Introduction

Phytoplasmas in the subgroup 16SrI-D of aster yellows are a group of microorganisms with no cell wall (Lee et al. 1993; Win et al. 2012). They range in size from 0.1 to 0.8 μm and have a 580–2200-kb circular DNA genome with a G + C content of less than 28 % (Hogenhout et al. 2008; Hoshi et al. 2007). Infected plants display a range of symptoms including witches’ broom, proliferation of axillary buds, and short internodes (Bayliss et al. 2005; Namba 2011).

Paulownia witches’ broom (PaWB) caused by a phytoplasma was first recognized in 1967 (Doi et al. 1967). Since then, the Paulownia-phytoplasma interaction has been of interest to researchers, but progress has been slow because phytoplasma are difficult to culture in vitro. The detection of a phytoplasma plasma-encoded protein and elongation factor EF-Tu (Lin et al. 2009; Wang et al. 2010), and information about the expression of host genes, such as those encoding plant hormones and proteins involved in cell wall biosynthesis and degradation, photosynthesis and carbohydrate metabolism, plant defense, plant-pathogen interaction, circadian rhythm, phytoplasma virulence, amino acid metabolism, and nitrogen metabolism have been reported so far (Cao et al. 2014a, b; Fan et al. 2014; Liu et al. 2013; Mou et al. 2013).

Paulownia tomentosa belongs to the family Paulowniaceae and is one of the fast growing trees native to China. Because it is a rich source of biologically active secondary metabolites, it is used in Chinese herbal medicine (Schneiderová and Šmejkal 2014). P. tomentosa has now been introduced into many other countries as a decorative ornamental tree (Erbar and Gülden 2011). To enrich the gene expression information associated with morphological changes in Paulownia plantlets after PaWB infection, here we investigated the expression patterns of genes that encode proteins involved in folate, fatty acid synthesis, and plant hormone signal transduction pathways. As reported previously, methyl methanesulfonate (MMS) at suitable concentrations could help Paulownia plants recover from the symptom of PaWB (Cao et al. 2014a). Hence, in this study, we performed a transcriptome analysis to determine the differences in gene expression in healthy P. tomentosa plantlets, diseased plantlets, and plantlets treated with 20 mg·L−1 MMS. The results will help in understand the mechanisms of Paulownia-phytoplasma interactions.

Materials and methods

Plant material treatment and RNA isolation

Healthy and diseased tissue culture plantlets of P. tomentosa were obtained from the Institute of Paulownia, Henan Agricultural University, Zhengzhou, China. The plantlets were cultured for 30 days on 1/2 MS medium (Murashige and Skoog 1962) before being clipped. The 1.5-cm-long terminal buds were clipped from the diseased plantlets and transplanted into 100-mL triangular flasks with 1/2 MS culture medium containing 0 and 20 mg·L−1 MMS. The same-sized terminal buds of the healthy plantlets were also transplanted into 1/2 MS medium without MMS, as the control. For each MMS concentration (0 and 20 mg·L−1) and for the control, three terminal buds were transplanted into one flask and a total of 20 flasks was used for each treatment. Each treatment was performed in triplicate.

The cultivation period for all these plantlets was 30 days for the MMS-treated plantlets, which was divided into two stages: first cultured at 20 °C in a darkroom for 5 days, and then transferred to a chamber at 25 ± 2 °C and 130 µmol·m−2 s−1 light intensity with a 16/8 h light/dark photoperiod for 25 days. The 1.5-cm long terminal buds were sheared from the healthy plantlets (PHP), diseased plantlets (DP), and the diseased plantlets treated with 20 mg·L−1 MMS (DP-20), and immediately frozen in liquid nitrogen and stored at −80 °C.

DNA isolation and phytoplasma detection

Total DNA was extracted from ten individual terminal buds of PHP, DP, and DP-20 using the cetyl trimethylammonium bromide (Beijing Chemical Co., Beijing, China) method as described by Zhang et al. (2009). The PaWB phytoplasma was detected by nested-PCR as described by (Lee et al. 1993). The PCR and agarose gel electrophoresis were performed according to the method of Fan et al. (2007).

