GC/MS analyses of candidate type II sex pheromones in adult H. armigera

Using gas chromatography (GC) and GC-mass spectrometry (GC/MS), we identified sex pheromone components in the crude extracts from pheromone glands (PG) and abdomen (Ab) of adult H. armigera. The type I sex pheromones, including Z11-16:Ald, Z9-16:Ald, and Z11-16:OH, were identified in the female pheromone gland, as previously investigated (Fig. 1A) [34]. Additionally, two candidate type II sex pheromones, 3Z,6Z,9Z-21:H (Retention time 17.3 min) and 3Z,6Z,9Z-23:H (Retention time 20.5 min), were conclusively identified in H. armigera for the first time, based on their retention time and MS spectra with synthetic standards (Fig. 1A-C). Although there was a slight difference in intensity of fragment ions (relative abundance), the major electron ionization ions of 3Z,6Z,9Z-21:H and 3Z,6Z,9Z-23:H were identical to those of synthetic 3Z,6Z,9Z-21:H and 3Z,6Z,9Z-23:H, respectively (Fig. 1B, C). However, 3Z,6Z,9Z-21:H was exclusively found in the male pheromone gland and abdomen, while 3Z,6Z,9Z-23:H was primarily detected in the abdomen of adult females (Fig. 1A).

To accurately quantify the content of these two chemicals, we established a linear regression of each component’s content (x) ranging from 1 ng to 100 ng, plotted against the integrated area (y) in GC/MS (Fig. S1). Consequently, the content of 3Z,6Z,9Z-21:H is significantly higher in the abdomen (125.46 ± 26.28 ng) than in the pheromone gland (16.54 ± 3.35 ng) of adult males (Student’s t-test, P = 0.008, Fig. 1D). In contrast, 3Z,6Z,9Z-23:H is much more abundant in the abdomen (268.96 ± 36.66 ng) than in the pheromone gland (10.77 ± 3.74 ng) of adult females (Student’s t-test, P < 0.001, Fig. 1E).

To further investigate the detection of the two candidate type II sex pheromones by male and female moths, we conducted the electroantennogram (EAG) experiments. The results revealed that both 3Z,6Z,9Z-21:H and 3Z,6Z,9Z-23:H elicited strong EAG responses in both male and female moths. As the concentrations of these two components increased from 100 µg to 1 mg, the EAG responses exhibited a significant dose-dependent increase (Fig. 1F). However, electrophysiology responses did not show a significant difference between sexes in response to the same concentration of each chemical, although adults displayed stronger responses to 1 mg of 3Z,6Z,9Z-21:H (Fig. 1F). In contrast, 100 µg of Z11-16:Ald, Z9-16:Ald, and Z9-14:Ald elicited stronger EAG responses in males compared to females, as expected (Fig. 1F).

Fig. 1

Gas chromatograph/mass spectrometry analyses of extracts in adult Helicoverpa armigera. A Left, total ion chromatogram (TIC) of female pheromone gland (PG, teal), female abdomen (Ab, rust), male pheromone gland (PG, purple), male abdomen (Ab, navy blue), and two standards (red). Right, structure and formula of five sex pheromones identified from pheromone glands and abdomens of adult H. armigera. B Mass spectra of synthetic 3Z,6Z,9Z-21:H (up) and the corresponding compound from male abdomen extracts (down). C Mass spectra of synthetic 3Z,6Z,9Z-23:H (up) and the corresponding compound from female abdomen extracts (down). D Average amount of Z3,Z6,Z9–21:H in individual pheromone gland and abdomen of male adults. Data presented are mean ± SEM (n = 5, 7 biological replicates), and statistical analysis was conducted by a two-tailed unpaired Student’s t-test. E Average amount of Z3,Z6,Z9–23:H in individual pheromone gland and abdomen of female adults. Data presented are mean ± SEM (n = 6, 9 biological replicates), and statistical analysis was conducted by a two-tailed unpaired Student’s t-test. F Relative EAG responses to sex pheromones in males (green) and females (purple). Type I pheromones were used in a concentration of 10 µg µL− 1, and type II pheromones were used in concentrations of 10 µg µL− 1 and 100 µg µL− 1. Data are shown as mean ± SEM (n = 20 biological replicates), and were analyzed by one-way ANOVA followed by two-side Tukey ’s post hoc test for multiple comparisons with SAS 9.2 for Windows. Different letters indicate significant differences (P < 0.05)

