Manduca sexta moths were either reared in our lab or pupae were kindly provided by Monika Stengl (University of Kassel). For rearing, we used an artificial diet (156.25 g soy flour, 144.38 g rye flour, 144.38 g wheat flour, 85 g agar-agar, 4.5 ml rapeseed oil, 1.45 g calcium carbonate, 1.45 g sodium chloride, 4.8 g methyl paraben, 4.8 g sorbic acid, 14.5 g ascorbic acid, 38.5 g milk powder, 32 mg nicotinic acid, 16 mg riboflavin, 7.5 mg thiamine, 7.5 mg pyridoxine, 7.5 mg folic acid, 0.64 mg biotin). We kept the larvae in a climate chamber with a 14 h:10 h light:dark cycle, a temperature of 26°C (light cycle) or 24°C (dark cycle), and at 60% relative humidity. Male and female pupae were sorted and moved to separate climate chambers with a 16 h:8 h light:dark cycle, a temperature of 25°C, and at 60% relative humidity (light cycle) or 70% relative humidity (dark cycle). Eclosing adults were collected daily and held in individual brown paper bags (17 cm × 26 cm) in the pupal climate chambers.

For virgin samples, males and females remained in separate chambers until tissue collection. For mated samples, males and females were paired 2 days after eclosion during early scotophase and observed every 2 h for successful mating. Tissues from virgin and mated moths (Supplementary Table 1) were collected 3 days after eclosion during the scotophase. Moths were first cooled down at 4°C for about 15 min to decrease movements, then tissues were dissected with microscissors, collected in 2 ml microcentrifuge tubes or 15 ml falcon tubes (forewings), immediately immersed in liquid N₂, and stored at −80°C until further processing. The tissues (except forewings) were lysed and disrupted in RLT buffer (Qiagen, Germany¹) using Tissue Lyser LT (Qiagen, Germany) with a steel bead (diameter: 5 mm) at 50 oscillations per second for 15 min. The forewings, due to their large size, were ground to a fine powder in liquid N₂ using a mortar and a pestle. Total RNA was extracted from the lysed tissues using RNeasy Mini Kit (Qiagen, Germany) and RNase-free DNase Set (Qiagen, Germany) was used for the DNase digestion step. The concentration and purity of the extracted total RNA were checked first on a NanoDrop One (Thermo Fisher Scientific²). Three samples for each experimental group, leading to six ovipositor samples (3 virgin female, 3 mated female), and 12 samples for each of the remaining tissues (3 virgin female, 3 mated female, 3 virgin males, 3 mated males) were then run on 2,100 Bioanalyzer (Agilent Technologies, Inc.³) to confirm RNA integrity. Based on initial standardization assays, 200 ng total RNA from each tissue was used for the NanoString gene expression assay.

We used the nCounter XT CodeSet gene expression assay (NanoString Technologies, Inc., United States⁴). The custom CodeSet (Zhang et al., 2022) contained 268 probes targeting 71 ORs, 29 IRs, 49 GRs, 47 odorant binding proteins, 5 pickpocket, 3 sensory neuron membrane proteins and 62 candidate reference gene transcripts (Supplementary Table 2). The probes were designed using the reference coding sequences of 74 ORs, 21 IRs, and 45 GRs (Koenig et al., 2015) that were submitted to M. sexta OGS2.0 (Kanost et al., 2016). Four isoforms of MsexGR11, six isoforms of MsexGR15 and 8 IRs that were identified later (MsexIR1.1, 7d.2, 7d.4, 85a, 100a, 100d, 100f, and 100h) (Liu et al., 2018) were included. Some of the receptor sequences had high homology with sequences of their duplicates, prohibiting the design of unique probes. Therefore, these receptors had to be excluded (MsexOR23, MsexGR15D, 25, 26, 27, and 36). For all final probes, we performed NCBI BLAST search against M. sexta reference mRNA. Except for a few cases with duplicate receptors for which the probes could cross-target, our probes did not have off-targets. The recent de novo M. sexta genome assembly (Gershman et al., 2021) indicated that MsexOR84 and MsexOR89 might be the same gene, and similarly MsexOR42 and MsexOR66 might be the same gene. Therefore, we provisionally named these receptors MsexOR84/89 and MsexOR42/66.

We followed the standard protocol from nCounter XT Gene Expression Assay User Manual (MAN-10023-11, page 16⁵). All reagents and buffers were purchased from NanoString Technologies, Inc., United States. An attenuation mix (Eurofins Genomics, Germany⁶) to suppress counts of MsexABPx, MsexOBP1, MsexOBP5 and MsexOBP6 was used as these genes had very high expression levels causing high binding densities for the antennal samples. Briefly, a master-mix of 42 μl Reporter CodeSet, 28 μl Reporter-Plus reagent and 70 μl nCounter SPRINT hybridization buffer was created. For each hybridization reaction we combined 10 μl master-mix, 5 μl total RNA (200 ng), 1 μl attenuation-mix and 3 μl mixture of Capture ProbeSet and Capture-Plus reagent. Hybridization was done at 65°C for 22 h, after which 16 μl Merck water was added to the sample. The entire volume was loaded on nCounter SPRINT Cartridge (NanoString, United States) and processed on nCounter SPRINT Profiler (NanoString, United States).

