Introduction

Flavescence dorée is an economically important disease of grapevine caused by a phytoplasma (FDp) belonging to the 16SrV ribosomal group, subgroups -C and -D (EFSA Panel on Plant Health PLH 2014). The disease is widespread in most of the grapevine growing areas of Europe, and FDp is considered as an A2 Quarantine pest and EU candidate priority pest (EFSA et al. 2019). In the absence of resistant Vitis vinifera genotypes (Eveillard et al. 2016; Ripamonti et al. 2021), infected plants show severe leaf and vascular symptoms (Rizzoli et al. 2022), leading to bunch drying and in some cases to plant death (EFSA et al. 2020). The phytoplasma is efficiently transmitted to grapevine mainly by the Nearctic leafhopper Scaphoideus titanus Ball (Hemiptera: Cicadellidae) (Chuche and Thiéry 2014). The persistent propagative transmission is driven by molecular interactions between vector proteins and FDp membrane proteins (Trivellone et al. 2019a; Arricau-Bouvery et al. 2021), and specific phytoplasma proteins are associated to epidemic FDp strains transmitted by S. titanus (Malembic-Maher et al. 2020). Other epidemiological cycles involving different vectors (e.g. Dictyophara europaea, Orientus ishidae, Oncopsis alni, Allygus spp., Malembic-Maher et al. 2020) and spontaneous vegetation (among which Ailanthus altissima, Clematis vitalba, Corylus avellana, Alnus spp., EFSA Panel on Plant Health PLH et al. 2016; Mehle et al. 2019) contribute in maintaining the natural inoculum of this Palearctic phytoplasma. Sudden FD outbreaks from the wild compartments of cultivated vineyards may then be further propagated by the S. titanus population within the vineyard (Rizzoli et al. 2021). Flavescence dorée management relies conventionally on control of vector population through insecticide applications and roguing of infected plants to reduce inoculum loads and planting of phytoplasma-free grafted cuttings (Oliveira et al. 2019).

Phytoplasmas are obligate intracellular bacteria colonizing the plant phloem and several organs of their insect vectors. Few complete phytoplasma genomes and several partial drafts highlight the reduced genome size and lack of crucial metabolic pathways (e.g. ATP synthesis, de novo synthesis of nucleotides), which may result from their endo-parasitic life style (Oshima et al. 2013; Debonneville et al. 2022). Despite this genome reduction, phytoplasmas encode several effectors that can be secreted into the host environment and manipulate the host behavior to favour the infection cycle (Tomkins et al. 2018), although differences in the effector repertoires exist, at least in the case of FDp (Debonneville et al. 2022). Moreover, FDp also encodes a specific set of restriction nucleases that has not been described for other phytoplasmas, while it shows a low level of genome plasticity (Debonneville et al. 2022). Functional studies on phytoplasma genomes are hampered by difficulties in establishing axenic cultures and shortage of fully annotated complete genomes. The present work aims at addressing this gap of knowledge, by exploring the transcriptional profiles of several FDp genes, belonging to different functional metabolic categories, during infection of different plant and vector hosts. Defining a core list of essential phytoplasma genes during host parasitic process is indeed crucial to facilitate axenic cultivation conditions and possibly disrupt the transmission cycle, paving the way for innovative strategies of disease management. In the absence of routine axenic cultivation protocols, functional studies of FDp genes are achievable by exploiting both natural and laboratory pathosystems, namely consisting of V. vinifera infected by S. titanus and Vicia faba infected by Euscelidius variegatus (Caudwell et al. 1972), respectively. These host/pathogen associations are driven by co-accomodation (sensu Brooks 1979), due to the quite recent introduction of S. titanus in the Palearctic area (Bertin et al. 2007; Papura et al. 2012). Indeed, the high efficiency of S. titanus in transmitting the Palearctic FDp to grapevine caused the epidemic development of the disease in Europe with serious damage to viticulture. The very limited ecological opportunity for the pathogen to be transmitted to broad bean by E. variegatus under natural conditions also may lead to a low level of co-evolution (or co-adaptation) among FDp and these laboratory hosts. Consistently, FDp has a detrimental effect on insect fitness of both vector species (Bressan et al. 2005a, b) and is perceived as a pathogen, since vector immune responses are activated upon FDp infection (Galetto et al. 2018; Gonella et al. 2019).

This work explores the transcriptional response of FDp during infection of the natural and experimental hosts. The results describe the pathogen’s ability to respond differently to each host, and also provide hints on some host-independent transcriptional responses, suggesting potential roles for some FDp genes with so far unknown functional roles.

Materials and methods

Phytoplasma isolate, insect vectors, and host plants

Flavescence dorée phytoplasma isolate (FDp, 16SrV-C) was identified in a vineyard at Cocconato d’Asti, Piedmont (north-western Italy) (45°04’58.4’’ N 8°03’21.1’’ E, N-S orientation), and transmitted to broad bean, Vicia faba L., by the natural vector S. titanus previously fed on infected grapevines; FDp was then maintained on broad bean by insect transmission, using the laboratory vector Euscelidius variegatus (Galetto et al. 2014).

