An efficient direct screening system for microorganisms that activate plant immune responses based on plant–microbe interactions using cultured plant cells

Microorganisms that activate plant immune responses have attracted considerable attention as potential biocontrol agents in agriculture because they could reduce agrochemical use. However, conventional methods to screen for such microorganisms using whole plants and pathogens are generally laborious and time consuming. Here, we describe a general strategy using cultured plant cells to identify microorganisms that activate plant defense responses based on plant–microbe interactions. Microbial cells were incubated with tobacco BY-2 cells, followed by treatment with cryptogein, a proteinaceous elicitor of tobacco immune responses secreted by an oomycete. Cryptogein-induced production of reactive oxygen species (ROS) in BY-2 cells served as a marker to evaluate the potential of microorganisms to activate plant defense responses. Twenty-nine bacterial strains isolated from the interior of Brassica rapa var. perviridis plants were screened, and 8 strains that enhanced cryptogein-induced ROS production in BY-2 cells were selected. Following application of these strains to the root tip of Arabidopsis seedlings, two strains, Delftia sp. BR1R-2 and Arthrobacter sp. BR2S-6, were found to induce whole-plant resistance to bacterial pathogens (Pseudomonas syringae pv. tomato DC3000 and Pectobacterium carotovora subsp. carotovora NBRC 14082). Pathogen-induced expression of plant defense-related genes (PR-1, PR-5, and PDF1.2) was enhanced by the pretreatment with strain BR1R-2. This cell–cell interaction-based platform is readily applicable to large-scale screening for microorganisms that enhance plant defense responses under various environmental conditions.


Results
Isolation of bacteria from the interior of B. rapa var. perviridis. We isolated bacteria from the interior of B. rapa var. perviridis grown by organic farming without the use of pesticides. Roots, stems, and leaves were cut into small pieces and surface-sterilized using appropriate concentrations of sodium hypochlorite and ethanol 24 , as described in the Materials and Methods. After surface-sterilization, each tissue sample was rinsed with sterile water and placed on NBRC802 and ISP2 agar plates. The water used for rinsing was also spread onto each medium as a control. When no microorganisms appeared on medium for the control, that is, the surface had been sterilized, colonies that formed around the tissues were selected as putative endophytes. Using this isolation procedure, a total of 31 bacterial strains were isolated, of which 10 and 20 strains were derived from roots and stems, respectively, and 1 strain was derived from a leaf (Table 1). Taxonomic identification of these strains was performed based on 16S rDNA sequencing, and a phylogenetic tree of the sequences was constructed (Fig. 1). We found a variety of cultivable bacteria in the microbiome. These bacteria belonged to 9 different genera: Bacillus, Brevibacterium, Glutamicibacter, Arthrobacter, Paenarthrobacter, Agrobacterium, Delftia, Pseudomonas, and Stenotrophomonas (Table 1 and Fig. 1). Four strains (BR2L-1, BR3S-2, BR3S-8, and BR3S-10) exhibited low identity (< 96%) to previously reported sequences of typical strains, indicating that these strains might constitute new genera or species. The isolated bacteria were divided into 3 phyla, Firmicutes, Actinobacteria, and Proteobacteria (Fig. 1). Excluding potential human pathogenic bacteria (two Stenotrophomonas strains), the isolated bacteria were analyzed further.

ROS production induced by interaction between bacteria and cultured plant cells.
We first examined interactions between the isolated bacteria and cultured plant cells by monitoring ROS production. Although microbial components such as lipopolysaccharides and several metabolites reportedly induce ROS production in cultured plant cells 25,26 , few reports have examined ROS production induced by intact microbial cells 27 . Tobacco BY-2 cells were incubated with each strain of isolated bacteria, and ROS production was monitored using a chemiluminescence assay with luminol. Most of the bacteria (19 strains) had no effect on BY-2 www.nature.com/scientificreports/ cells during co-incubation, based on ROS production (Fig. S1a). Interestingly, however, 10 strains (BR1R-2, BR1R-5, BR2R-4, BR2S-3, BR2S-6, BR3S-3, BR3S-7, BR3S-8, BR3S-10, and BR3S-11) induced ROS production after approximately 80 min of co-incubation (Fig. S1b). These results suggest that intact bacteria can induce ROS production by plant cells via interactions. In order to determine whether the ROS was produced by the bacteria or cultured plant cells, the assays were repeated using cells killed by autoclave treatment. Incubation of autoclaved plant cells with intact Delftia sp. BR1R-2 cells resulted in no detectable ROS production. In contrast, incubation of intact plant cells with autoclaved bacteria resulted in a biphasic increase in ROS production. The first peak in ROS generation occurred after 40 min and was followed by a second peak that reached a maximum at approximately 160 min (Fig. 2), resembling the temporal pattern of cryptogein-triggered ROS production in tobacco BY-2 cells 18,19 . These results clearly demonstrate that bacteria act on BY-2 cells to induce ROS production. It is interesting to note that www.nature.com/scientificreports/ co-incubation with intact bacteria resulted in only one peak in ROS production, whereas co-incubation with autoclaved bacteria resulted in a biphasic increase in ROS. During co-incubation with intact strain BR1R-2 cells, some factor(s) derived from the bacteria might have scavenged ROS produced by the BY-2 cells. Considered collectively, these data indicate that this experimental system is useful for evaluating interactions between bacteria and cultured plant cells.

