Background

Huanglongbing (HLB) and Tristeza are destructive and globally distributed citrus diseases, and are responsible for tremendous economic losses to citrus industries worldwide [8, 60]. HLB is strongly associated with ‘Ca. Liberibacter asiaticus’ (CaLas), a gram-negative and phloem-limited member of the α-Proteobacteria. CaLas has a global distribution and can be transmitted through grafting and by the citrus psyllid, Diaphorina citri [8]. Tristeza is caused by Citrus tristeza virus (CTV), a member of the Closteroviridae with a single-stranded and positive sense RNA-genome of 19.3 kb [16]. Like CaLas, CTV replicates in the phloem cells and may also be transmitted by grafting and by several aphids, most notably Toxoptera citricidus [60].

HLB has been studied for more than 100 years. Early observations of what is likely to have been HLB were made in the late 1880s by growers in Chaoshan, China and in the Pearl River Delta area of Guangdong in China. The earliest unambiguous report of HLB and association of the disease with the citrus psyllid was however made from India. Since CaLas has not been cultured in vitro, Koch’s postulates have not been completed. The first CaLas genomic sequence (Psy62) was obtained through multiple displacement amplification [22]. Genomes of additional six CaLas strains and three CaLam strains have been successively sequenced and are available at NCBI.

Considerable research has focused on host responses to infection. Transcriptome studies of host responses to HLB were conducted in leaves [24], fruits [47, 56], stems and roots [5, 92], as well as in resistant and susceptible varieties [3, 10, 26, 54]. Differentially expressed genes (DEGs) and altered metabolic pathways were described. Pathways related to sugar and starch metabolism, cell wall metabolism, secondary metabolism, hormone signaling, photosynthesis, phloem transport and ion uptake were found to be affected in citrus plants infected by CaLas.

Strains of CTV vary from very mild, which do not cause notable damage to citrus hosts, to severe strains that induce symptom patterns that include leaf yellowing, stiffening, and cupping. These symptoms are very similar to those associated with alterations in sucrose metabolism in the phloem and decline of roots [35, 41] and collapse of phloem tissue [29, 60] seen in HLB. Previously we have analyzed the transcriptome of sweet orange, Citrus sinensis, singly inoculated with CaLas-B232, mild CTV-B2 or severe CTV-B6 [30]. CTV-B2 does not induce visible symptoms in sweet orange while both CTV-B6 and CaLas-B232 are lethal. We found that the circadian rhythm system of C. sinensis and ionic balances were perturbed by all three pathogens but to differing degrees. Defense responses related to cell wall modification, regulation of transcription, hormone metabolism, secondary metabolism, kinases and both biotic and abiotic stress were activated by all three pathogens but with different patterns [30]. These results provided some valuable insight into the similarity and differences of the responses of C. sinensis to CaLas and CTV. Both CTV and CaLas are globally distributed, and are present simultaneously in most locations. Therefore, co-infection by strains of CTV and CaLas is now a common situation. In the current study, transcriptome profiles of C. sinensis co-infected with CTV-B2/CaLas-B232 (B2/B232) and CTV-B6/CaLas-B232 (B6/B232) were analyzed to identify differentially expressed gene (DEGs) and unveil how mild CTV-B2 and severe CTV-B6 affect the genome-wide gene expression changes of C. sinensis infected with CaLas-B232.

Results

40–45 million reads/sample were obtained and mapped to the C. sinensis reference genome [85], and about 60% of reads mapped successfully to the healthy, CTV-B2/CaLas-B232 and CTV-B6/CaLas-B232 sweet orange transcriptomes (Additional file 1: Figure S1). Differentially expressed transcripts (DETs) for CTV-B2/CaLas-B232 and CTV-B6/CaLas-B232 were identified by comparison with the transcriptome of the healthy controls. The number of up- or down regulated DETs varied for each co-infection (Additional file 2: Figure S2), and the overall alteration of the transcriptome was much greater in the presence of CTV-B6 than CTV-B2. We observed 402 and 420 transcripts induced and repressed respectively in response to CTV-B2/CaLas-B232, while 813 and 1009 transcripts were induced and repressed, respectively in response to CTV-B6/CaLas-B232. We compared the expression of the sweet orange genome in response to single infection and in combination after co-infection. We found that 29 transcripts were differentially regulated in response to CTV-B2 alone and in combination with CaLas-B232, while about twice that number were differentially regulated in response to CaLas-B232 alone and in combination with CTV-B2 (Fig. 1a and b). The number of transcripts that were co-regulated in response to CaLas-B232 alone and in combination with CTV-B6 was greater than the number that were co-regulated in response to CTV-B6 alone and in combination with CaLas-B232 (Fig. 1c and d).

