Signaling Cross-Talk between Salicylic and Gentisic Acid in the ‘Candidatus Phytoplasma Solani’ Interaction with Sangiovese Vines

“Bois noir” disease associated with ‘Candidatus Phytoplasma solani’ seriously compromises the production and survival of grapevines (Vitis vinifera L.) in Europe. Understanding the plant response to phytoplasmas should help to improve disease control strategies. Using a combined metabolomic and transcriptomic analysis, this work, therefore, investigated the phytoplasma–grapevine interaction in red cultivar Sangiovese in a vineyard over four seasonal growth stages (from late spring to late summer), comparing leaves from healthy and infected grapevines (symptomatic and symptomless). We found an accumulation of both conjugate and free salicylic acids (SAs) in the leaves of ‘Ca. P. solani’-positive plants from early stages of infection, when plants are still asymptomatic. A strong accumulation of gentisic acid (GA) associated with symptoms progression was found for the first time. A detailed analysis of phenylpropanoids revealed a significant accumulation of hydroxycinnamic acids, flavonols, flavan 3-ols, and anthocyanin cyanidin 3-O-glucoside, which are extensively studied due to their involvement in the plant response to various pathogens. Metabolomic data corroborated by gene expression analysis indicated that phenylpropanoid biosynthetic and salicylic acid-responsive genes were upregulated in ‘Ca. P. solani-positive plants compared to -negative ones during the observed period.


Introduction
Phytoplasmas are plant pathogenic bacteria in the class Mollicutes transmitted by phloem sap-feeding insects. They have a broad range of hosts among plant species worldwide, including many economically important crops such as grapevine, fruit trees, and ornamental plants [1,2]. In Europe, 'Candidatus Phytoplasma solani' ('Ca. P. solani'), taxonomic subgroup 16SrXII-A [3] is associated with "Bois noir" (BN) disease, which can lead to serious losses of grape clusters [4]. BN is endemic in the Euro-Mediterranean area, and it is characterized by a disease cycle, including insect vectors and many herbaceous plants as phytoplasma reservoirs [5,6]. Phytoplasma are still poorly characterized plant pathogens because of their low concentration in plants and the difficulties of in vitro cultivation.

Plant Symptoms
Leaves from 'Ca. P. solani'-positive and -negative plants were sampled in four periods, from the late spring to late summer (May, July, August, and September) of 2020. In May and July, samples were collected from 'Ca. P. solani'-positive plants showing symptoms of severity class 0 (no symptoms observed). On the other hand, in August, 'Ca. P. solani'-positive plants were classified as belonging to severity class 2 (the symptoms were mild), and lastly, in September, the plants selected for analysis were classified as belonging to severity class 3 (more than three shoots with reddening leaves) [39]. Therefore, samples collected in May and July from 'Ca. P. solani'-positive plants were considered as asymptomatic, while those collected in August and September were considered as symptomatic. Leaves collected from 'Ca. P. solani'-negative plants were classified as class 0 (no symptoms observed) during all sampling stages. Table 1 reports the phenolic compounds identified by negative ionization mode using HPLC/MS/TOF in leaves of 'Ca. P. solani'-positive or -negative plants. All compounds were detected, although in different amounts, in both positive and negative plants.

No.
Compound

Identification of Phenylpropanoid Compounds
Because many of the phenylpropanoids are detected in glycosylated form, to confirm and quantify each compound identified, extracts from leaves collected from 'Ca. P. solani' -positive and -negative plants were digested by β-glucosidase. The typical chromatogram is presented in Figure 1. Table 2 shows the compounds identified after the enzymatic treatment: mainly two dihydroxybenzoic acids, the 2,5 dihydroxybenzoic acid (or GA), the 2,3 dihydroxybenzoic acid, (compounds 1B and 2B, respectively), and salicylic acid (compound 8B, ohydroxybenzoic acid).

