Phytotoxin synthesis genes and type III effector genes of Pseudomonas syringae pv. actinidiae biovar 6 are regulated by culture conditions

The kiwifruit bacterial canker (Pseudomonas syringae pv. actinidiae; Psa) causes severe damage to kiwifruit production worldwide. Psa biovar 6 (Psa6), which was isolated in Japan in 2015, produces two types of phytotoxins: coronatine and phaseolotoxin. To elucidate the unique virulence of Psa6, we performed transcriptomic analysis of phytotoxin synthesis genes and type III effector genes in in vitro cultivation using various media. The genes related to phytotoxin synthesis and effectors of Psa6 were strictly regulated in the coronatine-inducing mediums (HS and HSC); 14 of 23 effector genes and a hrpL sigma factor gene were induced at 3 h after transferring to the media (early-inducible genes), and phytotoxin synthesis genes such as argD of phaseolotoxin and cfl of coronatine were induced at 6 and 12 h after transferring to the media (late-inducible genes). In contrast, induction of these genes was not observed in the hrp-inducing medium. Next, to examine whether the changes in gene expression in different media is specific to Psa6, we investigated gene expression in other related bacteria. For Psa biovar 1 (Psa1), biovar 3 (Psa3), and P. s. pv. glycinea (Psg), no clear trends were observed in expression behavior across various culture media and incubation times. Therefore, Psa6 seems to exert its virulence efficiently by using two phytotoxins and effectors according to environmental changes. This is not seen in other biovars and pathovars, so it is thought that Psa6 has acquired its own balance of virulence.


INTRODUCTION
Kiwifruit belong to the family Actinidiaceae and the genus Actinidia. They include Actinidia deliciosa, a species whose flesh is green, and A. chinensis, a species whose flesh is yellow or red. In Japan, kiwifruit seedlings were first introduced from New Zealand in the 1970s and are now popular as edible fruit (Ushiyama, 1993). However, around 1980, kiwifruit bacterial canker broke out and caused serious damage to kiwifruit production (Takikawa, Serizawa & Ichikawa, 1989). Since then, this disease has resulted in economic and productive damage in kiwifruit orchards throughout Japan.
The pathogen that causes kiwifruit bacterial canker is Pseudomonas syringae pv. actinidiae (Psa). It has been further classified into biovar 1, 2, 3, 5, and 6, reflecting its physiological and genetic diversity (Chapman et al., 2012;Fujikawa & Sawada, 2016;Fujikawa & Sawada, 2019;Sawada & Fujikawa, 2019a;Sawada & Fujikawa, 2019b). In Japan, Psa is highly diverse, with significant differences in the types or presence of phytotoxins produced among biovars (Sawada & Fujikawa, 2019a;Sawada & Fujikawa, 2019b). In general, the plant pathogenic bacteria group, P. syringae, is highly differentiated in pathogenicity, and it is known to have more than 60 different pathovars with different host plants and symptoms (Sawada, 2014). However, of all known P. syringae pathogens, only Psa biovar 6 (Psa6) has been reported to produce two phytotoxins (coronatine and phaseolotoxin) simultaneously. Phytotoxins are known to be important factors in the virulence of plant pathogens because they have a significant effect on plants, even at low concentrations (Bender, 1999;Strange, 2007). Coronatine, a phytotoxin produced by Psa6, is structurally similar to the plant hormone jasmonic acid (Geng et al., 2014). It has been reported to inhibit host defense responses by disrupting hormonal signals and reopening closed stomata (Geng et al., 2014). Phaseolotoxin, another phytotoxin produced by Psa6, has been shown to induce halos around lesions in leaves by inhibiting the function of ornithine carbamoyltransferase, which is an enzyme involved in arginine biosynthesis in host plants (Mitchell, 1976;Patil et al., 1976;Tamura et al., 2002). These phytotoxins have been extensively studied, and the conditions under which they are induced have been clarified in some bacteria. For example, P. syringae pv. glycinea (Psg), which is a causal pathogen of soybean (Glycine max), produces coronatine when cultured in HS and HSC medium (Hoitink & Sinden, 1970;Palmer & Bender, 1993), and P. syringae pv. phaseolicola, which is a causal pathogen of the common bean (Phaseolus vulgaris), produces the highest amounts of phaseolotoxin at 18 • C (Nüske & Fritsche, 1989).
In addition to the two phytotoxins, Psa6 produces other virulence factors, type III effector proteins. Effectors are known to suppress or interfere with the plant's defense response, increasing the pathogen's virulence (Jones & Dangl, 2006). The type III secretion system is a specialized molecular machine of gram-negative bacteria that allows the effector proteins to be injected directly into the plant host cells (Coombes, 2009). Effector genes of Psa6 in the early stages of infection can be classified into several groups based on differences in expression patterns (Fujikawa & Sawada, 2019). Although the core genome sequences of Psa are almost the same among biovars, the presence and combination of phytotoxins and effector genes are diverse. Therefore, the diversity of effectors and phytotoxins may be due to accessory genomes encoded by transposons and other mobile genetic elements (Sawada & Fujikawa, 2019a;Sawada & Fujikawa, 2019b).
It is important to understand the survival strategies, evolution, and natural adaptations of Psa. In particular, determining the expression behavior of the two phytotoxin synthesis genes and effector genes of Psa6 can provide important information. We used RNA sequencing (RNA-seq) and reverse transcription-quantitative PCR (RT-qPCR), to perform transcriptomic analysis of these genes when cultured under phytotoxin-inducing or effector-inducing conditions.