The raw data from the gene expression assay (see Data Availability Statement) was processed using nSolver4.0 (NanoString, United States). Quality control for mRNA data was done on each sample using default parameters for nCounter SPRINT Profiler according to NanoString Gene Expression Data Analysis Guidelines (MAN-C0011-04⁷). The parameters were Imaging QC: 75; Binding Density QC: 0.1 – 1.8; Positive Control Linearity QC: 0.95; Positive Control Limit of Detection QC: 2 standard deviations. Due to high expression levels of chemosensory genes in the antennae, those samples had a slightly higher binding density than the default threshold of 1.8. We visually inspected the raw data from antennae following instructions from the user manual and confirmed that the data was robust and usable. However, the raw counts of positive spike-in controls in antennae were lower than those of the other tissues (Supplementary Figure 1). To avoid normalization flags, we normalized the antennal samples separate from the samples of the other tissues. For antennae, background subtraction was done first using the raw counts of the 8 negative control probes (mean + 2 standard deviations). After that, two normalization steps were performed, first using the geometric mean counts of the 6 external positive control probes, and second, using the geometric mean counts of three endogenous reference genes that were selected based on their coefficient of variation (CV < 30%). The endogenous reference genes for antennal samples were melastatin, smoothened, and dunce. For the remaining tissue samples, background subtraction and positive control normalization was done similar to the antennae. The endogenous reference genes used for the second normalization were melastatin, smoothened and two G protein genes (Gbeta13F and concertina) with a CV < 25%. After these two normalization steps, the minimum normalized count for each sample was defined as “background,” and any chemosensory receptor gene with counts above this background in at least 2 of the 3 samples was considered to be expressed in that experimental group. The normalized data are available online (see section “Data availability statement”).

The double-normalized counts (referred to as “counts” in the Results and Discussion) were log₂ transformed and used for making heatmaps and for statistical analyses using R (version 4.1.2)⁸ and Rstudio.⁹ The heatmaps were generated using the libraries heatmaply and RColorBrewer and later edited on Adobe Illustrator CS5. For identifying sex and mating state dependent gene expression in a given tissue (except ovipositor), we performed a nested ANOVA for each gene with sex as the main factor and mating state as the nested factor and a significance level of 0.05. The test was done separately for each tissue only for those genes that were expressed in at least one of the four experimental groups for that tissue. We then calculated the false discovery rate (FDR) using the Benjamini-Yekutieli procedure with the sets of p-values for the two factors for each tissue. Finally, a gene was considered to be significantly differentially expressed between sexes or mating states if FDR < 0.05 and log₂ fold-change ≥ 2. Ovipositor samples from virgin and mated females were compared with an unpaired t-test. For a p-value < 0.05 and a log₂ fold-change = 2 between both groups, a gene was considered to be differently expressed.

We measured the expression of chemosensory receptors in different tissues (antennae, labial palps, proboscis, forelegs, midlegs, hindlegs, forewings, hindwings) of female and male M. sexta moths, both before and after mating. In addition, we examined the expression of chemosensory genes in the ovipositor of virgin and mated females. The probe set targeted 71 OR transcripts including the co-receptor MsexORCo, 29 IR transcripts including the co-receptors MsexIR8a, MsexIR25a, MsexIR76b, and 49 GR transcripts (see section “Materials and methods” for details). Out of these 149 chemosensory receptors, transcripts for two ORs, seven IRs and 25 GRs were not detected in any of the investigated tissues, that is, counts of these transcripts were in the range of background counts (see section “Materials and methods” for details, Table 1 and Supplementary Table 3). On the other hand, three ORs, three IRs and two GRs were detected in each of the tissues tested.

We identified several genes with tissue-specific expression (Table 1). We also found sex-biased expression of several ORs and one GR (FDR corrected p < 0.05, nested ANOVA), whereas the expression of IRs seemed to be independent of sex in the investigated tissues (FDR corrected p > 0.05, nested ANOVA, Table 2). In contrast, the mating status of female or male moths had no statistically significant effect on the expression level of any chemosensory receptor in any tissue (FDR corrected p > 0.05, nested ANOVA).

In total, in the male and female antennae, we detected transcripts of 63 out of the 71 ORs (89%, Figure 1A). Twelve of these antennal ORs had a sex-biased gene expression (Table 2). Sixteen out of the 29 IRs (55%) and 13 out of the 49 GR transcripts (27%) were also expressed in the antennae (Figures 1B,C).

ORs. The OR co-receptor MsexORCo had high counts for the antennal samples (Figure 1A). A high expression of MsexORCo was expected as its co-expression in all OR expressing neurons is considered necessary for OR-mediated odor detection (Sato et al., 2008) and knocking out MsexORCo reduces or completely abolishes antennal, antennal lobe and behavioral responses toward plant odors (Fandino et al., 2019).

We could confirm previous findings that MsexOR1, MsexOR4, MsexOR51, and MsexOR83 had a male-biased expression in the antennae (Grosse-Wilde et al., 2010; Koenig et al., 2015). MsexOR1 detects the major female sex-pheromone bombykal (Wicher et al., 2017), which, together with MsexOR4 and MsexOR51, belongs to the expanded pheromone receptor clade (Koenig et al., 2015). Likewise, the fourth male-biased receptor MsexOR83 belongs to a newly discovered lineage of pheromone receptors (Bastin-Héline et al., 2019). Hence, all three orphan, male-biased receptors might be involved in detecting minor components of the female-produced pheromone blend, containing in total 12 components (Kaissling et al., 1989; Kalinova et al., 2001). MsexOR1, 51, and 83 had a male-specific antennal expression, whereas MsexOR4 transcripts were also detected in female antennae. Such expression of a pheromone receptor in females is seen in other moth species (Bengtsson et al., 2012; Walker et al., 2016), suggesting that these females can detect their own pheromone (Holdcraft et al., 2016). M. sexta female antenna is known to only respond to the minor pheromone component Z11-hexadecanal (Kalinova et al., 2001). However, in male M. sexta, MsexOR4-expressing olfactory sensory neurons are housed in the most abundant type of sensilla in the male antenna, the long trichoid sensilla (Grosse-Wilde et al., 2010), and neurons in these sensilla do not respond to Z-11-hexadecanal (Kalinova et al., 2001). Therefore, it seems unlikely that MsexOR4 is the receptor for this minor pheromone component. One OR from the pheromone receptor clade, MsexOR15, is female-specific, a rare feature for this subfamily (Bastin-Héline et al., 2019). Its absence of expression in male antennae negates the possibility of MsexOR15 being the Z-11-hexadecanal receptor, as males can detect this component with neurons housed in short trichoid sensilla (Kalinova et al., 2001). MsexOR38, which belongs to the same group as MsexOR83 (Koenig et al., 2015) has an unbiased expression in males and females, and might therefore be the putative receptor for Z-11-hexadecanal. These ORs of the pheromone receptor clade may also detect pheromones of sympatric hawkmoths, such as M. quinquemaculata and M. rustica, thus ensuring reproductive isolation (Daimon et al., 2012). However, in the codling moth, C. pomonella, an OR belonging to the pheromone clade detects a plant volatile (Bengtsson et al., 2014), opening the possibility that some of the uncharacterized pheromone clade members of M. sexta might have a non-pheromonal function.

