Evolutionary neuroecology of olfactory-mediated sexual communication and host specialization in Drosophila – a review

Sex pheromone communication: a ‘curriculum vitae for mating’

Sex pheromones have a large impact on the life of an animal and represent a very detailed curriculum vitae to engage in copulation. They are the first and arguably the best-studied olfactory communication signals, from their production to detection and elicited behaviors (Dickson, 2008; Yu et al., 2010; Dahanukar & Ray, 2011). Sex pheromones are produced by members of almost all orders of animals, from worms (Jeong et al., 2005) and insects (Wyatt, 2017) to fish (Clarke et al., 1991), reptiles (Schoralkova et al., 2018), birds (Caro et al., 2015), and mammals (Goodwin et al., 1979). Animals use various mechanisms for releasing sex pheromones into the environment, e.g., they have been reported in the saliva of pigs, in urine and tears of mice, in vaginal fluids of hamsters, and in anal glands of insects (Brennan, 2010; Wyatt, 2017).

Although closely related species tend to produce similar sex pheromones, as a result of shared biosynthetic pathways that were present in common ancestors (Figure 1), they gain specificity by blending compounds in a species-specific concentration ratio. Aphids provide a fascinating example, as most species utilize the same four chemicals (nepetalactone and three isomers of nepetalactol), but each has its unique sex pheromonal blend due to a specific concentration ratio (Pickett et al., 1992; Hardie et al., 1994). Besides the various ratios, an enormous diversity of sex pheromones could be achieved by utilizing simple modifications (e.g., reduction and oxidation) to an existing sex pheromone (Lebreton et al., 2017; Borrero-Echeverry et al., 2021), which can explain the presence of different, but biosynthetically related (Figure 1), components in closely related species (Symonds & Elgar, 2008).

Contrary to the concept of species specificity, the same molecules could be used as a sex pheromone in different species, if: (1) they live in non-overlapping geographical regions, i.e., allopatric species (Wyatt, 2017), (2) the release time of sex pheromones is segregated to avoid heterospecific attraction (Ishikawa et al., 1999; Cardé & Haynes, 2004), or (3) the animals are unlikely to mate, as is the case for elephants and several moth species that utilize the same sex pheromone (Wyatt, 2014).

Animals often synthesize sex pheromones in a single form of two stereoisomers, i.e., enantiomers. Enantiomers are two compounds with the same molecular formula and order of atoms, but with different spatial orientations of the atoms, similar to the right and left human hands. For example, the mammalian pheromone 3,4-dehydro-exo-brevicomin has two chiral centers (RR and SS enantiomers); only the RR enantiomer is produced and detected by male mice (Novotny et al., 1995). Moreover, the biologically ‘wrong’ enantiomer sometimes acts as an inhibitor of the physiological activity of the ‘right’ component. It is, therefore, not surprising that racemic mixtures of pheromones (1:1 mixture of both enantiomers) often result in poor behavioral responses (Leal et al., 1998). An interesting example is provided by two Japanese beetles species, Popillia japonica Newman and Anomala osakana Sawada, that produce different enantiomers of γ-lactone. The two species can accurately distinguish between the two signals, because one of the enantiomers is an attractant whereas the other acts as an inhibitor, and vice versa for the other species (Leal et al., 1998).

Sex pheromones induce hard-wired and innate behaviors with immediate (i.e., releaser effects, such as courtship rituals) and long-lasting (i.e., primer effects, such as phase changes in locusts) consequences. However, the interpretation of the meaning of sex pheromones could vary among individuals due to context, age, sex, or other factors including internal or mating state (Wyatt, 2014). For example, the urinal male-specific pheromone of mice induces aggression in males, attracts females, induces estrus in mature females, and accelerates maturation of young females (Novotny, 2003). Similarly, in the vinegar fly, 11-cis-vaccenyl acetate (cVA), a male-specific sex pheromone that transfers to females during copulation, triggers many behavioral outputs (Figure 2). The compound induces sexual receptivity in virgin females (Bartelt et al., 1985; Ha & Smith, 2006; Kurtovic et al., 2007), but not in mated females (Lebreton et al., 2014; Das et al., 2017). On the other hand, it suppresses courtship and elicits aggression in males (Jallon et al., 1981; Ejima et al., 2007; Wang & Anderson, 2010). Furthermore, cVA works as an oviposition cue (Dumenil et al., 2016) and aggregation pheromone for both sexes (Bartelt et al., 1985), and induces long-range attraction behavior that could be modulated by starvation (Lebreton et al., 2015).