Screening for microorganisms that prime plant immune responses based on plant-microbe interactions using cultured plant cells.
We established an experimental system using intact bacteria and cultured plant cells to screen for microorganisms that prime plant immune responses. Cryptogein-induced ROS production in tobacco BY-2 cells was employed as a marker for the screening (Fig. S2). Buffer containing BY-2 cells was inoculated with culture solution of each isolated bacterial strain and incubated for 4 h. After the co-incubation, the cells were collected and suspended in fresh buffer to remove ROS scavengers and other bacteria-derived metabolites. Cryptogein, as an elicitor of plant immune responses, was then added to the buffer, and ROS production was monitored by chemiluminescence. We used Delftia sp. BR1R-2 to validate the screening system (Fig. 3). Incubation of only plant cells or bacteria resulted in low ROS production after cryptogein addition. In contrast, pre-incubation of BY-2 cells with BR1R-2 cells resulted in greatly enhanced cryptogein-induced ROS production. The amount of ROS produced by BY-2 cells after BR1R-2 treatment was three times that produced by BY-2 cells after mock treatment. These results indicate that strain BR1R-2 is suitable for priming the immune responses of BY-2 cells.
We also confirmed that these strains (with the exception of Bacillus sp. BR2S-4) enhanced ROS production in Arabidopsis T87 cells triggered by the plant immune response elicitor flg22, a 22-amino acid peptide derived from flagellin that is known to induce ROS production 28 (Figs. S5 and S6). These 7 strains were selected as candidate microorganisms for priming plant immune responses and then subjected to the second screening using whole plants.
Biocontrol activity of selected microorganisms. We examined the ability of the selected bacteria to enhance disease resistance using whole Arabidopsis plants. Plants were inoculated with each strain of selected bacteria by immersing the root tip of 7-day-old seedlings in the bacterial cell culture solution. After cultivation for an additional 7 days, we observed that plants inoculated with each of the bacterial strains were able to www.nature.com/scientificreports/ grow (Figs. S7 and S8). Plating extracts of surface-sterilized bacteria-inoculated plants on NBRC802 or ISP2 agar medium revealed that the bacteria colonized the interior of the Arabidopsis plants (Fig. 4). The number of bacteria ranged from 10 5 to 10 9 colony forming units (CFU) per gram of Arabidopsis, depending on the bacterial strain. Inoculation with Delftia sp. BR1R-2 or Arthrobacter sp. BR2S-6 did not affect plant growth, but inoculation with the other 5 strains resulted in a significant reduction in plant growth (Figs. S7 and S8). We also confirmed that strains BR1R-2 and BR2S-6 colonized the stems and leaves (Fig. S9), indicating that these strains spread from the roots to the aerial tissues of Arabidopsis as endophytes.
The promising endophytes Delftia sp. BR1R-2 and Arthrobacter sp. BR2S-6 were then tested for their ability to enhance disease resistance. Pseudomonas syringae pv. tomato DC3000 and Pectobacterium carotovorum subsp. carotovorum NBRC 14082 were used as hemibiotrophic and necrotrophic bacterial pathogens, respectively. Arabidopsis seedlings treated with each endophyte were cultivated for 7 days, and the plants were then challenged with P. syringae pv. tomato DC3000. After cultivation for an additional 3 days, we observed that mock-treated  www.nature.com/scientificreports/ plants exhibited severe disease symptoms of chlorosis (Fig. 5). In contrast, plants treated with strains BR1R-2 and BR2S-6 exhibited significantly less-severe disease symptoms compared with mock-treated plants (Fig. 5).