Fig. 1
figure 1

Differentially expressed transcripts in sweet orange responding to single and co-infection by CTV-B2, CTV-B6 and CaLas-B232. a venn diagram of differentially expressed transcripts (DEGs) in sweet orange responding to CTV-B2 and CTV-B2/CaLas-B232; b venn diagram of DEGs in sweet orange responding to CaLas-B232 and CTV-B2/CaLas-B232; c venn diagram of DEGs in sweet orange responding to CTV-B6 and CTV-B6/CaLas-B232; d venn diagram of DEGs in sweet orange responding to CaLas-B232 and CTV-B26CaLas-B232

Full size image

The DETs were functionally placed in different Gene Ontology (GO) biological process (BP) categories. The largest numbers of genes induced by both CTV-B2/CaLas-B232 and/or CTV-B6/CaLas-B232 were associated with metabolic processes, followed by cellular processes, localization, biological regulation and responses to stimulus. Genes repressed by CTV-B2/CaLas-B232 and CTV-B6/CaLas-B232 also had a similar distribution within GO categories (Fig. 2). Many important metabolic pathways were heavily perturbed by CTV-B2/CaLas-B232, especially by CTV-B6/CaLas-B232 (Fig. 3a and c) in the young leaves that were sampled prior to the development of symptoms. These included metabolism of carbon compounds such as starch/sucrose and amino acids. Pathways related to photosynthesis, plant-pathogen interactions, hormone metabolism, signal transduction, circadian rhythm and ribosomal structure were also perturbed (Fig. 3a, c and Additional file 3: Tables S1 and Additional file 4: Table S2). The changes in gene expression are consistent with the symptoms that developed in mature leaves (Fig. 3b and d). Sixteen months after the inoculation of CaLas and 13 months the inoculation of CTV, the trees with the most and least severe symptoms (Fig. 4) were selected and compared to the self-inoculated healthy controls. Trees confirmed to be infected with both pathogens, either CTV-B2/CaLas-B232 or CTV-B6/CaLas-B232, were all smaller than the healthy controls. Trees infected with CTV-B6/CaLas-B232 (Fig. 4b) were also much weaker and showed much more severe symptoms, including chlorosis, stunting and leaf curling than plants infected with CTV-B2/CaLas-B232 (Fig. 4a).

Fig. 2
figure 2

Categorization of differentially expressed transcripts in Citrus sinensis infected with CTV-B2/CaLas-B232 or CTV-B6/CaLas-B232 based on gene ontology. a and c, Induced transcripts; b and d, Repressed transcripts; B2/232, CTV-B2/CaLas-B232; B6/232, CTV-B6/CaLas-B232

Full size image
Fig. 3
figure 3

Metabolic overview in young leaves and subsequent symptoms in response to CTV-B2/CaLas-B232 and CTV-B6/CaLas-B232. a and b, CTV-B2/CaLas-B232; c and d, CTV-B6/CaLas-B232

Full size image
Fig. 4
figure 4

General appearance of sweet orange co-infected with CTV-B2/CaLas-B232 or CTV-B6/CaLas-B232. Pictures were taken 16 months after inoculation with CaLas and 13 months after inoculation with CTV. a, Trees co-infected with CTV-B2 and CaLas-B232; b, Trees co-infected with CTV-B6 and CaLas-B232; 1 and 2, Most and least severe symptom expression for each treatment; 3, Self-inoculated healthy control