RNA isolation and construction of cDNA libraries

Total RNA was extracted from ten individual terminal buds of PHP, DP, and DP-20 using TRIzol reagent (Invitrogen, Carlsbad, CA, USA) following the manufacturer’s protocol. The total RNA was treated with DNase I (RNase-free) to remove any contaminating DNA. The RNA quantity was assessed by the OD260/280 and OD260/230 ratios using a NanoDrop 2000 spectrophotometer (Thermo Scientific, Wilmington, DE). The integrity of the RNA was assessed by 1 % agarose gel electrophoresis. The high-quantity and high-integrity RNA was used to construct the cDNA libraries. Briefly, magnetic beads with oligo (dT) were used to isolate the poly (A)-containing mRNA from the total RNA. Then the mRNA was sheared into smaller fragments that were used as templates to synthesize the first-strand cDNA with SuperScript II reverse transcriptase (Life Technologies, Carlsbad, CA). Second-strand cDNA was synthesized using RNase H and DNA polymerase I. The resultant double-stranded cDNA was purified and washed with EB buffer for end reparation. Next, a single A (adenine) nucleotide was added to the 3′ ends and adapters were connected to the 5′ ends. The suitable fragments were used as templates for the PCR amplification. The PCR products were purified on agarose gel to obtain the final cDNA libraries, which were then quantified using an Agilent 2100 Bioanalyzer and ABI StepOnePlus Real-Time PCR System (ABI, New York, NY). The cDNA libraries were sequenced on an Illumina HiSeq™ 2000 platform (Illumina, San Diego, CA) by Beijing Genomics Institute (BGI) (Shenzhen, China) following the manufacturer’s standard cBot and sequencing protocols.

Unigene assembly and annotation

High-quality clean transcriptome sequencing reads were obtained after removing reads with adaptor sequences, more than 5 % unknown nucleotides, and low quality reads. The reads in the PHP, DP, and DP-20 libraries were assembled into all-unigene using the Trinity software (release-20121005) (Grabherr et al. 2011). The all-unigene sequences were aligned using BLASTX (E value < 1.0E-5) against the NCBI non-redundant (NR) protein database (release-20121005) (Altschul et al. 1997), the Swiss-Prot protein database (release-2012_08), the cluster of orthologous groups (COG) database (release-20090331), and the Kyoto Encyclopedia of Genes and Genomes (KEGG) Pathway database (Release 63.0). BLASTN (E value < 1.0E-5) was used to align the all-unigene sequences to the NCBI nucleotide (NT) database (release-20121005) (Xiang et al. 2010). All-unigene sequences that did not align to any of the above databases were processed using ESTScan (Iseli et al. 1999). To assign functions of the all-unigene that shared more than 70 % similarity with sequences in the NR database, the gene ontology (GO) (release 2012-10-01) functional annotations for the aligned NR sequences were retrieved using the WEGO software (Ye et al. 2006). The COG database was used to classify the assigned functional annotations of the all-unigene, and the KEGG Pathway database was used to determine the metabolic pathways that the all-unigene might be involved in.

Differential expression of genes among the three libraries

To assess the changes of gene expressions between DP vs. PHP and DP-20 vs. DP, the fragments per kilobase of transcript per million mapped reads (FPKM) method (Mortazavi et al. 2008) was used to calculate the expression levels as:

$${\text{FPKM}} = \frac{{10^{6} C}}{{{{NL} \mathord{\left/ {\vphantom {{NL} {10^{3} }}} \right. \kern-0pt} {10^{3} }}}}$$

where FPKM is the expression of unigene, C is the number of fragments that aligned uniquely to unigene, N is the total number of fragments that aligned uniquely to all-unigene, and L is the number of bases in the coding sequence of unigene.

The differentially expressed genes (DEGs) between any two samples according to Audic and Claverie (1997).

