The bean bug R. pedestris adults were collected in soybean fields, Nanjing, East China, in the summer of 2020. They were reared in tents (30 × 30 × 30 cm) in an incubator (25°C, LD = 16:8 h), where soybean seedlings (3–5 weeks old) and seeds (variety Lindou 10) were provided as food.

Thirty adults (male or female, 10-d old) or fourth-instar nymphs were anesthetized on ice and subsequently dissected to obtain salivary glands (Figure 1). The RNA was extracted by the TRIzol Total RNA Isolation Kit (Takara, Dalian, China), following the manufacturer’s instructions. The quality of extracted RNA was verified by the Agilent 2100 Bioanalyzer (Agilent Technologies, CA, United States). Polyadenylated RNA (mRNA) was purified from the total RNA by using oligo(dT) magnetic beads and then the total mRNA was fragmented into short sequences in the presence of divalent cations at 94°C for 5 min. The cleaved RNA was transcribed, and the second-strand cDNA was obtained. After end-repair and adaptor ligation, the products were PCR-amplified and purified using Ampure XP Beads (Agencourt Bioscience, MA, United States) to create the cDNA library.

The library was sequenced on the Illumina sequencing platform and the raw data were generated using Solexa GA pipeline 1.6. Low quality reads were removed, and the rest sequences were assembled using Short Oligonucleotide Analysis Package (SOAP) de novo software (Li et al., 2008), and then clustered by TGICL v2.0.6 to gain unique genes (Pertea et al., 2003). The clean reads of the transcriptomes have been deposited to SRA database with the accession number of PRJNA690963.

The sequences of unigenes were searched in one of four databases to obtain their annotations, including the NR database (NCBI¹); the Gene Ontology (GO²), the KEGG Orthology (KEGG³) and the EuKaryotic Orthologous Groups (KOG⁴).

TransDecoder.LongOrfs was used to extract the long open reading frames (ORFs). The ORFs were blasted in the SwissProt⁵ and Pfam databases⁶ by Diamond Blastp and Hmmscan, respectively. The coding sequences (CDSs) were extracted from the transcripts by TransDecoder 3.0.1 (Kim et al., 2015), and then the predicted proteins were obtained. Then, the SignalP 5.0⁷ was used to test whether sequences have secretion signal peptides or not (Armenteros et al., 2019), while the TMHMM 2.0⁸ was used to check the transmembrane areas of sequences (Krogh et al., 2001).

The predicted proteins with secretion signal peptides and simultaneously without transmembrane areas are likely to be secreted by salivary glands into saliva (Nielsen, 2017), and therefore with a relatively high potential in modulation of plant defenses. We had paid attention to the genes with secretion signals (about 192 individuals, Supplementary Table 1) in the top 1,000 expressed genes of the transcriptomes, and 18 genes were selected for testing their expression levels in different tissues (see below). In addition, the amino acid sequences of most reported effectors in sap-feeding arthropods were compared to the predicted proteins in the male transcriptome, and 19 proteins had a relatively high similarity (≥30%) with the effectors. Their expression levels were also compared between different tissues of R. pedestris.

Riptortus pedestris saliva was collected in a Petri dish (2 cm × 11 cm) whose open was covered by two layers of Parafilm with 2 ml sterile sucrose solution (2.5% in water) as food in between (Figure 1). The Parafilm was previously sterilized by 75% ethanol solution. The sucrose solution was prepared with aseptic water and filtered through a 0.22 μm syringe filter (Millipore, MA, United States) for the removal of microorganisms. Ten individuals (males, females or fourth-instar nymphs) were put in each Petri dish and the collection lasted 24 h. The collection was repeated 30 times. In total, 300 individuals were used. After collections, the sucrose solutions of each Petri dish were combined (about 60 ml) and concentrated by ultrafiltration (3-kDa, Amicon Ultra-4 Centrifugal Filter Tube, Millipore; 5,000 g, 4°C, 30 min). The proteins were dissolved in 200 μl of SDT buffer (4% sodium dodecyl sulfate; 1 mM DTT and 100 mM Tris-HCl) and then were incubated in warm water for 15 min.