Screening and characterization of the receptor for candidate type II sex pheromone in H. armigera

Until now, only one receptor, ObruOR1, in the winter moth, O. brumata, and its orthologue, AsegOR3, in A. segetum, have been identified as receptors for polyunsaturated polyenes without epoxy or other functional groups [27]. To determine the receptors responsible for the recognition of candidate type II sex pheromones in H. armigera, we constructed a phylogenetic tree to identify potential candidates. The phylogenetic tree revealed that ObruOR1 and AsegOR3 clustered within the traditional pheromone receptor clade (highlighted in light pink) (Fig. 2A). Furthermore, HarmOR11 was identified as the orthologous gene of ObruOR1 and AsegOR3 (Fig. 2A), making it the most promising candidate receptor for type II sex pheromones. Sequence alignment results indicated that HarmOR11 shares 55.71% and 79.22% of amino acid identity with ObruOR1 and AsegOR3, respectively (Fig. 2B).

Subsequently, we conducted functional characterization of HarmOR11 in the Xenopus oocyte heterologous expression system, coupled with the TEVC technique. The results revealed that oocytes co-expressing HarmOR11/Orco exhibited robust TEVC responses to 3Z,6Z,9Z-21:H at a concentration of 10− 4 mol, while they did not respond or exhibited only very weak responses to the other three polyenes (Fig. 3A, B). Water-injected oocytes showed no responses to any of the tested chemicals (Fig. 3A). Importantly, the responses of HarmOR11/Orco-injected oocytes to 3Z,6Z,9Z-21:H were concentration-dependent, with a threshold lower than 10− 7 mol, and an EC50 value of 1.637 × 10− 5 mol (Fig. 3C, D).

Fig. 2

Phylogenetic analysis of odorant receptors in Lepidopteran moths and sequence alignment of HarmOR11 and its orthologous genes. A The phylogenetic tree was constructed using RAxML version 8 with the Jones–Taylor–Thornton (JTT) amino acid substitution model [44]. Node support was assessed using a bootstrap method based on 1000 replicates. The phylogenetic tree was constructed using a total of 183 OR sequences from five Lepidoptera species. Dendrograms were created and colour labeled with FigTree v1.4 software. Harm, Helicoverpa armigera (dark teal); Bmor, Bombyx mori (blue); Slit, Spodoptera littoralis (green); Obru, Operophtera brumata (red); Aseg, Agrotis segetum (teal). The conserved Orco clade was marked in light blue, and the pheromone receptor clade and OR11 subclade were marked in light pink and coral, respectively. B Identities and sequence alignment of amino acid sequences of HarmOR11, SlitOR11, AsegOR3, and ObruOR1. The α-helix structures of these proteins are highlighted with red dashed boxes. The amino acids involved in ligand binding in HarmOR11 are marked according to the conservation among OR11 lineage

Fig. 3

Responses of Xenopus oocytes co-expressing HarmOR11/Orco to type II pheromones. A Inward current responses of HarmOR11/Orco co-expressing Xenopus oocytes exposed to four type II pheromones with a concentration of 10− 4 mol. The water-injected oocytes were used as a control. B Response profiles of HarmOR11/Orco to four type II pheromones with a concentration of 10− 4 mol. Data are shown as mean ± SEM (n = 16 biological replicates). C HarmOR11/Orco co-expressing Xenopus oocytes stimulated with various concentrations of 3Z,6Z,9Z-21:H. D Dose response curves of HarmOR11/Orco to 3Z,6Z,9Z-21:H. Error bars indicate the SEM (n = 7 biological replicates). The EC50 value was 1.637 × 10− 5 mol