To detect the DEGs, the p value threshold in multiple hypothesis testing and analyses was determined by manipulating the false discovery rate (FDR) value (Benjamini and Yekutieli 2001). Two standards (|log2 Ratio| ≥ 1 and FDR ≤ 0.001) were used to judge the significance of gene expression differences. The KEGG enrichment analysis was performed by applying the hypergeometric test to find significantly enriched KEGG pathways among the DEGs compared with the whole transcriptome background of P. tomentosa. The Q value (similar to p value) was calculated as:

$$P = 1 - \sum\limits_{i = 0}^{m - 1} {\frac{{({}_{i}^{M} )(_{n - i}^{N - M} )}}{{(_{n}^{N} )}}}$$

where N is similar as the total number of genes with KEGG annotation; n is the number of DEGs in N; M is the total number of all genes that were annotated to a certain KEGG pathway; and m is the number of DEGs in M. After applying the Bonferroni correction to the calculated p value, we chose a corrected p value of ≤0.05 as the threshold, and KEGG pathways with q ≤ 0.05 were defined as significantly enriched KEGG pathways for the DEGs.

Quantitative real-time PCR (qRT-PCR) analysis

To validate the expression levels of the DEGs in the PHP, DP, and DP-20 libraries, 20 DEGs that were predicted to be involved in the Paulownia-phytoplasma interaction were chosen. First-strand cDNAs of all were synthesized using an iScript cDNA synthesis kit (Bio-Rad, Hercules, CA). The qRT-PCR primers were designed using the Beacon Designer 6.0 software (Premier Biosoft International, Palo Alto, CA). The cDNAs were then amplified on a Bio-Rad CFX96TM Real-Time System (Bio-Rad). The PCR cycles were as follows: 95 °C for 1 min, followed by 40 cycles of 95 °C for 10 s and 55 °C for 15 s. 18S rRNA served as the internal reference gene. The results were analyzed using the 2−ΔΔCt method (Livak and Schmittgen 2001). Each gene was analyzed in three replicates. All the primers used for the qRT-PCR are showed in Online Resource 1: Table S1. Statistical analyses were performed using SPASS 19.0 (SPASS, Inc., Chicago, IL).

Results

Detection of phytoplasma in P. tomentosa plantlets

The nest-PCR results showed that phytoplasma 16S rDNA fragments (1.2 kb) were detected in the DP and DP-20 libraries. Morphological changes were observed in DP-20 compared with DP (Fig. 1); that is, the color of the plantlets changed from light yellow to green, the sizes of the leaves changed from small turn to normal, the seta gradually appear again, and the short internodes became normal; however, very tiny axillary buds were found in the base of the DP-20 plantlets (Fig. 2). These observations indicated that the PaWB symptoms were attenuated with the 20 mg·L−1 MMS treatment.

Fig. 1
figure 1

Detection of phytoplasma in P. tomentosa diseased plantlets with MMS treatment. The phytoplasma 1.2 kb fragment was detected in the diseased plantlets, and the same weakened fragment was also detected in the diseased plantlets with 20 mg·L−1MMS treatment, while no fragment was detected in the healthy plantlets. 1 diseased plantlets. 2 diseased plantlets with 20 mg·L−1MMS treatment. 3 healthy plantlets. 4 double distilled water. M: DNA Marker

Full size image
Fig. 2
figure 2

Morphological changes of P. tomentosa diseased plantlets with MMS treatment. When the phytoplasma infected healthy plantlets, the plantlets showed the diseased morphology, but this morphology can attenuate with 20 mg·L−1MMS treatment. a healthy plantlets; b diseased plantlets; c diseased plantlets with 20 mg·L−1MMS treatment

Full size image

Transcriptome sequencing and de novo assembly

The transcriptome sequencing produced 48,006,640 (PHP), 75,244,384 (DP), and 45,932,476 (DP-20) clean reads. The N percentages were 0.08, 0.11, and 0.04 % in PHP, DP, and DP-20, respectively, and the GC content was 46.59, 46.55, and 45.64 %, respectively (Online Resource 1: Table S2). The clean reads were assembled into 126,285 (PHP), 134,791 (DP), and 147,094 (DP-20) contigs with average length of 368, 367, and 351 nt, respectively. The N50s n each of the contigs were 750 bp (PHP), 78 5 bp (DP), and 667 bp (DP-20). The contigs were further assembled into 77,844 (PHP), 79,456 (DP), and 90,969 (DP-20) unigene with mean lengths of 780, 848, and 866 nt, and N50s of 1513, 1618, and 1664 bp, respectively. Finally, a total of 85,545 unique all-unigene with a mean length of 1127 nt and N50 of 1833 bp were obtained from among the unigene from the three samples (Online Resource 1: Table S2). The length distributions of these all-unigene are shown in Online Resource 2: Figure S1.

