Four different species of Photorhabdus were obtained from the Korean Agricultural Collection (KACC; Rural Development Administration, Jeonju, Republic of Korea) with accession numbers KACC91042 for P. temperata temperata (Ptt), KACC11928 for P. luminescens kayaii (Plk), KACC12284 for P. temperate thracensis (Pt), and KACC12282 for P. luminescens akhurstii (Pla). Xenorhabdus hominickii (Xh; Park et al., 2017) and Xenorhabdus nematophila (Xn; Park and Kim, 2000) were obtained from our frozen culture stock. These bacteria were cultured in tryptic soy broth (TSB, Difco, Sparks, MD, United States) for 72 h at 28°C in a shaking incubator at 180 rpm.

Larvae of S. exigua were collected from onion fields in Andong, South Korea and fed with an artificial diet (Goh et al., 1991) at 27 ± 1°C. Under these conditions, they underwent five larval instars (L1–L5). A sugar solution (10%) was used for feeding adults.

Four GXPs were synthesized and prepared by AnyGen Co., Ltd. (Gwangju, Republic of Korea). Their sequences were cyclo[D-Leu/L-Leu/D-Val/L-Leu/D-Phe] for GXP-A, cyclo[D-Leu/L-Leu/D-Leu/L-Leu/D-Phe] for GXP-B, cyclo[D-Leu/L-Leu/D-Val/L-Leu/D-Leu] for GXP-C, and cyclo[D-Leu/L-Leu/D-Leu/L-Leu/D-Leu] for GXP-D. In addition, a stable isotope-labeled cyclic pentapeptide of GXP-A was synthesized as cyclo[D-Leu/L-Leu*/D-Val/L-Leu/D-Phe] (Leu*: ¹³C6, ¹⁵N) for GXP-A*. The purities of the synthetic peptides were over 97% based on the quality control using MALDI-TOF and LC-UV analyses by the manufacturer (Supplementary Figure S1). The peptides were prepared with dimethylsulfoxide (DMSO) at a concentration of 1 mM. The stock solutions were stored in the freezer and diluted with DMSO to prepare working solutions. Anticoagulant buffer (ACB) was prepared with 186 mM of NaCl, 17 mM of Na₂-EDTA, and 41 mM of citric acid. Its pH was then adjusted to 4.5 with HCl. 4′,6-Diamidine-2′-phenylindole dihydrochloride (DAPI) and MTT (3-[4,5-dimethylthiazole-2-yl]-2,5-diphenyl tetrazolium bromide) were purchased from Sigma-Aldrich (Seoul, Republic of Korea). Alexa Fluor 488, as a dye of fluorescein isothiocyanate (FITC), was purchased from Thermo Fisher Scientific (Waltham, MA, United States). Phosphate-buffered saline (PBS) was prepared with 100 mM of phosphate and adjusted to a pH of 7.4 with 1 N NaOH.

The whole genome sequence of P. temperata temperata Meg1 was retrieved from GenBank (accession number: GCA_000722995.1). Ten NRPS genes were predicted from the genome using NCBI gene annotation. Functional modules for each NRPS and resulting products were predicted using antiSMASH bacterial version software.¹ Based on the predicted products of the NRPS genes, a pentapeptide-producing NRPS gene was predicted as the gxpS gene of P. temperata temperata. To support the prediction, a phylogenetic analysis was performed with gxpS orthologs from four other bacterial species: P. luminescens laumondii, P. luminescens akhurstii, Photorhabdus thracensis, and Ptt, with accession numbers of AXG48275.1, QXF34648.1, AKH65701.1, and JGVH01000001.1, respectively. The phylogeny analysis was performed by MEGA6 program.² Bootstrap values were obtained with 1,000 repetitions to support branching and clustering. A physical map of the annotated 10 NRPS loci was constructed by their alignment to a full genome of P. luminescens laumondii TT01 (GenBank accession number: NC_005126) with the BlastN searching machine of NCBI³ (Supplementary Table S1).

The cultured bacteria were centrifuged to obtain cell pellets, to which 1 mL of TRIzol reagent (Invitrogen, Carlsbad, CA, United States) was added. The pellets were then subjected to the procedure described in the manufacturer’s manual. Each extracted RNA was resuspended in 50 μL of diethyl pyrocarbonate, and its concentration was determined using a spectrophotometer (NanoDrop, Thermo Fisher Scientific). The RNA purity was confirmed by the absorbance ratio (260/280 nm) greater than 1.8. The cDNA synthesis was carried out using 400 ng of the purified RNA and an RT PreMix (iNtRON Biotechnology, Seoul, Republic of Korea) containing random primers. No genomic DNA contamination in the purified RNA was confirmed by RT-PCR (see below) using RNA extract as template rather than cDNA.