MATERIALS & METHODS
Bacterial strains, media, and culture conditions Psa6 MAFF 212134, Psa1 MAFF 302145, Psa3 MAFF 212104, and Psg MAFF 301683 were used as the representative strains. These strains can be obtained from NARO Genebank (https://www.gene.affrc.go.jp/index_en.php). For the pre-culture, each strain was scraped from LB (Bertani, 1951) solid medium and was transferred to new LB liquid media. The cultures were then shaken overnight. After shaking, the pre-cultures were centrifuged at 6,000×g for 5 m, then each pellet was resuspended with water to OD (660 nm) = 0.6. Two milliliters of resuspension were inoculated into 15 mL of respective test media and the cultures were shaken at 160 rpm at 27 or 18 • C. As test media, LB medium, HS medium, HSC medium, and hrp-inducing medium were used (Table 1). HS and HSC media have been reported to activate coronatine production of Psg (Hoitink & Sinden, 1970;Palmer & Bender, 1993). On the other hand, hrp-inducing medium has been reported to induce the expression of hrp genes and effector genes of Psg (Huynh, Dahlbeck & Staskawicz, 1989). Additionally, it has been reported that cultures at 18 • C, compared to 27 • C, induce the production of phytotoxins such as coronatine and phaseolotoxin (Nüske & Fritsche, 1989;Palmer & Bender, 1993). After incubation at 0, 3, 6, and 12 h, one mL of each culture was taken and used for extraction of DNA or RNA.

DNA and RNA extraction
DNA was extracted using the DNeasy mini kit (QIAGEN, Hilden, Germany) and RNA was extracted using the RNeasy mini kit (QIAGEN). DNA was used to confirm primer design, and RNA was used for RNA-Seq and RT-qPCR.

RNA-Seq
RNA-Seq was performed according to our previous study (Fujikawa & Sawada, 2019). Briefly, total RNA was extracted from the 3 h culture in HS or HSC media at 27 • C or 18 • C, were treated with the RiboMinus Transcriptome Isolation Kit for bacteria (Thermo Fisher Scientific Inc., Waltham, MA, USA), and rRNAs were almost depleted from the total RNAs. The RNA-Seq library was constructed from sample RNAs using the Ion Total RNA-Seq Kit v2 (Thermo Fisher Scientific Inc.) and Ion Xpress RNA-Seq barcode (Thermo Fisher Scientific Inc.). Subsequently, the RNA-Seq templates were prepared using the Ion PGM Hi-Q View OT2 Kit (Thermo Fisher Scientific Inc.) on an Ion OneTouch 2 system. The templates were sequenced with an Ion PGM Hi-Q View Sequencing Kit (Thermo Fisher Scientific Inc.) and a 318 Chip Kit v2 (Thermo Fisher Scientific Inc.) on an Ion PGM next generation sequencer. Sequence data were analyzed as RNA reads using the CLC Genomics Workbench ver. 11 (QIAGEN) and the statistical software R ver. 3.5.1 with various packages (edgeR, limma, and gplots) according to a previous report (Fujikawa & Sawada, 2019). Namely, using high-quality sequence reads with the reference genome of Psa6 MAFF 212134 (accession no. MSBW00000000), the reads per kilobase of exon per million mapped reads (RPKM) values were obtained and normalized. Then, using R software, the gene expression values relative to the 27 • C HS medium were calculated to For the RT-qPCR analysis, reference gene candidates were selected according to the following conditions: (1) low FC (fold change) value, which means that expression is not strongly affected by culture conditions (log 2 FC= ≤ ± 0.2), and (2) high and stable CPM (counts per million) value (sum of log 2 CPM). When genes meeting the above conditions were found, primers were designed and used for RT-qPCR analysis.