We also found that the density of strain DC3000 in Arabidopsis decreased to 0.9% and 7.4% in plants treated with strains BR1R-2 and BR2S-6, respectively, compared with mock-treated plants (Fig. S10). Similarly, although plants challenged with P. carotovorum subsp. carotovorum NBRC 14082 exhibited soft rot, pretreatment with strains BR1R-2 and BR2S-6 enhanced the disease resistance of Arabidopsis plants (Fig. 5). These results confirm that the microorganisms selected using the present screening system enhance the resistance of Arabidopsis plants to two different specific pathogens. The biocontrol effects of strain BR1R-2 were more pronounced than those of strain BR2S-6 under the experimental conditions used in this study (Fig. 5), and therefore was further validated.
Effects of colonization by Delftia sp. BR1R-2 on the expression of defense-related genes. In order to examine the mechanism by which Delftia sp. BR1R-2 enhances disease resistance in Arabidopsis, the expression patterns of various defense-related genes (PR-1, PR-5, and PDF1.2) in the aerial tissues of Arabidopsis plants were analyzed using reverse transcription-quantitative polymerase chain reaction (RT-qPCR). Activation of defense responses via the SA signaling pathway is accompanied by expression of PR-1 and PR-5, whereas PDF1.2 is a marker of the JA/ET signaling pathway 29,30 . We first evaluated the effects of BR1R-2 colonization on gene expression (Fig. 6, gray and yellow bars). RT-qPCR analysis revealed that colonization by strain BR1R-2 We then evaluated the effects of BR1R-2 colonization on pathogen-induced gene expression in Arabidopsis. Plants grown from BR1R-2-and mock-treated seedlings were challenged with P. syringae pv. tomato DC3000 (Fig. 6a, blue and 33,34 , to our knowledge, this is the first report describing the diversity of bacteria isolated from the interior parts of B. rapa var. perviridis. We isolated 9 Bacillus, 5 Pseudomonas, and 2 Stenotrophomonas strains. Bacteria of these genera have frequently been recovered from Brassicaceae plants.
In addition, we isolated 2 Brevibacterium, 3 Glutamicibacter, 3 Arthrobacter, and 2 Paenarthrobacter strains, as well as 1 Delftia strain. Interestingly, few reports have described bacteria of these genera in the microbiome of Brassicaceae. The number of bacterial strains isolated from roots and leaves were relatively small. This might be attributed to harsh conditions used for the surface sterilization in this study.
In this study, we developed a novel system for screening for microorganisms that activate plant immune responses based on plant-microbe interactions using cultured plant cells. The bacteria isolated from the interior of B. rapa var. perviridis plants were examined using the screening system with cryptogein-induced ROS production in tobacco BY-2 cells as a marker. A total of 8 bacterial strains were selected using this screening system (Figs. 3 and S3b). Interestingly, although 4 of these 8 strains (BR1R-3, BR2S-4, BR3S-1, and BR3S-9) did not induce BY-2 cells to produce ROS in the absence of cryptogein (Fig. S1), the 4 strains did enhance cryptogeininduced ROS production in the cultured plant cells. We also found that 7 of the 8 bacterial strains enhanced flg22-induced ROS production in Arabidopsis T87 cells (Fig. S6). Thus, using this screening system, bacteria belonging to a variety of genera within 3 phyla were selected as candidate microorganisms for priming plant immune responses. In this study, we subjected bacteria after 24-h cultivation to the assays in order to rapidly screen many bacterial strains. On the other hand, since cultivation time affects the growth phase, detailed examination of the time would be needed to optimize plant immunity-activating potential of each strain. It should also be noted that the developed method cannot select microorganisms that activate plant immune responses without enhancement of elicitor-induced ROS production.