Apart from MsexOR15, we found seven more female-biased ORs. These include those previously identified as female-biased (MsexOR5, 6, 15, and 87; Koenig et al., 2015), and four newly identified female-biased ORs (MsexOR17, 26, 33, and 40). MsexOR6, like MsexOR15, was not only female-biased but also female-specific for the antennae. MsexOR5 and 6 are encoded by two gene duplicates orthologous to the female-specific Bombyx mori linalool receptor BmorOR19 (Anderson et al., 2009; Koenig et al., 2015). Therefore, one of these ORs might be responsible for the enantioselective response to (+)-linalool of the lateral large female glomerulus (latLFG) in the female M. sexta antennal lobe (Reisenman et al., 2004) and for the female moth’s oviposition preference for (+)-linalool (Reisenman et al., 2010).

It is interesting to note that we have two times more female-biased than male-biased ORs. All male-biased ORs belong to pheromone receptor lineages; however, seven of the eight female-biased ORs are from different clades of potential plant volatile receptors (Guo et al., 2021). These ORs might be important in the context of the female-specific behavior of searching an appropriate oviposition site directed by host-plant volatiles. According to Bastin-Héline et al. (2019), another possible function of these polyphyletic, female-biased receptors might be the detection of male-produced pheromones, such as those presumably emitted from male hairpencils in M. sexta (Grant and Eaton, 1973; Birch et al., 1990).

The MsexORCo-associated glomeruli on the dorsal surface of the M. sexta antennal lobe (Fandino et al., 2019) respond to volatile headspace collections of native, nectar-rich flowers, and host and non-host plant foliage (Bisch-Knaden et al., 2022). This suggests that ORs might be responsible for the antennal detection of a large variety of floral and vegetative odors in M. sexta (Fraser et al., 2003; Spaethe et al., 2013). Ligands can be predicted based on homology to other de-orphanized Lepidopteran ORs for highly conserved OR clades. Such an approach has led to identification of the floral odor phenyl acetaldehyde as a ligand for MsexOR67 (Guo et al., 2021). The second identified plant volatile-detecting OR, MsexOR35, detects α-copaene, a volatile abundant in the headspace of the beetle-infested host-plant Datura wrightii that attracts ovipositing M. sexta (Zhang et al., 2022).

Overall, our results for the antennal expression of ORs largely agree with a previously performed RNA sequencing analysis, especially for the sex-biased ORs (Koenig et al., 2015), providing support for using the NanoString assay as a valid method for the study of insect chemosensory gene expression (Supplementary Figure 2).

IRs. Several IRs of the conserved antennal IR group (Croset et al., 2010; Yin et al., 2021) were expressed in the M. sexta antennae. These included the IR co-receptor genes MsexIR8a, 25a, 76b which had the highest counts among all IRs expressed in the antennae (Figure 1B). MsexIR8a and MsexIR25a are required for native responses of antennal coeloconic sensilla to carboxylic acids and amines, respectively (Zhang J. et al., 2019). For other antennal IRs, functions can be predicted based on D. melanogaster orthologs. MsexIR76b is ortholog to DmelIR76b, which is involved in both smell and taste of polyamines (Hussain et al., 2016) and also mediates gustatory detection of low salt concentrations and sour substances (Zhang et al., 2013; Chen and Amrein, 2017). Similarly, MsexIR31a, together with the co-receptor IR8a, is likely involved in acid detection, while MsexIR41a, when co-expressed with the co-receptor MsexIR25a or MsexIR76b, might be involved in detecting amines (Silbering et al., 2011; Hussain et al., 2016). IRs can also have non-chemosensory roles. Based on homology to Drosophila IRs, MsexIR21a, MsexIR93a and MsexIR25a could have cool-temperature-sensing functions (Knecht et al., 2016; Ni et al., 2016). Alternatively, MsexIR93a and MsexIR25a might be involved in hygrosensation (Knecht et al., 2016) by neurons present in the styliform sensilla of the antenna (Dahake et al., 2022).

We also found expression of nine Lepidoptera-specific IRs in the antennae (Koenig et al., 2015; Yin et al., 2021). Seven of these, MsexIR1.1, MsexIR1.2, MsexIR4, MsexIR75p.1, MsexIR75p.2, MsexIR75q.1, and MsexIR87a putatively encode IRs nested within the antennal IRs and are considered to have an olfactory function (Yin et al., 2021). All except MsexIR87a belong to a monophyletic group of putative acid-detectors (Hou et al., 2022). This group includes two acid-sensing IRs of Agrotis segetum, AsegIR75q.1, which is tuned specifically to octanoic acid and AsegIR75p.1, which has a broader tuning and detects acids, aldehydes and alcohols. Two Lepidoptera-specific IRs, MsexIR7d.3 and MsexIR100d, lie among the divergent IRs (Yin et al., 2021; Hou et al., 2022), which are defined as IRs that are not expressed in the antennae but in other tissues (Croset et al., 2010; Yin et al., 2021). However, the expression pattern of MsexIR7d.3 resembles that of its A. segetum ortholog AsegIR7d.3, which is highly expressed in the antennae (Hou et al., 2022). The function of receptors in the IR7d subfamily is unknown; however, the phylogenetically related DmelIR7a is expressed in neurons of the gustatory organs of larval and adult D. melanogaster (Croset et al., 2010), suggesting a gustatory function for these receptors.