Functional annotations of the all-unigene

To predict the functions of the all-unigene, we matched 85,545 all-unigene sequences first to the NR database, and then to the NT, Swiss-Prot, COG, GO, and KEGG databases (E value < 1.0E-5). We found that 60,328 (70.5 %) all-unigene matched known sequences in at least one of these databases (Table 1). Among them, 58,423 (68.3 %) all-unigene matched NR sequences, 52,631 (61.5 %) matched NT sequences, and 38,000 (44.4 %) matched Swiss-Prot sequences. In addition, 24,158 (28.2 %) all-unigene were mapped to the COG database, 48,326 (56.5 %) were assigned GO annotations, and 35,367 (41.3 %) were assigned to KEGG pathways. The remaining approximately 30 % of all-unigene that did not map to any of the databases may represent new genes that have not yet been identified in other plant species.

Table 1 Annotation of all-unigene of the transcriptome of P. tomentosa
Full size table

The distributions of the E values, sequence similarity, and species similarity of the all-unigene sequences that mapped to NR sequences are shown in Fig. 3. To increase the robustness of the annotations, the 24,158 all-unigene that mapped to the COG database were classified into 25 COG subgroups based on the COG classification (Fig. 4). Among them, the R subgroup “general function prediction only” was the largest group (8568; 35.5 %), followed by the K (4445; 18.4 %), L (4226; 17.5 %), and T (3545; 14.7 %) subgroups of “transcription”, “replication”, “recombination and repair”, and “signal transduction mechanisms”, respectively. The Y subgroup “nuclear structure” (5; 0.02 %) was the smallest subgroup (Online Resource 1: Table S3).

Fig. 3
figure 3

Nr annotation of all-unigene of P. tomentosa. Nr annotation of all-unigene includes e value distribution, similarity distribution and species distribution

Full size image
Fig. 4
figure 4

COG function classification of all-unigene of P. tomentosa. The all-unigene of P. tomentosa that mapped to the COG database were classified into 25 COG subgroups

Full size image

Based on the GO annotations, 48,326 all-unigene were assigned GO terms under the three main GO categories: biological process, molecular function, and cellular component. Among the GO terms under biological process, “cellular process” (31,520; 65.2 %) was assigned to the largest number of all-unigene (Online Resource 1: Table S4); under cellular component, “cell” (39,081; 80.9 %) and “cell part” (39,081; 80.9 %) were assigned to the largest number of all-unigene; and under molecular function, “catalytic activity” (23,261; 48.1 %) was assigned to the largest number of all-unigene. Based on the KEGG pathway annotations, 35,367 (41.3 %) all-unigene were predicted to be involved in 128 biological process; “metabolic pathways” (7890; 22.3 %), was the largest group, and “betalain biosynthesis” (2, 0.01 %) was the smallest group (Online Resource 1: Table S5), (Fig. 5).

Fig. 5
figure 5

GO classification of all-unigene of P. tomentosa GO classification main includes biological process, molecular function, and cellular component

Full size image

DEGs among the three P. tomentosa libraries

We compared the gene expression profiles in the DP versus PHP and DP-20 versus DP libraries to identify DEGs among the three samples (Online Resource 1: Table S6). A total of 2540 DEGs were detected; 2412 were down-regulated in DP versus PHP and up-regulated in DP-20 versus DP, and 128 were up-regulated in DP versus PHP and down-regulated in DP-20 versus DP (Fig. 6). The 2540 DEGs were assigned to 119 pathways in the KEGG database (Online Resource 1: Table S7). Among these DEGs, we identified several DEGs that may be related closely to PaWB disease, including genes encoding folylpolyglutamate synthase (FPGS) and bifunctional dihydrofolate reductase-thymidylate synthase (BDHFR-TS) that may be associated with the folate synthesis pathway. In addition, DEGs encoding acetyl-CoA carboxylase beta subunit (ACCase), beta-ketoacyl-ACP synthase II (KASII), 3-oxoacyl-[acyl-carrier-protein] reductase (KR), and acyl-ACP thioesterase (TE) that are related to the biosynthesis of fatty acid were altered after phytoplasma infection. DEGs associated with plant hormone signal transduction also were identified, namely genes encoding histidine kinase SHK278 (CRE1), histidine-containing phosphotransfer protein 5 (AHP), transcription repressor KAN1 (B-ARR), type-a response regulator (A-ARR), protein phosphatase (PP2C), serine/threonine-protein kinase SRK2B isoform 1 (SnRK2), ABA responsive element binding factor (ABF), interleukin-1 receptor -associated kinase 4 (BAK1), LRR receptor-like serine/threonine protein kinase (BRI1), serine/threonine protein kinase At4g35230 (BSK), serine/threonine protein kinase (BIN2), mid1- complementing activity 2 (CYCD3), and ubiquitin-protein ligase (TGA).