For RT-PCR, the cDNAs were amplified in specific target genes using Taq polymerase (GeneAll Biotechnology Co., Ltd., Seoul, Republic of Korea) with gene-specific primers (Supplementary Table S2). The PCR reaction was initiated with an initial denaturation at 95°C for 5 min. The subsequent 35 amplification cycles were performed at 95°C for 1 min, 53–55°C for 1 min, and 72°C for 1 min. The PCR reaction was finalized with an additional extension step at 72°C for 10 min. The PCR product was analyzed by 1% agarose gel electrophoresis to determine the presence of the amplified product. For RT-qPCR, the Power SYBR Green PCR Master Mix (Toyobo, Osaka, Japan) was used with gene-specific primers under the guidelines of Bustin et al. (2009). Quantitative analysis was performed using the comparative CT method (Livak and Schmittgen, 2001). All experiments were independently replicated three times.

RNA was extracted at various time points for each bacterial culture to measure the expression level of gxpS during the bacterial growth. Wild-type P. temperate temperata stock was streaked on a TSB plate, and a single colony was cultured in 10 mL TSB medium at 180 rpm and 28°C for 18–20 h. The culture broth was then transferred to 1 L of TSB medium and incubated at 180 rpm and 28°C. For each of the designated time points (0, 12, 24, 36, 48, 60, and 72 h), 5 mL of bacterial culture was obtained for RNA extraction. Additionally, 1 mL of bacterial culture was used to measure the optical density (OD) at 600 nm at each time point. Each time point was replicated with different bacterial culture flasks.

To monitor the bacterial growth and gxpS expression in S. exigua, 2 μL (2 × 10⁶ colony-forming unit (CFU)/mL) of freshly cultured bacteria was injected into the hemocoel of each fifth instar (L5) larva of S. exigua using a microsyringe (Hamilton, Reno, NE, United States). After incubation for 2, 4, 6, 12, and 18 h at room temperature, hemolymph was collected. The collected 50 μL of hemolymph was spread onto TSB plates. After 18 h culture at 28°C, the number of colonies was counted. In addition, total RNAs were extracted from the larvae injected with the same amount of bacteria used for the bacterial growth analysis. An experimental unit for the RNA extraction was a whole body at each time point. This experiment was independently repeated three times using different larvae. Extracted RNAs were used for RT-qPCR as described above.

Test bacteria were cultured in 1 L of TSB for 72 h at 28°C with a shaking speed of 180 rpm. The bacterial culture broth was centrifuged at 12,500× g for 20 min at 4°C, and the resulting supernatant was mixed with 1 L of ethyl acetate. Using a separate funnel, the organic phase was collected. The aqueous phase was used two times to extract the organic phase, as described above. The combined 3 L extract of the organic phase was dried using a rotary evaporator (N-1110 Eyela, Tokyo, Japan) at 30°C. The resulting dried pellet that contained the metabolites was weighed and resuspended with DMSO to a concentration of 100 ppm.

The organic extract, as described above, was mixed with an internal standard, GXP-A*, and dissolved in DMSO to a final concentration of 5 μM. The samples were filtered through a disposable membrane filter unit (PVDF, 0.2 μm pore size, Thermo Fisher Scientific), followed by 1:1 dilution with methanol prior to LC-MS analysis. For the calibration curves, six-point calibration solutions of four GXP mixtures were prepared from 0 to 100 μM with DMSO. The LC-MS/MS analysis was performed using a Prominence 20AD series HPLC (Shimadzu, Japan) coupled with a 5600+ TripleTOF electrospray ionization triple quadrupole-time of flight mass spectrometer (AB Sciex, Framingham, MA, United States). For determination of the GXPs, analytes were injected into a KINETEX F5 column (2.6 μm, 100 × 3 mm i.d., Phenomenex, Torrance, CA, United States) with a binary gradient consisting of 20 mM ammonium formate and 80/20 of acetonitrile/20 mM ammonium formate for the mobile phases A and B, respectively, both with 0.1% formic acid. Separation was obtained at a flow rate of 400 μL/min with a linear gradient elution from 95:5 (A:B) to 20:80 (A:B) for 7 min, followed by a 4 min re-equilibration to the initial condition. The injection volume was 2 μL, and the ion transitions (m/z) were as follows: GXP-A (586.4 > 473.3), GXP-B (600.4 > 487.3), GXP-C (552.4 > 439.3), GXP-D (566.4 > 453.2), and GXP-A* (593.30 > 480.3). The source voltage and temperature were 5.5 kV and 500°C, respectively. Other ionization and fragmentation parameters were carefully optimized by monitoring the MS signal prior to sample analysis.