Sex-dependent IR expression has been observed in other moth species. For instance, both female- and male-biased IRs have been identified in B. mori (Yin et al., 2021). In contrast, we did not find any significant effect of sex on IR expression in the antennae of M. sexta, although MsexIR8a-mediated detection of acids from larval feces is crucial for oviposition deterrence behavior, which prevents larval crowding (Zhang J. et al., 2019). This female-specific behavior seems to be independent of differential expression of the IRs investigated in our study.

GRs. Consistent with a previous report, we detected antennal expression of genes of two putative CO₂-receptors MsexGR2 and MsexGR3, sugar receptor MsexGR6 and bitter receptors MsexGR41 and MsexGR42 (Koenig et al., 2015). However, we could not confirm the antennal expression of the fructose receptor genes MsexGR9.1, 9.2 and 10.2. In addition, we found transcripts of seven more bitter receptors and another putative sugar receptor MsexGR8, whose ortholog BmorGR8 detects the sugar alcohol inositol, which is ubiquitously found in plants (Zhang et al., 2011).

MsexGR41 had the highest counts among all GRs expressed in the antennae (Figure 1C) and encodes a highly conserved and unduplicated putative bitter GR in the Lepidopteran order. The ortholog of MsexGR41 in the silkmoth (BmorGR63) is broadly expressed in different chemosensory tissues of adults and larvae (Guo et al., 2017). Similarly, MsexGR41 transcripts were detected not only in the adult and larval antennae but also in non-chemosensory tissues such as gut, fat bodies, brain and gonads (Koenig et al., 2015), suggesting additional roles for this receptor. Functional studies of bitter receptors in Lepidopteran species are sparse (reviewed in Xu, 2020). Moreover, predictions based on sequence similarity might be difficult since bitter receptors, like ORs, are highly divergent and often go through lineage-specific expansions within Lepidoptera (Koenig et al., 2015).

Although putative gustatory sensilla chaetica are present on the antennae of M. sexta (Lee and Strausfeld, 1990), the functional roles of antennal gustatory neurons are unknown. In the moths Heliothis virescens and Agrotis ipsilon, sugar and bitter detecting gustatory neurons in the antennae influence the proboscis extension behavior (Jørgensen et al., 2007; Hostachy et al., 2019). Similarly, antennal GRs might also influence proboscis extension in M. sexta, a behavior which is dependent on both olfactory and visual cues (Raguso and Willis, 2002) and MsexORCo function (Fandino et al., 2019).

In Drosophila, co-expression of the two CO₂ receptors, DmelGR21a and DmelGR63a, in the fly antennae is required for CO₂ detection (Jones et al., 2007). Similarly, expression of two putative CO₂ receptor genes, MsexGR2 and MsexGR3, suggests that the M. sexta antennae could also detect CO₂; a function that is not reported so far. In moths, however, CO₂ detection is generally assigned to labial palps. In fact, the CO₂ sensitive glomerulus in the antennal lobe in M. sexta is targeted by sensory neurons in the labial pit organ of the labial palp (Guerenstein et al., 2004a). In contrast to Drosophila, Lepidopteran species have three GRs (GR1, GR2 and GR3) in the CO₂ receptor subfamily (Robertson and Kent, 2009). Functional analysis in Helicoverpa armigera revealed that co-expression of HarmGR1 and HarmGR3 is necessary for CO₂ reception (Ning et al., 2016). Assuming that the functions of these GRs are conserved, the absence of expression of MsexGR1, the ortholog of HarmGR1, might actually indicate the inability of M. sexta antennae to detect CO₂. MsexGR2 and MsexGR3 might be involved in olfaction as GRs of the CO₂ receptor family were shown to be activated or inhibited by various odorants in D. melanogaster and Aedes aegypti (MacWilliam et al., 2018; Kumar et al., 2020).

In contrast to the antennae, transcripts of only 17 ORs each were detected in labial palps and proboscis (Figure 2A). The labial palps in addition expressed six IRs and seven GRs, whereas in the proboscis 11 IRs and nine GRs were expressed (Figures 2B,C).

ORs. Of the 17 ORs expressed in each mouthpart, only seven ORs were common between both tissues, indicating that labial palps and proboscis have different chemosensory functions (Figure 2A). MsexORCo expression was consistently detected in male and female labial palp samples at high levels, whereas in the proboscis, it seemed absent. In M. sexta, like in several other moth species, the labial palps are so far known only to detect CO₂ (Guerenstein et al., 2004a; Ning et al., 2016). The expression of several ORs, like the potential linalool-receptors MsexOR5 and MsexOR6 (Grosse-Wilde et al., 2010), and more importantly, MsexORCo, strongly indicates that the labial palps of M. sexta might have an olfactory function beyond the detection of CO₂. We found expression of 13 ORs in labial palps of mated females compared to only five ORs in virgins. Although this difference was not significant according to the nested ANOVA we applied, it might reflect a slightly increased olfactory sensitivity of the palps to foraging- or oviposition-associated volatiles in females following mating. For the proboscis, expression of MsexORCo in male M. sexta has been demonstrated previously using RT-PCR and immunohistochemistry methods (Haverkamp et al., 2016). In our study, however, we detected MsexORCo transcripts only in one out of three samples each of virgin and mated males, thus not fulfilling our criterion for gene expression (see section “Materials and methods” for details). MsexORCo is expressed only in a single neuronal cell in a single sensillum at the tip of the proboscis (Haverkamp et al., 2016). As we extracted RNA from the entire length of the proboscis and not only from the tip, such cell-specific expression might have gone undetected. This sensillum responds strongly to benzyl acetone (the major component of N. attenuata flowers) but in addition detects other typical floral odors such as cis-jasmone, linalool, geraniol, nerol and benzyl alcohol (Haverkamp et al., 2016; Bisch-Knaden et al., 2022). MsexOR84/89, MsexOR64 and MsexOR77 were highly expressed in the proboscis and might be candidate receptors involved in the detection of these floral odors. Another intriguing feature was expression of one of the male-specific antennal pheromone receptors, MsexOR51, in the proboscis of female virgin M. sexta. This suggests that females might have multiple mechanisms of detecting their own pheromone, and that MsexOR51 might not really be “male-specific.”