Euscelidius variegatus were originally collected in Piedmont region of Italy and continuously reared on oat (Avena sativa L.) plants from seed, inside plastic and nylon cages in growth chambers at 20–25 °C with a L16:D8 photoperiod. Scaphoideus titanus has one generation per year and its continuous rearing under controlled conditions is not feasible. To obtain S. titanus specimens for the present work, two-year-old grapevine canes bearing leafhopper eggs were collected in Veneto (Italy) vineyards during winter and kept at 5 ± 1 °C. To allow coordinated egg hatching, grapevine branches were caged inside insect-proof screen houses in a glasshouse with natural light and temperature ranging from 20 to 25 °C. Potted grapevine cuttings and healthy broad bean plants from seed were introduced in the screen house to feed the newly hatched nymphs.

Broad bean plants were all grown from seed in greenhouses at 20–25 ◦C with a L16:D8 photoperiod. Naturally infected grapevine plants (Vitis vinifera L. ‘Chardonnay’) were located in the same commercial vineyard at Cocconato d’Asti, Piedmont. The vineyard was managed with integrated pest management and regularly treated with mandatory insecticides against S. titanus according to the annual directives of the Regional Phytosanitary Service.

To evaluate phytoplasma gene expression profiles in V. faba, experimental plants were inoculated with FDp by E. variegatus as infective vector. About 100 E. variegatus nymphs were fed on infected broad beans for an acquisition access period (AAP) of 14 days and were then transferred on oat (immune to phytoplasma) for a 14-day latency period (LP). Six V. faba plants were caged together and exposed to about 50 infective insects for a 7-day inoculation access period (IAP) and were then treated with insecticide. Leaf samples were collected from the broad beans plants at 20, 30 and 40 days post inoculation (dpi) for nucleic acid extraction.

To evaluate the phytoplasma gene expression profile in V. vinifera, ten symptomatic plants were collected from the same vineyard as described above at early July (according to Roggia et al. 2014), analyzed to confirm FDp presence by PCR and exclude Bois noir phytoplasma (BNp, 16SrXII-A) infection as detailed below and then the same plants were re-sampled in mid August and beginning of September of the same season. In the end, the transcription analysis was completed on leaf samples collected in July, August and September from the same six grapevine plants.

To evaluate the phytoplasma gene expression profile in insect vectors, FDp-infected E. variegatus and S. titanus were used. About 100 nymphs of each species were collected from healthy colonies reared as described above, caged for a 14-day AAP on FDp-source broad beans, and then maintained on healthy plants (oats for E. variegatus and grapevines for S. titanus). About 10 insects of each species were collected at 21, 28, and 35 days post acquisition (dpa) for nucleic acid extraction.

The sanitary status of all source plants was confirmed by symptom observation and PCR diagnosis as previously described.

Extraction of nucleic acids

Extraction of total DNA and RNA from the same sample unit was carried out according to Pacifico et al. (2015).

Insect samples collected at different dpa were stored at − 80 °C before DNA and RNA extraction. Both total DNA and total RNA were extracted from single insects. A few liquid nitrogen drops were spilled into a 1.5-ml tube containing a single leafhopper and the insect was then quickly crushed using a sterile micropestel in 200 µl of TE buffer (10 mM Tris, 1 mM EDTA) prepared with diethyl pyro- carbonate (DEPC) (0.1%) water. The resulting homogenate suspension was rapidly divided for DNA and RNA extraction; 100 µl was added to 400 µl of 3% CTAB buffer for DNA extraction, whereas 100 µl was added to 400 µl of TRIzol reagent (Invitrogen, USA) for RNA extraction following the manufacturer’s instructions.

Plant samples (about 200 mg of leaf veins), collected at different dates, were pooled and divided into 100-mg aliquots stored at -80 °C before DNA and RNA extraction. Total DNA was extracted from 100 mg of plant leaf veins using a modified cetyltrimethyl ammonium bromide (CTAB) procedure originally described in Daire et al. (1997), and the final DNA pellet was dissolved in 50 µl of sterile double-distilled water (ddH2O). Total RNA from the remaining 100 mg aliquot was extracted using TRIzol reagent (Invitrogen, USA) following the manufacturer’s instructions.

Total RNA samples, extracted from both insects and plants, were treated with RNase-free DNase I (Life Technologies, Italy) in the supplied buffer to avoid residual DNA contamination. Following the digestion, the DNase was inactivated by phenol-chloroform extraction according to the manufacturer’s instructions. RNA was finally suspended in 30 µl of RNase- free 0.1% DEPC treated water. Nucleic acid extracts were analyzed in a NanoDrop spectrophotometer to evaluate the concentration and purity and stored at -80 °C.

