The P. fluorescens RG11 strain, which was previously isolated from the roots of the V. vinifera Red Globe grape cultivar, was used in this study. The root surfaces were sterilized to ensure that bacteria were isolated from internal tissues. A preliminary test showed that RG11 was able to produce melatonin in vitro and P. fluorescens was identified using morphological and biochemical characteristics according to Bergey’s Manual of Determinative Bacteriology (Krieg and Holt, 1984) and using a 16S rRNA sequencing analysis. The P. fluorescens 16S rDNA gene sequence (GenBank accession no. KY172955) was amplified using the universal bacterial primer pairs 27F/1492R in accordance with previously described conditions (Bai et al., 2002), and a neighbor-joining dendrogram generated by the program MEGA 6.06 indicated close similarity to P. fluorescens (Figure 1).

The RG11 strain was maintained at 4°C on nutrient agar slants (3 g L^-1 beef extract, 10 g L^-1 tryptone, 5 g L^-1 NaCl, and 20 g L^-1 agar; pH 7.4). To prepare inoculum cultures, cells from one colony were transferred to 50 mL nutrient broth medium and cultivated for 12–16 h at 30°C with an agitation of 150 rpm to a final OD₆₀₀ of approximately 1.0. The cultures were then centrifuged at 6,000 × g for 10 min, re-suspended, and standardized to 1 × 10⁸ cells mL^-1 in 0.9% sterilized saline solution using a Petroff–Hausser counting chamber.

To characterize melatonin biosynthesis in RG11, 1 mL of standardized bacterial inoculum was added to individual 100-mL brown bottles that contained 50 mL of nutrient broth (3 g L^-1 beef extract, 10 g L^-1 tryptone, and 5 g L^-1 NaCl; pH 7.4) with 200 mg L^-1 15N double-labeled L-tryptophan. The cultures were incubated in a rotary shaker at 28°C with a rotational speed of 150 rpm in the dark under aerobic conditions. Samples were collected every 6 h for viable cell counts using the plate counting method and appropriate dilutions. A 1-mL aliquot of the bacterial culture was centrifuged, and the supernatant was diluted 1:1 with methanol. The resulting mixture was passed through a 0.22-μm filter, and was then used for the detection and quantification of ¹⁵N-tryptamine, ¹⁵N-5-hydroxytryptophan, ¹⁵N-serotonin, ¹⁵N-N-acetylserotonin, and ¹⁵N-melatonin using UPLC-MS/MS. The concentrations of these compounds were determined using an eight-point calibration curve of non-isotopic labeling standards as described by Jiao et al. (2016). The experiment was repeated in triplicate.

Vitis vinifera ‘Red Globe grape’ plantlets, obtained from an in vitro culture system, were grown in sterilized 500-mL glass bottles filled with 1/2 MS modified basal salt mixture (PhytoTechnology Laboratories, Shawnee Mission, KS, USA), which contained 3% sucrose, 0.6% agar, and 0.2 mg L^-1 indolebutyric acid. Plantlets were grown at 28°C with a with a 16-h light/8-h dark cycle for 6 weeks. Plantlets of approximately 8 cm in height and with six leaves were selected and standardized to four 5-cm adventitious roots using sterilized scissors. The roots of some plantlets were immersed in the RG11 bacterial inoculum for 1 min (E+), whereas others were immersed in 0.9% sterile saline solution (E-). All plantlets were transferred to 500-mL culture bottles that contained 150 g sterile nutrient soil (Pindstrup, Ryomgaard, Denmark) and 40 mL nutrient-rich water that was prepared from 1/2 MS. The bottles were randomly placed in a controlled chamber at 26°C with 70% humidity under a 16-h light/8-h dark cycle, and plantlets were irrigated with 5 mL distilled water every 2 days.