Subsequently, DTT was added into protein samples to a concentration of 100 mM, and then the samples were boiled for 5 min. After ultrafiltration (3-kDa; 14,000 g, 25°C, 10 min), 100 μl iodoacetamide (IAA) buffer (100 mM IAA in UA buffer) was used to dissolve the proteins, and then the samples were incubated at room temperature for 30 min in darkness. After ultrafiltration (3-kDa) again, the samples were washed with 100 μl UA buffer (8 M urea, 150 mM Tris-HCl, pH 8.0) twice, and then washed with 100 μl NH₄HCO₃ buffer (25 mM, Sigma) twice. Finally, the proteins were digested overnight in 4 μg of trypsin (Sigma) in 40 μl NH₄HCO₃ buffer (25 mM) at 37°C. The digested peptides were collected by ultrafiltration (3-kDa) and were dissolved in 40 μl NH₄HCO₃ buffer (25 mM).

The digested peptides were separated by Thermo Scientific Easy nanoLC 1000 that was equipped with a C18 column (Thermo Scientific Acclaim PepMap100, 100 μm × 2 cm). Buffer A (0.1% formic acid in water) including 5% buffer B (84% acetonitrile and 0.1% formic acid in water) were used as the mobile phase for gradient separation. The sample was uploaded onto the column at a flow rate of 0.3 μl/min. Subsequently, the column was eluted by a linear gradient of buffer B at a flow rate of 0.25 μl/min (0–50 min, concentration increasing from 0 to 35%; 50–55 min, 35 to 100%; and finally pure buffer B maintained for 5 min).

The eluted peptides were analyzed by the Q-Exactive mass spectrometer (Thermo Fisher Scientific, United States). Full MS scans were acquired in the Orbitrap mass analyzer over the range m/z 300–1800 with a mass resolution of 70000 (at m/z 200). The twenty most intense peaks with charge state ≥2 were fragmented in the higher-energy collisional dissociation (HCD) with a normalized collision energy of 30% (the isolation window was 2 m/z), and tandem mass spectra were acquired in the Orbitrap mass analyzer with a mass resolution of 17,500 at m/z 200. For all detections, the dynamic exclusion time was set to 60 s.

Proteins were identified and annotated by using Mascot 2.2 to search UniProt (see footnote 5) with the restriction to R. pedestris data. The following parameters were used: trypsin was selected as the enzyme; two missed cleavage sites were allowed; 20 ppm mass tolerances for MS and 0.6 Da for MS/MS fragment ions; oxidation was a variable modification; carbamidomethyl was a static modification.

The relative expression of selected genes (68 genes) in different tissues of R. pedestris males, including salivary glands, mid-guts, fat bodies and testes, were compared. Those are 31 proteins found in male saliva, 18 genes that exist in the top 1,000 transcripts and 19 genes (shown in Table 1) that are homologs to reported effectors. First, the total RNA of each tissue (30 individuals) was extracted by the TRIzol Total RNA Isolation Kit (Takara, Dalian, China). The first strand cDNA was synthesized from RNA by using the HiScript III RT SuperMix qPCR kit (Vazyme, Nanjing, China). Then, real time quantitative PCR (RT-qPCR) was performed on a QuantStudio 5 Real-Time System (Thermo Fisher Scientific, United States) by using the Top Green qPCR SuperMix kit (TransGen Biotech, Beijing, China). The reaction program started with an initial denaturation step at 95°C for 30 s, and then 40 cycles including two steps per cycle, 95°C for 5 s and 60°C for 34 s, were performed. The gene-specific primers were designed by using the Primer Premier 5.0 software. To evaluate the primers, the cDNA concentrations were either unchanged, or further diluted by 4, 16, or 64 times. When amplification efficiencies ranged from 90–110%, and the R² values were over 0.99 in the regression analysis, the primers were selected. Three biological replicates and three technical replicates were applied. The individual efficiency–corrected calculation method was used to compare the fold changes in expression levels of genes in mid-guts, testes and fat bodies, related to that in salivary glands (Rieu and Powers, 2009; Rao et al., 2013). Two housekeeping genes RpEF-1 and actin were used as reference genes (Lee et al., 2019). The primers and the result of the regression analysis of each gene were listed in the Supplementary File 1.