Sex pheromone processing: turning the chemical codes into electrical signals

Insects detect odors with their antennae – specifically the third antennal segment (the funiculus) – and maxillary palps (Figure 3). Olfaction starts when odor molecules diffuse into the pores of hair-like structures called sensilla, which cover the surfaces of these olfactory organs (Hansson, 1999; Menini, 2010). Each sensillum houses a different number of olfactory sensory neurons (OSNs). Once an odor binds to the cognate chemoreceptor, which is located on the surface of the ciliary membranes of OSNs, the responsive OSN transforms this chemical information into electric signals that are conveyed further to the primary olfactory neuropil (Figure 3). In insects, this neuropil is located in a region called the antennal lobe (AL), which is analogous to the olfactory lobe in crustaceans or the olfactory bulb in vertebrates (Ache & Young, 2005; Homberg et al., 1989). The OSNs send their axons to the AL (Figure 3) to innervate discrete spherical subunits of this neuropil − the olfactory glomeruli.

The number of glomeruli varies widely across animal species and correlates to the number of odorant receptors (ORs), e.g., 62 ORs and 52 glomeruli in D. melanogaster, 170 ORS and 165 glomeruli in honeybees, around 380 ORs and 400 glomeruli in ants, and ca. 1000 ORs and 1800 glomeruli in the olfactory bulb of mice (Galizia et al., 1999a,b; Schachtner et al., 2005; Maresh et al., 2008; Zube et al., 2008; Krieger et al., 2010; Braubach et al., 2012; Grabe et al., 2016). In both invertebrates and vertebrates, each glomerulus is innervated by a group of OSNs that express the same chemoreceptor(s) (i.e., OSNs with identical odor response profiles) (Mombaerts et al., 1996; Gao et al., 2000; Vosshall et al., 2000). The glomeruli are connected by a network of local interneurons (LNs), which represent an early processing step for shaping the efferent olfactory information into the higher brain centers [reviewed in Grabe & Sachse (2018) and Yang et al. (2019)]. The LNs’ arborization patterns diversify their anatomical structure, with some LNs innervating only a few glomeruli (so-called patchy LNs) and others targeting many, if not all, glomeruli (so-called panglomerular LNs) (Chou et al., 2010; Seki et al., 2010; Liou et al., 2018; Mohamed et al., 2019). Two types of LNs are distinguished by the impact of the signal and use of neurotransmitters: inhibitory and excitatory neurons (Shang et al., 2007; Silbering et al., 2008).

After processing by LNs, afferent olfactory information is conveyed from the AL to higher brain centers via second-order neurons, so-called olfactory projection neurons (PNs, analogous to mitral/tufted cells in vertebrates). The PNs send their dendrites to the AL glomeruli and their axons to higher-order brain centers. The PNs reveal a clear spatial segregation with respect to their response, to either pheromones or food odors (Jefferis et al., 2007), attractive amines or aversive acids (Min et al., 2013), and opposing hedonic valences and odor intensity (Strutz et al., 2014). The PNs convey information through the inner, middle, and outer antennocerebral tract (iAct, mAct, and oAct) (Stocker et al., 1990) to the mushroom body (MB) and the lateral horn (LH) (Figure 3) (Vosshall & Stocker, 2007). The mushroom body is analogous to the piriform cortex in mammals (Su et al., 2009) and is thought to be a center for olfactory learning and memory (Davis, 1993; Heisenberg, 2003; Fiala, 2007; Waddell, 2013), whereas the lateral horn shares many similarities with the mammalian amygdala (Miyamichi et al., 2011; Sosulski et al., 2011) and is presumed to mediate innate behavior (Heimbeck et al., 2001). However, recent studies have documented many roles of the MB in innate behaviors (Bracker et al., 2013; Lin et al., 2014; Lewis et al., 2015) and of the LH in modulation of the learning processing of the MB (Dolan et al., 2019). The LH output neurons project their axonal terminals to many other brain regions (Tanaka et al., 2004; Frechter et al., 2019) and then to descending neurons that will ultimately activate the motor system, to enable the animal to perform odor-guided behaviors.