Endophytes are generally preferable as biocontrol agents due to their inherent ability to stably colonize in the interior of plants. Characteristics such as motility, adhesion, and cell-wall degradation activity are reportedly required for such colonization 35,36 . We confirmed that 7 bacterial strains selected using the proposed screening system were capable of colonizing the interior of Arabidopsis plants (Fig. 4). The number of bacteria colonizing plants was relatively high (Fig. 4). The medium used here contained 10 g/l sucrose as a carbon source for the plant, which is likely a good carbon source for the bacteria as well. However, 5 of these 7 strains caused a significant reduction in plant growth (Figs. S7 and S8). This growth inhibition was not correlated with the number of bacteria colonizing the plants (Figs. 4 and S8). One possible explanation is that these strains induce defense responses too strongly. Strong induction of defense responses in plants is often accompanied by cell cycle arrest or growth inhibition 21,37,38 . In contrast, 2 of the 7 bacterial strains, Delftia sp. BR1R-2 and Arthrobacter sp. BR2S-6, colonized the interior of Arabidopsis plants without inhibiting their growth (Figs. S7 and S8). These two endophytes endowed Arabidopsis with resistance to both hemibiotrophic and necrotrophic bacterial pathogens (Fig. 5). Therefore, strains BR1R-2 and BR2S-6 could be useful biocontrol agents. Strain BR1R-2 is the first bacterium of the genus Delftia shown to function as a biocontrol agent and exhibited more pronounced biocontrol effects on Arabidopsis than strain BR2S-6 ( Fig. 5).
Delftia sp. BR1R-2 was further examined in order to elucidate the mechanism by which it enhances pathogen resistance in Arabidopsis. Nonpathogenic bacteria reportedly enhance disease resistance by stimulating plant defense-related genes, as described above. Here, we investigated the expression of PR-1 and PR-5, which are generally involved in the SA signaling pathway, and the expression of PDF1.2, which is involved in the JA/ET signaling pathway. Colonization by strain BR1R-2 induced the expression of all three genes (Fig. 6). These results suggest that strain BR1R-2 simultaneously activates the SA and JA/ET signaling pathways in Arabidopsis and that the resulting expression of defense-related genes provides resistance to two different pathogens. The biocontrol activity of most nonpathogenic bacteria involves stimulation of either pathway (primarily the JA/ET signaling pathway), whereas the number of bacteria that activate both pathways is limited 39 . For example, defense responses mediated by the rhizobacterium Bacillus cereus AR156 are dependent on both pathways 39 . Furthermore, the expression of PR-1, PR-5, and PDF1.2 induced by the pathogens in the present study was enhanced by pretreatment with strain BR1R-2 (Fig. 6). These results indicate that strain BR1R-2 enhances the pathogen resistance of Arabidopsis by priming its immune responses.
In conclusion, we described a general strategy for exploring the potential of microorganisms to activate plant immune responses based on plant-microbe interactions using cultured plant cells. The value of this strategy was demonstrated by identifying novel plant immunity-activating bacteria, Delftia sp. BR1R-2 and Arthrobacter sp. BR2S-6. The developed method using cultured plant cells enables rapid direct screening of microorganisms for plant immunity-activating potential, thus reducing the number of samples subjected to laborious assays using whole plants (Fig. 7). Therefore, this approach should be readily applicable to large-scale screening for plant immunity-activating microorganisms from a variety of environments.

Isolation and identification of bacteria from the interior of B. rapa var. perviridis. Brassica rapa
var. perviridis plants were grown by organic farming without the use of pesticides at the Suzuki Farm (Tachikawa, Tokyo, Japan) and collected between May and July 2017. The plants were separated into roots, stems, and leaves. The plant tissues were then washed with running tap water and aseptically sectioned into 1-cm fragments. These fragments were surface-sterilized by dipping in 5% sodium hypochlorite for 3 min, followed by 70% ethanol for 2 min, after which they were rinsed with sterile water for a few minutes, according to a previous report 24 , with some modifications. Each fragment was further cut and placed onto NBRC802 or ISP2 agar medium and incubated at 30 °C for approximately 1 month. The final rinse water was also plated onto each medium to confirm the effectiveness of the surface sterilization. After incubation, single-colony isolation was repeated for colonies formed around the tissues. NBRC802 medium contained (per liter) Hipolypepton (10 g), Bacto yeast extract (2 g), and MgSO 4 ·7H 2 O (1 g) (pH 7.0). ISP2 medium contained (per liter) Bacto yeast extract (4 g), Bacto malt extract (10 g), and glucose (4 g) (pH 7.3). Taxonomic identification of isolated bacteria was performed based on the 16S rDNA sequence. The DNA was amplified from colonies by polymerase