The statistical analyses on RT-qPCR data were carried out by using SigmaPlot 14 with one-way ANOVA tests. A Holm-Sidak post hoc analysis was used for pairwise comparisons. When the expression levels in four tissues (salivary glands, mid-guts, testes and fat bodies) were fitted with a normal distribution, the comparisons were performed in one run. Otherwise, pairwise comparisons were conducted by each pairs, and the normality always passed. Different lowercase letters above the bars in the Figure 2 indicate that there are significant differences (P ≤ 0.05).

Male R. pedestris migrate to soybean fields earlier than females during the flowering period, and then they will release pheromone and possibly induce plants to release volatiles for attracting females and nymphs (Endo et al., 2011; Xu et al., 2021). The release of male pheromone is stimulated by feeding (Morishima et al., 2005). So males may excrete more salivary proteins when feeding on newly located plants to overcome a relatively intact immunity. In addition, adults express some genes specifically by salivary glands, as opposed to nymphs (Huang et al., 2021a), which may also contribute to more proteins found in adult saliva than in nymph saliva. However, a strong variance sometimes occurs among replicates, when proteomes in saliva were analyzed in hemipterans, as reported in other papers (Carolan et al., 2009, 2011; Huang et al., 2018).

In the 814 proteins with secretion signals in the male transcriptome, many of them are probably used for digesting proteins and lipids, as also suggested by Huang et al. (2021a), including 112 proteases (peptidases) and 41 lipases (esterases). Since R. pedestris prefers to feed on bean pods in nature, the enzymes are possibly applied to digest proteins and oils in beans. Similar results were obtained from studies on other seed-feeding bugs, as well as predator bugs (Soyelu et al., 2007; Bigham and Hosseininaveh, 2010; Zibaee et al., 2012). Extra-oral digestion seems to be important for many stink bug species (Miles and Taylor, 1994). In laboratory, R. pedestris is normally reared on dry soybean seeds and water supply (Takeshita and Kikuchi, 2017), indicating the extra-oral digestion is a primary process of feeding. In addition, enzymes for sugar digestion were also found, including 6 α-amylases and other glucosidases. The enzymes appeared to be less abundant than proteinases in the salivary glands, as also found in other pod-feeding bugs (Soyelu et al., 2007).

In the male proteome, we found an α-glucosidase (UniProt ID: R4WDP5) and a proteinase (R4WQ74), and the both enzymes are highly expressed in mid-guts (Figure 2A). We also found several cathepsin L enzymes in the saliva of males and females (Table 2 and Supplementary Table 3), which are normally expected to occur in lysosomes and never leave the cells. However, the enzymes are often secreted by digestive systems in insects and act as cysteine proteinases (Terra and Ferreira, 2005). The cathepsins L in R. pedestris saliva normally have the secretion signals and are probably used for extra-oral digestion.

Catalases, glutathione peroxidases and peroxiredoxins are oxidoreductases that are well recognized for degrading ROS and maintaining redox homeostasis in the damaged plant cells (Petrova and Smith, 2014; Sharma et al., 2014; Chaudhary et al., 2015; Dong et al., 2020). A catalase (R4WNB5) existed in male saliva, and the enzyme was abundantly expressed in fat bodies (Figure 2). A peroxiredoxin (MW625814) and a glutathione peroxidase (MW625813) were produced by salivary glands in a relatively high amount (Figure 2B). These enzymes may also play an important role in suppressing the first-line defense of plants (Sharma et al., 2014; Dong et al., 2020).

