The larvae of S. exigua were reared using the method of Goh et al. (1990) at 25 ± 1°C, with which the larvae underwent five instars (“L1–L5”). For the immune challenge, Escherichia coli Top10 (Invitrogen, Carlsbad, CA, United States) were cultured with the method described previously (Hasan et al., 2019). Bacterial cells were resuspended in distilled water to have 4.5 × 10⁴ cells/μl for treatment. For a pathogenicity test, Bacillus thuringiensis subsp. aizawai (Bt) and Xenorhabdus hominickii ANU101 (Xh) were cultured with the method described previously (Mollah et al., 2020a).

An anticoagulant buffer (ACB) was prepared with 186 mM of NaCl, 17 mM of Na₂EDTA, and 41 of mM citric acid, and then the pH was adjusted to 4.5 with HCl. Phosphate-buffered saline (PBS) was prepared with 100 mM of phosphoric acid, and the pH was adjusted to 7.4. The transfection reagent Metafectene Pro was purchased from Biontex (Plannegg, Germany). Salicylic acid (2-hydroxybenzoic acid: SA, ≥99%) was obtained from Sigma-Aldrich (Seoul, Korea) and dissolved in dimethylsulfoxide.

A sequence (Se-DSP1) of S. exigua homologous to human HMGB1 (GenBank accession number: CAG33144.1) was obtained from a transcriptome (GenBank accession number: GAFU01017850.1). Its open reading frame (ORF) was deposited at GenBank (accession number: MK737894). Phylogenetic and domain analyses were performed using the MEGA6 and Clustal W programs from the European Molecular Biology Laboratory's European Bioinformatics Institute (EMBL-EBI) (www.ebi.ac.uk). Bootstrapping values were obtained with 1,000 repetitions to support branches. Protein domains were predicted using the SMART (http://smart.embl-heidelberg.de/) search program.

Total RNA extraction, complementary DNA (cDNA) preparation, and quantitative reverse transcription (RT-q)PCR were performed following the methods described previously (Mollah et al., 2020a,b). The synthesized first-stranded cDNA was used as a template for PCR amplification with 35 cycles of 94°C for 30 s, 58.8°C for 30 s, and 72°C for 30 s after an initial heat treatment at 94°C for 2 min with the gene-specific primers listed in Supplementary Table S2.

The DNA template was amplified with the gene-specific primers of Se-DSP1 (Supplementary Table S2) containing a T7 promoter sequence (5′-TAATACGACTCACTATAGGGAGA-3′) at the 5′ end. Double-stranded RNA (dsRNA) was prepared following the method described previously (Mollah et al., 2020b). A viral gene, CpBV302, was used as a control dsRNA (“dsCON”).

For a cellular immune assay, hemocytic nodules were assessed with the method described previously (Mollah et al., 2020a). For humoral immune response, the expression levels of six antimicrobial peptide (AMP) genes (Supplementary Table S2) were assessed with the RT-qPCR method as described above.

The ORF of Se-DSP1 was cloned into a pET expression vector (Invitrogen, Carlsbad, CA, United States). After overexpression with isopropyl β-D-thiogalactopyranoside, the recombinant protein containing a 6× His tag was purified with the method described previously (Mollah et al., 2021). Resulting proteins contained partial degraded fragments, which were removed by gel filtration chromatography with Sephadex G-100 resin (Sigma-Aldrich, Seoul, Korea). To detect the binding of Se-DSP1 to SA, a thermal shift assay (Simeonov, 2013) was performed with a Protein Thermal Shift (Applied Biosystem, Foster City, CA, United States) dye kit according to the instructions of the manufacturer. Briefly, a reaction mixture consisted of 5 μl of a protein thermal shift buffer, 2.5 μl of protein thermal shift dye (8×, prepared by kit), 10 μl (500 ng) of purified protein (Se-DSP1), and 2.5 μl of ligands at different final concentrations (0, 5, 10, 15, and 20 μM). A melting curve experiment type was set up using the StepOne Real Time PCR System (Applied Biosystems, Foster City, CA, United States) with a continuous ramping mode (2.96°C/min), at which two thermal extremes were 25°C for 2 min at the initial step and 99°C for 2 min at the final step. Melting temperatures resulting from the experiment were plotted with SigmaPlot 10.0 (Systat Software, San Jose, CA, United States). A dissociation constant (Kd) was calculated using the ligand binding equation category.