After 20 days, 12 plantlets from each group (E+ and E-) were randomly selected for root length, root fresh weight (FW), lateral root number, plant height, and chlorophyll content measurements. The chlorophyll content of fully expanded leaves was analyzed using a chlorophyll ELISA kit, according to the manufacturer’s instructions (Lvyuan, Beijing, China). Both E+ and E- groups were divided into two subgroups (at least 84 plantlets each), and were watered either normally or with 20 mL of 80 mM NaCl solution. Roots were collected from 12 randomly selected plantlets from each subgroup between 9:00 AM and 10:00 AM daily. The roots were ground into powder in liquid nitrogen in individual mortars, and 0.5 g samples were extracted with 2 mL methanol as previously described (Boccalandro et al., 2011). The extracts were mixed with 2 mL ultrapure water, centrifuged, passed through a 0.22-μm filter, and stored in amber vials for the detection and quantification of 5-hydroxytryptophan, tryptamine, serotonin, N-acetylserotonin, and melatonin with UPLC-MS/MS. Other samples were used for the determination of MDA content and H₂O₂ accumulation.

To investigate whether the grape genotype could influence bacterial colonization and melatonin production, RG11 was used to inoculate the roots of three additional grape cultivars, including Riesling, Chardonnay, and Cabernet Sauvignon. We used the procedure described above with the following two modifications: (i) after measuring the growth attributes of each plantlet, the whole root was collected to assess bacterial colonization ability; and (ii) after exposure to 80 mM NaCl stress for 4 days, the plantlet roots were sampled between 9:00 AM and 10:00 AM for subsequent UPLC-MS/MS analyses. To assess bacterial colonization ability, roots were surface-sterilized with 70% ethanol for 3 min, soaked in sodium hypochlorite (3% available chlorine) for 2 min, and rinsed three times with sterile water. Then, 0.5 g of the root samples were ground and homogenized in 1.5 mL ice-cold phosphate-buffer saline solution using sterile quartz sand in individual mortars. The homogenates were serially diluted, spread, and incubated on Luria–Bertani agar at 30°C for 2–3 days to determine viable bacterial counts.

After exposure to 80 mM NaCl stress for 4 days, 10 g of root samples from E- Red Globe, Riesling, Chardonnay, and Cabernet Sauvignon plantlets were collected, washed three times with sterile H₂O, ground, and homogenized in 20 mL sterile H₂O using sterile quartz sand in individual mortars. The homogenates were centrifuged at 6000 × g for 10 min, and the supernatants were filtered through 0.22-μm membrane filters (Millipore, Billerica, MA, USA). The extraction sterility was tested by plating 100 μL of extract on nutrient agar and incubating at 37°C for 24 h. A 0.25-mL sample from each standardized bacterial culture (1 × 10⁸ cells mL^-1) was used to inoculate 25 mL nutrient broth containing 200 mg L^-1 L-tryptophan and 20% (v/v) sterile root extract. The cultures were incubated as described above and sampled every 6 h to measure melatonin production.

Fresh root samples were ground into a powder in liquid nitrogen, and H₂O₂ accumulation was quantified using a hydrogen peroxide assay kit (Beyotime, Shanghai, China), according to the manufacturer’s instructions. In brief, 0.1 g of root powder was homogenized in 1 mL phosphate buffer (50 mM; pH 6.0) at 4°C. The supernatant (50 μL) was mixed with 100 μL of test solution and placed at 30°C for 30 min. The absorbance was measured with a spectrometer at 560 nm and calibrated to a standard curve generated using known H₂O₂ concentrations.

In addition, 0.5 g of root powder was extracted with 2 mL of 10% (w/v) TCA to determine the MDA content of the grape roots as previously described (Zhao et al., 2013) with some modifications. After centrifugation at 8,000 × g for 10 min, 0.2 mL of the supernatant was added to the same volume of 0.5% (w/v) thiobarbituric acid in 20% (w/v) TCA, and was then heated at 100°C for 20 min. Reactions were stopped on ice. After centrifugation at 8,000 × g for 10 min, the absorbance was measured with a spectrometer at 440 nm, 532 nm, and 600 nm, and the MDA content (nmol g^-1) was calculated as follows: [6.45 × (A532 – A600) – 0.56 × A450] × V/W, where V (mL) is the volume of the tissue extract, and W (g) is the FW.

All standard reference materials were purchased from Sigma-Aldrich (St. Louis, MO, USA), and 15N double-labeled L-tryptophan was obtained from Cambridge Isotope Laboratories (Andover, MA, USA). The reagents, including methanol and formic acid (HPLC grade), were purchased from Merck (Darmstadt, Germany). Stock solutions were prepared by dissolving 10 mg of each standard in 1 mL of methanol under dim light conditions, and the stocks were then stored at -80°C to prevent degradation. Fresh working solutions were prepared in a 50:50 (v/v) solution of methanol:water.