Phytoplasma detection and quantification

The presence of FDp in E. variegatus, S. titanus, V. faba, and V. vinifera samples was verified by qPCR using the protocol described in Pelletier et al. (2009). The DNA extracts from 6 infected plants of both species and from 5 to 7 infected insects of both species for each sampling date (a total of 18–20 samples for each of the four species) were diluted in ddH2O at 1 ng/µl and used to measure the absolute number of phytoplasma genome units (GU) per nanogram of host DNA (Marzachí and Bosco 2005). In particular, 1 µl of diluted DNA samples was used as a template for qPCR in a 10 µL volume mix, containing 1X iTaq Universal Sybr Green Supermix (Bio-Rad, USA) and 300 nM of each primer targeting phytoplasma 16 S rRNA genes. In parallel, 1 µl of diluted DNA samples was also used as a template for qPCR in a 10 µL volume mix, containing 1X iTaq Universal Sybr Green Supermix (Bio-Rad, Hercules, CA, USA) and 300 nM of the appropriate primer pairs targeting host 18 S rRNA genes. All primer pairs used for qPCR are listed in Supplementary Table 1. CYS2Fw/Rv primer pair (Marzachí and Bosco 2005) was used to quantify phytoplasma presence, whereas, to quantify host DNA, Vitis18SF1/R1 (Roggia et al. 2014) and MqFw/MqRv (Marzachí and Bosco 2005) primer pairs were used to measure plant and insect nucleic acids, respectively. Samples were run in triplicate in a CFX Connect Real-Time PCR Detection System (Bio-Rad, USA). Cycling conditions were: 95 °C for 3 min, and 40 cycles at 95 °C for 15 s, and 60 °C for 30 s of the annealing/extension step. The specificity of the PCR products was verified by melting curve analysis for all samples. No-template controls were always included in each plate. The absolute quantification of phytoplasma cells was achieved by comparing the quantification cycles (Cqs) of the samples with those of four serial dilutions of pOP74 plasmid, containing a fragment of the phytoplasma 16 S rRNA gene (Marzachí and Bosco 2005). One fg of pOP74 contains 194 molecules of plasmid, with each containing a single copy of the 16 S rRNA gene. Because this gene is present in two copies in phytoplasma genomes, 1 fg of pOP74 corresponded to 97 FDp GU. For the absolute quantification of host DNA, four quantities of total healthy host DNA diluted in ddH2O (50, 5, 0.05 and 0.005 ng) were used. The final concentration was expressed as FDp GU/ng of host DNA.

Selection of phytoplasma target genes

Flavescence dorée genes selected for this study are listed in Table 1. The 11 genes were selected on the basis of the previous description of RNA-Seq profile of FDp in grapevine (Abbà et al. 2014), to ensure quantifiable expression of the selected genes. In particular, the target genes were selected taking into account three criteria: (1) a wide range of in silico expression levels, from the highest (contig12) to the lowest (rpoD), (2) possible involvement in host-phytoplasma interactions and (3) annotated as hypothetical protein. Eleven target genes were selected: contig12 as the most expressed transcript, one known surface-exposed protein (imp) (Rossi et al. 2019), two highly-expressed hypothetical proteins (Comp83, Comp115), a protein known to regulate the excretion of virulence factors in other bacteria (SpoVG) (Schulthess et al. 2011), a protein involved in the secretion pathway (FtsY) (Oswald et al. 2021), two proteolytic enzymes potentially contributing to virulence (YsdC and TldD) (Allali et al. 2002; Badger et al. 2005; Dutoit et al. 2012; Tsibulskaya et al. 2017), a protein involved in the defense mechanisms against oxidative stress (OsmC) (Checroun and Gutierrez 2004; Lang et al. 2015), the RNA polymerase sigma factor RpoD (Miura et al. 2015) and an ABC transporter (CoABC). For contig12, the selected target sequence was CP097583.1:597856–598,553, as in Abbà et al. (2014). The gene was later shown to be present in two copies (Accession numbers: CP097583.1:597856–598,553 and CP097583.1:180770–181,467) in the FDp genome (Debonneville et al. 2022). Under this circumstance, primer pair contig12Fw3/contig12Rv2 (Supplementary Table 1) amplified both of them, despite the presence of 2 internal mismatches in the reverse primer sequence.

Table 1 List of Flavescence dorée genes selected for the study, grouped by their functional category
Full size table