Phospholipase A2 enzyme activity was measured with a PLA₂ Assay Kit (Cayman Chemical, Ann Arbor, MI, United States) using diheptanoyl thio-phosphatidyl choline for the secretory phospholipase A2 (sPLA₂) substrate and arachidonyl thio-phosphatidyl choline for the cellular PLA₂ (cPLA₂) substrate after sample preparation as described previously (Vatanparast et al., 2018). Briefly, the L5 larvae of S. exigua were injected with dsDSP1 and dsCON (1 μg/larva) using a microsyringe (Hamilton, Reno, NV, United States), incubated at room temperature for 24 h, and then challenged with E. coli (10⁴ cells/larva). At 8 h post injection of the bacteria, fat body was collected. The reaction volume consisted of 10 μl of fat body extract, 10 μl of 5,5-dithio-bis-(2-nitrobenzoic acid), 5 μl of the assay buffer, and 200 μl of the substrate. The change in absorbance at 405 nm was monitored with three biologically independent replications.

For the Western blot, hemolymph samples were collected from the L5 larvae of S. exigua. After centrifugation with 800 × g for 3 min at 4°C, the supernatant plasma samples were obtained and used for the analysis. A polyclonal antibody against rat HMGB1 (NB100-2322; Abcam, Waltham, MA, United States) was used after 5,000 times of dilution. α-Tubulin was used as a reference protein and detected by a polyclonal antibody (GTX112141; GeneTex, Irvine, CA, United States) after 1,000 times of dilution. The primary antibodies were detected with an anti-rabbit immunoglobulin G (IgG)–alkaline phosphatase secondary antibody (WP20007; Sigma-Aldrich, Seoul, Korea) after incubation with a substrate (BICP/NBT; Sigma-Aldrich, Seoul, Korea). This antibody showed a specific cross-reactivity with Se-DSP1 (Supplementary Figure S1).

An immunofluorescence assay (IFA) was performed according to the method described previously (Mollah et al., 2021). Briefly, hemocytes and fat body tissues were collected from the L5 larvae of S. exigua. Furthermore, Se-DSP1 was observed with the HMGB1 antibody. The cytoplasm and nucleus were stained with Alexa Fluor 555 phalloidin and 4′,6-diamidino-2-phenylindole (DAPI, 1 μg/ml) (Thermo Fisher Scientific, Rockford, IL, United States), respectively. Subsequently, cells were observed under a fluorescence microscope (DM2500; Leica, Wetzlar, Germany) at 400× magnification under microscope parameter conditions (Supplementary Table S3).

To determine Bt virulence, the L4 larvae were subjected to feeding assays using a leaf-dipping method. Briefly, a piece of cabbage leaf (3 × 3 cm) was soaked in 250 ppm of a Bt suspension containing different concentrations of SA for 5 min. The treated leaves were then provided to larvae to allow feeding for 24 h. To determine Xh virulence, the L4 larvae were subjected to injecting assays. Each larva was injected with 100 live bacterial cells containing different concentrations of SA using a microsyringe (Hamilton, Reno, NV, United States). The treated larvae were then incubated for another 4 days under rearing conditions, and mortality was monitored. Each replication consisted of 10 larvae. Each treatment was replicated three times.

Two lines of Solanum esculantum, “M82” and “IL2-2,” were used. Seeds were obtained from the Tomato Genetics Resource Center at the University of California (Davis CA, United States) (http://tgrc.ucdavis.edu/). For the bioassays, the L2 larvae of S. exigua were reared on 4-week-old tomato plants for 7 days with a 16-h light/8-h dark period at a temperature of 25 ± 2°C. The larvae were then used to assess immune responses indicated by nodule formation and AMP expression, as described above. The larvae were also used to determine bacterial virulence, as described above.

Approximately 300 mg of leaf tissues were collected from each line of tomato plants, and then extracted and analyzed for SA levels, as described previously (Bowling et al., 1994). The SA concentration was determined by high-performance liquid chromatography (HPLC) with an ARH-601 organic acid column (100 × 6.5 mm; Transgenomic Inc., Omaha, NE, United States) run at 45°C in.01 N of H₂SO₄ and a flow rate of 0.6 ml/min.