Quantitative detection was conducted using a triple quadrupole UPLC-MS/MS (Agilent, Santa Clara, CA, USA). In the solvent system, eluent A was composed of Milli-Q water containing 0.1% (v/v) formic acid, and eluent B was methanol. Analyte separation was accomplished using an Agilent ZORBAX Eclipse XDB-C18 Rapid Resolution HT column (1.8 μm, 3.0 mm × 50 mm) at 42°C with linear elution gradient protocols of 0–6 min for 5–55% B and 6–15 min for 55–100% B at a flow rate of 0.2 mL min^-1. Then, 100% B was kept constant for 2 min, and the column was re-equilibrated for 5 min using the initial solvent composition. The injection volume was 1 μL. Quantitation was determined using the multiple reactions monitoring mode under unit mass-resolution conditions (L-tryptophan m/z⁺ 205→188; ¹⁵N-L-tryptophan m/z⁺ 207→189; tryptamine m/z⁺ 161→144; ¹⁵N-tryptamine m/z⁺ 163→145; 5-hydroxytryptophan m/z⁺ 221→204; ¹⁵N-5-hydroxytryptophan m/z⁺ 223→205; serotonin m/z⁺ 177→160; ¹⁵N-serotonin m/z⁺ 179→161; N-acetylserotonin m/z⁺ 219→160; ¹⁵N-N-acetylserotonin m/z⁺ 220→161; melatonin m/z⁺ 233→174; and ¹⁵N-melatonin m/z⁺ 235→175).

The expression levels of the grapevine tryptophan decarboxylase gene (VvTDC1) and the serotonin N-acetyltransferase gene (VvSNAT) in grape roots, previously predicted by Byeon et al. (2014a) and Jiao et al. (2016), were analyzed using quantitative real-time PCR (qRT-PCR). Root tissues were washed three times with sterile H₂O, and RNA was extracted using the cetyl trimethylammonium bromide method (Reid et al., 2006). The RNA quantity was determined by measuring the A260 and A280 (NanoDrop^® ND-1000; Thermo Scientific, Wilmington, DE, USA). RNA was reverse transcribed into cDNA using the PrimeScript RT reagent kit with gDNA Eraser (TaKaRa, Dalian, China). qRT-PCR was conducted using a Roche 480 light cycler system with SYBR Fast qPCR Mix (TaKaRa) under the following conditions: 95°C for 30 s; followed by 40 cycles at 95°C for 5 s, 60°C for 10 s, and 72°C for 15 s. The melting curve analysis was performed at 95°C for 5 s, 60°C for 1 min, 95°C continuously, and then 50°C for 30 s. The primers used in this study are listed in Table 1. EF1-α was used as an internal reference to calculate the relative expression of each gene. All qRT-PCR reactions were performed in triplicate.

A correlation analysis based on a simple linear regression was performed on the assayed variables at the 95% confidence level between the relative expression of VvTDC1 and VvSNAT and the MDA content. Data were expressed as the average of 3–12 replicates ± standard deviation. One-way analysis of variance (ANOVA) was performed in conjunction with Tukey’s test to detect melatonin intermediates in the E- roots of the four grape cultivars, and Student’s t-test was used to identify significant differences (P < 0.05) in bacterial colonization between treatments. All statistical analyses were performed using SPSS 19.0 (IBM, Armonk, NY, USA).