Full size image

Validation of RNA sequence data by RT-qPCR

Gene expression analyses were performed with RT-qPCR to verify RNA sequence data. In total, 60 genes involved in different biological processes were selected, including genes encoding transporters ZIP1, ZIP5, ZIP11, COPT1, PHT2;1 and OPT7, transcription factors (TF) WRKY33, WRKY46, WRKY70 and MYB77, hormone metabolism proteins LOX3, GA2OX8, TPS21, phloem protein PP2-B15, light reaction proteins LHCA4 and LHCB6, cytochrome P450 members CYP83B1 and CYP94B1, light signaling related ELIP1, lipid metabolism protein LP1, an F-box family protein, and CCT-motif family proteins related to regulation of transcription PCL1 and CBF4. Primer information is provided in Additional file 5: Table S3. Similar expression patterns were revealed by both techniques for these genes. Spearman’s rho values of 0.91 for CTV-B2/CaLas-B232 and 0.87 for CTV-B6/ CaLas-B232 (Fig. 5) and low variation within the replicated samples (Additional file 6: Figure S3) confirmed the reliability of RNA sequence data.

Fig. 5
figure 5

Correlation of estimates of differential gene expression by RT-qPCR and RNA-seq. FC: Log2 Fold Change

Full size image

Photosynthesis, carbohydrate and amino acid metabolism

We found that many DETs associated with the light reactions of photosynthesis, Calvin cycle and photorespiration were repressed by CTV-B2/CaLas-B232 and/or CTV-B6/CaLas-B232 (Fig. 6). In photosystem II these genes included chlorophyll binding proteins (LHB1B1, LHCA1 and LHCA2), light harvesting complex (LHCB3, LHCB5, LHCB6 and LHCA4) and calcium ion binding proteins (PPL1 and PSBQ). In photosystem I subunits (PSAD-1, PSAF, PSAO, PSAD-2, PSAH-1, PSAL, PSAK, PSAP and PSAO), rubisco activase (RCA) and plastid transcriptionally active14 (PTAC14) were repressed (Additional file 3: Table S1 and Additional file 4: Table S2). The repression of genes involved with the synthesis of tetrapyrrole (Additional file 7: Figure S4) and the light harvesting complex of the chloroplast is consistent with reduced capacity for photosynthesis, and is consistent with the symptoms that subsequently developed in the plants (Fig. 3). Symptoms observed following co-infection by CTV-B6 and CaLasB232 were similar to those produced by either pathogen alone (Additional file 8: Figure S5).

Fig. 6
figure 6

Photosynthetic system of Citrus sinensis perturbed by co-infection with CTV-B2/CaLas-B232 and CTV-B6/CaLas-B232

Full size image

Metabolism of amino acids including aspartate, proline, arginine, methionine, cysteine, asparagine, tyrosine and tryptophan was also disrupted by CTV-B2/CaLas-B232 and CTV-B6/CaLas-B232. Both nitrogen (N) and sulfur (S) assimilation are linked with amino acid metabolism. For example, SERAT1;1 (serine acetyltransferase1;1) and SERAT2;2 involved in cysteine synthesis were induced by both CTV-B2/CaLas-B232 and CTV-B6/CaLas-B232 (Additional file 3: Table S1 and Additional file 4: Table S2). Glutamate synthase (NADH) (GLT1), involved in N-metabolism, was repressed by both CTV-B2/CaLas-B232 and CTV-B6/CaLas-B232. Intriguingly, several genes associated with S-assimilation were exclusively activated by CTV-B6/CaLas-B232 rather than CTV-B2/CaLas-B232. These included APS1 (ATP sulfurylase 1), APS2 and APR1 (APS reductase 1). AKN1 (APS kinase) was repressed by CTV-B6/CaLas-B232 but was not affected by CTV-B2/CaLas-B232 (Additional file 3: Table S1 and Additional file 4: Table S2).

Cell wall

The primary cell wall of a land plant is composed of the major carbohydrates cellulose, hemicelluloses and pectin. Genes encoding pectin methyl esterase were generally repressed, whereas genes related to cellulose synthesis were induced by CTV-B6/CaLas-B232.

Transport and heavy metal ions

The transportation system of the host plants was heavily disturbed by CTV-B6/CaLas-B232 and CTV-B2/CaLas-B232. Genes encoding sugar transporters were up-regulated in young leaves in response to co-infection. Transportation of metals was impaired by CTV-B2/CaLas-B232 and/or CTV-B6/CaLas-B232 as inferred from the suppression of iron, copper and zinc transporter genes (Additional file 3: Table S1 and Additional file 4: Table S2). The expression of phloem protein PP2-B15 was highly up-regulated in response to CTV-B6/CaLas-B232 (Additional file 4: Table S2), but no alteration was observed in response to CTV-B2/CaLas-B232.