cDNA synthesis and mRNA quantification

For absolute quantification of phytoplasma mRNAs during insect vector and plant infection, standard curves were produced using serial dilutions of recombinant plasmids carrying a fragment of the corresponding genes obtained in (Abbà et al. 2014). In the absence of a phytoplasma endogenous control mRNA, the expression level of each pathogen transcript was correlated to the bacterial population (FDp GU) measured by qPCR for each experimental sample (100 mg of plant material, or single insect). Standard curves were constructed by linear regression analysis of the Cq value of each standard dilution replicate over the log of the number of plasmid copies present in each sample. Data acquisition and analysis were handled by the use of CFX Manager software, version 3.0, which automatically calculates the Cq values and the parameters of the standard curves. For each sampling date, cDNA was synthesized from total RNA (500 ng) using a High Capacity cDNA reverse transcription kit (Applied Biosystems, USA) according to the manufacturer’s instructions, diluted 1:4 in ddH2O and stored at -80 °C in sterile microtubes until the qPCRs were performed. The qPCR mix contained 1 µl of diluted cDNA, 1X iQ SYBR green Supermix (Bio-Rad, USA), 300 nM of each primers (Supplementary Table 1) in a final volume of 10 µl. Reaction conditions were as follows: 5 min at 95 °C and 45 cycles of 15 s at 95 °C, followed by 1 min at the optimized annealing and extension temperatures (Supplementary Table 1). On each plate, samples were run in duplicate together with four 100-fold serial dilutions of the corresponding standard plasmid (from 10E + 08 to 10E + 02 target copy number/µl). The use of DNA standard curves for transcript quantification allowed comparisons of expression data from different genes (Gomes et al. 2005; Florindo et al. 2012). Complete qPCR mix with total RNA and ddH2O instead of cDNA were used as negative controls in each plate. Reactions were carried out in a CFX Connect real-time PCR detection system (Bio-Rad, USA) supported by CFX Manager software, version 3.0. Melting curves were produced at the end of the PCR to assess the reaction specificity.

An absolute quantification of bacterial transcripts was therefore performed. For each phytoplasma gene, an expression index (EI) was calculated, indicating the transcript copy number per phytoplasma GU at each sampling date and in each infected host, according to the guidelines published for cultivable bacteria (Vandecasteele et al. 2001; Stenico et al. 2014).

Data analysis

To compare the phytoplasma population sizes at different times in each host species, analysis of variance (ANOVA) was performed on log(x + 1) transformed data, followed by Sidak test for multiple comparisons.

The variation of the Expression Index of the analyzed FDp genes between host species and different sampling dates was analyzed by Principal Component Analysis (PCA) (vegan R package, Oksanen et al., 2022). An overall PCA was performed considering the four host species together (insects + plants). Given the large variation in EI of FD between insects and plants, two PCAs were carried out separately for insect and plant hosts to allow a better graphical separation of differences in Expression Index values of FD.

The Expression Index of the analyzed genes between insect host species was modelled by ANOVA (response variable log(x + 1) transformed), with Gene, Host species, Sampling Date, and the interaction terms Gene × Organism and Sampling Date × Organism as fixed effects. The Expression Index of the analysed genes between the two plant hosts (i.e. V. faba and V. vinifera) was analyzed separately by ANOVA (response variable log(x + 1) transformed), with Gene, Sampling Date and their interaction as fixed effects. Multiple comparisons were performed with Tukey’s HSD test. The Expression Index of the analysed FDp genes during infection of the laboratory system (E. variegatus and V. faba) was analysed by ANOVA (response variable log(x + 1) transformed), with Gene and Host species as fixed effects. Effect size of fixed effects in ANOVA models, measuring the proportion of the variation in dependent variable that is associated with membership of the different groups defined by the experimental factors, was calculated using generalized eta squared formula (ges, Olejnik and Algina 2003). For multiple comparisons Sidak test was performed. All statistical analyses were conducted using R software v. 3.6.2 (R Core Team 2019), using multiple packages (emmeans, tidyverse, vegan).

Results

Phytoplasma load

All analyzed samples were FD-infected (not shown). Phytoplasma loads in the different hosts are presented in Fig. 1a (raw data in Supplementary Table 2), expressed as FDp GU/ng of host DNA. Phytoplasma loads were significantly different among hosts and, in some cases, among sampling dates (ANOVA, Organism: F = 6.42, df = 3, p = .001; Organism × Sampling date: F = 9.38, df = 6, p < .001). In E. variegatus, FDp load ranged between 1.72E + 01 and 1.80E + 04, and no significant differences were recorded among the sampling dates. In S. titanus, FDp load ranged from 2.74E + 02 to 3.10E + 04, and significantly increased from 21 to 28 dpa (1st and 2nd sampling dates; t = 2.79, p = .007). In plants, FDp load ranged between 3.84E + 01 and 5.80E + 05 (V. faba) and from 8.56E + 02 to 6.14E + 04 (V. vinifera), and it increased significantly in V. faba only between 20 and 30 dpi (1st and 2nd sampling dates; t = 7.7, P < .001) and remained constant afterwards. Indeed, the phytoplasma infecting V. faba was in the log phase between the 1st and 2nd sampling dates, whereas in the case of V. vinifera, the stationary phase (Pacifico et al. 2015) was already reached at the 1st sampling date (July).

Expression of the different phytoplasma gene categories during insect and plant infections

The Expression Index of the analyzed genes within their functional categories in the different hosts overtime is described in Fig. 1b (raw data in Supplementary Table 2). All genes were more expressed in insects than in plants. Indeed, all genes were actively transcribed in all samples at the three sampling dates in S. titanus and in most of the E. variegatus samples. On the other hand, the expression of all selected genes, except for contig12, was at the lower limit of the quantification system in several samples during plant infection (Supplementary Table 2).