The different melatonin synthetic pathways in animals and plants are presented in Figure 2A, and the chemical structures and chromatograms of six unlabeled standards (40–60 ng mL^-1) and their corresponding ¹⁵N-metabolites are presented in Figures 2B,C. If tryptophan hydroxylase and tryptophan decarboxylase (or enzymes with similar activity) were present in P. fluorescens RG11, ¹⁵N-tryptophan could be converted to ¹⁵N-5-hydroxytryptophan (m/z⁺ 223→205) and ¹⁵N-tryptamine (m/z⁺ 163→145). A peak was obtained at m/z⁺ 223→205 with an identical retention time as that of the unlabeled 5-hydroxytryptophan standard, indicating that the compound was ¹⁵N-5-hydroxytryptophan. However, after RG11 was incubated with ¹⁵N-tryptophan, no peak was detected at the correct retention time for ¹⁵N-tryptamine at any time point of m/z⁺ 163→145 (Figure 2C). These findings suggest that 5-hydroxytryptophan might be a key intermediate in the melatonin biosynthesis pathway of P. fluorescens RG11. Similarly, the ¹⁵N-serotonin, ¹⁵N-N-acetylserotonin, and ¹⁵N-melatonin were all detected and confirmed using the reference compounds.

All of the ¹⁵N-metabolites were detectable 6 h post-incubation, and their levels showed a progressive increase over time (Figure 3). ¹⁵N-5-hydroxytryptophan, ¹⁵N-serotonin, and ¹⁵N-N-acetylserotonin were found at relatively higher concentrations that reached maximum values of 18.06 ± 1.14 ng mL^-1, 8.28 ± 0.65 ng mL^-1, and 8.66 ± 0.82 ng mL^-1 at 36 h post-incubation, respectively. The concentrations of ¹⁵N-melatonin reached a maximum value of 1.32 ± 0.12 ng mL^-1 at 30 h post-incubation and declined slightly thereafter. When the results were expressed in ng 10^-11 viable cells, the production capacity for all metabolites peaked at 6 h post-incubation (cell number: 9.06 log 10 CFU mL^-1) and subsequently declined with the increasing cell density (final cell number: 11.52 log 10 CFU mL^-1).

The effects of P. fluorescens RG11 on the levels of endogenous melatonin and its intermediates in the roots of Red Globe plantlets under salt stress are shown in Figure 4. The results show that the production of melatonin and its intermediates first increased in the roots of both E+ and E- plantlets and then decreased over time. The levels of 5-hydroxytryptophan, N-acetylserotonin, and melatonin in the roots of E+ plantlets were higher than those in the roots of E- plantlets, and the opposite trend was observed for levels of tryptamine and serotonin. The roots of E+ plantlets produced approximately 19.92–26.01% higher melatonin than those of E- plantlets between day 2 and day 6 of the salt stress treatment. The maximum value of melatonin in the roots of E+ plantlets (401.31 ± 37.78 pg g^-1 FW) was found on day 4 of the salt stress treatment, whereas the maximum melatonin value of the roots of E- plantlets (326.66 ± 23.40 pg g^-1 FW) was observed on day 5 of the salt stress treatment. Similar trends were also observed for 5-hydroxytryptophan and N-acetylserotonin, since their levels in the roots of E+ plantlets were 13.21–24.69% and 12.40–23.84% higher, respectively, compared with those in the roots of E- plantlets. These results indicate that the biosynthesis of melatonin in RG11-colonized plants responded earlier to salt stress, and the melatonin levels increased much more than E- plants. Conversely, the levels of tryptamine and serotonin in the roots of E+ plantlets were lower than those in the roots of E- plants, with a rate of decline of 12.10–21.98% and 6.23–15.54% between day 2 and day 6 of the salt stress treatment, respectively.

The levels of MDA and H₂O₂ were similar in the roots of E+ and E- plantlets, but strongly increased after the salt stress treatment in all plantlets (Figure 5). However, E+ plantlets had levels that were 15.07–20.76% and 14.29–27.59% (P < 0.05) lower in MDA and H₂O₂, respectively, compared to E- plantlets between day 2 and day 6 of the salt stress treatment.

The expression profiles of VvTDC1 and VvSNAT in the roots of salt-treated plantlets are shown in Figure 6A. The qRT-PCR analysis demonstrated that the expression of VvTDC1 and VvSNAT first increased in the roots of all plantlets, and then decreased on day 5 or day 6 of the salt stress treatment. Changes in the expression of VvTDC1 and VvSNAT showed that the response of E+ plantlets to salt stress was slower and weaker compared to that of E- plantlets. However, the levels of endogenous melatonin in the roots of E+ plantlets were higher than those in the roots of E- plantlets.