Fig. 6
figure 6

Differentially expressed genes’ analysis of P. tomentosa. A differentially expressed genes in diseased plantlets versus healthy plantlets up-regulated and down-regulated in diseased plantlets with 20 mg·L−1MMS treatment versus diseased plantlets. B differentially expressed genes in diseased plantlets versus healthy plantlets down-regulated and up-regulated in diseased plantlets with 20 mg·L−1MMS treatment versus diseased plantlets

Full size image

QRT-PCR analysis

To confirm the results of the transcriptome sequencing, 12 PaWB-related genes were selected randomly for qRT-PCR (Fig. 7). The results showed that the relative expressions of 11 of the 12 selected genes were significantly up-regulated in DP versus PHP and down-regulated in DP-20 versus DP, while one of the genes was significantly down-regulated in DP versus PHP and up-regulated in DP-20 versus DP. These expression patterns are similar to the expressions predicted from the transcriptome data analysis, indicating that the Illumina sequencing data were robust enough to be used to assess transcriptomic changes associated with the morphological changes that were observed in the PaWB plantlets.

Fig. 7
figure 7

qRT-PCR analysis of P. tomentosa selective DEGs. A relative expression of peroxidase 31 (Per31). B relative expression of metallothionein-like protein 1 (MetP1). C relative expression of carotenoid cleavage dioxygenase 4 (CarCD). D relative expression of polyphenol oxidase (PolO). E relative expression of multicopper oxidase. F relative expression of cryptochrome 2 (Cry2). G relative expression of dihydrofolate reductase (DihR). H relative expression of HSP90 (Hsp90). I relative expression of flavonoid 7-O-methyltransferase (FlaOM). J relative expression of caffeoyl-CoA O-methyltransferase (CCOMT). K relative expression of chalcone isomerase (ChaI). L relative expression of disease resistance protein 1-like protein b (DisRPb). DP-20 diseased plantlets with 20 mg·L−1 MMS treatment, PHP phytoplasma-free plantlets. The different letters within a gene repression level indicate significant difference, while the same letters within a gene repression level indicate no significant differences (p < 0.05)

Full size image

Discussion

Phytoplasmas, microbes without cell walls, have to exclusively inhabit nutrient-rich plant phloem tissues in their hosts. At the same time, the hosts launch a series of measures against phytoplasma infection. It has been reported that phytoplasma invasion induced callose deposition in the sieve plates of sieve cells in host plants, which can block the spread of the phytoplasma through the vascular system of plants (Hren et al. 2009). However, callose deposition will reduce the transport of sugar, phytohormones, ions, and metabolites and affect the host’s metabolism (Sugio et al. 2011). Indeed, it has been reported that phytoplasma infection modulated the membrane system, modified the cell wall, altered the cell cycle and cell division, and altered membrane ionic permeability. These changes initiated the synthesis and release of secondary messengers, changed the expression patterns of related stress-induced genes, influenced carbohydrate metabolism, and changed cell wall biosynthesis and degradation. The expression of genes such as BAK1, MEKK1, and WRKY29 have been associated with plant-pathogen interactions and cytokinin (CK) and abscisic acid (ABA) biosynthesis, and genes such as TOC1 and LHY/CCA1 have been reported to influence circadian rhythm (Fan et al. 2014; Liu et al. 2013; Mou et al. 2013). Phytoplasma infection was reported to repress the expression of several genes in the photosynthesis process, and genes related to defense and energy metabolism were found to be differentially expressed (Liu et al. 2013; Mou et al. 2013), indicating that the expression of these gene was related closely to the smaller leaves and shorter internode symptoms that have been observed in PaWB-infected plants.