RT-qPCR
For RT-qPCR analysis, cDNA was synthesized from total RNA of each culture sample using the PrimeScript II 1st strand cDNA Synthesis Kit (Takara Bio, Shiga, Japan) with random hexamers to yield a total volume of 20 µL. Then, 180 µL of nuclease-free water was added to bring the total volume to 200 µL. The synthesized cDNA was subjected to RT-qPCR using TB Green Premix Ex Taq II (Takara Bio). The PCR conditions were set as follows: denaturation at 96 • C for 10 m; a cycle of 96 • C for 15 s, 58 • C for 30 s, and 72 • C for 1 m, repeated 40 times; then 96 • C for 15 s, 58 • C for 1 m, and 95 • C for 15 s to form a melting curve. DNA of Psa6 was used as a positive control and nuclease-free water was used as a negative control. Then, based on the results of RNA-Seq (described above) we selected the genes that had similar expression levels across all culture conditions as reference genes for each bacterium. According to the method of Smith, Lovelace & Kvitko (2018), RT-qPCR results were processed using the following equation: where NRQ is the normalized relative quantity, E is amplification efficiency of qPCR, GOI is gene of interest, and RG is reference gene To solve the NRQ ( Ct value), a geometric mean was calculated from the RT-qPCR results of three reference genes as shown by the denominator equation, and Ct of each sample was calculated based on it. Then, Ct for each condition was calculated on the basis of the RT-qPCR result for 0 h inoculation under each condition. Moreover, the results of RT-qPCR analysis were clustered using the unweighted pair group method with arithmetic mean (UPGMA) and output to heat maps using R (ver. 3.5.1) software with two packages (Heatplus, and viridis).
To confirm the validity of the designed primers, the extracted DNA was subjected to PCR using Takara Ex Taq Hot Start Version (Takara Bio) with the following conditions: denaturation at 95 • C for 10 min; a cycle of 95 • C for 30 s, 65 • C for 30 s, and 72 • C for 1 m, repeated 40 times; and extension at 72 • C for 7 m. It was confirmed that respective amplicons were single bands by electrophoresis (Fig. S1), and each melt peak of single band using melt curve analysis was also measured (Table S1). In addition, each amplification efficiency of primer sets was calculated (Table S1). These appropriate primer sets were used for RT-qPCR analysis.