IRs. All IRs expressed in the labial palps were also detected in the proboscis (Figure 2B). In both mouthparts, the IRs with the highest counts were those encoding co-receptors, MsexIR25a and MsexIR76b, and the Lepidoptera-specific IR MsexIR7d.3, which were expressed in each of the tested tissues (Table 1). The IR-co-receptor MsexIR8a, on the other hand, was not detected in the mouthparts. This is contrary to a previous report where both MsexIR8a and MsexIR25a were shown to be expressed in the proboscis by using RT-PCR (Haverkamp et al., 2016). We detected MsexIR8a transcripts in only one sample of a virgin female, indicating that MsexIR8a is expressed in very low levels in the proboscis, and PCR amplification would be a better technique for detecting this expression. The proboscis expression of genes of three IRs from the “expanded acid-sensing cluster” (MsexIR1.2, MsexIR4 and MsexIR75q.1) suggests that the proboscis could detect volatile carboxylic acids (Hou et al., 2022); however, based on the functional characterization of IRs from this cluster in the turnip moth A. segetum, such acid-sensing would require expression of the co-receptor IR8a.

Several IRs that were not detected in the antennae, were expressed in the mouthparts. These are MsexIR7d.2, a duplicate of MsexIR7d.3 expressed in the antenna, and MsexIR68a, whose Drosophila ortholog is involved in hygrosensation (Knecht et al., 2017). Additionally, the proboscis expressed three more Lepidoptera-specific IRs (MsexIR100a, MsexIR100f, and MsexIR143; Liu et al., 2018) with unknown functions.

GRs. Four GRs were commonly expressed between labial palps and proboscis (Figure 2C). High counts of the ubiquitously expressed MsexGR41 (Table 1) were found in both mouthparts. A distinct and high expression of all three putative CO₂ receptor genes, MsexGR1, 2 and 3, was observed only in the labial palps, the organ known for CO₂ detection in M. sexta (Guerenstein et al., 2004a). In the noctuid moth H. armigera, a similar high expression of the orthologous GRs, HarmGR1, 2, and 3, was reported for the labial palps (Guo et al., 2018). In the proboscis, like in the antennae, transcripts of MsexGR2 and MsexGR3, but not of MsexGR1 were detected. Therefore, CO₂-detection in M. sexta seems to be restricted to the labial palps.

All other GRs expressed in the mouthparts belong to the clade of bitter receptors (Koenig et al., 2015) and, surprisingly, no sugar or fructose receptors were detected. However, as the proboscis is considered the major gustatory organ, we expected to find expression of sugar or fructose receptors. In H. armigera, for example, sugar receptors are expressed in the proboscis (Guo et al., 2018). Reiter et al. (2015) showed that gustatory neurons in the proboscis of M. sexta can discriminate various tastants including sucrose and other sugars, strongly suggesting the presence of sugar receptors in the proboscis. However, gustatory neurons are mainly found in the distal one-third of the proboscis (Reiter et al., 2015) and the expression level of these sugar receptors might be very low.

We detected 19, 18, and 9 OR transcripts in fore-, mid- and hindlegs respectively (Figure 3). In addition, 11, 10, and 7 IRs, and 9, 10, and 5 GRs were expressed in fore-, mid- and hindlegs, respectively.

ORs. Nine ORs were common among all three leg pairs (Figure 3A). The ORs with the highest counts in all leg pairs were MsexOR6 and the ubiquitous MsexOR84/89 (Table 1). The putative linalool receptor gene MsexOR6, which was highly expressed and female-specific in the antennae, was also highly expressed and female-specific in all three leg pairs. Transcripts of its paralog MsexOR5, however, was not detected in the legs. Based on our criteria (see section “Materials and methods”), the obligatory co-receptor MsexORCo was only expressed in forelegs of mated females. MsexORCo was, however, detected in one foreleg sample each of virgin females and mated males. We also note that there are 15 ORs expressed in the mated female versus only eight in virgin females (see discussion of a similar trend in female labial palps). It could be possible that expression of ORs including MsexORCo is upregulated in female forelegs after mating and that the forelegs might be involved in the assessment of oviposition substrates via olfaction. Transcripts of putative pheromone receptor gene MsexOR83, which is male-specific in the antennae, were expressed in the virgin male midlegs. This suggests that in addition to the antennae, legs may also detect pheromones.

Our results indicate a potential olfactory function of the legs of M. sexta. A class of chemosensory sensilla found on the M. sexta legs, the scattered blunt sensilla, has been hypothesized to have olfactory function due to the presence of wall pores (Kent and Griffin, 1990). Both male and female legs possess an equal number of these sensilla; however, their olfactory function still needs to be tested.

IRs. Six IRs were common among all leg pairs (Figure 3B). The transcripts of IR co-receptor genes MsexIR25a, MsexIR76b were detected in all leg pairs, whereas expression of acid-sensing co-receptor MsexIR8a seemed to be absent in leg tissues. However, we detected expression of IRs that potentially code for IRs tuned to acids based on their homolog’s functions: MsexIR4, MsexIR75p.2, and Msex75q.1 (Ai et al., 2010; Rytz et al., 2013; Hou et al., 2022). This suggests that like in the mouthparts, MsexIR8a expression might be very low in the legs. Alternatively, these receptors might have a different function in the legs of M. sexta. In fact, some A. segetum homologs AsegIR75p.2 and AsegIR75q.2 did not respond to any acid, aldehyde or alcohol stimulation when co-expressed with AsegIR8a in Xenopus oocytes (Hou et al., 2022), either indicating that their volatile ligands were not tested, or that they have a non-olfactory function. Like in the labial palps and proboscis, the Lepidoptera-specific MsexIR7d.2 and the putative gene for humidity sensor MsexIR68a are expressed in all leg pairs (Knecht et al., 2017). MsexIR93a, which could have both hygro- and thermosensory roles, was detected only in the forelegs (Knecht et al., 2016).