Phenolic Compounds Accumulate in 'Ca. P. solani'-Positive Plants
To evaluate the changes in phenylpropanoid contents with regard to BN health status, a quantitative analysis was carried out using chemical standards.
In the chromatogram (Figure 1), the peak 1B corresponds to gentisic acid, resulting from its coelution with standard 2,5 DHBA under the same chromatographic conditions and by comparing its retention time with those previously reported [50]. Figure 2 shows the mass spectra of the compound obtained by ESI in negative ion mode and their UV/Vis peaks. The mass spectrum from the total ion current chromatogram showed a main fragment, which corresponded with the molecular weight of gentisic acid. The mass spectrum also showed an ion at m/z 108, probably arising from the decarboxylation of status, a quantitative analysis was carried out using chemical standards.
In the chromatogram (Figure 1), the peak 1B corresponds to gentisic acid, resulting from its coelution with standard 2,5 DHBA under the same chromatographic conditions and by comparing its retention time with those previously reported [50]. Figure 2 shows the mass spectra of the compound obtained by ESI in negative ion mode and their UV/Vis peaks. The mass spectrum from the total ion current chromatogram showed a main fragment, which corresponded with the molecular weight of gentisic acid. The mass spectrum also showed an ion at m/z 108, probably arising from the decarboxylation of gentisic acid. The relative intensities of the mass peaks from plant extracts precisely matched those of the standard 2,5 DHBA. A comparison of the characteristics of compound no. 8B with the MS-spectrum of SA (as standard) showed a perfect match, thus confirming the identity of the two compounds ( Figure 3). A comparison of the characteristics of compound no. 8B with the MS-spectrum of SA (as standard) showed a perfect match, thus confirming the identity of the two compounds ( Figure 3).  Similarly, peak no. 2B was coeluted with standard 2,3 DHBA, and showed a retention time identical to the peak of the compound found by Zhang et al. [37] ( Figure S1).
To investigate the accumulation of phenolic acids in free and conjugated forms, we carried out a time-course quantitative analysis, comparing leaf extracts collected from 'Ca. P. solani'-positive or -negative plants and with different symptom levels according to the different seasons ( Figure 4).  Similarly, peak no. 2B was coeluted with standard 2,3 DHBA, and showed a retention time identical to the peak of the compound found by Zhang et al. [37] ( Figure S1).
To investigate the accumulation of phenolic acids in free and conjugated forms, we carried out a time-course quantitative analysis, comparing leaf extracts collected from 'Ca. The statistical analysis between Ca. P. solani-negative and Ca. P. solani-positive leaves was carried out using a multiple t-test (FDR = 5%) and significant differences are marked by asterisks: * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001. Values are reported as means and standard deviation of five harvested samples (n = 5 'Ca. P. solani'-negative and 'Ca. P. solani'-positive plants, respectively), each measured in three technical replicates.
We followed the analysis pattern of free GA and SA along the disease progression ( Figure 4), which revealed that the presence of free GA was very low in spring and early summer in leaves from 'Ca. P. solani'-negative plants, similarly to that in the positive ones. However, free GA underwent a slight increase in positive plants in August (0.27 µg/g FW), while in September, it reached a level of about 44 times higher than in negative plants ( Figure 4A).
The free SA basal content, in leaves collected from negative plants, was constant throughout the entire growth season, with an average value of about 0.23 µg/g FW. In positive plants, the SA level increased from spring to early summer compared to healthy ones, and in September, free SA increased roughly threefold, reaching the GA content in the same sample ( Figure 4B).
In 'Ca. P. solani'-positive plants, a progressive accumulation of conjugated GA was observed, while in 'Ca. P. solani'-negative plants, its level remained nearly constant from late spring to early summer, but tripled in August, and finally returned to the initial level in September ( Figure 4A). Conversely, in 'Ca. P. solani'-positive plants, the conjugated GA content gradually increased from May to August, in September, when disease symptoms The statistical analysis between Ca. P. solani-negative and Ca. P. solani-positive leaves was carried out using a multiple t-test (FDR = 5%) and significant differences are marked by asterisks: * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001. Values are reported as means and standard deviation of five harvested samples (n = 5 'Ca. P. solani'-negative and 'Ca. P. solani'-positive plants, respectively), each measured in three technical replicates.
We followed the analysis pattern of free GA and SA along the disease progression ( Figure 4), which revealed that the presence of free GA was very low in spring and early summer in leaves from 'Ca. P. solani'-negative plants, similarly to that in the positive ones. However, free GA underwent a slight increase in positive plants in August (0.27 µg/g FW), while in September, it reached a level of about 44 times higher than in negative plants ( Figure 4A).
The free SA basal content, in leaves collected from negative plants, was constant throughout the entire growth season, with an average value of about 0.23 µg/g FW. In positive plants, the SA level increased from spring to early summer compared to healthy ones, and in September, free SA increased roughly threefold, reaching the GA content in the same sample ( Figure 4B).
In 'Ca. P. solani'-positive plants, a progressive accumulation of conjugated GA was observed, while in 'Ca. P. solani'-negative plants, its level remained nearly constant from late spring to early summer, but tripled in August, and finally returned to the initial level in September ( Figure 4A). Conversely, in 'Ca. P. solani'-positive plants, the conjugated GA content gradually increased from May to August, in September, when disease symptoms were pronounced, reaching a high value (about 200 µg/g FW), which was about 25-fold higher than -negative plants and more than 100-fold higher than the GA free amount in the same sample ( Figure 4A,C).
The data on conjugated SA ( Figure 4D) showed a constant low level in 'Ca. P. solani'negative plants (class = 0) for the entire analysis period, and a slightly higher level (average value about 4 µg/g FW) in leaves of 'Ca. P. solani'-positive plants from May (asymptomatic plants, class = 0) to August. This was the period when the leaves showed mild BN disease symptoms (class = 2), reaching a maximum level of 14 µg/g FW (about three times higher than the basal level) in September, when the plants showed evident disease symptoms (class = 3). Interestingly, in September, in the leaves of positive plants, the level of conjugated GA was more than nine times higher than conjugated SA.
The other SA catabolite, 2,3 DHBA, was not found in free form and, after β-glucosidase digestion, it was undetected in the leaves of both 'Ca. P. solani'-positive and -negative plants in the late spring and in early summer. On the other hand, it was found in 'Ca. P. solani'-positive samples at a low level in August (about 2.9 µg/g FW) and increased, about threefold, in infected plants in September ( Figure 5A). In 'Ca. P. solani'-negative plants, levels of 2,3 DHBA in mid and late summer were very low compared to positive plants. Flavan-3-ols, catechin and epicatechin, identified in all analyzed samples, maintained throughout the seasons approximately the same concentration in the leaves of negative plants. In May and July, no significant differences between 'Ca. P. solani'-negative andpositive plants were detected, whilst infected plants showed, in August, a 2-fold and 1.5- It is known that hydroxycinnamic acids and monocyclic phenylpropanoids are involved in the interactions between pathogens and host plants. When methanol extracts from 'Ca. P. solani'-positive and -negative grapevine plants were digested with β-glucosidase, p-coumaric and ferulic acid were clearly detected. Both p-coumaric and ferulic acid showed a similar trend during seasonal growth ( Figure 5B,C); the increase in hydroxycinnamic acids occurred in parallel with the development and severity of symptoms. The p-coumaric acid levels ( Figure 5B) measured in late spring and early summer were either low or absent in both negative and positive plants, whereas in August, the amount increased for both positive and negative plants but without a significant difference. In the last growth stage, p-coumaric levels showed a consistent increase (up to about 100 µg/g FW) in 'Ca. P. solani'-positive plants, whilst in 'Ca. P. solani'-negative plants, the levels decreased approximately to 5 µg/g FW. Ferulic acid started to accumulate in August in both 'Ca. P. solani'-positive and -negative plants, and it reached a peak (33 µg/g FW) only in positive plants in September. In 'Ca. P. solani'-negative plants, ferulic acid was undetectable ( Figure 5C).
Flavan-3-ols, catechin and epicatechin, identified in all analyzed samples, maintained throughout the seasons approximately the same concentration in the leaves of negative plants. In May and July, no significant differences between 'Ca. P. solani'-negative and -positive plants were detected, whilst infected plants showed, in August, a 2-fold and 1.5-fold higher content of catechin and epicatechin, respectively, than negative plants. A significant increase (about 3.8-folds) was found in extracts of leaves from 'Ca. P. solani'positive plants for catechin and about threefold for epicatechin in September ( Figure 6A,B).
Flavonols are typical compounds of grapevine leaves. We found three main forms of conjugated quercetin: the most represented molecule was quercetin 3-O-glucuronide ( Figure 6C). Its trend was similar throughout the growth season in both 'Ca. P. solani'negative and -positive plants. The quercetin glucoside level ( Figure 6D) was similar in negative plants compared to positive ones with a maximum value recorded in August and September for both health statuses. The quercetin 3-O-rhamnoside content ( Figure 6E) was the lowest and was the only compound with higher levels in spring in negative plants compared to positive plants. In May, it was twofold higher in 'Ca. P. solani'-negative plants compared to -positive ones, whereas in early and mid-summer, the amount decreased and then increased again in late summer. The trend in 'Ca. P. solani'-positive plants was more linear, with low levels in spring and early summer and a significant increase in mid and late summer ( Figure 6D). In addition, quercetin 3-O-rhamnoside reached similarly high levels for both health statuses in late summer, increasing earlier in 'Ca. P. solani'-positive plants, as it reached a high level (3-folds higher than negative plants) from mid-summer.
Another key flavonol, kaempferol, was predominantly present as kaempferol 3-Oglucoside ( Figure 6F Among anthocyanins, we identified cyanidin 3-O-glucoside in the leaf extracts. As shown in Figure 6H, there was no significant change in its content with regard to health status from spring to mid-summer, whereas in line with complete leaf reddening, its concentration reached a peak value (about 58 µg/g FW) in 'Ca. P. solani'-positive plants.
glucoside ( Figure 6F,G), showing an increase only in 'Ca. P. solani'-positive plants in late summer. Kaempferol 3-O-glucuronide instead maintained a relatively constant amount during all sampling periods, regardless of the health status.
Among anthocyanins, we identified cyanidin 3-O-glucoside in the leaf extracts. As shown in Figure 6H, there was no significant change in its content with regard to health status from spring to mid-summer, whereas in line with complete leaf reddening, its concentration reached a peak value (about 58 µg/g FW) in 'Ca. P. solani'-positive plants. The statistical analysis between Ca. P. solani-negative and Ca. P. solani-positive leaves was carried out using a multiple t-test (FDR = 5%) and significant differences are marked by asterisks: * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001. Values are reported as means and standard deviation of five harvested samples (n = 5 'Ca. P. solani'-negative and 'Ca. P. solani'-positive plants, respectively), each measured in three technical replicates. The statistical analysis between Ca. P. solani-negative and Ca. P. solani-positive leaves was carried out using a multiple t-test (FDR = 5%) and significant differences are marked by asterisks: * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001. Values are reported as means and standard deviation of five harvested samples (n = 5 'Ca. P. solani'-negative and 'Ca. P. solani'-positive plants, respectively), each measured in three technical replicates.