In V. faba, where an intense FDp multiplication occurred (from 1st to 2nd sampling dates, FDp load increased by 200-folds; Fig. 1a), expression of the analyzed genes generally increased during the first two sampling dates, while it decreased upon entry into the stationary phase. On the other hand, expression of the analyzed genes generally remained constant during the infection of all the other hosts. Genes of the same category were transcribed at similar levels within the same host, except for comp83 and osmC, both belonging to Stress response. In this category, comp83 was constantly more expressed than osmC, especially in insects (Fig. 1b).

Fig. 1
figure 1

Phytoplasma loads (panel a, Genome units of FDp/ng of host DNA) and Expression Index (log 10 transformed) of the selected genes (panel b) at the different sampling dates (1, 2, 3, as detailed in Materials and methods) and for the different hosts (Euscelidius variegatus, Scaphoideus titanus, Vicia faba, Vitis vinifera). In panel b, mean values and confidence intervals (95%) are presented. Genes within the same functional categories share the same color code. A red line is drawn at 1E-02 on the Y axes, to improve readability of the Expression Indexes for the different hosts.

Full size image

PCA analysis of the Expression index of all analyzed genes at all sampling dates (Fig. 2) separated the FDp transcription profiles in insect from those in plant hosts along the first axis (PC1). The host plants were mainly represented by V. faba, due to the constant expression at all time points of FDp genes in V. vinifera (Supplementary Table 2). FDp transcription profiles in the two insect species were separated mainly along the second axis (PC2). The first principal component accounted for most of the variability in FDp transcription profiles (92.48%), and was highly correlated to the expression of two FDp genes with unknown function (comp115, contig12), one gene of the ‘Stress response’ category (comp83), and also of imp and ysdC within ‘Membrane’ and ‘Protein metabolism and secretion’ groups, respectively.

Fig. 2
figure 2

Principal component analysis (PCA) plot of the Expression Index of the analysed genes at all sampling dates for the different hosts (Euscelidius variegatus, Scaphoideus titanus, Vicia faba, Vitis vinifera). Clusters were grouped with ellipses. The host plants were mainly represented by V. faba, due to the constant expression at all time points of FDp genes in V. vinifera. The new condensed PCA variables explained 92.28% (Dim1, x axis) and 4.12% (Dim2, y axis) of variability.

Full size image

The Expression Index of the same FDp genes was also highly correlated to first axis (PC1) of a PCA analysis including insects only (Fig. 3). The two insect hosts differed mainly in the expression of comp115 and imp, with S. titanus showing on average a higher Expression Index than E. variegatus for these genes.

Fig. 3
figure 3

Principal component analysis (PCA) plot of the Expression Index of the analysed FDp genes at all sampling dates for the different insect hosts (Euscelidius variegatus, Scaphoideus titanus). Clusters were grouped with ellipses. The new condensed PCA variables explained 86.23% (Dim1, x axis) and 7.21% (Dim2, y axis) of variability.

Full size image

According to a PCA including only the stationary phase of plant infection (V. faba: 3rd sampling date; V. vinifera: all sampling dates, Fig. 1a), 66.25% of the Expression Index variability was accounted by PC1, separating the two plant species (Fig. 4). The Expression Index of contig12 and imp genes were directly correlated to PC1 and on average higher in V. faba (3rd sampling date only). Expression of CoABC (‘ABC transporters’) was instead correlated to PC2 (7.2% of total Expression Index variability) and on average higher in V. vinifera than in V. faba.

Fig. 4
figure 4

Principal component analysis (PCA) plot of the Expression Index of the analysed FDp genes at the stationary phase of the infection for the plant hosts Vicia faba (sampled at 40 dpi), and Vitis vinifera (all sampling dates). Clusters were grouped with ellipses. The new condensed PCA variables explained 66.25% (Dim1, x axis) and 19.67% (Dim2, y axis) of variability.

Full size image

Expression of the different phytoplasma genes during insect infection

In insects, the analysis of the overall expression detected significant differences between the analyzed genes, the insect species, and the sampling dates, ranked according to the effect size (ges), as well as an interaction between the insect species and the sampling date (Table 2). Post hoc comparisons to explore significant differences in mean expression of FDp genes within each insect species (Supplementary Table 3) indicated that contig12 was the most expressed gene in E. variegatus, followed by comp115 and comp83. In S. titanus, both contig12 and comp115 showed the highest expression levels, followed by comp83 and the imp genes. The same analysis to explore significant differences in the mean expression of each FDp gene between the two insect species (Supplementary Table 4) showed that the ‘Unknown’ category genes (comp115, contig12), the ‘Membrane’ imp and the ‘Protein metabolism and secretion’ ysdC genes were more transcribed in S. titanus than in E. variegatus.