To fully understand the molecular mechanisms behind the morphological changes that have been observed in Paulownia tomentosa plantlets after phytoplasma infection, high-throughput mRNA sequencing and de novo assembly were conducted to found the relation of expression levels of some genes involved in folate, fatty acid synthesis and signal transduction of plant hormones with the morphological changes of the plantlets infected with phytoplasma.

DEGs related to folate synthesis

Folate is an essential B vitamin that functions as a cofactor for enzymes in one-carbon metabolism in plants and plays an important role in the biosynthesis of nucleic acids, amino acids, pantothenate, and photorespiration (Anukul et al. 2010; Basset et al. 2005). In plants and insects, it has been reported that the infecting phytoplasma harbor genes (folk, folP, folC, and folA) that encode enzymes involved in folate biosynthesis (Oshima et al. 2013), thereby allowing the phytoplasma to adapt to different environments by regulating gene expression in their hosts (Oshima et al. 2004).

In the present study, genes encoding FPGS and BDHFR-TS were significantly up-regulated in DP versus PHP and down-regulated in DP-20 versus DP. These two enzymes are involved in the folate synthesis pathway. FPGS catalyzes the conversion of 7,8-dihydropteroate to 7,8-dihydrofolate (DHF), which can be converted to folate and 5,6,7,8-tetrahydrofolate (THF) by BDHFR-TS. In this process, THF-L-glutamate and THF-polyglutamate are also produced. In the phytoplasma, tetrahydrofolate was found to be synthesized by a combination of the four phytoplasma folate biosynthesis-related enzymes and FPGS (Rébeillé et al. 1997); therefore, we speculated that the high expression of an FPGS-encoding gene in the PaWB-infected P. tomentosa plantlets would provide a good source of folate for phytoplasma survival. Nevertheless, successful phytoplasma might consume a lot of folate, which might result in a lack of folate in their host and trigger the disease symptoms. Additionally, folate deficiency in the host could decrease the ratio of S-adenosylmethionine to S-adenosylhomocysteine and affect DNA methylation reactions (Balaghi and Wagner 1993). BDHFR-TS, which is involved in the biosynthesis of tetrahydrofolic acid, is the carrier of 1-carbon transfer reactions that are essential for DNA methylation (Maejima et al. 2014). Hence, the higher expression of FPGS and BDHFR-TS may be related to the changes in DNA methylation in Paulownia plantlets after phytoplasma infection, which is consistent with our previous results (Cao et al. 2014a, b; Zhai et al. 2010).

DEGs related to fatty acid synthesis

Fatty acids are the main building blocks for the phospholipid components of cell membranes and are determinants of intracellular communication (Prieschl and Baumruker 2000). Phytoplasma cannot synthesize fatty acids and, therefore, have to import them from their hosts (Bertaccini and Duduk 2010; Oshima et al. 2004). Phytoplasma infection was reported to modulate the plant cell membrane system by not only altering the membrane ionic (H+, Ca2+ and K+) permeability and changing the fatty acid composition, but also by releasing noxious reactive oxygen species (ROS), which is an important cell signal in phytoplasma-infected plants (Fan et al. 2014; Mou et al. 2013).