Drosophila as a model to study the evolution of sex pheromone communication

Since 1903, when William Ernest Castle started to use D. melanogaster as a research organism in the laboratory, five Nobel prizes in physiology or medicine have been awarded to nine scientists for their vinegar fly-based discoveries: (1) In 1933, to Thomas Hunt Morgan for his pioneering discovery that the ‘chromosome plays a key role in heredity’ (Morgan, 1910). (2) In 1946, to Hermann Joseph Muller for the discovery that the ‘mutation rate of genes increases by x-ray radiation’ (Muller, 1930). (3) In 1995, to Edward B Lewis, Christiane Nüsslein-Volhard, and Eric F Wieschaus for understanding the genetic control of embryonic development (Lewis, 1978; Nüsslein-Volhard & Wieschaus, 1980). (4) In 2011, to Jules A Hoffmann for discoveries concerning the activation of innate immunity (Lemaitre et al., 1996). And (5) in 2017, to Jeffrey C Hall, Michael Rosbash, and Michael W Young for uncovering the molecular mechanisms that control the circadian rhythm (Bargiello et al., 1984; Zehring et al., 1984; Siwicki et al., 1988; Hardin et al., 1990; Liu et al., 1992; Vosshall et al., 1994; Price et al., 1998).

All of these discoveries were possible due to the ease of manipulating fly genomes by unique and sophisticated genetic tools, which are capable of activating or silencing the gene(s) of interest (Senturk & Bellen, 2018). Besides our ever-increasing bounty of knowledge and tools, and the ability to extrapolate to related species, their wealth of ecological diversity has made Drosophila spp. an excellent model to elucidate the genetic and neural correlates of their evolution. The genus Drosophila includes more than 1500 described species that live in quite diverse environments (Markow & O'Grady, 2005; Jezovit et al., 2017). Drosophilids’ various ecological niches and species-specific behaviors and neural circuits, as well as the available genome sequences and experimental tools, have made insect sexual communication a valuable model to study evolutionary neuroecology.

Sex pheromone signaling in D. melanogaster is well understood, from sensory signals and their cognate receptors to neurons and their connectivity (Auer & Benton, 2016). Therefore, sex pheromone signaling in drosophilid flies represents an ideal system to address the evolutionary basis of various sexual traits. Moreover, the small number of pheromone-sensing ORs compared to other ORs tuned to general odors (Couto et al., 2005) makes a comprehensive evolutionary study feasible. Out of the 52 OSN classes in D. melanogaster (Grabe et al., 2016), only four express OSNs tuned to pheromones, which are localized in a particular sensillum type (trichoid sensillum) (Couto et al., 2005). The OSNs housed in trichoid sensilla are activated exclusively by olfactory sex pheromones and consist of two classes (at1 and at4). Whereas at1 sensilla are located on the ventro-distal region of the funiculus and house one OSN, at4 sensilla are present on the dorso-distal region and house three OSNs (Clyne et al., 1997; van der Goes van Naters & Carlson, 2007; Datta et al., 2008; Dweck et al., 2015; Lin & Potter, 2015). The pheromone-tuned OSNs and their corresponding pheromones are Or67d and Or65a/b/c, which are both tuned to cVA, Or47b, tuned to methyl laurate, and Or88a, tuned to methyl laurate, methyl myristate, and methyl palmitate (Kurtovic et al., 2007; van der Goes van Naters & Carlson, 2007; Dweck et al., 2015).

Reproductive isolation through host specialization

Colonization of a new host could directly lead to divergence in the olfactory communication systems through many ways (Figure 4). First, host shift has severe impact on the biology of animals, e.g., body size, shape, and color, as well as population density (Nosil et al., 2003; Messina, 2004). These changes, therefore, may impact communication signals, mate-searching strategies, and sexual preferences (Etges, 1998). For example, searching for mates in less dense environments could lead to the adoption of long-range sexual signals. Second, new hosts could provide new ways of signal transmission and perception, resulting in differential sexual signaling (Funk et al., 2002), a process called sensory drive (Endler & Basolo, 1998). Sensory drive concerns how sexual traits and preferences coevolve in response to new environments. Lastly, host volatiles may be perceived and preferred as background or as components of the sex pheromone blend. For example, exposure to a particular host plant while mating has a strong impact on the subsequent reproductive preferences of male moths (Proffit et al., 2015; Zakir et al., 2017).