Table 2 Results of the ANOVA analyses of the overall expression of the selected FDp genes in the two vectors Euscelidius variegatus and Scaphoideus titanus, at three sampling times post infection. Sum of squares in the numerator (SSn), sum of squares in the denominator (SSd), degrees of freedom in the numerator (DFn), degrees of freedom in the denominator (DFd), F-value (F), p-value (p), Generalized Eta-Squared measure of effect size (ges). * indicates significant effects at p < .05
Full size table

Expression of the different phytoplasma genes during plant infection

The Expression Index of the FDp genes during plant infection was analyzed separately for V. faba and V. vinifera (Fig. 4; Table 3), as low variability characterized the latter species according to the previous PCA analyses (Fig. 2, Supplementary Table 2). Furthermore, sampling dates differed between the two plant species, to ensure a correct representation of the FDp cycle within the herbaceous lab-host and the woody natural one, and phytoplasmas infecting the two hosts were at different stages of their growth cycle (Fig. 1a). In V. faba, the analysis detected significant differences between the analyzed genes, the sampling dates, ranked according to the effect size (ges), as well as an interaction between the gene and the sampling date (Table 3). Post hoc comparisons to explore significant differences in mean expression of FDp genes within each sampling date (Supplementary Table 5) indicated that ysdC gene (Protein metabolism and secretion) was the most expressed at the first sampling date, while at the second sampling, contig12 (Unknown) was the most expressed, followed by imp (Membrane) which was significantly more transcribed than osmC (Stress), ftsY (Protein metabolism and secretion), and rpoD (Regulative). No differences were detected among the Expression Index of the different genes at the third sampling date, confirming entry into the stationary phase. The same analysis to explore significant differences in the mean expression of each FDp gene among the three sampling dates (Supplementary Table 6) showed that ysdC (‘Protein metabolism and secretion’) was upregulated at the first date, while high expression of contig12 (‘Unknown’), comp83 (‘Stress’), imp (‘Membrane’) characterized the second sampling date. In V. vinifera, neither genes, nor sampling dates had significant effects on the Expression Index of the selected FDp genes at the three sampling times.

Table 3 Results of the ANOVA analyses of the overall expression of the selected FDp genes in V. faba at the three sampling times post infection. Sum of squares in the numerator (SSn), sum of squares in the denominator (SSd), degrees of freedom in the numerator (DFn), degrees of freedom in the denominator (DFd), F-value (F), p-value (p), Generalized Eta-Squared measure of effect size (ges). * indicates significant effects at p < .05
Full size table

Comparison of the transcriptional response of the selected FD phytoplasma genes during infection of the Vicia faba – Euscelidius variegatus pathosystem.

The Expression Index of the FDp genes during infection of either the plant or the insect host of the laboratory pathosystem evidenced significant differences in the expression profiles of the different genes with significant interactions between gene and organism (Table 4). In particular, the Expression Index of each gene during the FDp infection of the two hosts (all sampling dates for E. variegatus, 2nd sampling dates for V. faba,) was compared (Fig. 5, Supplementary Table 7). The 2nd sampling date for V. faba infection was selected for the analyses as it showed the highest FDp Expression Index for most genes (Fig. 1b). Both genes of the ‘Unknown’ category, and comp83 of the ‘Stress response’ category were significantly more transcribed in the insect host compared to the plant, while all other genes shared similar transcription profiles. In particular, osmC (‘Stress Response’) and ftsY (‘Protein metabolism and secretion’) were overexpressed during E. variegatus infection, although the trend was not significant. Similarly, the overexpression patterns during V. faba infection of the remaining ‘Protein metabolism and secretion’ genes (ysdC and tldD) as well as that of all genes within the ‘Regulative elements’ category (spoVG and rpoD) were not significant (Fig. 5).

Table 4 Results of the ANOVA analyses of the Expression Index of the FDp genes during infection of the plant (Vicia faba) or the insect (Euscelidius variegatus) hosts of the laboratory pathosystem. Sum of squares in the numerator (SSn), sum of squares in the denominator (SSd), degrees of freedom in the numerator (DFn), degrees of freedom in the denominator (DFd), F-value (F), p-value (p), Generalized Eta-Squared measure of effect size (ges). * indicates significant effects at p < .05
Full size table
Fig. 5
figure 5

Differential expression (log2DE) of mean Expression Index of FDp genes (insect versus plant) during infection of Euscelidius variegatus (all sampling dates) and Vicia faba (sampled at 30 dpi). Positive log2DE values indicate overexpression in the insect host, while negative ones indicate overexpression in the plant host. Genes within the same functional categories share the same color code