In our study, four genes involved in the fatty acid synthesis pathway were significantly up-regulated in DP versus PHP and down-regulated in DP-20 versus DP. In DP, the up-regulated DEGs associated with this pathway included genes encoding ACCase, KASII, KR and TE. In plants, ACCase is believed to be a major determinant of the overall rate of fatty acid synthesis. ACCase catalyzes the reaction between bicarbonate and acetyl-CoA to form malonyl-CoA, the precursor of malonyl-ACP, which is the carbon donor for the subsequent elongation reactions (Rismani-Yazdi et al. 2011), and also activated the expression of KASII and KR. KASII is believed to take part in the elongation of C16:0-ACP to C18:0-ACP, but KR catalyzes the reduction of this production of the elongation of the fatty acid (Slabas et al. 1992). Interestingly, this reduction reaction is reversible, the interaction of these two enzymes has been reported to further increase the synthesis of octanoyl-ACP, decanoyl-ACP, and hexa-deconoyl-ACP, the precursor substrates of octanoic acid, decanoic acid, and hexadecanoic acid, respectively. It has been reported that these substrates may be catalyzed by TE to break the fatty acid elongation cycle and release free fatty acids, such as octanoic acid, decanoic acid, and hexadecanoic acid, and induce the fatty acid degradative pathways (Zhang et al. 2011), which enhanced leakage of cytoplasmic solutes and damaged the plant membranes (McKersie et al. 1989). Moreover, because phytoplasma possess an enriched type III secretion system and ATP-binding cassette transporter system (Maejima et al. 2014; Oshima et al. 2004), the phytoplasma virulence factors may be transported easily through the damaged cell membranes, thereby enhancing the susceptibility of the plantlets to phytoplasma infection. At the same time, the gene encoding the cell membrane receptor, LRR receptor-like serine/threonine protein kinase (BRI1), was also up-regulated in DP. It has been reported that BRI1 can recognize a pathogen and trigger pathogen-associated molecular patterns that can bind the phytoplasma effector (EF-TU) and induce witches’ broom in infected plants (Schwessinger and Zipfel 2008; Wang et al. 2010). Hence, the formation of witches’ broom as a result of PaWB infection may be associated with changes in the expression levels of the genes that are involved in the release of free fatty acids.

DEGs related to plant hormone signal transduction

Plant hormones play important roles in regulating plant developmental processes and signaling networks in response to biotic and abiotic stresses (Bari and Jones 2009; Murmu et al. 2014). The disruption of hormonal balance has been related to abnormal morphology such as stunting and yellowing (Ćurković Perica 2008; Ehya et al. 2013; Leljak-Levanic et al. 2010). Hoshi et al. (2009) discovered that a phytoplasma virulence factor TENGU could induce proliferation and dwarfism by interfering with the expression of an auxin-related gene. Sugio et al. (2011) found phytoplasma infection reduced the production of jasmonic acid (JA), a hormone that plays fundamental roles in plant defense, which might decrease the hosts’ resistance and create an appropriate environment for phytoplasma survival. However, in a previous study, we identified genes involved in CK and ABA synthesis that were up-regulated in phytoplasma-infected Paulownia, whereas genes related to auxin were down-regulated. The expression of zeatin-related genes also was enhanced after phytoplasma infection (Fan et al. 2014; Liu et al. 2013; Mou et al. 2013). These observations suggested that changes in the plant hormones may play important roles in the formation of phytoplasma-mediated symptoms in the host.

In this study, we identified several genes in the multiple plant hormone signal transduction network that were markedly up-regulated in DP versus PHP and down-regulated in DP-20 versus DP. In DP, the up-regulated DEGs associated with this network included CRE1, AHP, B-ARR, A-ARR, PP2C, SnRK2, ABF, BAK1, BRI1, BSK, BIN2, CYCD3, and TGA.

Cytokinins are a class of plant hormones that influence cell proliferation, apical dominance, and leaf senescence (Ferreira and Kieber 2005). Alterations in the CK balance may help the pathogen to invade the stele region and vasculature (Moreau et al. 2014). Of the DEGs in the hormone signal transduction network, CRE1, AHP, B-ARR, and A-ARR may be involved in cytokinin signal transduction, and were differentially expressed in DP. CRE1 encodes a cytokinin receptor, a membrane histidine kinase that contains an extracellular sensing CHASE domain. CRE1 was reported to initiate the signal of phosphorylation, and regulate the osmoregulation, photosensitivity, and microbial pathogenesis (Bilwes et al. 1999; Inoue et al. 2001), thereby playing an important role in cell signal transduction. As reported in our previous studies (Cao et al. 2014a, b), CRE1 also may activate the expression of AHP. AHP encodes a histidine phosphotransfer protein that mediates the transfer of phosphoryl groups from the receptor kinases to the response regulators of nucleus, such as B-ARR and A-ARR (Ferreira and Kieber 2005). Over-expression of AHP can inhibit hypocotyl elongation of seedlings, showing the characteristic phenotype of cytokinin-hypersensitivity (Suzuki et al. 2002), and induce the expression of B-ARR and A-ARR in the cell nucleus. B-ARR usually implicated as DNA-binding transcription factors in the phosphorelay-mediated cytokinin signal transduction network (Ishida et al. 2008). A-ARR can negative regulation of CKs signaling, and have a role in regulating the homeostasis of the CK response (Gonzalez-Rizzo et al. 2006; Yokoyama et al. 2007), may be the production of large numbers of axillary shoots were implicated in the specific interaction between B-ARR and A-ARR.