SA and Biosynthesis of Its Derivatives
With the data obtained on SA accumulation and the increase in its GA derivative, we investigated the expression of genes such as ICS (E.C. 5.4.4.2) and DMR6 (E.C. 1.14.11) (Figure 9) (in addition to PAL, whose expression profile is presented above in Figure 7), which are responsible for SA and GA syntheses, as indicated by several authors [15,41,42,51,52].
The ICS gene was upregulated in 'Ca. P. solani'-positive plants for the full growth season (Figure 9). The expression of DMR6 was already high in positive plants (Figure 9) in spring and reached its highest level of overexpression in July (about 14-fold change), maintaining the upregulation in August (about 5-fold change) and September (about 2fold change). At the transcriptional level, compared with healthy plants, phytoplasma infection significantly upregulated the expression of genes coding for PAL and C4H in the mid and late summer (August and September) (Figure 7).
For CHS1 and CHS2, we registered an overexpression of about a 14-and 38-fold change in positive plants compared to negative ones in August (Figure 7), respectively, while CHS3 transcript was upregulated (about 4-fold change) in September.
The two isoforms of F3H (F3H1 and F3H2) coding flavanone-3-hydroxylase involved in dihydrokaempferol synthesis showed a similar behavior. In May and July, F3H1 and F3H2 transcripts showed no significant changes, but in August, both genes reached the highest expression (about a 12-and 60-fold change, respectively); in September, the F3H2 transcript overexpression in positive plants was approximatively a 5-fold change.
F'3H coding a flavonoid-3 -hydroxylase showed no significant change in expression with regard to health status.
A high expression level of the FLS gene, involved in the biosynthesis of three main flavonols, quercetin, kaempferol, and myricetin, were observed ( Figure 8) in Ca. P. solanipositive plants from mid-summer with a 9-fold change in August and a 19-fold change in September.
DFR coding a NADPH-dependent dihydroflavonol reductase for leucoanthocyanidins synthesis showed a high expression, 22-and 43-fold (Figure 7), respectively, in August and September in positive plants.
Among the "late genes", specific for the anthocyanin's biosynthesis, LDOX, which is specific for the conversion of leucoanthocyanidins to anthocyanidins, was overexpressed in September in positive plants (about a 3-fold change) (Figure 8). We also registered an increase in the UF3GT transcript, coding a 3-O-glucosyltransferase, which is involved in the synthesis of cyanidin-3-O-glucoside. The gene was upregulated in leaves of 'Ca. P. solani'positive plants only in September (Figure 8). The LAR gene, involved in proanthocyanins (flavan 3-ols) synthesis, was upregulated by 3-fold in July in 'Ca. P. solani'-positive plants and even more in mid-summer (16-fold), while in late summer, although overexpressed, the transcript level decreased (4-fold) (Figure 8).