Gene expression of Psa6 under different culture conditions using RNA-Seq analysis
RNA-Seq data of Psa6 was archived in NCBI GEO (accession no. GSE149743). RNA-Seq of Psa6 revealed the behavior of 5,796 genes (Table S2). The expression values from the HS medium at 27 • C were used to calculate the expression values from the HS medium at 18 • C and the HSC medium at 27 and 18 • C. Seventy-eight genes had an FDR value lower than 0.1 (Table 1). In contrast to the HS medium at 27 • C, 26 genes were induced in the HSC medium at 18 • C, including the type III chaperone ShcN protein gene and the type III effector (translocator) hrpK. In particular, hrpK had a large effect size (0.96). However, there was no significant induction of other type III effector genes. In addition, neither the induction nor the suppression of genes involved in toxin production was observed. In contrast to HS medium at 27 • C, 17 genes were suppressed in both the HSC medium and the 18 • C condition; however, no genes thought to be involved in virulence were significant (Table 2). Similarly, no virulence genes were significantly induced or suppressed under the different culture conditions. The HSC medium is an improvement of the HS medium and induces coronatine production (Palmer & Bender, 1993). However, we found no significant change in the expression of phytotoxin genes. This suggests limited coronatine synthesis in the 3 h incubation of Psa6. Incubation of P. syringae at 18 • C can induce phytotoxin production compared with incubation at 27 • C (Nüske & Fritsche, 1989;Palmer & Bender, 1993), but we did not observe significant induction of phytotoxins under those conditions (Table 2). This suggests that temperature in a 3 h incubation may have little effect on phytotoxin production for Psa6. To clarify the role of temporal changes in the behavior of virulence genes, such as type III effector genes and phytotoxin synthesis genes, we examined the behavior of gene expression over time using RT-qPCR analysis. Because the relative expression values in RT-qPCR analyses are calculated using constantly expressed reference genes, the selection of appropriate reference genes is important. Based on the conditions described in the Materials and Methods section, we selected five candidates among a total of 5,796 genes (Table S1). From these five genes (50S ribosomal protein L13, bifunctional glutamine synthetase adenylyltransferase/deadenyltransferase, tRNA dihydrouridine (20/20a) synthase dusA, transcriptional regulator ftrA, and microcin ABC transporter ATP-binding protein), three genes for each bacterium were used as reference genes. For these genes, expression level was high and stable (expression change was low), regardless of the media used. Thus, we designed primers for the RT-qPCR analysis based on the sequences of the following reference genes: 50S ribosomal protein L13, bifunctional glutamine synthetase adenylyltransferase/deadenyltransferase, and tRNA dihydrouridine (20/20a) synthase dusA for Psa6; 50S ribosomal protein L13, bifunctional glutamine synthetase adenylyltransferase/deadenyltransferase, and transcriptional regulator ftrA for Psa1 and Psa3; and 50S ribosomal protein L13, bifunctional glutamine synthetase adenylyltransferase/deadenyltransferase, and microcin ABC transporter ATP-binding protein for Psg (Table S1). Since these reference genes are a group of genes with stable expression and housekeeping, it is expected that strains other than Psa6 have similar expression tendencies. In addition, it is expected that internal control can be guaranteed by the geometric mean of 3 kinds of genes in order to avoid the bias of expression of each gene.