GRs. Among the three leg pairs, four GRs were commonly expressed (Figure 3C). Once again, MsexGR41 had the highest counts in each of them. Due to the crucial gustatory function of a moth’s legs, especially for oviposition, we expected high expression of several GRs. However, we detected transcripts of fewer GRs in the legs than in the antennae. A similar, unexpectedly low GR expression was also seen in the legs of S. littoralis (Koutroumpa et al., 2021). Interestingly, we found expression of more GRs in the legs of mated females, than in virgin females or in males. Expression of a bitter receptor transcript, MsexGR15C, was detected only in the fore- and midlegs of mated females, and in the midlegs we found a statistically significant, female-biased expression of this GR. This is interesting as there is evidence of sexual dimorphism in the chemosensory sensilla on the legs of M. sexta. A class of putative contact chemosensilla, the “spine-associated clustered sensilla”, is more abundant on the females’ fore-, mid- and hindlegs than on the males’ legs (Kent and Griffin, 1990). The female might have subclasses of these sensilla that express female specific receptors. Transcripts of sugar receptors MsexGR5 and MsexGR6 (Koenig et al., 2015) were detected in the fore- and midlegs of both sexes but not in the hindleg samples.

As in the antennae and proboscis, GRs from the CO₂ receptor group, MsexGR2 and MsexGR3, were expressed. While MsexGR2 had a consistent and high expression in males and females, MsexGR3 was only detected in mated female mid- and hindlegs. As these receptors are known for detecting odorants rather than tastants, they could be associated with the scattered blunt sensilla on the legs.

Overall, fore- and midlegs expressed about two times more GRs than the hindlegs. After arriving at a potential hostplant via multisensory cues such as olfaction, vision, CO₂, humidity, and temperature, a gravid female M. sexta touches the leaves with its fore- and midlegs while still hovering in front of the plant. Such direct contact is important for the stimulation of egg laying behavior in hawkmoths (Yamamoto et al., 1969; Mechaber et al., 2002) as well as in butterflies (Staedler et al., 1995), and allows the assessment of primary and secondary plant metabolites to decide whether the plant is suitable for oviposition. Both the expressed sugar and bitter receptors might be involved in this behavior. The spines on the moth’s legs may help them to pierce through the leaf surface to allow these metabolites to come in contact with the gustatory sensilla, such as the spine-associated clustered sensilla (Kent and Griffin, 1990). In another Lepidopteran species, Cydia pomonella, primary metabolites, like plant surface sugars can influence egg laying decisions (Lombarkia and Derridj, 2008).

We detected expression of 16 ORs, eight IRs and five GRs in forewings and 16 ORs, nine IRs and four GRs in hindwings (Figure 4).

ORs. Of the 16 expressed ORs, 14 were common in fore- and hindwings (Figure 4A). In both wing pairs, MsexOR84/89 transcripts had the highest counts. However, MsexORCo transcripts were not detected in any of the samples. Similar expression of ORs but not ORCo was reported for the wings of the geometrid moth Ectropis obliqua (Ma et al., 2016). Such expression of ORs without that of ORCo in the wings, suggests that ORs may have more functions than the detection of odor molecules. Alternatively, some ORs might function without ORCo as in the case of basal insect species that lack the ORCo gene (del Mármol et al., 2021). MsexOR83, whose transcripts were specifically detected in antennae and midlegs of males, was in addition expressed in wings of both males and females. Therefore, moth wings might detect female pheromones. For females, this information might be useful to regulate pheromone production or release. In moths, pheromone perception with wings has not been reported to the best of our knowledge. However, an IR in gustatory sensilla on the wings of Drosophila is required for normal mating behavior in both males and females and is thought to detect pheromonal cuticular hydrocarbons (He et al., 2019).

IRs. Except for one additional IR (MsexIR100f) in the hindwings, we detected expression of the same set of IR transcripts in both wings (Figure 4B). MsexIR7d.3 transcripts had the highest counts in the wings like in other tissues. In both pairs of wings, two IR co-receptor genes, MsexIR25a and MsexIR76b, were expressed, but the third one MsexIR8a was not. The expression of MsexIR25a and MsexIR76b might indicate olfactory sensing of amines, or in case of MsexIR76b, gustatory detection of salts, sour acids and amino acids (Zhang et al., 2013; Hussain et al., 2016; Knecht et al., 2016, 2017; Chen and Amrein, 2017). Of special interest is the expression of MsexIR21a in fore- and hindwings of both sexes. The only other tissue where this gene was expressed was the antennae. MsexIR21a is potentially involved in thermosensation (cool sensing) as reported for its ortholog DmelIR21a (Knecht et al., 2016; Ni et al., 2016).

GRs. Three GRs that were detected in almost all tissues described so far, MsexGR2 and MsexGR3 of the CO₂ subfamily, and MsexGR41, were expressed in both wing pairs, with MsexGR41 having the highest counts (Figure 4C). Three more bitter receptor transcripts were detected. No putative sugar or fructose receptors (Koenig et al., 2015) were found in the wings. So far, no chemosensory sensilla have been reported on the wings of M. sexta. GR expression was also found in the wings of E. obliqua including a putative sugar receptor gene (Ma et al., 2016). In Drosophila, functional gustatory sensilla on the wings are important for sugar and bitter detection eliciting food exploration behavior and grooming (Raad et al., 2016; Yanagawa et al., 2019).

We detected 30 OR, 15 IR and 14 GR transcripts in the ovipositor (Figure 5). This is a higher number of receptor genes than was found to be expressed previously (Klinner et al., 2016). This difference might be due to different sample collection methods applied in both studies. As opposed to the previous study, where the last four abdominal segments were taken, our sample consisted only of the anal papillae at the last abdominal segment that bears the sensilla. Hence, it is possible that we were able to detect more genes expressed in chemosensory neurons housed in these sensilla because potential receptor transcripts were more concentrated in our samples.