Abscisic acid content has been positively related with the degree of disease, and high ABA content was found to promote the formation of witches’ broom (Fan et al. 2014; Hegele and Bangerth 1998; Liu et al. 2013; Mou et al. 2013). In this study, PP2C, SnRK2, and ABF were implicated in ABA signal transduction. These genes were up-regulated in the phytoplasma-infected Paulownia plantlets compared with their expression in DP-20 and PHP. PP2C was found to be an essential component of the ABA signaling pathway, and the over-expression of a PP2C isoform was reported to activate cell proliferation and expansion, especially in vascular tissues and guard cells (Sugimoto et al. 2014). PP2C also was shown to mediate the inactivation of SnRK2 via dephosphorylation (Umezawa et al. 2009). However, SnRK2 was not repressed by the over-expression of PP2C in the phytoplasma infected Paulownia plantlets, but it can directly phosphorylate ABF in response to activate ABA-responsive genes, which can induce stomatal pore closure (Kobayashi et al. 2005). Thus, the formation of witches’ broom may be related to the up-regulation of ABF associated with ABA signal transduction.

Brassinosteroid (BR) plays an important role in plant development. In this study, genes annotated as BAK1, BRI1, BSK, BIN2, and CYCD3 may be implicated in BR signal transduction. BRI1 and BAK1 are the main membrane receptor-like kinases in BR signaling pathway. BAK1, a member of the somatic embryogenesis receptor kinase family, has been described as a signal transducer that can recognize microbial pathogens and induce the expressions of defense-related genes (Uematsu et al. 2005). BRI1 usually acts as a cell recognition receptor that can trigger the recognition of pathogen-associated molecular patterns (Schwessinger and Zipfel 2008), which can bind the phytoplasma effector (EF-TU) and promote proliferation and dwarfism in the infected plants (Dodds and Rathjen 2010). Activation of BRI1 can also initiate the expressions of the BSK kinases, which is the substrates of BRI1 kinase, which can activate downstream BR signal transduction. However, not all of these genes are activated in such situations, for example, BSU1 was not activated by the up-regulation of BSK, while BIN2, one kinase that should be repressed by BSU1 showing up-regulation in the DP, and BIN2 has been reported to repress BR-responsive gene expression (Kim et al. 2011). Interestingly, BIN2 was found to indirectly target the host’s DNA by phosphorylation, and can induce the expression of CYCD3, which can induce cell division. Thus, the different expression of BAK1, BRI1 and CYCD3 might be related to the development of witches’ broom symptoms.

The present study showed that plant hormone changes may be one reason for the PaWB, combining the previous findings (Cao et al. 2014a, b; Fan et al. 2014; Liu et al. 2013; Mou et al. 2013). It can be speculated that the occurrence of PaWB is a complex process that may be regulated by multiple hormone signaling processes. Based on these results, this study suggested that the expression levels of CRE1, AHF, B-ARR, ABF, and CYCD3 may be closely related to PaWB. Further studies are required to elucidate the contributions of these genes to the morphological changes of Paulownia in response to phytoplasma.

Conclusions

The aim of the present study was to reveal the mechanisms behind phytoplasma pathogenicity based on a transcriptome sequencing analysis. A total of 2540 DEGs were obtained among the three P. tomentosa libraries. The functions of these DEGs were enriched in 119 KEGG pathways including folate and fatty acid biosynthesis and the plant hormone signal transduction pathway, which might play important roles in the occurrence of PaWB. By analysis of the functions of the DEGs in these pathways, we believed that folate deficiency, accumulation of free fatty acids, and disruption of plant hormones related to signal transduction might be closely related to the morphological changes of Paulownia plantlets in response to phytoplasma infection.

Author contribution statement

Conceived and designed the experiments: G.F. Performed the experiments: Z.Z. Analyzed the data: M.D. Contributed reagents/materials/analysis tools: X.C. Wrote the paper: G.F and X.C.