Dehydrogenases may regulate plant defense signaling and detoxify plant toxic compounds (Sharma et al., 2014). For example, glucose dehydrogenases were found in the saliva of some aphid species and the activities of the enzymes were corresponding to their virulence (Carolan et al., 2011; Nicholson et al., 2012; Sharma et al., 2014). Several dehydrogenases (R4WD44, R4WCW6, R4WEC6, R4WIH8, and R4WQZ0) were found in the saliva of males and females. In addition, R. pedestris produced two glucose dehydrogenases (MW561670 and MW625827) in a higher amount in salivary glands (Figure 2B). The functions of these enzymes remained to be confirmed in R. pedestris.

Like the brown planthopper, Nilaparvata lugens (Stål) (Huang et al., 2016), R. pedestris secret leucyl aminopeptidases (R4WCJ9) in saliva (Table 2). The enzymes cleave defense peptides (e.g., hormones and neuropeptides) at N-terminus, especially leucine residues. In addition, an aminopeptidase (MW625818) in transcriptome was also found to be specific in salivary glands and testes of R. pedestris. The enzymes were considered to be essential in defending aphids against plant lectins (Nicholson et al., 2012). Metalloproteases, in contrast, possibly cleave peptides at the C-terminal end (Carolan et al., 2011). In aphids and thrips, they are able to counteract host defenses, by degrading plant defense proteins (Carolan et al., 2009, 2011; Stafford-Banks et al., 2014; Wang et al., 2015b). The metalloprotease (MW625833) of R. pedestris appeared to be a Zn-metallocarboxypeptidase, and it was abundantly expressed in salivary glands. These enzymes have a great potential in degrading defense proteins of host plants.

The chitooligosaccharidolytic beta-N-acetylglucosaminidase (NAGase, R4WE69) is a chitinase. The enzyme was highly expressed in salivary glands of R. pedestris and found in male saliva (Tables 1, 2). Plants NAGases act as an antifungal compound by hydrolyzing N-glycans of polysaccharides and glycoproteins (Altmann et al., 1999). Therefore, insects may also use NAGases for inhibiting fungal infection during feeding on plants (Nicholson et al., 2012; Sharma et al., 2014). In addition, NAGases in the saliva of sap-sucking herbivores possibly affect plant immunity by the interaction with NAGases of host plants (Nicholson et al., 2012; Sharma et al., 2014).

Phloem sieve elements respond to the feeding by piercing-sucking insects by quickly inducing calcium flux which possibly triggers the occlusion of sieve elements and increases the related plant defenses (Will et al., 2009). However, the calcium-binding proteins in saliva possibly reduce the reaction, which guarantees a continuous feeding (Will et al., 2007; Ye et al., 2017; Tian et al., 2021). Indeed, several types of calcium-binding proteins were found in the proteomes and transcriptomes (Tables 1, 2), and one of them (MW625835) had been found to be highly expressed in salivary glands of R. pedestris.

Trialysins have been found in saliva of the hematophagous bug Triatoma infestans (Klug) (Amino et al., 2002). The protein may lyse cells of both animals and microorganisms, indicating it plays an important role in interaction with hosts (Amino et al., 2002). Similarly, two trialysins (A0A1B4X9A5 and A0A1B4X9A9) were found in the saliva of the bean bug (Table 2). And the A0A1B4X9A5 was abundantly produced by salivary glands (Figure 2A).

We also found several mucin-like proteins in the top 1,000 expressed genes of the transcriptomes that were commonly with secretion signals, and most of them were expressed in a relatively higher amount in salivary glands (Table 1 and Figure 2). Similar result was found in N. lugens that secreted mucin-like proteins into both watery and gelling saliva (Huang et al., 2016). One of their functions was to form a developed salivary sheath and increase the adaptation of the brown planthoppers to rice plants (Huang et al., 2017). In the laboratory, we observed many salivary sheaths on soybean seeds after fed by R. pedestris under an optical microscope. The sheaths are normally white tubes with helical curves and with variable lengths. Whether mucin-like proteins also contribute the formation of salivary sheaths in R. pedestris needs further studies.