Similarly, in D. melanogaster, food odors (vinegar) boost the perception of the male sex pheromone – cVA – in virgin females, but not in males nor in mated females (Das et al., 2017). Although vinegar does not directly activate glomerulus DA1 – a specific glomerulus that responds to cVA (Kurtovic et al., 2007) – vinegar odors potentiated the cVA response on the PNs projecting to this glomerulus through electrical synapses between excitatory LNs and PNs in the antennal lobe. In line with these findings, other food odors (phenyl acetic acid and phenyl acetaldehyde) induce courtship in male flies through the ionotropic receptor (IR84a) (Grosjean et al., 2011). Vinegar is a complex blend that consists of many odorants, of which acetic acid is the most abundant compound (Becher et al., 2010). Although acetic acid alone is sufficient to elicit attraction in flies, it failed to potentiate the cVA response in combination with cVA, indicating the vinegar blend is necessary to mediate this synergism. Notably, the synergistic pattern, which was observed on the electrophysiological level, has been translated into increased female receptivity (Das et al., 2017). Interestingly, males showed neither synergistic effects nor behavioral responses in the presence of vinegar and cVA (Das et al., 2017). In behavioral assays, fed virgin Drosophila females were more attracted to the blend of cVA and vinegar than to vinegar alone, whereas males were not (Lebreton et al., 2014). Obviously, female flies benefit from not only focusing on male odors when looking for a mating partner, but also keeping the odors of potential food sources in mind.

Drosophila spp. live in a wide range of habitats across all climatic conditions, from deserts and caves to mountains and forests (Markow & O'Grady, 2005; Jezovit et al., 2017). In these environments, they feed and breed on diverse hosts such as decaying fruits, slime fluxes, mushrooms, and flowers, as well as frog spawn. On the one hand, many drosophilids share the same habitat (Markow & O'Grady, 2005; Jezovit et al., 2017); to reduce competition, they might allocate more energy to a distinct sensory modality (Keesey et al., 2019). On the other hand, many drosophilids specialize on a single host resource, such as Drosophila mojavensis Patterson on cacti, and Drosophila erecta Tsacas & Lachaise on Pandanus fruit. Another notable drosophilid that has received particular attention with respect to its chemoreception is Drosophila sechellia Tsacas & Bachli, which is endemic to the Seychelles archipelago and has evolved extreme specialism for noni fruits (Morinda citrifolia L.), which are toxic for most (if not all) other drosophilids, including its sibling species Drosophila simulans Sturtevant and D. melanogaster (Legal et al., 1992).

The extreme specialism that D. sechellia has evolved compared to its close cousins did not only result in an increased metabolic tolerance of noni fruit toxins (R'Kha et al., 1991; Jones, 1998), but also in changes in courtship and mate choice (Cobb et al., 1989; Coyne, 1992), oviposition (Moreteau et al., 1994; Amlou et al., 1998), and pupariation site selection (Erezyilmaz & Stern, 2013). Recently, these shifts were shown to be correlated with changes at the periphery of the fly’s olfactory system, i.e., whereas the odorant receptor OR22b is lost, the flies exhibit an increased number of neurons expressing OR22a (Auer et al., 2020). However, the many behavioral shifts raise the possibility that the peripheral changes of olfactory coding are accompanied by novel central projection patterns. Indeed, circuit tracing experiments revealed the presence of increased sensory pooling onto interneurons and novel branches of PNs (Auer et al., 2020). Consistent with these findings, Or22a displays substantial intra- and interspecific nucleotide and copy number variation in many drosophilids (Guo & Kim, 2007; Nozawa & Nei, 2007; Aguade, 2009; Goldman-Huertas et al., 2015). In addition, the number of Or22a neurons has also expanded in D. erecta, a specialist on Pandanus fruit (Linz et al., 2013). Obviously, OR22a is an important receptor when it comes to host specificity. However, despite the importance of the Or22a pathway, several other olfactory channels play a role in noni attraction. These include Or85c/b neurons, which have conserved physiological properties among drosophilids but increased in number in D. sechellia, and Ir75b neurons, which have changed in both function and number while preserving central projections (Prieto-Godino et al., 2017). Such observations indicate that neural pathways may adapt in distinct ways, possibly reflecting diverse selection pressures and roles that enable animals to specialize to a unique host, and thus limit encountering heterospecific flies.