SA and Biosynthesis of Its Derivatives
With the data obtained on SA accumulation and the increase in its GA derivative, we investigated the expression of genes such as ICS (E.C. 5.4.4.2) and DMR6 (E.C. 1.14.11) (Figure 9) (in addition to PAL, whose expression profile is presented above in Figure 7), which are responsible for SA and GA syntheses, as indicated by several authors [15,41,42,51,52]. In addition, we have analyzed the expression of plant pathogenesis-related genes PR1, PR2, and PR5 and their activator NPR1 (Figure 10). The results highlighted that the NPR1 gene was still upregulated in spring and maintained its upregulation until August. The ICS gene was upregulated in 'Ca. P. solani'-positive plants for the full growth season (Figure 9). The expression of DMR6 was already high in positive plants (Figure 9) in spring and reached its highest level of overexpression in July (about 14-fold change), maintaining the upregulation in August (about 5-fold change) and September (about 2-fold change).

Expression Patterns of SA-Dependent Defense-Related Genes
In addition, we have analyzed the expression of plant pathogenesis-related genes PR1, PR2, and PR5 and their activator NPR1 (Figure 10). The results highlighted that the NPR1 gene was still upregulated in spring and maintained its upregulation until August.

Discussion
Many studies on plant response to phytoplasma infection have analyzed the late growing season, when the disease symptoms are well evident, analyzing the metabolome and the expression of targeted genes [12,53]. Instead, in our study, we covered a complete growing season to evaluate both the early response of grapevine to 'Ca. P. solani' and the late response when disease symptoms appear.
As previously stated, typical BN symptoms on Sangiovese grapevines in central Italy are generally not evident in spring and become visible in the late summer [13]. In fact, our experiments confirmed that the leaves of 'Ca. P. solani'-positive plants showed no symptoms in spring and in early summer (May and July). On the other hand, weak symptoms, such as reddish bands, appeared along the main veins in August, and gradually covered large leaf areas in September, when plants showed severe symptoms such as discoloration of veins and the reddening of laminas. A similar behavior was observed in 'Nebbiolo', which is less susceptible to the Flavescence dorée phytoplasma. In this variety, symptoms are evident only in mid-summer, while in 'Barbera', a susceptible cultivar, severe symptoms are already manifest in early summer [15]. Moreover, in 'Manzoni Bianco', a BN medium-susceptible cultivar, the symptoms on the leaves are delayed in late summer, with yellowing near the veins [24].
In our work, the analysis of the metabolites highlighted the presence of different phenolic compounds, regardless of health status, including SA and GA. Comparing free and conjugated forms of both phenolics, most were in the conjugated form, in accordance with previous works regarding other compatible interactions [54,55].
In our study, plant immunity seems to be exclusively related to SA and GA, because