The circadian system

The plant circadian clock regulates expression of as much as one-third of the Arabidopsis genome [71], including carbohydrate metabolism and defense responses [33, 37, 69]. The clock is composed of interlocking transcription-translation feedback loops. These were greatly disrupted by CTV-B2/CaLas-B232 and CTV-B6/CaLas-B232 (Fig. 7).

Fig. 7
figure 7

Disturbances in the circadian system by CTV-B2/CaLas-B232 and CTV-B6/CaLas-B232. Red boxes, down- regulated genes; Green boxes, up-regulated genes

Full size image

Hormones and control of defense responses

The metabolism of abscisic acid, auxins, brassinosteroids, cytokinins, ethylene, gibberellins, jasmonate and salicylic acid were all perturbed by CTV-B6/CaLas-B232 and to a lesser extent by CTV-B2/CaLas-B232 (Table 1; Additional file 9: Figure S6). A large number of transcription factors (TF) were affected by the pathogens, including AP2/EREBP, basic helix-loop-helix (bHLH), C2C2 (Zn), MYB, WRKY, bZIP, PHOR1 and Pseudo ARR (PRR) (Additional file 3: Tables S1 and Additional file 4: Table S2). An overwhelming majority of genes regulated by AP2/EREBP (APETALA/ETHYLENE RESPONSIVE ELEMENT BINDING PROTEIN) were down-regulated in response to CTV-B2/CaLas-B232 and CTV-B6/CaLas-B232. These included ethylene-responsive element binding factor (ERF13 and ERF38) and the C-repeat-binding factors (CBF4, CBF1 and CBF2). MYB14, MYB73 and MYB77 were all repressed in response to both CTV-B2/CaLas-B232 and CTV-B6/CaLas-B232. Ethylene signaling and ethylene mediated defense responses were activated in response to CTV-B2/CaLas-B232 and CTV-B6/CaLas-B232 (Additional file 3: Table S1 and Additional file 4: Table S2), as well as to CTV-B2, CTV-B6 and CaLas-B232 separately.

Table 1 Responses of hormones to infection by CTV-B2, CTV-B6 and CaLas-B232
Full size table

Alteration of ribosomal composition

Ribosomal proteins play crucial roles in cell division, growth and metabolism in plants. Several members of ribosomal protein families were down-regulated in host plants in response to dual infection by CTV-B2/CaLas-B232 but many more by CTV-B6/CaLas-B232 (Table 2).

Table 2 Transcripts encoding ribosomal proteins were differentially regulated in Citrus sinensis in response to infection by CTV-B2/CaLas-B232 and CTV-B6/CaLas-B232
Full size table

Discussion

Differentially expressed genes and altered metabolic pathways of hosts provide a broad perspective of the host response to infection by CaLas [2, 54, 92] and by CTV [14, 49, 86]. We have compared changes in transcription in sweet orange plants in response to mild and severe strains of CTV and to CaLas-B232 when inoculated separately, which provided an overview of how susceptible hosts respond to CTV and CaLas [30]. Because CTV and CaLas are globally distributed, it is very common to have trees infected with both pathogens. Thus single pathogen studies, while essential, do not provide all of the needed data. In this study we have collected and analyzed transcriptome data from sweet orange plants co-infected with CTV-B2 and CaLas-B232 or CTV-B6 and CaLas-B232. In preliminary experiments we found that trees became PCR-positive for CTV much more quickly and more reliably than for CaLas (data not shown), so that simultaneous inoculations with the two pathogens was not effective. We therefore inoculated trees with CaLas-B232 first, and after the trees became positive for CaLas by qPCR (about 3 months after of inoculation) and we inoculated the trees with CTV-B2 or CTV-B6. In this way we observed the host response to infection by both pathogens compared with each pathogen separately [30]. The alteration of the transcriptome in response to co-infection by both pathogens was much larger than in response to infection by any single pathogen [30]. Some genes showed similar expression patterns in both circumstances but others showed different, and even opposite expression patterns (Additional file 3: Table S1 and Additional file 4: Table S2). Although all trees died eventually, trees co-infected with CTV-B2 and CaLas-B232 were slower to develop symptoms and survived longer than trees co-infected with CTV-B6 and CaLas-B232 (Fig. 4). Defense responses in trees may have been activated by CTV-B2 to provide limited protection against CaLas-B232, while CTV-B6 and CaLas-B232 acted synergistically to rapidly kill the trees.