Reproductive isolation through mate selection

Sexual traits involved in communication between the sexes often provide an instant contribution to reproductive isolation (Blair, 1955; Bridle & Ritchie, 2001). Upon search for a mate, females rely on honest signals – including visual, acoustic, and chemical signals – to find high-quality males (Bateman, 1948; Grafen, 1990). Therefore, sexual selection has led to the evolution of sexual signals in males and the cognate sensory systems in females for many animal species. Compared to other sensory information, olfactory sex pheromones provide the first cue to recognize a conspecific individual and can include much valuable information, such as its sex, age, body size, reproductive status, and health condition (Martin et al., 2007; Nielsen, 2017). These olfactory cues not only provide information on the sender, but may also have negative or positive effects on the receiver. For example, in mice, male scent induces estrus in females, whereas female scent suppresses ovarian activity in other females (Whitten, 1959, 1966; Nielsen, 2017). Once they are attracted by sex pheromones, males and females engage in species-specific courtship rituals, by which both partners ensure their reproductive success.

Sex pheromones and courtship behaviors of drosophilid flies differ quantitatively and qualitatively (Spieth, 1952; Symonds & Wertheim, 2005) and therefore represent an attractive model to identify the genetic basis of phenotypic evolution. These species-specific behaviors represent a pre-mating isolation barrier, especially between sympatric species (Spieth, 1952). When meeting at the host, flies identify conspecific mating partners through integration of various sensory modalities including vision, audition, olfaction, and gustation (Markow & O'Grady, 2005), which leads to either increased or inhibited courtship behaviors. Fruitful insights have been gained regarding the evolution of sensory modalities that underlie the interspecific variations in drosophilid courtship behavior (Manoli et al., 2005; Ding et al., 2016; Tanaka et al., 2017; Seeholzer et al., 2018; Ahmed et al., 2019). Sexual behaviors in drosophilids consist of a series of actions and reactions that vary qualitatively and quantitively between species (Spieth, 1952) and result in species and sexual discrimination. At first, males orient to and follow the females. Subsequently, males tap the females’ bodies with their forelegs, followed by wing spreading and fanning for vibrational song production. Males follow the females by extending their proboscis to lick the females’ genitalia and then attempt copulation. Obviously, the male has to invest to convince the rather choosy female. Interestingly, female choosiness varies even within species. When population size is small, i.e., when potential mates are rare, females that are too choosy may not find any suitable males, whereas females that are less choosy might still reproduce, although with less satisfying males (de Jong & Sabelis, 1991; Martínez-Ruiz & Knell, 2017).

One remarkable drosophilid that splits into four geographically isolated and ecologically distinct populations, and represents a model of incipient speciation and host adaptation, is D. mojavensis (Etges, 2019). The northern subspecies D. m. wrigleyi and D. m. mojavensis utilize the prickly pear cacti in Santa Catalina Island (CA, USA) and the red barrel cactus in the Mojave Desert (CA, NV, AZ, and UT, USA), respectively. The southern subspecies D. m. sonorensis and D. m. baja breed and feed on the organ pipe cactus in the mainland Sonoran Desert (Mexico and AZ and CA, USA) and the agria cactus in Baja California (Mexico), respectively (Ruiz et al., 1990). Phylogenetic analyses revealed that the two northern and the two southern subspecies cluster with each other (Allan & Matzkin, 2019). These closely-related subspecies exhibit many differences, including in their morphological features (Pfeiler et al., 2009), neurophysiological responses (Crowley-Gall et al., 2016; Date et al., 2013; Nemeth et al., 2018), genomic (Matzkin, 2014) and transcriptomic (Matzkin & Markow, 2013) characteristics, and behavioral traits (Newby & Etges, 1998). Experimental reciprocal crosses between these allopatric subspecies revealed various pre- and post-zygotic isolation barriers (Zouros & Dentremont, 1980; Krebs & Markow, 1989; Knowles & Markow, 2001).