HarmOR11 mutants severely reduced the EAG responses to the candidate type II sex pheromones

To confirm whether HarmOR11 is the receptor responsible for detecting the candidate type II sex pheromones in vivo, we employed the CRISPR/Cas9 technique to knock out the HarmOR11 gene. We designed a single-guide RNA (sgRNA) targeting exon 3 with a conserved sequence of 5’-TCTATCTCAGAAATATAGAGAGG-3’ to target the gene (Fig. 4A). The nucleotide sequence of the sgRNA was highly specific, particularly the 12 bp sequence (seed region) near the PAM site, which could not be fully matched to other genomic regions, indicating a very low risk of off-target effects. PCR amplification of the fragment carrying the target site from the genomic DNA (gDNA) of each injected G0 moth revealed several types of somatic mutations at the targeted loci of the HarmOR11 gene (Fig. 4A). Ultimately, we retained the fourth mutation strain due to its largest population. In this strain, a 13-bp short DNA fragment was inserted into the genome before the PAM sequence (Fig. 4A-C). This mutation resulted in a frameshift at codon 108 and introduced a premature stop codon, leading to the production of a truncated protein consisting of 110 amino acids (Fig. 4B).

Given that the both candidate type II sex pheromones elicited strong electrophysiological responses in male and female moths (Fig. 1F), we aimed to confirm whether the HarmOR11 mutant affected the responses to these pheromones by conducting EAG assays. The results revealed that the EAG responses to 100 µg and 1 mg of 3Z,6Z,9Z-21:H were significantly reduced in male mutants (Fig. 4D). There was no significant difference in the responses to 100 µg of 3Z,6Z,9Z-23:H between wild type and mutated males (Fig. 4D). As expected, the HarmOR11 mutants did not alter the EAG responses to type I sex pheromones, including Z11-16:Ald, Z9-14:Ald, and Z11-16:OH (Fig. 4D).

A similar situation was observed in females. The EAG responses to 100 µg and 1 mg of 3Z,6Z,9Z-21:H were all significant reduced in female mutants. However, the responses to 100 µg of 3Z,6Z,9Z-23:H and several plant volatiles, including (−)-β-Pinene, cis-Jasmone, Benzaldehyde, Salicylaldehyde, Geraniol, and (Z)-2-Hexen-1-ol, remained unchanged in HarmOR11 mutated female (Fig. 4E).

Fig. 4

HarmOR11 mediated electrophysiology responses to type II pheromones. A HarmOR11 possesses nine exons (black rectangle) in the genome, and a signal guide RNA (sgRNA) target (in red) located in exon 3 was chosen for CRISPR/Cas9-directed gene editing. PAM site (AGG) was marked in purple, and four types of somatic mutations at the targeted loci were obtained. B The mutation with a 13-bp DNA fragment insertion caused a frameshift at codon 108 and introduced a premature stop codon, giving rise to a truncated protein of 110 amino acids. C The two sequencing chromatograms indicated two genotypes, wild type (WT, top) and HarmOR11 mutant (bottom), respectively, and the insertion fragment was shown with a rust dushed frame. D Relative EAG responses to three type I and two type II pheromones in males of WT (green) and HarmOR11 mutation (teal). Type I pheromones and 3Z,6Z,9Z-23:H were used in a concentration of 10 µg µL− 1, and 3Z,6Z,9Z-21:H was used in concentrations of 10 µg µL− 1 and 100 µg µL− 1. Data are shown as mean ± SEM (n = 25 biological replicates), and statistical analysis was conducted by a two-tailed unpaired Student’s t-test. ns, P > 0.05; ***, P < 0.001. E Relative EAG responses to six plant volatiles and two type II pheromones in WT (purple) and HarmOR11 mutation (rust) females. Plant volatiles and 3Z,6Z,9Z-23:H were applied in a concentration of 10 µg µL− 1, and 3Z,6Z,9Z-21:H was used in concentrations of 10 µg µL− 1 and 100 µg µL− 1. Data are shown as mean ± SEM (n = 22 and 30 biological replicates), and were analyzed by a two-tailed unpaired Student’s t-test. ns, P > 0.05; *, P < 0.05; **, P < 0.01