ORs. The highest counts were found for MsexOR84/89 transcripts, followed by MsexOR35 and MsexOR75 (Figure 5A). Our results furthermore, included the three ORs that were previously reported to be expressed in the ovipositor using RNA-sequencing, MsexOR9, MsexOR26, and MsexORCo (Klinner et al., 2016). The expression of MsexORCo in our NanoString assay was detected only in the virgin female ovipositor as it was the case for the RNA sequencing data in our previous study (Klinner et al., 2016). However, MsexORCo expression was confirmed using RT-PCR and western blot analysis for both mating states (Klinner et al., 2016). Hence, MsexORCo might be expressed in both mating states, but the expression level is most probably lower in mated than in virgin females. This indicates that OR-mediated olfactory function of the ovipositor might be involved in courtship rather than in oviposition behavior. MsexOR17, MsexOR26, and MsexOR87 that were female-biased on the antennae, were expressed in the ovipositor as well and might be involved in detection of mating or oviposition related cues.

Our result of low expression of ORCo and higher expression of several ORs is similar to reports from other moth species. Transcriptome sequencing together of pheromone gland and ovipositor of Spodoptera frugiperda revealed the expression of SfruORCo and 11 SfruORs, of which the transcript level of SfruORCo was the lowest (Sun et al., 2022). Furthermore, in the ovipositor of Helicoverpa assulta, the expression level of a tuning OR gene, HassOR31, is higher than that of HassORCo (Li et al., 2020). HassOR31 is co-expressed with HassORCo in some, but not all sensilla on the ovipositor. Heterologous expression in Xenopus oocyte show that HassORCo is necessary for olfactory detection of host-plant volatiles by HassOR31, therefore cells where HassORCo is not co-expressed with HassOR31 might use this OR in a non-olfactory function. One such example is CpomOR1, the pheromone receptor of C. pomonella, whose expression in the ovipositor has been suggested to play a role in the process of egg production and maturation (Garczynski et al., 2017). In H. virescens, HvirOR13, the receptor for its major pheromone component (Z)-11-hexadecenal was detected in the ovipositor, along with the pheromone binding protein PBP2, which is needed for the function of the pheromone receptor. Pheromone receptor and binding protein were proposed to be involved in a potential feedback mechanism (Widmayer et al., 2009). As the putative pheromone receptor gene MsexOR83 was expressed in the ovipositor of M. sexta, it might accordingly be involved in the regulation of pheromone emission or in other non-chemosensory cellular functions.

IRs. Similar to the proboscis, legs and wings, among the expressed IRs, MsexIR7d.3 transcripts again had the highest counts (Figure 5B). Expression of all known IR co-receptors (MsexIR8a, 25a, and 76b) were detected in the ovipositor, and also of most previously reported IRs (MsexIR4, 7d, 68a, and 75d; Klinner et al., 2016). In fact, MsexIR75d expression was found exclusively in the ovipositor in our study. Based on homology with Drosophila, we predict that MsexIR75d might be involved in the detection of pyrrolidine or other amines (Silbering et al., 2011; Rytz et al., 2013), in line with the strong activation of sensilla on the ovipositor upon stimulation with pyrrolidine (Klinner et al., 2016). MsexIR8a is involved in acid detection (Zhang J. et al., 2019) and sensilla at the ovipositor of M. sexta also detect acids, especially hexanoic acid (Klinner et al., 2016). Acids play a key role in oviposition deterrence of M. sexta (Zhang J. et al., 2019). As the ovipositor is the only tissue apart from the antennae where we find MsexIR8a expression, the role of the ovipositor in acid-induced oviposition deterrence would be worth investigating. We also found expression of other IR8a-associated putatively acid-sensing IR genes: MsexIR1.1, MsexIR4, and MsexIR75q.1 (Hou et al., 2022). Non-chemosensory IRs such as MsexIR68a and MsexIR93a were also expressed in the ovipositor. We found a very clear expression of a Lepidoptera-specific IR, MsexIR143, in the ovipositor. The only other tissue where this IR was expressed is the male proboscis.

GRs. MsexGR3 and MsexGR41 transcripts had the highest counts among all GRs in the ovipositor (Figure 5C). In agreement with a previous report, the putative genes for CO₂ receptors MsexGR2 and MsexGR3 were expressed in the ovipositor (Klinner et al., 2016). The gene for sugar receptor MsexGR6 and several bitter receptors (Koenig et al., 2015) were expressed in addition, and might be involved in detecting sugars and other plant metabolites that can be used for evaluating leaf quality for oviposition.

Of the chemosensory tissues we examined, the ovipositor expressed the highest number of GRs. At the anal papillae of the M. sexta ovipositor, there are blunt tipped sensilla, which have both a terminal pore and wall pores, indicating olfactory and gustatory functions of the neurons residing in them (Klinner et al., 2016). Single sensillum recordings (Klinner et al., 2016) have demonstrated only an olfactory function so far. In the moth S. littoralis, however, gustatory responses of ovipositor sensilla to salts, sugar and bitter compounds have been shown (Seada et al., 2016). It is possible, that gravid female moths, after landing on a potential host plant, use the spines on their legs to scratch the leaf surface and expose essential sugars and bitter secondary plant metabolites. Sensing these substances with sensilla on both the leg and the ovipositor might help the females to make an oviposition choice. As more GRs are present in the ovipositor than in the legs (Figure 6), gustation by the ovipositor might have the most important contribution to this decision-making in M. sexta.

In our results, the antennae and the ovipositor of M. sexta expressed every co-receptor known so far: MsexORCo, MsexIR8a, MsexIR25a, and MsexIR76b (Table 3), indicating that all pathways of odor detection take place in these organs. The labial palps and the forelegs express all co-receptors except MsexIR8a. The striking expression of MsexORCo in the labial palps suggests that these might be the second most important olfactory organs in Manduca, analogous to the maxillary palps in Drosophila. In the noctuid moth S. frugiperda labial palps detect other volatiles in addition to CO₂ (Chen et al., 2021), although the response profile indicates IR-mediated olfaction. Transcripts of two IR co-receptor (MsexIR25a and MsexIR76b) were detected in all nine tissues, similar to the expression patterns observed in other insects and even mollusks. These receptors are expressed in neurons of gustatory organs all over the body (Croset et al., 2010; van Schooten et al., 2016; Liu et al., 2018; Zhu et al., 2018). However, the expression of the MsexIR25a and MsexIR76b in peripheral tissues may not only indicate chemosensory activity as these genes could be involved in hygro- and thermosensation.