Leveraging the short evolutionary time scale between the closely related D. mojavensis populations, we elucidated the evolution of the reproductive isolation barriers among these populations and the underlying sensory mechanisms (Khallaf et al., 2020). We identified four male-specific acetates of which three were exclusively produced by the northern subspecies. This dramatic change in chemical profile was surprising given the short divergence time, ca. 0.25 million years, between D. mojavensis subspecies (Matzkin, 2014; Etges, 2019). The four populations were feeding on similar artificial media, suggesting that the differential male-specific compounds are not a consequence of nutrition but are rather genetically determined. When checking for the function of these novel compounds, we realized that R-(Z)-10-heptadecen-2-yl acetate (R-HDEA) induces female receptivity, whereas (Z,Z)-19,22-octacosadien-1-yl acetate (OCDA) reduces male courtship. Together, these two substances fulfil the same roles that are described for cVA in D. melanogaster. Our results support the findings of Chin et al. (2014), who showed that D. m. wrigleyi males avoid courting females perfumed with ejaculatory bulb extracts of mature males (which, according to our analyses, contain OCDA). By contrast, similar extracts from immature males (which do not produce OCDA) did not elicit this behavior (Chin et al., 2014). We provided direct evidence that, in the northern subspecies, the dedicated R-HDEA-sensing neurons and auditory cues collaborate to mediate mate recognition. However, the southern subspecies rely on auditory cues only (Etges et al., 2006) and rather ignore R-HDEA for mate recognition. As all D. mojavensis flies can detect R-HDEA (which is only present in the northern subspecies), the striking change of R-HDEA-induced behaviors among D. mojavensis subspecies implies the existence of differences in the central processing pathways (Khallaf et al., 2020). These findings are reminiscent of sexually dimorphic behavioral responses to cVA in D. melanogaster (Ruta et al., 2010; Kohl et al., 2015), where both males and females detect cVA, but interpret this in different ways.

Most drosophilids are gregarious in mating, except the Hawaiian species, whose males display lekking behavior, i.e., they select and defend a unique territory on their host (Spieth, 1952; Carson et al., 1970). Similar to Hawaiian species, we demonstrated that D. m. wrigleyi males advertise their presence by pulsating a droplet of anal secretions, which contains the volatile sex pheromone, R-HDEA, in close proximity to females to enhance their receptivity (Khallaf et al., 2020). Such a trait is supportive to the observed sexual behaviors of D. mojavensis in nature, where males conquer undamaged areas next to feeding sites on cactus and attract females to this spot (O'Grady & Markow, 2012). Similarly, other members of the Drosophila genus show species-specific courtship behaviors (Spieth, 1952), which might include nuptial gift donation (Steele, 1986), partners’ song duet (O'Grady & Markow, 2012; LaRue et al., 2015), or territorial dating (O'Grady & Markow, 2012).

Divergence of the chemosensory genes could be accompanied by a rewiring of the central circuitry (Ding et al., 2016; Seeholzer et al., 2018). Indeed, neurons of D. mojavensis expressing the pheromone receptor Or65a do not innervate to the same glomerulus (DL3) as in D. melanogaster. Furthermore, despite expressing homologues of Or65a, the corresponding pheromone circuits of D. mojavensis and D. melanogaster govern different behaviors: the Dmoj-Or65a circuit detects R-HDEA and mediates female receptivity and conspecific recognition, whereas the homologous circuitry in D. melanogaster detects cVA and suppresses courtship and aggression behaviors in males (Ejima et al., 2007; Liu et al., 2011), as well as attraction towards cVA in mated females (Lebreton et al., 2014). These differences also suggest that the target domains in the higher brain centers of both circuits have changed. Notably, the processing pathway of a comparable pheromonal circuitry, between D. melanogaster and D. simulans (Seeholzer et al., 2018), has changed during less than one-tenth of the divergence time between D. melanogaster and D. mojavensis (ca. 40 million years ago). Thus, divergent functions of a homologous neural circuitry during a longer divergence time could be presumed.