The first 600 bp of the gxpS open reading frame (ORF) of P. temperata temperata was amplified with a forward primer (CATATGATGAAAGACAGTATTACCAG), which contained an Nde I restriction site (underlined) along with a start codon, and the reverse primer (CTGCAG CATGATATACGCCGGCCCGG), with a Pst I restriction site (underlined). The resulting PCR product was cloned into a pCEP plasmid (Bode et al., 2015) and transformed using E. coli S17. For conjugation using E. coli S17 as a donor to P. temperata temperata, both bacterial cells were grown in Luria–Bertani (LB) broth until reaching an optical density of 0.6–0.7 at 600 nm. Transformed cells (20 uL) of E. coli S17 were co-cultured with P. tempera temperata (60 uL) and spread onto an LB agar plate followed by incubating at 30°C. The cultured bacterial colonies exhibiting a red color (characteristic of P. temperata temperate colonies) were streaked onto selective LB agar containing kanamycin and incubated at 30°C for 2 days. Single colonies were analyzed by PCR with specific primers for the mutant P. temperata temperata (Supplementary Table S2). Each end of the resulting 5,823 bp product was sequenced to confirm the mutant (Supplementary Figure S2).

The mutant stock was streaked on TSB plates with the addition of 100 ppm kanamycin and cultured for over 18 h at 28°C. Subsequently, a single colony was cultured in 10 mL of TSB with 100 ppm kanamycin at 180 rpm and 28°C for 18–20 h. The cultured broth was added to 1 L of TSB and cultured until the OD at 600 nm reached 0.6. At this point (−10 h after the initial culture), l-arabinose (40% stock) was added to a final concentration of 0.2% to induce overexpression. The culture was further incubated at 180 rpm and 28°C for 72 h, including preculturing time, before the arabinose addition. For the uninduced bacterial culture, the same media and conditions were used for culturing the mutant bacteria, except for the addition of l-arabinose.

Hemocytic nodules, which are formed in response to the bacterial infection, were assessed using 3 days-old L5 larvae of S. exigua. The larva was injected with 1 μL of overnight-grown E. coli (5 × 10⁷ cells/mL) and 1 μL of bacterial extracts (100 ppm). A microsyringe (Hamilton, Reno, NV, United States) was used for the infection by thrusting the needle into the larval hemocoel through the proleg. The injected larvae were then incubated at 25°C for 8 h. After incubation, the larvae were dissected to count the melanized nodules under a microscope (Stemi SV11, Zeiss, Jena, Germany) at 50× magnification. Each treatment was replicated three times and consisted of five larvae per replication.

Hemolymph (approximately 250 μL) was collected from L5 larvae of S. exigua by cutting the proleg and mixing it with 350 μL of ice-cold ACB. The mixture was then subjected to centrifugation at 1,000× g for 2 min, and 400 μL of the supernatant was removed. The remaining hemocyte pellet was resuspended in TC-100 insect culture medium (HyClone, Daegu, Republic of Korea). A total of 10 μL of the reaction mixture, consisting of 9 μL of the hemocyte suspension and 1 μL of the bacterial extract (100 ppm), was mounted onto a glass slide. After incubation at room temperature under darkness for 30 min, hemocytes were observed under a phase contrast microscope (DM2500, Leica, Wetzlar, Germany) at 400× magnification. The spread hemocytes were characterized by cytoplasmic extensions beyond the cell boundary. Each treatment was replicated three times using independent hemocyte preparations. In each replication, 100 hemocytes were randomly selected for counting the spread hemocytes.

Hemolymph was collected from L5 larvae of S. exigua and separated into hemocytes and plasma, as described above. Phenoloxidase (PO) activity in the plasma was measured using L-3,4-dihydroxyphenylalanine (DOPA) as a substrate. The 200 μL reaction mixture consisted of 10 μL of plasma, 10 μL of DOPA, 2 μL of bacterial metabolites (100 ppm), and 178 μL of PBS. The absorbance (ABS) of the reaction mixture was assessed at 495 nm using a VICTOR multi-label plate reader (PerkinElmer, Waltham, MA, United States). PO activity was quantified as ABS/min/mL. Each treatment was replicated three times.