Apart from these two IR-coreceptors, 13 more receptors were expressed ubiquitously in all or almost all tissues (Table 1). The expression of three of these receptors (MsexOR64, MsexOR75, and MsexOR77) would require further validation due to possible cross-hybridization of the probes with duplicate receptor transcripts. For the remaining 12 receptors our probes did not have any off-targets. MsexOR84/89 was broadly expressed even in tissues where no ORCo was detected, suggesting a possible non-olfactory function of this gene. The ortholog of this gene in the silkmoth, BmorOR23, is expressed only in the larvae, but its function remains unknown (Tanaka et al., 2009; Koenig et al., 2015). However, expression of some ORs in several tissues including non-chemosensory ones was also detected in other moth species like Spodoptera littoralis (Walker et al., 2019). MsexIR7d.2 and MsexIR7d.3 belong to the Lepidoptera-specific IR7d group and homologs of their genes in other moth species have a broad expression in several tissues in adult and larval stages (Liu et al., 2018; Zhu et al., 2018). However, their paralog MsexIR7d.4 was not detected in any of the tissues investigated here. MsexGR41, similar to its ortholog BmorGR63, is expressed in different chemosensory tissues (Guo et al., 2017).

In contrast to the broad expression of several genes, expression of 25 GRs were not detected in any of the tissues under investigation. We hypothesize that the low expression levels of GRs underlay the underrepresentation of this gene family in our expression data which has also hampered the detection of these genes in transcriptomes and RNA in situ hybridizations (Clyne et al., 2000; Koutroumpa et al., 2021). Another aspect is the spatially restricted distribution of cells expressing different receptors, for example, the segregation of MsexORCo and MsexIR8a expression in the proboscis (Haverkamp et al., 2016). Extracting RNA from smaller segments of each organ and increasing the amount of input RNA could also aid in detection of lowly expressed genes. Alternatively, these genes could be expressed in internal organs such as the brain or gut, where they might be involved in cell signaling acting as transmembrane receptors for various molecules. In addition, many of these not expressed GRs in adults could be larvae-specific, as larval feeding behaviors are strongly influenced by gustation (Glendinning et al., 2000; Zhang Z. J. et al., 2019).

Among all the tissues investigated, we found the highest number of expressed chemosensory receptors in the antennae (92), followed by the ovipositor (59, Figure 6). The hindlegs, in contrast, showed the lowest number of chemosensory receptor genes, particularly the least number of ORs compared to all other tissues, indicating that hindlegs might have a minor role in chemoreception, a phenomenon also seen in mosquitoes (Yang et al., 2020). In female Ae. aegypti, electrophysiological responses from different tissues (antennae, mouthparts, wings and tarsal segments of legs) to seven insect repellents and attractants showed that all these tissues can detect odors with different capacities. The hindlegs were the least chemoreceptive as they responded only to one odor, triethylamine. This amine response is likely due to an IR25a or IR76b pathway, as these receptors are known to be broadly expressed and amine-detecting in many insects.

Furthermore, the antennae had the highest number of exclusively expressed genes, with 24 ORs, 4 IRs and 3 GRs (Table 1). One antennal GR is a putative inositol receptor gene, MsexGR8 (Zhang et al., 2011). In M. sexta caterpillars inositol, an alcohol sugar that occurs in host plants (Nelson and Bernays, 1998), promotes feeding and helps overcome the inhibitory effects of bitter compounds (Glendinning et al., 2000), whereas, in adults, this GR may help identify host plants. Remarkably, MsexGR1 showed specific expression in the labial palps, which differs from the broad expression patterns of the other two candidate CO₂ receptor genes MsexGR2 and MsexGR3. If MsexGR1 is necessary for CO₂ perception, then no tissue other than the labial palps might be able to detect it.

We found sex-dependent expression of chemosensory receptor genes only in the antennae (12 ORs) and legs (1 OR and 1 GR), the two organs for which sexual dimorphism has been described. In M. sexta, similar to other moth species, the antenna is sexually dimorphic with respect to its shape as well as the size, number and type of its sensilla (Sanes and Hildebrand, 1976; Keil, 1989; Shields and Hildebrand, 1999). For all three pairs of legs, females have more spine-associated sensilla than males (Kent and Griffin, 1990). No sexual dimorphism is observed in the palps (Kent et al., 1986), and none has been reported for proboscis or wings.

However, sex-biased receptors in one tissue may have a sex-independent expression in other tissues. Examples are the expression of the “male-specific” putative pheromone receptor gene MsexOR51 in the proboscis of virgin females and the expression of the “female-biased” MsexOR17 in the legs of males. A receptor could thus have multiple functions, being involved in certain sex-specific roles when expressed in one tissue and a role common for both sexes when expressed in another tissue.

Few studies have shown plasticity in insect chemosensory receptor gene expression dependent on mating state. In female D. suzukii, for example, several ORs are upregulated after mating (Crava et al., 2019). In the moth Dendrolimus punctatus, both up- and down-regulation have been reported for ORs and other chemosensory genes in both males and females (Zhang et al., 2017). In H. armigera, two antennal IRs are downregulated after mating in females (Liu et al., 2018). Such differential expression of chemosensory receptors can modulate sensitivity to essential cues and lead to changes in behavior and physiology.

For M. sexta, post-mating changes of odor-evoked behavioral and physiological responses have been described. Host plant volatiles induce more often a directed upwind flight and abdomen curling in mated than in virgin female M. sexta (Mechaber et al., 2002). In addition, some plant odor-evoked activation patterns in the antennal lobe of female M. sexta change after mating (Bisch-Knaden et al., 2022). However, we did not find significant differences between virgin and mated moths in the expression levels of ORs in the antennae or other tissues, suggesting that other mechanisms are responsible for the observed state-dependent changes in physiology and behavior.