Discussion
Many studies on plant response to phytoplasma infection have analyzed the late growing season, when the disease symptoms are well evident, analyzing the metabolome and the expression of targeted genes [12,53]. Instead, in our study, we covered a complete growing season to evaluate both the early response of grapevine to 'Ca. P. solani' and the late response when disease symptoms appear.
As previously stated, typical BN symptoms on Sangiovese grapevines in central Italy are generally not evident in spring and become visible in the late summer [13]. In fact, our experiments confirmed that the leaves of 'Ca. P. solani'-positive plants showed no symptoms in spring and in early summer (May and July). On the other hand, weak symptoms, such as reddish bands, appeared along the main veins in August, and gradually covered large leaf areas in September, when plants showed severe symptoms such as discoloration of veins and the reddening of laminas. A similar behavior was observed in 'Nebbiolo', which is less susceptible to the Flavescence dorée phytoplasma. In this variety, symptoms are evident only in mid-summer, while in 'Barbera', a susceptible cultivar, severe symptoms are already manifest in early summer [15]. Moreover, in 'Manzoni Bianco', a BN medium-susceptible cultivar, the symptoms on the leaves are delayed in late summer, with yellowing near the veins [24].
In our work, the analysis of the metabolites highlighted the presence of different phenolic compounds, regardless of health status, including SA and GA. Comparing free and conjugated forms of both phenolics, most were in the conjugated form, in accordance with previous works regarding other compatible interactions [54,55].
In our study, plant immunity seems to be exclusively related to SA and GA, because jasmonic acid, which is known to be involved in the plant response to phytoplasmas [26], was not detected through HPLC analysis in any sample (data not shown). This therefore suggests that, in the interaction between the grapevine cv. Sangiovese and 'Ca. P. solani', a key role is played by SA and its derivatives, such as GA.
'Ca. P. solani' seems to induce a high level of conjugated SA forms in infected leaves at the beginning of the growth season, which was maintained constant until the onset of severe disease symptoms, when the conjugated SA reached the maximum accumulation ( Figure 4). GA in glycosylated form was constitutively present in the leaves of negative plants in all growth stages. However, in the leaves of positive plants, it gradually increased to approximatively 200 µg/g FW, which is about 13 times higher than conjugated SA when symptoms are more pronounced.
As reported in the literature, this behavior may be related to the required fine-tuning homeostasis of SA in plants. In addition to upstream regulation, an active SA level has been shown to be modulated through metabolic modifications, such as glycosylation, methylation, and hydroxylation [41]. SA hydroxylation is the major pathway for SA catabolism [31,37]. SA could therefore represent the early signaling molecule that triggers the plant response to pathogen. It was then catabolized by the hydroxylation reaction to its two principal derivatives, 2,3-DHBA and GA, which then became glycosylated ( Figure 11). increased to approximatively 200 µg/g FW, which is about 13 times higher than conjugated SA when symptoms are more pronounced. As reported in the literature, this behavior may be related to the required fine-tuning homeostasis of SA in plants. In addition to upstream regulation, an active SA level has been shown to be modulated through metabolic modifications, such as glycosylation, methylation, and hydroxylation [41]. SA hydroxylation is the major pathway for SA catabolism [31,37]. SA could therefore represent the early signaling molecule that triggers the plant response to pathogen. It was then catabolized by the hydroxylation reaction to its two principal derivatives, 2,3-DHBA and GA, which then became glycosylated ( Figure  11).  (1,2,5) genes. Sugar molecule is represented in blue.
We found that 2,3-DHBA was accumulated from mid-summer ( Figure 5A), in line with Zhang et al. [37,40]. These authors reported that SA 3-hydroxylase, which is responsible for 2,3-DHBA biosynthesis, was more specifically expressed at the mature and senescence stage, suggesting its key role in preventing SA overaccumulation. In our experiment, 2,3-DHBA was detected in conjugated form, supporting the idea that this metabolite represents an inactive form of SA. We found that 2,3-DHBA was accumulated from mid-summer ( Figure 5A), in line with Zhang et al. [37,40]. These authors reported that SA 3-hydroxylase, which is responsible for 2,3-DHBA biosynthesis, was more specifically expressed at the mature and senescence stage, suggesting its key role in preventing SA overaccumulation. In our experiment, 2,3-DHBA was detected in conjugated form, supporting the idea that this metabolite represents an inactive form of SA.
Most of GA (about 99%) was present in the leaves of positive plants as the conjugated form ( Figure 4C). On the other hand, the maximum content of GA free-form content was detected in September, but representing less than 1% of the total (free + conjugated), and negative plants accumulated very low levels of free GA during the growth season ( Figure 4A).
Although SA was also predominant in the conjugated form, in 'Ca. P. solani'-positive plants, the free form was present at a higher level than in the leaves of negative plants, and it quickly increased (about threefold) to a similar level to free GA in September in positive plants ( Figure 4B). This could be explained by considering that free SA basal levels vary in different plant species (in Arabidopsis, the level ranges from 0.24 to 1 µg/g FW; in Oryza sativa, from 0.01 to 37.19 µg/g FW) [40]. Plants generally maintain SA homeostasis by fine-tuning the balance between the biosynthesis and catabolism of SA to regulate biological functions, photosynthesis, and pathogen responses [56]. Therefore, upon pathogen infection, a small amount of SA produced in planta remains in a free state in order to activate the defense response signaling. However, most SAs are subjected to biological modifications to prevent toxic accumulation as shown in Arabidopsis, where SA was glycosylated and then translocated to vacuole for storage [31,37]. Conjugated SA is the most common inactive form of SA, which can be transformed back into the active form when plants are attacked by pathogens [30].
Our finding agrees with the work of Zhang et al. [40], who reported that in Arabidopsis, free 2,3-DHBA and GA were not detected by HPLC analysis, indicating that the levels of these free acids are much lower than free SAs. In addition, according to previous findings [55,57], the conjugation of phenolics with sugar may regulate the endogenous level of free phenolics to protect plants from the toxic effects of free phenolics.
Conjugated GA accumulates in response to different types of plant pathogen interactions in much higher levels than conjugated SAs [58,59], and although the enzymes that catalyze GA glycosylation are known, the enzymes that convert SA in GA are still unknown. Some glycosyltransferases have been identified due to their function in the glycosylation of 2,3-DHBA and GA in vitro [60]. Huang et al. [41] demonstrated that the glycosylation reaction catalyzed by UGT76D1 glycosyltransferase (E.C. 2.4.1) on DHBAs plays a key role in the plant innate immune response through the modulation of SA homeostasis because UGT76D1 overexpression lines show the upregulation of the SA-responsive PR1 and PR2 genes. The same authors also suggested that the increase in conjugated DHBAs activates a mechanism of a positive feedback loop to induce SA biosynthesis, providing a continuous flux toward SAs and DHBAs, which occurs with an enhanced plant defense response.
SA biosynthesis requires the primary metabolite chorismate, and occurs through two major enzymatic pathways, one involving the phenylalanine ammonia lyase (PAL) pathway, and another involving a two-step process metabolized by the enzyme isochorismate synthase (ICS), which converts chorismate to isochorismate [42]. Our data showed an upregulation of the PAL transcript in leaves of 'Ca. P. solani'-positive plants in late summer. On the other hand, ICS expression was higher in positive plants during all sampling stages (Figure 9), with a good correlation with SA accumulation during the growth season in the leaves of positive plants. In agreement with our results, several studies have revealed that the SA biosynthetic genes are upregulated in whole leaves of grapevines infected by 'Ca. P. solani' [23,25,26,35].
The importance of the two pathways varies in different species. In Arabidopsis, the ICS pathway seems to be more important for SA biosynthesis, while rice plants use the PAL pathway, and soybean activates both pathways [61]. For grapevine, we hypothesize ( Figure 11) that plants react to phytoplasma infection by already activating the ICS pathway in spring, even if no symptoms are evident, preferring the ICS pathway for SA synthesis. On the other hand, the PAL upregulation registered in infected plants in late summer is probably linked to phenylpropanoid biosynthesis.
Our results on the expression profile of DMR6 (Figure 9) showed an upregulation in infected plants already in spring, correlating with the high content of GA observed in 'Ca. P. solani'-positive plants. This suggests that GA could be synthesized from SA through a reaction catalyzed by the DMR6-like protein. In accordance with our data, DMR6 was shown to be upregulated early in BN-diseased and -recovered grapevine plants [25]. In addition, a statistical model of general plant pathology proposed by Rotter et al. [35] assumed that BN is linked to the differential expression of the DMR6 gene in infected and uninfected plants of 'Chardonnay' grapevine plants. In line with Prezelj et al. [34], this suggests that DMR6 represents a potential early marker gene in the diagnosis of BN grapevine disease. Zhang et al. [40] have proposed that DMR6, a 2-oxoglutarate-Fe (II) oxygenase, acts in Arabidopsis as a salicylic acid 5-hydroxylase (S5H), which converts SA in GA in vivo and in vitro. The same authors speculated that the role of DMR6 is to regulate SA homeostasis during the plant response to pathogens through a feedback mechanism induced by SA.
Although studies have highlighted its involvement in plant immunity through a direct and indirect role, the biological role of GA remains unclear. According to Belles et al. [50,58], GA represents a pathogen-inducible signal in addition to SA for the activation of plant defense, based on the accumulation of GA in conjugated form in the case of exocortis viroid (CEVd) or Prunus necrotic ringspot virus (PNRSV) infection in Cucumis sativus and Gynura aurantica. In addition, an increase in conjugated GA, although lower, was registered in cucumber inoculated with a low titer of Pseudomonas syringae pv. tomato.
Other authors [40,41,62] suggest that GA is involved in plant pathogen resistance by acting as a regulator of SA homeostasis.
To evaluate the probable role of GA in salicylic signaling function in response to phytoplasma attack, we analyzed the expression of the genes coding well-known plant pathogenesis-related proteins such as PR1, PR2, and PR5, which are considered as SA markers. We also analyzed the gene coding NPR1, a key regulator of SA-mediated signaling transduction, which is a useful molecular marker for the SAR response whose expression is salicylic acid responsive. The NPR protein could be considered a hub that controls the reprogramming of gene expression induced by SA, probably via interaction with other compounds [42,63].
Our results ( Figure 9) showed that NPR1 was upregulated in infected grapevines from early summer, before the symptoms' appearance, thus confirming that it represents the first activator of PR protein-mediated response. NPR1 should activate PR1, which seems to be involved in the "early" response, as it was upregulated in infected plants already in July and reached maximum expression in September when symptoms were well evident. Consistently with our data, PR1 were upregulated in 'Chardonnay', suggesting that in response to 'Ca. P. solani infection, the SA signaling pathway is triggered [24].
According to Dermastia et al. [25], who showed that PR2 and PR5 genes were upregulated in the 'Ca. P. solani'-infected grapevine cultivar 'Chardonnay', we found an upregulation of PR2 and PR5 in leaves of infected plants in the late summer, when symptoms appeared. Also, PR2 and PR5 genes are commonly considered as molecular markers for SA-dependent systemic-acquired resistance (SAR) signaling, and their expression is regulated by SA [27]. All these results suggest that 'Ca. P. solani' induces the SA-dependent response of infected Sangiovese grapevines, although we detected a higher conjugated GA level than conjugated SA in infected plants.
Therefore, in 'Sangiovese', as a probable regulator of the SA level, GA is somehow able to modulate the expression of NPR1 and pathogen-related genes, PR1, PR2, PR5, thus confirming that signaling mediated by SA and GA is responsible for the response to BN disease ( Figure 11).
We analyzed the phenylpropanoid content in both health statuses, as the flavonoid metabolism is involved in SA-related stress signaling, as reported by other authors [22]. The accumulation of flavonoids and the activation of genes involved in the flavonoid biosynthesis have also been found in multiple phytoplasma infected plants, such as paulownia, grapevine, jujube, and Mexican lime [64]. An increased flavonoid synthesis in phytoplasmainfected plants may be part of the plant's natural defensive response against pathogen infections. HPLC ESI/MS-TOF analysis identified some phenolic acids, known as hydroxycinnamic acids (HCAs) with a significant increase in p-coumaric and ferulic acid only in symptomatic 'Ca. P. solani'-positive plants. Their accumulation has been frequently observed as a result of pathogen infections [65,66]. A variation of the HCA content was also reported by Geny et al. [67] after fungal infection in grape berries. On the other hand, an accumulation of ferulic acid has been observed in Vitis vinifera cv. 'Chardonnay' BN-diseased plants in earlier phenological stages of shoot lignification [68]. The increase in p-coumaric level in September in extracts obtained from the leaves of positive plants, together with the high upregulation of the CH4 gene, explains the role of p-coumaric acid as a precursor of many flavonoid compounds.
Our results showed higher levels of catechin and epicatechin in 'Ca. P. solani'-positive plants in September. These results were complemented by transcript analysis, which underlined that in positive plants, the LAR transcript showed a consistent upregulation starting from July until late summer, for a longer period of time than indicated by Negro et al. [38].
Flavan-3-ols, including the monomeric catechin and polymeric proanthocyanidins (PAs), are major end products of the flavonoid biosynthetic pathway in many plant species [69]. They are involved in the protection against pathogen infection, as their biosynthesis is often induced by mechanical wounding and pathogen infections [70,71]. In poplar, their content increased in infected leaves; the transcript of LAR and anthocyanidine reductase (ANR) genes, involved in the last steps of biosynthesis, were upregulated upon infection [72].
An interesting result in relation to anthocyanins was that cyanidin 3-O-glucoside reached the highest level in 'Ca. P. solani'-positive plants in the last sampling period. Because one of the typical symptoms provoked by phytoplasma is the leaf redness, the anthocyanins accumulation is likely responsible for this phenomenon [73]. In addition, we found a high level of quercetin and kaempferol in positive plants in mid and late summer, respectively. These flavonols, which are a subgroup of flavonoids, are primarily synthesized from dihydroflavonols by FLS; the trend of these compounds correlates with an increase in FLS expression, in infected leaves. Similarly, in grapevines infected by viruses, flavanols synthesis was enhanced by the higher expression of FLS than healthy ones [74]. These results confirm the assumption [13,23,74] that, upon phytoplasma infection through the activation of the NPR1 regulator, SA signaling determines a reprogramming of genes expression, which results in a modulation of phenylpropanoid biosynthetic pathway genes, which are generally upregulated in infected plants that show symptoms.
Our data suggest that SA represents the early signaling molecule that triggers several pathways involved in the 'Sangiovese' physiological response to BN, culminating with disease development. This mechanism could provide Sangiovese with a partial resistance to BN disease, although it is unable to completely combat it.
We proposed a hypothetical working model for the signal mechanisms mediated by SA in the interaction of Sangiovese grapevine-phytoplasma during the growth season ( Figure 11). When plants are challenged by phytoplasma, this triggers plant hormone regulation, which significantly increases SA biosynthesis. SA immediately activates a signaling pathway in the host, mainly consisting of the reprogramming of different physiological mechanisms, through the induction of some phenylpropanoids biosynthetic and SA-responsive genes, such as NPR1, PR1, PR2, and PR5.
When SA is accumulated at a low basal level (May), a gene coding for SA hydroxylation, such as DMR6, was induced in infected leaves to prevent SA accumulation, thus increasing the GA content. Most of the GA produced was quickly glycosylated, and the continuous accumulation of glycosylated GA probably triggers an unknown positive feedback mechanism to synthetize SA, in line with the model proposed by Huang et al. [41]. This mechanism ensures a flux of SAs and DHBAs to maintain the activation of the plant response to phytoplasma, and at the same time, limiting SA toxicity.
In support of this hypothesis, a constant upregulation of ICS was detected during the growth season; moreover, although we did not analyze a grapevine glycosyltransferase, SA, GA, and 2,3-DHBA were found in the conjugate form throughout the experimental period, confirming the important role of glycosylation.