By interrogation with a D. melanogaster DSP1 sequence (GenBank accession number: NP_727960.1) as a query, Se-DSP1 was predicted from a transcriptome (GenBank accession number: GAFU 01017850.1). Its ORF consisted of 1,518 bp encoding 505 amino acids. Se-DSP1 shared 95.1, 77.1, 73, and 52.1% amino acid sequence similarities with the DSP genes of Helicoverpa armigera, Leptinotarsa decemlineata, D. melanogaster, and the HMGB1 of Homo sapiens, respectively. A phylogenetic analysis showed that Se-DSP1 was closely related to other lepidopterans within an insect cluster (Supplementary Figure S2A). Like mammalian HMGB1, Se-DSP1 comprised HMG boxes A and B and an acidic tail (Supplementary Figure S2B). Phosphorylation and disulfide linkage sites are frequently located in box A compared with other domains. In the two HMG boxes, human HMGB1 and Se-DSP1 exhibited higher sequence identities (69.9%) in box A than in box B (39.4%). Unlike mammalian HMGB1s, insect DSP1s possess additional N-terminal domains such as a coiled coil, low complexity, and RNase.

In all developmental stages, Se-DSP1 was expressed showing high expression levels at the L5 and adult stages. It had low expression levels at the pupal stage (Supplementary Figure S3A). At the L5 larval stage, Se-DSP1 was expressed in all the tested tissues such as hemocytes, fat body, gut, and epidermis (Supplementary Figure S3B). The basal expression levels of Se-DSP1 were significantly upregulated following the immune challenge with E. coli in hemocytes and the fat body (Supplementary Figure S3C). Such an upregulation was more acute in hemocytes than in the fat body.

In naïve larvae, Se-DSP1 was localized in the nuclei of the hemocytes (Figure 1A). Upon bacterial challenge, some Se-DSP1 proteins migrated to the cytosol. Similarly, Se-DSP1 was localized in nuclei of the fat body and isolated from naïve larvae, although some of them were detected in the cytosol after the immune challenge (Figure 1B). Interestingly, Se-DSP1 was detected in the plasma of the immune-challenged larvae, although it was slightly detected in the plasma collected from the naïve larvae (Figure 1C).

To analyze the role of Se-DSP1 in association with immune responses, its expression was manipulated by RNAi with gene-specific dsRNA (Figure 2A). Such an RNAi treatment significantly reduced Se-DSP1 expression levels by more than 90% in the hemocytes and fat body at 24 h. The reduced mRNA levels were confirmed by weaker bands in Western blotting against Se-DSP1. At such a reduced expression level of DSP1, cellular immune response was measured by counting hemocytic nodules (= a cellular immune response sequestering small pathogens with hemocytes and subsequent melanization) after the bacterial challenge (Figure 2B). The RNAi-treated larvae exhibited a significant (P < 0.05) impairment in cellular immune response based on nodule formation. Furthermore, the bacterial challenge upregulated the gene expression levels of six AMPs in the control larvae (Figure 2C). The reduction of AMP gene expression in the RNAi-treated larvae was significant (F = 997; df = 1, 144; P < 0.0001) compared with that in the control larvae.

Eicosanoids mediate various insect immune responses (Stanley and Kim, 2011). To understand the immune mediation of Se-DSP1, its activation of eicosanoid immune signaling was tested by measuring PLA₂ activities, catalyzing the committed step for eicosanoid biosynthesis, after RNAi treatment against Se-DSP1 (Figure 3). Two substrates, arachidonate phospholipid and non-arachidonate phospholipid, were used to measure PLA₂ enzyme activities. In particular, the PLA₂ activity obtained from the non-arachidonate phospholipid was designated as sPLA₂ activity. In contrast, cPLA₂ activity was obtained from PLA₂ activity using the arachidonate phospholipid. The immune challenge significantly (P < 0.05) increased both PLA₂ activities (Figures 3A,B). However, the larvae treated with RNAi specific to Se-DSP1 expression failed to induce PLA₂ activities. Such suppressed immune responses by RNAi treatment were significantly (P < 0.05) rescued by arachidonic acid (AA) or prostaglandin E₂ (PGE₂) in cellular (Figure 3C) or humoral immunity (Figure 3D).

A recombinant protein of Se-DSP1 was generated using a bacterial expression system. The recombinant protein was then purified using an affinity chromatography system containing nickel-nitrilotriacetic acid (Ni-NTA) resin and further cleared with a gel filtration chromatography system containing Sephadex G-100 (Sigma-Aldrich, Seoul, Korea) (Figure 4A). During the purification of the full-length Se-DSP1, a partial protein was also purified and confirmed by a Western analysis against a V5 epitope (Figure 4B). The truncated protein was assessed by liquid chromatography with tandem mass spectrometry (LC-MS/MS) and showed its partial structure deleting the N-terminal extension domain (Supplementary Figure S4). The injection of Se-DSP1 significantly increased the PLA₂ activity of the naïve larvae (Figure 4C). However, the partial Se-DSP1 failed to induce enzyme activity. The enzyme induction with full-length Se-DSP1 was prevented by the addition of an HMGB1 antibody recognizing Se-DSP1 (Figure 4C).