HarmOR11 mutants impaired the electrophysiology responses to 3Z,6Z,9Z-21:H in sensilla trichoid

Previous evidence suggested that HarmOR11 is expressed in neighboring ORNs with HarmOR13 within the same sensilla trichoid in males of H. armigera [39]. However, the cellular localization of HarmOR11 in the female antenna has not been determined. To address this, we performed in situ hybridization on cryosections of female H. armigera antennae. Longitudinal sections through female antennae were labelled using digoxigenin (DIG)-labelled HarmOR11 antisense RNA. The results indicated strong HarmOR11 hybridization signals restricted to the bases of sensilla trichoid in the female antenna (Fig. 5A, B).

To investigated whether the neurons expressing HarmOR11 respond to the candidate type II sex pheromone, we conducted single-sensillum recordings (SSR) using sensilla trichoid on adult antennae. Previously, HarmOR11 was reported to be expressed with HarmOR13 in neighboring ORNs within the same sensilla trichoid in male adults [39]. Therefore, we used Z11-16:Ald to identify the sensilla trichoid housing HarmOR11 and HarmOR13 neurons. We tested a total of 132 sensilla trichoid, and 83 of them were strongly activated by Z11-16:Ald, indicating the presence of HarmOR13 and HarmOR11. Remarkably, all these activated sensilla trichoid responded rapidly to 3Z,6Z,9Z-21:H but did not respond to 3Z,6Z,9Z-23:H, even when presented at a total amount of 1 mg (Fig. 5C). However, in HarmOR11 mutant males, we observed that the sensilla trichoid responding to Z11-16:Ald significantly impaired electrophysiology responses to 3Z,6Z,9Z-21:H (Student’s t-test, P < 0.001), while they had no effect on the response intensity to Z11-16:Ald (Student’s t-test, P = 0.667, Fig. 5C, E). Additionally, these sensilla in male mutants showed no response to 3Z,6Z,9Z-23:H, similar to the wild type (Fig. 5C).

In females, HarmOR11 also specifically expressed in sensilla trichoid (Fig. 5A, B). Therefore, we investigated the responses of the sensilla trichoid to 3Z,6Z,9Z-21:H in the female antenna. It is important to note that there was no chemical marker like Z11-16:Ald in males to locate the sensilla housing HarmOR11 neurons. Consequently, we swept the sensilla trichoid that responded to 3Z,6Z,9Z-21:H blindly. We tested a total of 153 sensilla trichoid, and 68 of them narrowly tuned to 3Z,6Z,9Z-21:H in a similar manner to that in males (Fig. 5D, E). Not surprisingly, these sensilla exhibited no response to 3Z,6Z,9Z-23:H (Fig. 5D, E). In contrast, we tested 113 sensilla trichoid in antennae of female mutants, but none of them exhibited specific responses to 3Z,6Z,9Z-21:H as observed in the wild type.