Temporal changes in the expression of virulence genes of Psa6 in RT-qPCR analysis
The expression of phytotoxin synthesis genes and type III effector genes of Psa6 was investigated with RT-qPCR. Normalization was performed using the reference genes selected above.
For the reference genes, the amplification efficiency (E) of target genes in qPCR was approximately 2.0 (Table S1) and three genes were used as the reference (n = 3). Thus, using Eq. (1) in the Materials and Methods section with each E of primer sets, the relative expression levels were calculated from the Ct value at a certain time and the Ct value at 0 h. Here, in order to indicate that the results of RNA-Seq and the results of RT-qPCR did not diverge significantly, the expression levels of the virulence genes in the RT-qPCR analysis for Psa6 and the expression levels of the same genes in the RNA-Seq analysis were compared      (Table S4). When the expression levels of the virulence genes from RNA-Seq were plotted on the vertical axis and those from RT-qPCR were plotted on the horizontal axis, both RNA-Seq and RT-qPCR results were correlated (Fig. 1). The plots were concentrated in the first and third quadrants (R 2 (coefficient of determination) value is about 0.6), although some genes were expressed differently. We concluded that the results from RT-qPCR can almost function as a representation of transcriptome analysis. The expression of phytotoxin synthesis genes and effector-related genes of Psa6 under various culture conditions with different incubation times was investigated using RT-qPCR (Table S5 & Data S1). From the results of RT-qPCR analysis, heat maps were output (Fig. 2). We found that phytotoxin synthesis genes and effector genes of Psa6 were induced in HS and HSC media ( Fig. 2A), which are known to induce coronatine production. In contrast, these genes were not induced in the hrp-inducing medium, which is known to induce hrp-dependent effector genes. The LB and hrp-inducing media were clustered together and were separate from the HS and HSC media. In the LB and hrp-inducing media cluster, most genes were suppressed. Moreover, in the hrp-inducing medium, no clear induction of the hrp L-dependent effector genes of Psa6 was observed. On the other hand, phytotoxin synthesis genes and effector genes were significantly induced in HS and HSC media. Additionally, incubation at 18 • C and 27 • C did not result in significant changes to gene expression; samples from each temperature were clustered separately. This suggests that culture medium composition and incubation time have a stronger effect on Psa6 cultivation than incubation temperature.  Intriguingly, the expression of virulence genes was generally separated into the following two groups: early-inducible genes and late-inducible genes (Fig. 2). This was more clearly observed in the heat map, excluding the conditions of hrp-inducing medium (Fig. 2B). The early-inducible genes (hrpL,avrD1,avrpto5,avrRpm1,hopAE1,hopAU1,hopAZ1,hopM1,hopN1,hopQ1,hopR1,hopS2,hopY1, hrpK, and argK ) were induced after 3 h of incubation but suppressed at 6 and 12 h. Most of them were thought to be hrpLdependent effector genes (Lam et al., 2014;Vencato et al., 2006). The early-inducible genes of Psa6 contains hrpL and several effector genes. It is known that the HrpL protein binds to the Hrp box promoter sequence on the upper region of protein-encoding sequences in several effector genes (Greenberg & Yao, 2004;Lam et al., 2014), then, we could also find the Hrp box promoter sequences in some early-inducible genes of Psa6 (Fig. S2). Therefore, HrpL regulation may be involved in the early-inducible genes of Psa6 in response to HS and HSC media. However, further research is needed on this detail. In contrast, the late-inducible genes (avrE1,avrRps4,hopAH1,hopAI1,hopAS1,hopD1,hopJ,hopZ3,cfl,cfa1,cfa5,cfa9,cmaD,cmaU,corP,corR,corS,and argD) were induced after 6 h of incubation. These genes included phytotoxin synthesis genes, except argK and some effector genes. Thus, when Psa6 is incubated under phytotoxin-induced conditions (HS and HSC media), induction of phytotoxin synthesis genes followed that of effector genes. In addition, phytotoxin synthesis genes were expressed for longer than the hrpL-regulated effectors.

Comparison of gene expression behavior between Psa6 and other bacteria
As described above, in Psa6, the effector genes are not induced in the hrp-inducing medium, and the induction of virulence genes in phytotoxin-inducing media is strictly controlled over time. It was unclear whether these behaviors of gene expression were Psa6 specific; thus, similar tests were performed on the other bacteria. The expressions of virulence genes in two biovars (Psa1 and Psa3) and one other species (Psg) were investigated using RT-qPCR (Table S6 & Data S2). Normalization was performed using the reference genes selected above. The expression level of virulence genes was compared across all bacteria. Heat maps were generated from the expression patterns of the effector genes shared by each bacterium (Fig. 3). Psa1 and Psa3 were nested in the heat map (Fig. 3A). In Psa6, the expression of effector genes was significantly induced compared with other bacteria. Furthermore, Psg was scattered in clusters of Psa1 and Psa3, and the clusters did not split due to differences in pathovars. When clustering was performed with only common effectors of both Psa1 and Psa3 (omitting Psa6 and Psg), medium composition had no effect on gene expression (Fig. 3B). Our results show that medium composition and incubation time did not significantly affect the expression of effector genes in Psa1, Psa3, or Psg.
We also compared phytotoxin synthesis genes of Psa6 with other bacteria. To compare the expression of coronatine synthesis genes in Psa6 and Psg, the expression levels of cfa1 and cmaD at 0, 3, and 6 h incubation were examined (Table 3). The results showed that incubation of Psa6 in HS and HSC media induced the expression of cfa1 and cmaD over time, but incubation in LB and hrp-inducing media significantly suppressed cfa1 and slowly induced the expression of cmaD. Therefore, the behavior of the coronatine synthesis genes of Psa6 was affected by the medium composition. On the other hand, Psg was not affected by medium composition. In Psg, there were no differences in gene expression across the different media, and only cfa1 was induced over time regardless of medium composition. There was no induction or suppression of cmaD. Additionally, we compared the expression of phaseolotoxin synthesis-related genes in Psa6 and Psa1 (Table 4). In Psa6, the expression of argK and argD was slowly induced by culture in HS and HSC media. In addition, the expression of argK was notably induced and argD was greatly suppressed in LB and hrp-inducing media. However, in Psa1, argK was induced in LB and hrp-inducing media, and argD remained expressed at a constant level without induction or suppression. In summary, the coronatine and phaseolotoxin synthesis genes in Psa6 differed in their response to medium composition and incubation time compared to other bacteria, with a particularly strong effect of medium composition.