Plant Samples and Phytoplasma Detection
Field surveys were conducted in a cv. Sangiovese vineyard (Vitis vinifera, L. 'Sangiovese' I-SS F9 A5 48) located in Greve in Chianti (Tuscany, central Italy) where 'Ca. P. solani'-positive and -negative plants were detected through multi-year monitoring. Leaf sampling was carried out at different periods according to symptom appearance from late spring to late summer (May, July, August, and September), following the gradual appearance of symptoms in positive plants. Sampling was always performed on the same plants all the time, collecting 10-15 leaves from five 'Ca. P. solani'-positive plants and five 'Ca. P. solani'-negative plants. In all sampling periods, the severity of symptoms was classified according to a grapevine reddening symptomatic scale from 0 to 3, as reported by Pierro et al. (2018): (i) symptom severity class 0 = plants with no symptoms, (ii) symptom severity class 1 = one shoot with mild leaf symptoms, (iii) symptom severity class 2 = two to three shoots with leaf symptoms, and (iv) symptom severity class 3 = more than three shoots with symptoms of reddening leaf and berry shrivel.
The collected leaves were stored at −20 • C until DNA extraction for phytoplasma detection or were lyophilized (Christ alpha 2-4 LSC plus, Osterode am Harz, Germany) for biochemical analysis. The DNA was extracted following the procedure described in Nicolì et al. [75]. Specific detection of 'Ca. P. solani' was carried out by amplification of 16S ribosomal DNA using a TaqMan assay following reaction conditions as described in Angelini et al. [76]. A threshold cycle of <37 was associated with the presence of 'Ca. P. solani'.
Both 'Ca. P. solani'-positive and 'Ca. P. solani'-negative plants were tested for some of the most common viruses of Vitis spp. (European Commission directive 2005/43/EC). Diagnostic tests (real-time PCR) were carried out for grapevine fanleaf virus, arabis mosaic virus, grapevine leafroll-associated virus 1, grapevine leafroll-associated virus 3, and Grapevine fleck virus [77][78][79]. Both healthy and infected samples were collected from plants that had negative results in all diagnostic tests. In addition, the protection of 'Ca. P. solani'-positive and 'Ca. P. solani'-negative plants was carried out according to common practices in the area, and sampled plants showed no symptoms related to Uncinula necator (Schw.) Burr., Plasmopara viticola (Berk. et Curt.) Berl. et de Toni, Botrytis cinerea Pers, and Guignardia bidwellii (Ellis) Viala and Ravaz.