Fig. 5

HarmOR11 mutant impairs the electrophysiology responses to type II pheromones in sensilla trichoid of adults. A, B In situ hybridization of HarmOR11 in female antennae of Helicoverpa armigera. In situ hybridization was performed with digoxigenin-labelled antisense RNA probes on longitudinal tissue sections of antennae. Signals were visualized using an anti-DIG antibody. A, the signal in short sensilla trichoid (st); B, the signal in long sensilla trichoid (st). C Representative traces of HarmOR11-expressing sensilla trichoid stimulated with 1 mg of pheromone components, including Z11-16:Ald, 3Z,6Z,9Z-21:H, and 3Z,6Z,9Z-23:H in WT and HarmOR11 mutant males. The red line represents the 0.3 s odor stimulation. D Representative traces of HarmOR11-expressing sensilla trichoid activated by 1 mg of 3Z,6Z,9Z-21:H and 3Z,6Z,9Z-23:H in WT and HarmOR11 mutant females. E Quantification of the mean responses to the indicated stimulus for the experiment shown in C and D. Data are plotted as mean ± SEM (n = 12–15 biological replicates), and statistical analysis was conducted by a two-tailed unpaired Student’s t-test. ns, P > 0.05; ***, P < 0.001

Molecular modeling and simulations uncover the molecular basis of selective recognition of different pheromone types

To elucidate the divergent recognition mechanisms allowing lineage-specific pheromone receptors to detect distinct sex pheromones, we compared the binding pockets of HarmOR11 and HarmOR13 using computational modeling. Though the exact stoichiometry remains unknown, previous studies put forth a hypothetical 2:2 complex between odorant receptors and Orco proteins [61]. Building on our previous modeling of HarmOR14b-HarmOrco and HarmOR16-HarmOrco heterotetramers [42], we constructed analogous complexes of HarmOR11-HarmOrco and HarmOR13-HarmOrco docking with their corresponding pheromone ligands, 3Z,6Z,9Z-21:H and Z11-16:Ald. We then carried out microsecond-long molecular dynamics simulations of the resulting ligand-receptor complexes to characterize the binding modes and key stabilizing interactions underpinning pheromone recognition (Fig. 6A).

Our MD simulations revealed stable binding of both 3Z,6Z,9Z-21:H and Z11-16:Ald in their respective pockets, albeit with distinct binding modes. Z11-16:Ald points towards the solvent with its aldehyde head group, while 3Z,6Z,9Z-21:H conceals its two heads deeply, likely due to its high hydrophobicity (Fig. 6B, C). Analysis of distance components (Distx, Disty, Distz) between the ligand center and the OR pocket center, as well as the angle (θ) between the ligand and the membrane normal, revealed a considerably broader conformational distribution for 3Z,6Z,9Z-21:H in HarmOR11 compared to Z11-16:Ald in HarmOR13 (Fig. 6C, S2, S3).

Further scrutiny of atomic details unveiled the significant stability of 3Z,6Z,9Z-21:H in HarmOR11, primarily attributed to L216 in transmembrane helix 4 (TMH4), F159 in TMH3, and I188 in extracellular loop 2 (Figs. 2B and 6D, S4). In the parallel investigation of the HarmOR13-Z11-16:Ald system, simulations indicated that the aldehyde group of Z11-16:Ald was solvent-exposed, while the long carbon chain established interactions with surrounding hydrophobic residues (Fig. 6B). Notably, this interaction was predominantly facilitated by F153 in TMH3, F214 in TMH4, and V335 in TMH6 (Fig. 6D, S4).

Additionally, our sequence alignment analysis highlighted L152, a specific residue in HarmOR11 located at one end of the crescent-shaped pocket (Fig, 2B, 6B), as a potential key factor influencing ligand selectivity. Notably, at the corresponding position in AsegOR3, which recognizes the same ligand (3Z,6Z,9Z-21:H), a similarly hydrophobic leucine is present (Fig. 2B). In contrast, in the pocket of ObruOR1, which recognizes the relatively less hydrophobic 1,3Z,6Z,9Z-19:H, this position is occupied by a polar threonine. These sequence distinctions likely play a role in determining ligand selectivity.