Expression patterns and conditions of the virulence genes in Psa6
Coronatine production in Psg PG 4180 is more active in cultures at 18 • C than 24 • C or higher (Palmer & Bender, 1993). In contrast, coronatine production in P. syringae pv. tomato (Pto) DC 3000 is temperature independent and constant (Braun et al., 2008). Additionally, seven strains of P. syringae, including Psg PG 4180, Psg 7a, and Pto PT 23.2, produce more coronatine in HSC medium than in HS medium (Palmer & Bender, 1993

Notes.
The expression level (log2) is indicated as a ratio relatively against 0h of each culture condition. Bold letters indicate that expression is induced (log2 > 1.0), and blue letters indicate that expression is suppressed (log2 < −1.0). coronatine in HSC medium (Palmer & Bender, 1993). Moreover, in Psg PG 4180, the CmaB protein, a synthesis protein of coronatine, decreases 4 h after the temperature is changed from 18 • C to 28 • C, suggesting that changes in gene expression can be observed at an earlier stage (Budde et al., 1998). However, the coronatine synthesis genes of Psa6 were induced later, regardless of temperature (Fig. 2). Thus, the coronatine synthesis genes of Psa6 may be regulated by different environmental conditions than Psg PG 4180. In addition, in the strain of Psg used in this study (MAFF 301683), which was isolated in Japan, the expression of the coronatine synthesis genes was induced regardless of medium

Notes.
The expression level (log2) is indicated as a ratio relatively against 0h of each culture condition. Bold letters indicate that expression is induced (log2 > 1.0), and blue letters indicate that expression is suppressed (log2 < −1.0). composition (Table 3). Conversely, in a different Psg strain, PG 4180, coronatine synthesis was induced in HS and HSC media (Palmer & Bender, 1993). Thus, depending on the type of Pseudomonas (pathovars and strains), coronatine production is thought to be affected by various environmental conditions. Moreover, despite Psa6 and Psa1 possessing a similar phaseolotoxin synthesis gene cluster (Fujikawa & Sawada, 2019), both expressed them differently (Table 4). This gene cluster is contained in the tox island in the genomes of Psa6 and Psa1. However, in Psa6, the ginABCD operon, which is located at the left end of the tox island and is involved in the excision/insertion of the tox island, was truncated, suggesting that the island of biovar 6 might have lost its mobility (Fujikawa & Sawada, 2019). At present, it is unknown whether the expression pattern of the phaseolotoxin synthesis genes is different due to differences in the ginABCD operon or whether it depends on another regulation system. This requires further investigation.
We found that Psa6 did not induce hrpL-dependent effectors in the hrp-inducing medium but was induced in HS and HSC media (Fig. 2). The hrp-inducing medium is an oligotrophic medium (Huynh, Dahlbeck & Staskawicz, 1989), but HS and HSC media contain rich sugars and minerals (Palmer & Bender, 1993). In Psa6, phytotoxin production and effector-related genes were induced similarly even with different concentrations of glucose (10 g/L in HS medium and 20 g/L in HSC medium) (Table 1). However, fructose (1.8 g/L) in the hrp-inducing medium was not significantly induced. This may be due to qualitative or quantitative differences in carbon sources. Additionally, it seemed that the FeCl 3 concentration (HS, HSC, and the hrp-inducing media) (Table 1) may not be involved in the expression behavior of phytotoxin synthesis genes and effector genes. In this study, HS and HSC media were found to induce the virulence genes of Psa6. These genes were divided into two groups according to how long it took for them to be induced.
The early-inducible genes included many effectors and hrpL. The effectors in this group are involved in plant immunity such as pathogen-associated molecular patterns (PAMPs) triggered immunity (PTI) and effector-triggered immunity (ETI) (Jones & Dangl, 2006).