Extraction of Phenylpropanoids
Samples (about 0.5 g of tissue) were ground in a pre-cooled mortar to a fine powder using liquid nitrogen and then homogenized in 1.5 mL water/methanol 40/60 v/v. The extracts were centrifuged for 15 min at 10,000× g to remove cellular debris. The supernatant for each sample was divided into two equal aliquots and vacuum-dried at 40 • C. An aliquot was resuspended in 900 µL of 50 M −3 sodium acetate buffer (pH 4.5) and 100 µL of water to analyze free SA and GA together with phenolic compounds. Another aliquot was resuspended in 900 µL of 50 M −3 sodium acetate (pH 4.5) and 100 µL of water containing 10 U of almond β-glucosidase (3.2.1.21) (14.3 U mg −1 , Sigma Aldrich, Milano, Italy) to remove sugars bound to phenolic compounds. The enzymatic reaction was incubated overnight at 37 • C and then stopped by adding 75 µL of water/perchloric acid 30/70 v/v to the incubation mixture and stored at 4 • C for 1 h. After centrifugation at 14,000× g for 15 min to remove polymers, the supernatants were extracted with 2.5 mL of cyclopentane/ethyl acetate 1/1 v/v. The organic upper phase was collected and dried at 40 • C. The residue was resuspended in 200 µL of methanol and filtered through 0.22 µm prior to HPLC analysis.