Photosynthesis carbohydrate and amino acid metabolism

The most abundant tetrapyrrole in plants is chlorophyll, the pigment that harvests light for photosynthesis. Genes related to tetrapyrrole synthesis were repressed (Additional file 7: Figure S4) in young leaves prior to symptom development in response to CTV-B2/CaLas-B232 and especially by CTV-B6/CaLas-B232. These genes included CHLI2 (magnesium chelatase I2), FC1 (ferrochelatase 1) and HEME2 (uroporphyrinogen decarboxylase) (Additional file 3: Table S1 and Additional file 4: Table S2). These pathways were also disrupted by infection with CaLas alone [2, 14, 30, 55]. Expression of some photosynthesis-related genes was increased in response to CTV-B6 and CaLas-B232 alone but no changes were found in response to CTV-B2 [30]. This indicates that the reduced capacity for photosynthesis as the host responds to the two pathogens is primarily in response to CaLas.

The HLB syndrome also includes impaired carbohydrate metabolism [2, 54, 55, 91], and similar effects have been reported in sweet orange trees infected with Spiroplasma citri, another phloem-limited bacterial pathogen [9]. Sucrose is transported from source to sink organs and starch is the storage reserve in plants. Metabolic pathways related to energy generation from carbohydrates, including glycolysis, the oxidative pentose phosphate pathway (OPP) and the tricarboxylic acid cycle (TCA), were all repressed by both CTV-B2/CaLas-B232 and CTV-B6/CaLas-B232 (Additional file 3: Table S1 and Additional file 4: Table S2). The OPP is the main source of NADPH. Both the respiratory TCA pathway and glycolysis are essential to produce energy and precursors of amino acids and secondary metabolites. The repression of these processes may limit plant growth and successful defense response as plant defense is energy intensive [7].

Amino acids are also involved in plant-pathogen interactions [74, 89]. Amino acid transporters are involved in plant defense responses by alteration of amino acid concentrations in balancing N pools. Our data strongly indicated that not only the metabolism of amino acids but also their related transporters were increased in response to dual infection, particularly by CTV-B6/CaLas-B232 (Additional files 3: Table S1 and Additional file 4: Table S2). Amino acid permeases (AAPs) are important for phloem transport of amino acids. AAP6, AAP7 and ProT1 (proline transporter1) were induced by CTV-B2/CaLas-B232; AAP6 and CAT8 (cationic amino acid transporter 8) were induced while AAP3 was highly repressed by CTV-B6/CaLas-B232. AAP3 transports the basic amino acids lysine and arginine [28]. AAP6 was expressed primarily in sink tissues [68] but also in xylem parenchyma in Arabidopsis and AAP6 was predicted to transfer amino acids from the xylem to the phloem [39, 63]. Some studies have found that the abundance of amino acids is correlated positively with the performance and behavior of phloem feeding insects: higher levels of amino acids increased aphid performance in potato [43, 66]. Both CaLas and CTV are transmitted by phloem feeding insects, and CaLas has to obtain nutrients such as amino acids and enzymes from the host due to its greatly reduced genome [22, 38]. The altered expression of transporter genes may result in a traffic jam of carbohydrates, amino acids and other substances in the phloem. The availability and abundance of nutrients and the composition of the amino acids in phloem would be correlated with the growth of CaLas and favor feeding by the insect vectors.