An MTT assay was conducted using Sf9 cells according to the previously described protocol (Boonsuepsakul et al., 2008). Sf9 cells were seeded into 96-well plates at a density of 1.2 × 10⁴ cells per well and incubated for 24 h at 28°C. The cells were then treated with different bacterial metabolites and further incubated for 24 h at 28°C. After adding 10 μL of MTT solution (5 mg/mL in PBS) to each well, the cells were cultured for an additional 8 h at 28°C. Viable cells produced purple formazan granules dissolved in 50 μL of DMSO. The ABS of the resulting solution was measured at 570 nm using a microplate reader (Victor Multi-label Plate Reader, PerkinElmer).

The in-situ Cell Death Detection Kit from Abcam (Cambridge, United Kingdom) was used to perform the dUTP nick end labeling (TUNEL) assay on hemocytes of S. exigua L5 larvae. Larvae were injected with 2 μL of bacterial metabolite (100 ppm) and incubated for 18 h at 25°C. To prepare the hemocyte suspension, hemolymph was obtained from five or six L5 larvae and diluted in 300 μL of ACB. The sample was incubated on ice for 30 min before being replaced with TC-100 insect cell culture medium. For the assay, a reaction mixture was prepared by combining 10 μL of hemocyte suspension with 1 μL of a 10 μM solution of 5-bromo-2′-deoxyuridine (BrdU) containing terminal deoxynucleotidyl transferase (TdT). After the reaction mixture was prepared, it was placed on a cover glass in a wet chamber. To fix the cells, 2% paraformaldehyde was added, and the sample was incubated for 15 min. The cells were then washed with PBS and permeabilized with 0.3% Triton-X in PBS for 2 min at room temperature. To block non-specific binding sites, the cells were incubated with 4% bovine serum albumin in PBS for 10 min. Mouse anti-BrdU antibody (diluted 1:15 in blocking solution) was added, and the cells were incubated for 1 h at room temperature. After washing out the unbound anti-BrdU antibody, the FITC-conjugated anti-mouse IgG antibody (diluted 1:300 in blocking solution) was added, and the cells were incubated for 1 h at room temperature. DAPI (diluted 1:1,000 in PBS) was added to visualize the nuclei, and the cells were incubated at room temperature for 5 min. The cells were then washed with PBS, and a mixture of glycerol and PBS (1:1) solution was added to the cells on the cover glass. The cover glass was placed onto a glass slide, and the cells were observed under a fluorescence microscope (DM2500, Leica, Wetzlar, Germany) in FITC mode.

For this bioassay, L4 larvae of S. exigua were injected with freshly cultured bacteria at 1.5 × 10² colony-forming unit (CFU)/larva using a microsyringe, as described above. Mortality was then measured at 72 h after the bacteria injection. Each treatment used 10 larvae and was replicated three times.

Virulence data were analyzed by arsine transformation (ANOVA). All assay data were analyzed using PROC GLM of the SAS program (SAS Institute Inc., 1989). All data were plotted as the mean ± standard error using Sigma Plot (Systat Software, Point Richmond, CA, United States). The least squared difference (LSD) test was used to compare the means with a type I error of 0.05.

A full genome (GenBank accession number: PRJNA217865) of P. temperata temperata Meg1 encodes 4,132 genes, in which 10 NRPS genes were predicted (Figure 1A). They were not clustered but located in different locations on the genome. Each NRPS gene had commonly adenylation, condensation, peptide carrier protein, and thioesterase domains (Figure 1B). However, their product peptides were predicted to be different. Among these NRPS genes, NRPS10 was predicted to produce a pentapeptide (=GXP) and was called GXP synthetase (gxpS) in this bacterium. This gxpS shared amino acid sequences with those of other gxpS orthologs and was distinct from other NRPS genes (Figure 1C). Furthermore, the gxpS of the bacteria showed the highest sequence homology with NRPS10 of P. luminescens laumondii TT01 (Supplementary Table S3), which was originally identified as gxpS (Nollmann et al., 2015).

The gxpS expression was monitored at different culture periods during bacterial growth in TSB (Figure 2A) and S. exigua (Figure 2B). When grown in TSB, the bacteria exhibited exponential growth after an initial 6 h lag phase and reached stationary phase after 24 h. The gxpS exhibited a basal expression level during the exponential bacterial growth phase but rapidly increased its expression during the stationary phase. In S. exigua, the bacterial growth profile was similar to that in TSB, except a faster bacterial growth rate. Interestingly, gxpS was expressed at the early exponential stage in S. exigua.

GXP production was assessed from the culture broth of P. temperata temperata using four different GXP standards (GXP-A to GXP-D). These four standards were well separated in the pentafluorophenyl (F5) stationary phase, and their chemical identities were confirmed from MS/MS analyses (Figure 2C). Under the same LC-MS/MS analytical conditions, the four GXPs were detected in the culture broth extract of P. temperata temperata (Figure 2D). GXP-A was the most abundant among the four GXPs in the bacterial extract, with a concentration of approximately 135 mg/L. This value was over 7-fold higher than those of the other three GXP components.