HPLC ESI/MS-TOF Analysis
The phenolic characterization and quantification of leaf extracts were performed using an Agilent 1200 High Performance Liquid Chromatography (HPLC) System (Agilent Technologies, Palo Alto, CA, USA) equipped with a standard autosampler and an Agilent Zorbax Extend-C18 analytical column (5 × 2.1 cm, 1.8 µm), as reported by Vergine et al. [80]. The HPLC system was coupled to an Agilent diode-array detector (wavelength 280 nm) and an Agilent 6320 TOF mass spectrometer equipped with a dual ESI interface (Agilent Technologies) operating in negative ion mode. Detection was carried out within a mass range of 50-1700 m/z. Accurate mass measurements of each peak from the total ion chromatograms (TICs) were obtained by an ISO Pump (Agilent G1310B) using a dual nebulizer ESI source that introduces a low flow (20 µL min −1 ) of a calibration solution containing the internal reference masses at m/z 112.9856, 301.9981, 601.9790, and 1033.9881, in negative ion mode. The anthocyanins were identified with the same method, but with positive ionization (detection wavelength 280 and 520 nm), using the internal reference masses at m/z 121.050873, 149.02332, 322.048121, and 922.009798, as reported by Aprile et al. [81].

RNA Extraction and qRT-PCR
Leaf tissues sampled from healthy and infected plants at four growth stages were frozen in liquid nitrogen, and total RNA was isolated from 0.1 g of samples using TRI-ZOL (Invitrogen, Carlsbad, CA, USA). cDNA synthesis was carried out using TaqMan ® Reverse Transcription Reagents (Applied Biosystems, Foster City, CA, USA) according to the manufacturer's standard protocol. Amplification reactions were performed using the Applied Biosystems ® QuantStudio ® 3 Real-Time PCR System. Each reaction consisted of 2 ng of cDNA, 12.5 µL of Power SYBR Green RT-PCR Master mix (Applied Biosystems), 5.0 M-6 forward and reverse primers, and ultrapure DNase/RNase-free water (Carlo Erba Reagents S.r.l.) in a total volume of 25 µL. The cycling conditions were as follows: 2 min at 50 • C and 10 min at 95 • C, followed by 45 cycles of 95 • C for 15 s, and 60 • C for 1 min. Melting curve analysis was performed after PCR to evaluate the presence of non-specific PCR products and primer dimers.
The primers (Supplementary Table S1) were designed with Primer Express Software 3.0 on the mRNA sequences obtained from the literature [13,23,34,74].
Quantitative real-time PCR was used for rapid and reliable quantification of mRNA transcription. However, selecting an appropriate reference gene is crucial for an exact comparison of mRNA transcription in different samples. Of the various genes reported in the literature, we used COX (cytochrome c oxidase, E.C. 1.9.3.1) as a reference gene, as reported by Bertazzon et al. [82].
For the relative quantification of gene expression, we calculated the fold changes (FCs) using the following formula: With the cut-off value of a 2-fold change in gene expression, we considered the upregulation in positive plants compared to negative plants.

Statistical Analysis
All data were reported as the mean ± SD in triplicate for each analyzed sample (n = 5, 'Ca. P. solani'-positive and -negative plants, respectively). The statistical analysis was performed using multiple t-tests (FDR = 5%) to highlight the differences between 'Ca. P. solani'-positive and -negative leaves for each physiological parameter analyzed. Statistical analyses were performed using GraphPad v. 6.01. A one-way ANOVA test was applied to expression gene data.

Conclusions
Our results confirm the previous evidence that phytoplasma interact with the SA pathway [36]. In addition, we identified a high level of conjugate gentisic acid in grapevine infected by Candidatus Phytoplasma solani.
The altered physiological conditions may also be a consequence of the phytoplasma effects on development and stress signaling pathways and of the interactions between them.
The biology of phytoplasmas and the actual defense mechanisms of plants are still unknown because the pathogens need plants and insects for survival in nature, which means that an "in vitro" cultivation is very difficult. However, the metabolomic and transcriptomic data that we obtained confirm that innate immunity, phytohormone signaling, and many phenylpropanoid compounds, which constitute a complex defense network in plants, are involved in the response of grapevine-to-phytoplasma infection.
Although SA is essential in the grapevine-phytoplasma interaction, it is not the exclusive signal, with GA appearing to play a role in enabling or modulating the grapevine response to phytoplasma infection. GA did not interfere with the biological effects of SA; however, it represents a component of the grapevine SA-dependent response, probably for the role to fine-tune the SA level.
Our results suggest a new point of view concerning the physiological mechanisms underlying phytoplasma-grapevine interactions aimed at improving disease control strategies.
Supplementary Materials: The following supporting information can be downloaded at: https://www.mdpi.com/article/10.3390/plants12142695/s1, Figure S1: Mass spectra of the 2,3 DHBA found in the sample (A) and of the chemical standard (B) and their UV/Vis absorption spectra of the peak 8B shown in Figure 1 after sample digestion with β-glucosidase (C); Table S1: Primer selected for the gene expression analysis.