Salicylic acid is known to bind to human HMGB1 and interrupt immune mediation (Choi et al., 2015). This suggests that it might antagonize the immune-mediating role of Se-DSP1, which is an insect homolog of human HMGB1. To test this hypothesis, different doses of SA were injected to larvae. The immune responses of the larvae were then assessed (Figure 5). Six AMP genes expressed in response to the bacterial challenge exhibited significant (P < 0.05) reductions in their expression levels after RNAi treatment using dsRNA specifically targeting the Se-DSP1 expression (Figure 5A). Among the six AMPs regulated by Se-DPS1, SA significantly (P < 0.05) prevented the upregulation of the expression of the AMP genes, except for attacin 1 (Att).

The immunosuppressive role of SA in immune responses was supported further by a mixture treatment of SA and entomopathogens. Nodule formation in response to the bacterial (X. hominickii) challenge was suppressed by SA injection in a dose-dependent manner (Figure 5B). The addition of SA to X. hominickii also enhanced bacterial pathogenicity by a hemocoelic injection assay (Figure 5C). The suppressive effect of SA was also observed after SA-feeding treatment. The addition of SA to Bacillus thuringiensis treatment significantly enhanced the bacterial pathogenicity after oral administration.

The immunosuppressive role of SA suggested that it might bind to Se-DSP1 and inhibit its immune mediation. In addition, Se-DSP1-like human HMGB1 (Choi et al., 2015) possesses highly conserved residues Arg and Lys to interact with SA (Supplementary Figure S5). To test the direct binding hypothesis, the purified Se-DSP1 protein was assessed for its binding affinity to SA via thermal stability assay (Figure 6A). When the concentration of SA was increased, the melting temperature (median dissociation temperature between SA and Se-DSP1 following denaturation) increased (see inset “full” figure in Figure 6A). In contrast, the partial Se-DSP1 did not show a thermal shift of the melting points in different SA concentrations (see inset “partial” figure in Figure 6A). The thermal shift curve of the full-length Se-DSP1 was used to estimate its dissociation constant (Kd) with SA: 1.28 ±.05 μM. The binding of SA to Se-DSP1 resulted in the inhibition of PLA₂ induction (Figure 6B). In addition, the inhibition of PLA₂ activity by SA treatment led to the suppression of cellular immune response assessed by nodulation (Figure 6C). The effect of the SA treatment was similar to that of the RNAi treatment against Se-DSP1 expression. Because SA inhibited Se-DSP1 and subsequent PLA₂ activity (see the diagram), the addition of AA significantly (P < 0.05) rescued the cellular immune response.

The immunosuppressive effect of SA on insect immunity suggested that plants might use SA to suppress the immune responses of phytophagous insects, leading the insects to be susceptible to entomopathogens. Before we tested the efficacy of SA in a plant host, an artificial diet containing SA was fed to S. exigua larvae and showed significant suppression in a cellular immune response (Figure 7A). Then, to test the role of SA in plant defense response against insects, an isogenic tomato line and its parental line were selected because of their different SA contents (Figure 7B). In particular, IL2-2 is an isogenic line generated by introgressing a segment of chromosome 2 from Solanum pennellii “LA716” to the S. lycopersicum “M82” variety by successive backcrosses (Eshed and Zamir, 1995). Furthermore, IL2-2 possesses a significantly higher amount of SA amount M82.

The larvae of S. exigua were fed with tomato leaves. Their immune responses were then assessed after a bacterial challenge. There was a significant (P < 0.05) difference in cellular immunity based on nodule formation between the larvae fed with M82 and those fed with IL2-2 (Figure 7C). The expression levels of the six AMP genes regulated by Se-DSP1 were then compared among the larvae fed with two different lines of tomato. The expression levels of the AMPs, except for gallerimycin, exhibited a significant (P < 0.05) reduction after the bacterial challenge when the larvae were fed with IL2-2 (Figure 7D).

To explore the role of SA in plant defense response by suppressing the insect immune system, the S. exigua larvae fed with either M82 or IL2-2 were treated with entomopathogens (Figures 7E,F). Both entomopathogenic bacteria (B. thuringiensis and X. hominickii) showed significantly different virulence levels against the larvae fed with either M82 or IL2-2. The larvae fed with IL2-2 were more susceptible to both bacterial pathogens compared with larvae fed with M82, indicating that endogenous SA in plants could effectively suppress insect immune defenses to two different entomopathogenic bacteria.