Full size image

Discussion

The transcriptional profile of selected Flavescence dorée phytoplasma genes was explored both in the natural and the experimental pathosystems, as the latter is usually preferred for functional studies on this phytoplasma (Caudwell et al. 1972; Arricau-Bouvery et al. 2018, 2021). While it is known that phytoplasmas dramatically alter their gene expression in response to the insect or plant infection cycle (‘host switching’, Oshima et al. 2011), overall pictures on the entity of this alteration for several phytoplasmas are hard to achieve even through the application of NGS techniques (e.g. RNAseq), due to both the low and erratic concentration of these pathogens within their hosts, and the absence, in most cases, of complete phytoplasma reference genomes. We showed that FDp is metabolically more active in insects than in plants, at least as far as the pool of selected genes for this study is concerned. This more active transcriptional response of FDp in insects suggests that different phytoplasma species may behave differently from a transcriptional point of view. However, it was previously shown that ‘Candidatus Phytoplasma asteris’ isolates are metabolically more active during infection of the plant, both globally (microarray analyses, Oshima et al. 2011), but also confirmed on a more restricted gene panel (Pacifico et al. 2015). Since FDp and ‘Ca. P. asteris’ are different species, they may show different transcriptional behavior during infection of the insect or plant hosts. The possible specific upregulation of genes in the vector may be achieved through the increase of the copy number of an extrachromosomal circular potential mobile unit (PMU1) with considerable higher expression in the insect than in the plant, as shown for a ‘Ca. P. asteris’ strain (Toruño et al. 2010). On the other hand, in the case of FDp, the absence of PMUs (Debonneville et al. 2022) indicates that other strategies must be invoked to support higher transcriptional activity in the vector. Indeed, the transcription analyses of several FDp ortholog genes encoding ATP-dependent zinc proteases (ftsH) has confirmed that most of them are overexpressed in the vectors, according to the different pathosystems (Jollard et al. 2019). In particular, FDp transcription of these ftsH orthologs designs complex patterns in the V. vinifera/S. titanus natural association or the V. faba/E. variegatus experimental system (Jollard et al. 2019). This highlights the complexity of the tritrophic plant/phytoplasma/vector interactions, and the intrinsic difficulty in addressing molecular switches of this pathogen. Low Expression Indexes recorded in V. vinifera samples compared to those obtained from lab-infected plants and insects may be explained by the different experimental conditions (e.g. natural field infection and sampling times) used for testing FDp gene expression in grapevine. Phytoplasma loads in V. faba strongly increased between the first and second samplings, indicating active bacterial multiplication during which the transcription of the selected genes was comparable to that registered in insects. Conversely, during the successive stationary phase, the Expression Indexes were similar in the two host plants.

Genes within the ‘Unknown’ category were highly expressed at all sampling dates during infection of insects and contig12 was also the most expressed at the second sampling date during V. faba infection. In the absence of sequence/structural hints on its possible function, the active transcription of comp115 during insect infection, and its overexpression in E. variegatus compared to V. faba in the laboratory pathosystem suggest a specific role for its cognate protein during insect infection. Moreover, the gene was more actively transcribed during S. titanus infection, suggesting that the phytoplasma may react differently to the two leafhoppers. On the other hand, contig12, the second gene of the ‘Unknown’ category, was highly and stably transcribed during insect infection and during the exponential phytoplasma growth in V. faba, suggesting an essential role during the pathogen life cycle. The gene is the most abundant FDp transcript in infected grapevines collected from the field, a member of the “ORF-less” intron group II (Abbà et al. 2014), and it is present in two copies in the FDp genome (Debonneville et al. 2022). These are exceptionally successful mobile genetic elements (Novikova and Belfort 2017), and the intense transcription in FD-infected grapevines has suggested that it may contribute to the genomic plasticity required for host-adaptive strategies. In particular, high transcription of contig12 in insects may be required for rapid metabolic switch upon inoculation into the plant, as both vectors were already infective from the 1st sampling date onwards.

Two genes were included in the ‘Stress response’ category: osmC and comp83. The conserved bacterial protein encoded by osmC is an osmotically induced hydroperoxide peroxidase involved in the detoxification of endogenously and host derived oxidative radicals during host–pathogen interactions (Saikolappan et al. 2011). On the other hand, comp83 encodes a protein of unknown function that contains an S1-like cold-shock domain, typical of major cold-shock proteins CspA and CspB in bacteria. During insect infection, the two FDp genes were expressed differently, and comp83 was always more transcribed than osmC, while during plant infection both genes were similarly transcribed. Major cold-shock proteins may contribute to the expression regulation of virulence and flagella-associated genes in Listeria monocytogenes (Eshwar et al. 2017), and promote competitiveness at different life stages of Rhizobium leguminosarum (Wheatley et al. 2020). A possible role for Comp83 protein to ensure quick adaptation to a new host following inoculation, as reported for other bacterial major cold-shock proteins (Evdokimova 2022), is also supported by its over-expression in the insect partner of the FD laboratory pathosystem.