Intriguingly, the AlphaFold2 predictions for the HarmOR11-Orco heterotetramer revealed two distinct conformational states of the critical gating residues F427 of OR11 and Q471 of Orco: a closed state with both residues directed towards the ion channel’s center (State1) and another presumed open state with both residues facing outward (State2, used for simulating the pheromone-bound state), as depicted in Fig. 7A. For the HarmOR13-Orco heterotetramer, all predictions pointed to a uniform closed state, leading us to forego further analysis for this heterotetramer. To further explore the molecular mechanism of channel activation by substrate binding, we conducted additional MD simulations starting from the closed State1 (Fig. 7B) and compared these with simulations of the pheromone-bound states originating from the presumed open State2 (Supplementary Table S3). These simulations showed stability, with State 2 maintaining a larger pore size with large separation of the F427 residues, and a closer arrangement of the Q471 residues (Fig. 7D).

This observation is reminiscent of the substrate binding induced conformational shifts observed in the recently resolved cryo-EM homotetrameric structures of gustatory receptors (Gr), such as Drosophila melanogaster Gr43a (DmGr43a) and B. mori Gr9 (BmGr9) in both close and open states [62,63,64], which facilitate ion passage by rotating helices and repositioning the side chains of hydrophobic (F422 in Gr43a, F444 in BmGr9) and hydrophilic residues (Q421 in Gr43a, Q443 in BmGr9) at the entrance of the channel (Fig. 7C). Notably, the position preceding F427 in HarmOR11 is occupied by a smaller residue, Ala, rather than the hydrophilic Gln. Despite this, we identified similar hydrophilic residues like Q471 in Orco. We hypothesize that the smaller size of Ala may facilitate channel opening. Thus, we propose that the ion channel opening mechanism in HarmOR11 might slightly differ from that in Gr43a, potentially involving rotation of the S7b helix and replacement of F427 with the shorter amino acid A426. Our sequence alignment of typical odorant receptors from six insects suggests that this pattern of replacing F427 with shorter side chains or substituting with hydrophilic amino acids could be somewhat universal (Fig. 7E). However, we did not observe complete ion passage events or spontaneous ion channel openings triggered by pheromone binding, likely because such processes exceed the microsecond time scale of our simulations.

Fig. 6

Molecular dynamics simulations reveal distinct binding modalities for pheromones in HarmOR11 and HarmOR13 complexes. A Structural models of (HarmOrco)2/(HarmOR11)2 with 3Z,6Z,9Z-21:H and (HarmOrco)2/(HarmOR13)2 with Z11-16:Ald. HarmOrco, HarmOR11, HarmOR13 are colored blue, orange, and cyan respectively. B Representative ligand binding modes derived from MD simulations, with density maps displaying pheromone conformational distributions and water molecules depicted as spheres. The purple sticks indicate HarmOR11 residue L152, a potential determinant of substrate selectivity. C Simulation trajectories and conformational distributions of ligand-receptor center distance components (Distx, Disty, and Distz) and ligand-membrane angle (θ). D Key receptor residues interacting with 3Z,6Z,9Z-21:H (HarmOR11, left) and Z11-16:Ald (HarmOR13, right) shown as sticks

Fig. 7

Key residues involved in channel opening as predicted by AlphaFold2 and observed in MD simulations. A AlphaFold2 predictions showing two conformational states of the ion channel in the HarmOR11-Orco heterotetramer. State1 is a closed state with F427 of HarmOR11 oriented towards the channel center, and State2 is a putative open state with F427 of HarmOR11 moving away from the center while Q471 of HarmOrco moving towards the center. B Representative conformations from MD simulations initiated from both State1 and State2. C Conformational transitions of key residues F422 and Q421 in the closed and open states of the gustatory receptor GR43a, as revealed by recently published cryo-EM structures (PDB ID: 8JM9 and 8X83). D Trajectories of the distances between the F427 residues of HarmOR11 and the Q471 residues, alongside the population of the pore size (calculated as the area between two F427 and two Q471 residues) from MD simulations of the HarmOR11-Orco heterotetramer in both the apo State 1 and the pheromone-bound State 2. E Sequence alignment of pheromone receptors highlighting key amino acids potentially related to ion channel opening