chain reaction (PCR) using two oligonucleotide primers, 9F 5′-GAG TTT GAT CCT GGC TCA G-3′ and 1541R 5′-AAG GAG GTG ATC CAGCC-3′. PCR was performed using KOD FX Neo polymerase (Toyobo, Osaka, Japan) according to the manufacturer's recommendations under the following conditions: 94 °C for 2 min, followed by 40 cycles of 98 °C for 10 s, 68 °C for 2 min, and 72 °C for 10 min. After purification, the amplified DNAs were sequenced by Eurofins (Tokyo, Japan). The sequences of the 5′-terminal region (ca. 500 bp) were determined for all strains except BR2R-1, for which the 3′-terminal region (ca. 500 bp) sequence was determined because the 16S rDNA sequence contains an insertion (ca. 300 bp) in the 5′-terminal region. The sequences were compared to those in the GenBank database using BLASTN (https:// blast. ncbi. nlm. nih. gov/ Blast. cgi). MEGA software (https:// www. megas oftwa re. net/) was used to align the sequences and construct a neighbor-joining phylogenetic tree. Yellow 2) cells were maintained by weekly dilution (1/100) with fresh Linsmaier and Skoog (LS) medium, modified according to previous reports 18,19 . The cells were maintained in the dark at 28 °C with aeration (shaking at 120 rpm). Suspensions of A. thaliana T87 cells were maintained by weekly dilution (2/100) with fresh Jouanneau and Péaud-Lenoël (JPL) medium 40 . The cells were maintained at 22 °C with aeration (shaking at 120 rpm) under a light intensity of 60-100 µE m −2 s −1 . Although we cannot provide the plant cell lines used in our laboratory, BY-2 and T87 cells are available from RIKEN BioResource Research Center in Japan.
Arabidopsis thaliana Columbia-0 was employed for whole-plant experiments. Seeds were surface-sterilized by dipping in 20% sodium hypochlorite for 10 min and then washed repeatedly with sterile water. After treatment at 4 °C in the dark for 2 days, sterilized seeds were sown in 1/2 Murashige and Skoog (MS) medium (Sigma-Aldrich, St. Louis, MO, USA) supplemented with 10 g/l sucrose and solidified with 3 g/l Phytagel (Sigma-Aldrich) in Petri dishes 41,42   . The plant cell suspension (60 g wet cell weight/l) was incubated at room temperature on a rotary shaker (120 rpm) for 3 h. Cells of each isolated bacterial strain were cultivated in liquid NBRC802 or ISP2 medium at 30 °C for 24 h and then added to the plant cell suspension. In this process, the bacterial cell culture solution was adjusted to an optical density at 600 nm (OD 600 ) of 0.8 using NBRC802 or ISP2 medium, and the solution was further diluted by a factor of 2 using ROS assay buffer. Then, 0.1 mL of the diluted solution (cells and extracellular components in a mixture of the medium and the buffer at a ratio of 50:50) was mixed with 1.9 mL of the plant cell suspension (60 g wet cell weight/l) in a well (3 mL) of a 6-well plate. In this experimental system, both cells and extracellular components produced by cells were subjected to the assays to evaluate plant-microbe interactions based on physical and chemical signals. After addition of the diluted solution of bacterial cell culture, the mixture was incubated at room temperature on a rotary shaker (120 rpm), and production of ROS was monitored using a chemiluminescence assay with luminol. The mixture was filtered, and the filtrate (10 μL) was added to Tris-HCl buffer (50 mM [pH 8.0], 150 µL), followed by the addition of luminol (Wako, Osaka, Japan; 1 mM, 25 μl) and potassium ferricyanide (6 mM, 25 µL). ROS-associated chemiluminescence was measured for 15 s using a luminometer (Centro LB 960, Berthold, Germany). Chemiluminescence was integrated and expressed as relative intensity 18,19 . Samples that exhibited relative chemiluminescence intensity more than twice as high as mock treatment were selected as positives (Fig. S1).

Measurement of cryptogein-induced ROS production in BY-2 cells after co-incubation with bacteria.
After cultivation in modified LS medium for 3 days, tobacco BY-2 cells were collected by centrifugation and suspended in ROS assay buffer. The bacterial cell culture solution was adjusted to OD 600 of 0.8 using NBRC802 or ISP2 medium (trypticase soy broth was used for strain PsJN), and the solution was further diluted by a factor of 10 using ROS assay buffer. Then, 0.1 mL of the diluted solution (cells and extracellular components in a mixture of the medium and the buffer at a ratio of 10:90) was added to the plant cell suspension (60 g wet cell weight/l, 1.8 mL) (Fig. S2). After co-incubation at room temperature on a rotary shaker (120 rpm) for 4 h, the cells were collected by centrifugation (1000 rpm, 3 min) and suspended in fresh buffer to remove ROS scavengers and other bacteria-derived metabolites. Cryptogein (6 µM, 0.1 mL), as a plant immune response elicitor, was then added to the solution. The mixture was incubated at room temperature on a rotary shaker (120 rpm), and production of ROS was monitored using a chemiluminescence assay with luminol, as described above. Samples that exhibited relative chemiluminescence intensity more than twice as high as mock treatment were selected as positives (Fig. S3).