The high transcription level of the imp gene supports its importance during phytoplasma infection of the host, as already suggested by the presence of clear positive selection acting on orthologs of this gene in different phytoplasma species (Kakizawa et al. 2009; Siampour et al. 2013). Indeed, phytoplasma Imp proteins, being in direct contact with the host cell, interact with plant (Boonrod et al. 2012) and insect proteins (Trivellone et al. 2019a). In particular, Imp interaction with plant actin suggests its involvement in phytoplasma movement within and between plant cells in the ‘Ca. P. mali’ pathosystem (Boonrod et al. 2012), and the high expression of this gene in the plant hosts, particularly V. faba, supports a similar role also for FDp. Imp transcripts were also abundant in insects, especially in S. titanus, although its function is hard to predict, as no definitive hints have emerged so far on the FD Imp interactome in vectors (Trivellone et al. 2019b).

Three genes were included in the ‘Protein metabolism and secretion’ category: two metalloproteases (ysdC and tldD) and a GTPase (ftsY). An interaction with bacterial inner membrane proteins is reported for tldD (Babu et al. 2018), and both metalloproteases are important for bacterial pathogenesis, as they may degrade host immune proteins and release amino acids to feed the bacteria. In the case of Mycoplasma hyopneumoniae, an M42 metalloprotease is present at the cell surface, even in the absence of a recognizable signal peptide, with the double role of an adhesin able to degrade host extracellular matrix to facilitate infection and feed the mycoplasma, at the same time (Robinson et al. 2013). In V. faba, the ysdC transcription peak during the earliest two sampling dates may support high phytoplasma amino-acid requirement during the exponential growth phase inside the protein-depleted phloem niche. Indeed, both ysdC and tldD proteases showed a trend of over-transcription in the plant counterpart of the laboratory pathosystem. Nutritional requirements may also explain the high transcription of ysdC in S. titanus during its active growth. FtsY encodes the signal recognition particle GTPase required for the co-translational membrane targeting of proteins and the secretion through the secY translocation pathway. In Bacillus subtilis it is inhibited under nutritional stress conditions (Czech et al. 2022). During FDp infection, the gene showed a trend of overexpression in the insect counterpart of the laboratory system, in contrast with its homolog ‘Ca. P. asteris’ gene (Oshima et al. 2011).

Two genes were included in the category ‘Regulative elements’: spoVG and rpoD. In Lysteria monocytogenes SpoVG interacts with RNA both as a protein and at its 5’ untranslated moiety, a feature that suggests its activity as posttranscriptional regulator in the control of several aspects of bacterial physiology (Burke and Portnoy 2016), while in other bacteria it regulates the excretion of virulence factors (Schulthess et al. 2011). The rpoD gene is the primary sigma transcription factor during bacterial exponential growth, and in phytoplasmas it mediates the transcription of housekeeping genes as well as virulence and host interaction-related genes (Miura et al. 2015). Its role as a reference gene in transcriptomic studies of several bacteria (Gomes et al. 2018; Koch et al. 2019) has been questioned in the case of Helicobacter pylori (De la Cruz et al. 2017), Xanthomonas arboricola (Petriccione et al. 2021) under specific environmental conditions, and for E. coli under oxygen (Phue and Shiloach 2005) or osmotic stress conditions (Ionescu and Belkin 2009). During FDp infection of all hosts, both genes were stably expressed, suggesting their function to ensure the basal phytoplasma metabolism, and therefore they are useful reference genes for future transcriptomic studies of these uncultivable bacteria. In the FDp laboratory pathosystem, the overexpression trend in planta of both genes contrasted with the insect-specific expression of ‘Ca. P. asteris’ rpoD and its regulated genes (Ishii et al. 2013), indicating that different phytoplasmas may exploit alternative strategies to cope with the ‘host switch’ encountered during their life cycles. Moreover, further work has confirmed that rpoD is the principal sigma factor in both insect and plant hosts (Miura et al. 2015).

CoABC is the ATPase component of the ABC membrane transporter system for cobalt in association with the cobalamin (vitamin B12) biosynthetic pathways. This enzyme participates in ensuring cobalamin uptake from the host, it acts as a cellular antioxidant in Leptospirillum spp. (Ferrer et al. 2016), and it is required during Listeria monocytogenes growth under cold stress conditions (Vásquez et al. 2022). Flavescence dorée CoABC is stably expressed during infection of plant and insect hosts, as also supported by its similar expression profile in the two counterparts of the laboratory pathosystem. Conversely, its ‘Ca. P. asteris’ homolog, is upregulated in planta (Oshima et al. 2011).

In conclusion, among the analysed genes, contig12, CoABC, comp83, and imp were expressed, although at different levels, during FD infection in all or almost all samples of each host, and their functions may therefore be considered essential for phytoplasma infection cycle. As no specific functions have been associated to contig12 and comp83, we can only speculate that cobalamin homeostasis and imp-mediated FDp-host interactions seem indispensable for successful phytoplasma growth in the host. On the other hand, expression of comp115 seems to be required for insect infection, as the gene is always transcribed at high levels in all insect samples, and also significantly more transcribed in the insect counterpart of the FDp pathosystem. Also in this case, lack of functional knowledge for this gene, hampers further explanation of its importance. Further functional genomics studies are required to fully describe phytoplasma response to their different hosts, and to possibly obtain their in vitro cultivation.