circular dichroism (CD) and nuclear magnetic resonance (NMR)) and ligand binding studies. The first such demonstration of pheromone binding to recombinant protein was for the A. pernyi PBP expressed in a baculovirus vector (Krieger et al., 1992), and this was followed by similar studies on the A. polyphemus PBP (Du et al., 1994; Du & Prestwich, 1995). Such studies have led to an elegant model for a PBP–pheromone complex activating a G-protein-coupled receptor containing recognition sites for both the presented pheromone and the presenting PBP (Prestwich et al., 1995). Recombinant GOBPs from M. sexta (Feng & Prestwich, 1997) and M. brassicae (Maïbèche-Coisnéet al., 1998a) have also been used to demonstrate that GOBPs can be tuned to a class of 'green' and floral odours. More recently, a recombinant protein representing a new class of carriers from T. molitor has been studied by NMR (Rothemund et al., 1999). This has revealed the tertiary structure of the protein and binding of the pheromone 4-methyl nonanol. There have also been considerable advances in studies of the recombinant PBP from B. mori and its ligand, bombykol. The crystal structure of the protein–ligand complex has been obtained (Sandler et al., 2000) and NMR has been used to characterize the pH-dependent equilibrium between two detected conformations of the PBP (Damberger et al., 2000). The advantage of cloned OBP cDNAs is that they can be subjected to site-directed mutagenesis to help define ligand binding specificity (e.g. Prestwich, 1996). However, further studies will be required to determine whether or not binding proteins play a major role in molecular recognition. Although different species have varying olfactory systems, the molecular receptors are invariably G-protein-coupled seven transmembrane proteins. The first report of cloning genes encoding such proteins from vertebrates came in the early 1990s where eighteen members of a multigene family were characterized from rats, with evidence for at least seventy different genes (Buck & Axel, 1991). A recent review concluded that, in the mouse and rat, the odourant receptor genes number 1000 or 1% of the genome (Mombaerts, 1999). For insects, the sequencing of the D. melanogaster genome has recently allowed screening of the database for similar proteins, which has again identified seven transmembrane domain proteins that are putative olfactory receptors encoded by large multigene families (Clyne et al., 1999; Gao & Chess, 1999; Vosshall et al., 1999). These are expressed in olfactory organs with different genes showing expression at different times during antennal development (Clyne et al., 1999), and sensory neurones expressing different complements of receptor genes (Vosshall et al., 1999). The finding of seventeen such genes from the just 15% of the then available D. melanogaster genome indicated that the total might be nearer 100–200 receptor genes (Gao & Chess, 1999). With the recent completion of the genome sequence (Adams et al., 2000), revised estimates should soon become possible. Rubin et al. (2000) have done a comparative analysis of the D. melanogaster genome with other eukaryotic genomes and concluded that, not only do the sequences of odourant receptors differ, but the processes generating the receptor gene families may have differed in each lineage. Parallel work on gustatory systems has also revealed at least 100 receptors (Clyne et al., 2000). It is interesting that these receptor genes were only found by novel database searches looking for sequences that might encode transmembrane domains, whereas previous attempts to find genes by homology with mammalian sequences had failed. This is a good example of protein function, conferred by its three-dimensional structure, being conserved but not the sequence itself, and provides a lesson in the shortcomings of homology cloning. Once an odour or pheromone has activated the olfactory receptors, it needs to be degraded, to prevent interference with the recognition system. Several detoxifying enzymes capable of this function are at high levels in olfactory systems (for review see Pelosi, 1996). In insects, these enzymes include esterases, oxidases and glutathione transferases. In D. melanogaster, a cDNA encoding a NADPH-cytochrome P450 oxidoreductase shows high antennal expression, suggesting a function in olfactory clearance (Hovemann et al., 1997). In M. sexta, a cDNA homologous to glutathione-S-transferase genes has been characterized, suggesting the dual function of protecting the olfactory system from harmful xenobiotics and inactivating aldehyde odourants, especially components of the sex pheromone (Rogers et al., 1999). Studies of how pheromone–receptor interactions and pheromone deactivation occur are now advancing. A recent model proposes that, in A. polyphemus, the pheromone interacts with the receptor whilst still being bound to the reduced form of the PBP. The receptor then catalyses the conversion of the reduced form of the PBP to the oxidized form, which deactivates the pheromone bound to the PBP (Kaissling, 1998). This model is at present unsubstantiated at the physiological or biochemical levels. Such complicated models are testable when the genes encoding the proteins involved have been cloned, and site-directed mutagenesis can be used to investigate how complexes form and deactivate. As discussed, much of the success in cloning genes and cDNAs encoding proteins involved in insect responses to semiochemicals has come about by the use of known sequences to clone related genes. However, this is clearly not possible if similar genes have not been cloned or the genes have diverged too far. Furthermore, we must always bear in mind that selection operates on the phenotype and not the genotype, and proteins with the same function may be unrelated at the sequence level. Many molecular techniques, available for cloning novel genes, have been used in studies of chemical ecology. For example, OBPs from D. melanogaster were cloned using subtraction libraries, whereby antennal-specific cDNAs were cloned and then putative OBP genes identified from their conserved cysteine pattern (Pikielny et al., 1994). We have used a similar approach to identify putative aphid OBP genes (L. M. Field & A. P. Liggins, unpublished). The techniques of differential display PCR and cDNA amplified fragment length polymorphisms (AFLPs) can also be used to isolate tissue-specific mRNAs, and this approach was successful in identifying the GST from M. sexta described above (Rogers et al., 1999). These techniques clearly offer opportunities for identifying genes which are associated with insect olfaction without a prior knowledge of the corresponding biochemistry. The general techniques of quantitative trait loci mapping have also been used to identify loci that influence the level of alarm pheromone in honeybees (Hunt et al., 1999). Olfactory mutants of D. melanogaster have proven useful in providing genetic evidence for the role of inositol-1,4,5-trisphosphate in olfactory signal transduction (HekmatScafe & Carlson, 1996), and to show that the sex-lethal gene regulates sexual behaviour and the synthesis of three major sex pheromones (Tompkins & McRobert, 1995). The D. melanogaster genome database will also allow the identification of gene sequences known to be associated with proteins involved in small molecule recognition. Sequence data from these genes could then be used to design degenerate oligonucleotides to isolate appropriate D. melanogaster mRNAs. These could be translated in a cell-free system and this library of proteins screened against further structured libraries of small lipophilic molecule metabolites. Thus, a functional link could be provided from the D. melanogaster genome through related proteins to molecules with which they interact, proving a new example for elucidating genomic function and a new database for exploiting protein molecular recognition. Overall, molecular techniques are now playing a major role in studies of insect chemical ecology and are likely to do so with increasing importance in the future. IACR-Rothamsted receives grant-aided support from the Biotechnology and Biological Sciences Research Council of the UK.