GXP-A production was assessed in different Photorhabdus and Xenorhabdus bacteria (Figure 2E). Although it was detected in all test bacteria, it was produced significantly (F = 19.80; df = 5, 12; p < 0.0001) higher amounts in Photorhabdus than those in Xenorhabdus. Among the Photorhabdus, P. temperata temperata produced 2- to 3-fold more GXP-A than that of the other Photorhabdus species.

To confirm the role of gxpS expression in the production of GXP in P. temperata temperata, its mutant was constructed by replacing the original bacterial promoter with an inducible arabinose promoter (Figure 3A). A partial (600 bp) open reading frame at the 5′ end was cloned into a pCEP vector with an arabinose promoter. The recombinant vector was transferred to the wild type of P. temperata temperata through conjugation and inserted into the bacterial genome by homologous recombination. The mutant was confirmed by long-range PCR, which produced 5,823 bp covering the insertion sites (Figure 3B). The PCR product was sequenced and showed the insertion sites in addition to the vector (Supplementary Figure S2). The mutant bacteria showed altered expression levels of gxpS according to the presence of the inducer, l-arabinose (Figure 3C). Without the inducer, the mutant showed a significantly (F = 15.55; df = 1, 4; p = 0.0169) lower level of gxpS expression compared to that of the wild type. In contrast, the addition of the inducer significantly (F = 24.49; df = 1, 4; p = 0.0078) upregulated gxpS expression compared to the wild type.

Inducible gxpS expression led to the modulation of GXP production in the mutant bacteria (Figure 3D). Without the inducer, the mutant produced a significantly (F = 97.49; df = 1, 4; p = 0.0006) lower GXP-A amount compared to that of the wild type. In contrast, the addition of the inducer significantly (F = 12.13; df = 1, 4; p = 0.0253) increased GXP-A production compared to the wild type (Figure 3E). Similar modulations were detected in the other three GXPs in the mutant bacteria.

The metabolite extracts of the wild-type and mutant bacteria contained different amounts of GXPs. All bacterial extracts significantly suppressed the cellular immune response, as measured by hemocyte-spreading behavior (Figure 4A). However, there was variation among the bacterial extracts, in which the mutant extract (−ARA) without the inducer was less potent in inhibiting the immune responses compared to the wild-type extract, while the mutant extract (+ARA) with the addition of the inducer became more potent. Similar phenomena were observed in the assessments of other immune responses as measured by nodule formation (Figure 4B) and PO activity (Figure 4C).

Both wild-type and mutant bacterial extracts exhibited cytotoxicity against insect cells (Figure 5). Using an MTT test, the bacterial extracts were applied to Sf9 cells, which resulted in significant cytotoxicity in a dose-dependent manner (Figure 5A). However, the cytotoxicity was varied among the bacterial extract, in which the mutant extract (−ARA) without the inducer exhibited less potent cytotoxicity compared to the wild-type extract, whereas the mutant extract (+ARA) with the addition of the inducer became more potent.

Cytotoxic activity was also detected in hemocytes through apoptosis. A TUNEL assay showed that the bacterial extracts induced DNA fragmentation (see BrdU-positive cells) in the hemocytes (Figure 5B). However, the apoptotic activity was varied among the bacterial extract, in which the mutant extract (−ARA) without the inducer exhibited less potent apoptotic activity compared to the wild-type extract, whereas the mutant extract (+ARA) with the addition of the inducer became more potent.

Bacterial virulence was assessed between the wild-type and the mutant bacteria against S. exigua by injecting the bacteria into the larval hemocoel (Figure 6). Arabonose is rarely detected in insects. Exogenous arabinose to the insect was not likely to be used by the injected bacteria. Thus, the experiment should assess the gxpS deletion mutant because no arabinose did not induce this gene expression. To rescue the mutant, this study used the exogenous addition of GXP-A. All bacteria, including the wild type, resulted in significant insecticidal activity within 3 days. However, the mutant showed significantly lower insecticidal activities compared to those of the wild type (Figure 6A). When the mutant was injected along with GXP-A at 10 μg/larva, the insecticidal activity was significantly increased in a dose-dependent manner (Figure 6B). GXP-A was produced in TSB at 140 mg/L. S. exigua larvae at fifth instar have −100 μL of hemolymph. Thus, 14 μg of GXP-A may be produced in the larvae infected with these bacteria. Thus, we used the test dose of 10 μg/larva.