Two serine protease inhibitors (R4WL96 and R4WCP6) were found in the salvia of males and females. The proteins have been found to be essential in regulation of host defenses by various hematophagous arthropods (Amino et al., 2001; Chmelař et al., 2017; Soares et al., 2018). Both enzymes of R. pedestris were highly expressed in salivary glands (Figure 2), and they are expected to be important in interaction with plant defenses.

The effector proteins of fungal pathogens are often small secreted cysteine-rich proteins (SSCPs) with less than 200 amino acid residues, and have a high cysteine content (>2%) (Stergiopoulos and de Wit, 2009). The strategy might also occur in insects. For example, Nl28 is a species-specific SSCP in N. lugens, which induced cell-death symptoms after the transient expression in Nicotiana benthamiana (Rao et al., 2019). The cysteines in the effectors often contribute the formation of disulfide bonds, thereby supporting effectors a specific structure (Saunders et al., 2012). Here, we also found three SSCPs (R4WDU9, R4WQ74, and MW625844) that were greatly expressed in salivary glands or mid-guts of R. pedestris (Figure 2). Therefore, R. pedestris might also use SSCPs to modulate host plant immunity, like fungi and the brown planthoppers.

Insect chemosensory proteins (CSPs) are well known for their functions in olfaction and gustation (Pelosi et al., 2005). However, some papers have found that the proteins are sometimes specifically expressed in salivary glands, and they trigger chlorosis and dwarf phenotypes of N. benthamiana after the transient expressions (Copenhaver et al., 2010; Rao et al., 2019). The MP10, a CSP of the green peach aphid Myzus persicae (Sulz.), activated the jasmonic acid and salicylic acid signaling pathways of N. benthamiana during feeding (Rodriguez et al., 2014; Mugford et al., 2016). Here, two CSPs (A0A2Z4HQ00 and MW625831) were found in saliva or transcriptomes of R. pedestris. However, their expression levels were not biased to a tested tissue (Supplementary Table 4).

Similar with CSPs, odorant binding proteins (OBPs) are also well recognized for the function in sensing odors (Zhu et al., 2019). Here, an OPB (A0A2Z4HQ32) was found in the male saliva (Table 2), and another OPB (MW625837) was largely produced in R. pedestris salivary glands (Figure 2B). How OBPs could act as effectors in herbivores is not yet understood. However, some OBPs are used by mosquitoes to scavenge host amines during feeding, which contributes to anti-inflammatory effect (Calvo et al., 2006, 2009). Since OBPs possibly have ligand-binding hydrophobic channels (Calvo et al., 2009), they may be used by herbivores to bind defense-related molecules of plants.

Carbonic andydrases are zinc metalloenzymes that catalyze the reversible hydration of carbon dioxide to bicarbonate. A carbonic andydrase (MW625839) was specifically expressed in salivary glands of R. pedestris (Figure 2B). Carbonic andydrases were also found in the watery saliva of a leafhopper and a planthopper species (Hattori et al., 2015; Huang et al., 2016). Silencing the gene resulted in lethality of N. lugens (Huang et al., 2016). How the enzymes help hemipterans feed on plants is not clear. They may play a protective role in the elevated CO₂ concentration during feeding (Huang et al., 2016).

Transcriptome analysis indicates that salivary glands of R. pedestris possibly produce a rich repertoire of proteins, in which many of them are possibly used to digest proteins and oils in beans. In addition, rich proteins were found in their saliva, and a high proportion of the proteins are not yet annotated, indicating knowledge on the salivary proteins of the pest is very limited. Therefore, the datasets reported here represent an important first step in identifying effectors in R. pedestris. In addition, a few elicitors of moth species are relatively small molecules that are not complete proteins, such as volicitin and inceptin peptides (Alborn et al., 1997; Steinbrenner et al., 2020). Those elicitors might also exist in heteropteran species, in which they have been ignored so far. The different kinds of elicitors and effector proteins are likely to work together in facilitating the feeding success of R. pedestris on soybeans.