Measurement of flg22-induced ROS production in T87 cells after co-incubation. Arabidopsis
T87 cells were cultivated in JPL medium for 3 days and then collected by centrifugation and suspended in ROS assay buffer. The plant cells were co-incubated with each strain of isolated bacteria as described for tobacco BY-2 cells. The plant immune response elicitor flg22 (final concentration, 1 µM) was then added to the buffer instead of cryptogein (Fig. S5). The luminol derivative L-012 (Wako; final concentration, 50 µM) was added to the buffer simultaneously. The mixture was incubated at room temperature on a rotary shaker (120 rpm), and ROS-associated chemiluminescence was measured for 0.5 s using a luminometer.  41,42 . After washing three times with sterile water, a pooled sample of 6 seedlings was homogenized in 5 mL of sterile water using a mortar and pestle. Subsequently, appropriately diluted samples were plated onto NBRC802 or ISP2 agar medium. After incubation at 30 °C for a few days, colonies formed on the plates were counted, and bacterial density was expressed as CFU per gram of plant fresh weight.
Pathogenic bacterial cell suspension (4 × 10 5 CFU mL −1 ; 40 mL) prepared in sterile water containing 0.025% Silwet L-77 (Biomedical Science, Tokyo, Japan) was dispensed into 1/2 MS agar medium containing 14-day-old Arabidopsis seedlings, and the plates were incubated at room temperature for 2 min 41,42 . After the pathogen cell suspension was removed by decantation, the seedlings on the plates were rinsed twice with sterile water. The plates were then sealed with 3 M Micropore 2.5-cm surgical tape ( www.nature.com/scientificreports/ To determine the growth of strain DC3000 in Arabidopsis, the aerial tissues of infected plants were sampled. The tissues were surface-sterilized by dipping in 5% H 2 O 2 for 2 min 41,42 . After washing twice with sterile water, a pooled sample of 5 seedlings was homogenized in 5 mL of sterile water using a mortar and pestle. Subsequently, appropriately diluted samples were plated onto MG agar medium containing rifampicin. After incubation at 28 °C for 2 days, colonies formed on the plates were counted, and bacterial density was expressed as CFU per gram of plant fresh weight. Gene expression analysis. The aerial tissues of Arabidopsis plants were sampled at 3, 9, and 24 h after pathogen infection and ground in liquid nitrogen using a mortar and pestle. Total RNA was isolated using an RNA extraction kit (NucleoSpin RNA Plus, Takara Bio, Shiga, Japan). Reverse transcription was performed using reverse transcriptase (ReverTra Ace qPCR RT Master Mix with gDNA Remover, Toyobo). The expression levels of defense-related genes were determined by quantitative PCR using Thunderbird SYBR qPCR Mix (Toyobo) and specific primer sets. The following primers were used: EF-1α, forward 5′-TGA GCA CGC TCT TCT TGC TTTCA-3′ and reverse 5′-GGT GGT GGC ATC CAT CTT GTTA-3′; PR-1, forward 5′-GTG GGT TAG CGA GAA GGC TA-3′ and reverse 5′-ACT TTG GCA CAT CCG AGT CT-3′; PR-5, forward 5′-TCG GCG ATG GAG GAT TTG AA-3′ and reverse 5′-AGC CAG AGT GAC GGG AGG AAC-3′; PDF1.2, forward 5′-TCA TGG CTA AGT TTG CTT CC-3′ and reverse 5′-AAT ACA CAC GAT TTA GCA CC-3′. Quantitative PCR was performed using a CFX Connect real-time system (BIO-RAD, Tokyo, Japan) according to the manufacturer's recommendations under the following conditions: 95 °C for 1 min, followed by 40 cycles of 95 °C for 15 s, 60 °C for 1 min, and 65 °C for 15 s. The specificity of the amplifications was verified by melting curve analysis of the PCR products at the end of each experiment. The relative expression level of each gene was normalized against the expression level of EF1α, and calculated using the ΔΔCt method 44 .