Regression analyses showed a higher linear correlation coefficient between the relative expression of VvTDC1 and the MDA content (R² = 0.9759 and 0.9404) in the roots of E+ and E- plantlets than that of VvSNAT (R² = 0.7574 and 0.6853; Figure 6B). The results indicate that the melatonin synthesis genes, especially VvTDC1, might be induced by salt stress, and that the transcript levels were highly correlated with stress-induced oxidative damage.

Different grape hosts were used to assess the root colonization capacity of RG11, as well as its ability to promote plant growth. Root colonization of the inoculated endophyte is considered a prerequisite for successful growth promotion. We found that at 20 days post-inoculation, this strain successfully colonized the roots of all grape cultivars, and it was able to establish endophytic populations within the different cultivars with a density of approximately 5.44–6.05 Log 10 CFU g^-1 FW.

Further, colonization with RG11 beneficially improved the growth of the grape plants as observed in the significant increase of growth attributes, although the extent of improvement varied between grape varieties (Table 2; Figure 7A). The increased ratios of root FW, total root length, and the number of lateral roots in relation to control plants ranged from approximately 25.98–38.85%, 18.51–44.41%, and 41.51–63.64%, respectively, and the highest ratio was found in the Red Globe cultivar. On the other hand, Cabernet Sauvignon displayed the highest increase in the ratio of plant height as well as in the chlorophyll content in leaves as compared to the other three varieties. Thus, it could be inferred that RG11 possesses a fairly broad plant growth promoting ability in different grape cultivars.

The maximum difference in the levels of melatonin in E+ and E- Red Globe plantlets was detected at approximately day 4 after the start of the salt stress treatment (Figure 4), so the roots of E+ and E- plantlets were collected at this time for UPLC-MS/MS (Figure 7B). No significant differences were identified in the levels of melatonin between E+ and E- plantlets within each cultivar. However, all plants responded to salt stress by synthesizing melatonin, and P. fluorescens RG11 colonization increased the up-regulation of endogenous melatonin levels in the E+ plantlets of all cultivars as compared to E- plantlets. The levels of melatonin were higher (61.11%) in the roots of E+ Cabernet Sauvignon plantlets than in the roots of E- plantlets. Similar trends were observed in Red Globe, Riesling, and Chardonnay, although the increase was relatively lower. Therefore, P. fluorescens RG11 colonization might enhance the synthesis of melatonin in the roots of grape plantlets, especially in Cabernet Sauvignon, under salt stress conditions.

To further understand the relatively higher increase in melatonin production by RG11 colonization in Cabernet Sauvignon than in the other three varieties, we analyzed the levels of melatonin biosynthesis intermediates in the E- roots of the four grape cultivars (Figure 8A). At day 4 of the salt stress treatment, the levels of melatonin intermediates in the E- roots were different among the four grape cultivars: 894.55–1459.84 ng g^-1 FW for tryptamine, 35.48–61.23 ng^-1 FW for 5-hydroxytryptophan, 228.47–406.27 ng^-1 FW for serotonin, and 82.87–126.65 ng^-1 FW for N-acetylserotonin. Cabernet Sauvignon had the highest levels of tryptamine, 5-hydroxytryptophan, serotonin, and N-acetylserotonin, and these respective levels were 10.42–63.19%, 4.10–72.58%, 33.11–78.07%, and 15.42–52.83% higher than those in the roots of Chardonnay, Riesling, and Red Globe. These results suggest that genetic traits of the cultivars may noticeably influence the levels of melatonin intermediates in grapes.

The addition of root extracts collected from different cultivars significantly enhanced the melatonin synthesis of RG11 in vitro compared to the control (Figure 8B). The melatonin levels initially increased, but then slightly decreased in all cultures. The highest melatonin levels were detected after 24 h post-inoculation in control cultures and 30 h post-inoculation in cultures with root extracts. At 30 h post-inoculation, the levels of melatonin in cultures with root extracts from Red Globe, Riesling, Chardonnay, and Cabernet Sauvignon were higher by 5.75-fold, 10.11-fold, 8.19-fold, and 13.42-fold, respectively, as compared to the control (1.14 ± 0.16 ng mL^-1). Noticeably, the root extracts of Cabernet Sauvignon had the strongest capacity to promote the melatonin synthesis of RG11 in vitro.