Cell wall

CSLB04 and CSLE1 associated with cellulose synthesis in cell walls were up-regulated while COBL7 (COBRA-like 7), which is thought to be required for properly oriented cell expansion [72, 73] was down-regulated in response to CTV-B6/CaLas-B232. In our earlier study, COBL7 was strongly up-regulated by CTV-B2 [30]. Genes related to cell wall degradation were generally up-regulated in response to both CTV-B2/CaLas-B232 and CTV-B6/CaLas-B232. Genes encoding cell wall proteins, such as HRGP (hydroxyproline-rich glycoprotein), RGP2 (reversibly glycosylated polypeptide 2) and LRR (leucine-rich repeat family protein) were up-regulated in response to CTV-B2/CaLas-B232 and/or CTV-B6/CaLas-B232 (Additional file 3: Table S1 and Additional file 4: Table S2). Genes related to cell wall modification were differentially regulated in response to CTV-B2/CaLas-B232 and CTV-B6/CaLas-B232. XTR6 (xyloglucan endotransglycosylase) was down-regulated in response to both CTV-B2/CaLas-B232 and CTV-B6/CaLas-B232 but was up-regulated in response to CTV-B2, CTV-B6 and CaLas-B232 when inoculated singly [30]. Expansin genes, EXP1, EXP3, EXP4, EXPB5, EXPL1, etc., were also differentially regulated by CTV-B2/CaLas-B232 and/or CTV-B6/CaLas-B232. EXLB3 was especially up-regulated by CTV-B6/CaLas-B232 (Additional file 3: Table S1 and Additional file 4: Table S2).

Transport and heavy metal ions

Genes encoding sugar transporters were up-regulated in young leaves in response to co-infection, as well as to CTV-B6 and CaLas-B232 alone [30]. The phloem-located SUC/SUT1 (sucrose-proton symporter2) is responsible for loading sucrose into sieve tubes [21, 75] and was induced in response to CTV-B2/CaLas-B232. Potassium ions also play a crucial role in the transport of photo assimilates in the phloem [34]. Potassium channel SKOR (cyclic nucleotide binding) was up-regulated in response to CTV-B6/CaLas-B232 (Additional file 4: Table S2). In Arabidopsis the loss of the potassium channel AKT2/3 impacted the loading of sugar into the phloem via the regulation of the phloem osmotic potential and the activity of sucrose transporters [18]. mRNAs for ATPases, CNG channels, ABC and nitrate transporters have been identified in sieve element sap of Ricinus and barley [20, 21]. These were also differentially expressed in response to CTV-B2/CaLas-B232 and CTV-B6/CaLas-B232.

FER1 (Ferric iron binding), radical SAM domain-containing protein, PCS2 (phytochelatin synthase 2) were down-regulated in response to CTV-B2/CaLas-B232 and/or CTV-B6/CaLas-B232 (Additional file 3: Table S1 and Additional file 4: Table S2). The over-expression of ferric reduction oxidase genes FRO6, FRO7 and PCS2 was found in response to CaLas-B232 but not to CTV-B2 or to CTV-B6 in the previous study [30]. This is consistent with iron deficiency in young leaves. 80% of leaf iron is located in the chloroplasts [78]. A shortage of iron leads to reduction of iron-containing proteins and most markedly to decreased production of chlorophyll. The deficiency of chlorophyll results in yellowing symptoms on leaves as observed on leaves infected with CaLas or CTV-B6 and also lowers the capacity for photosynthesis directly [1].

FP6 (Farnesylated protein 6) was induced by CTV-B2/CaLas-B232. FP3 and FP6 also possess the heavy metal binding MXCXXC domain and are involved in Cu(II) and Zn(II) transport in Arabidopsis roots [31]. HMA5 (Heavy metal ATPase) also contains the MXCXXC domain and functions in copper (Cu(II)) detoxification [4]. The expression of these metal handling proteins is consistent with deficiency of metal ions, which may be exacerbated by the dramatic decline of fibrous roots after infection with CaLas [35, 41]. Symptoms of zinc-deficiency are often observed on new leaves following infection by CaLas [8, 76, 84].

Up-regulation of zinc and copper transporter genes was also found in response to CTV-B6 and CaLas-B232 when inoculated alone [30, 25, 54]. The expressions of ZIP1, ZIP2, ZIP4, ZIP5 and ZIP11 were all induced by CaLas-B232 and CTV-B6/CaLas-232, while only ZIP5 and ZIP11 were induced by CTV-B2/CaLas-B232. ZIP6 was also induced by CTV-B6/CaLas-B232 and the expression level of these ZIPs was much higher in response to CTV-B6/CaLas-B232 than to CTV-B2/CaLas-B232 or CaLas-B232 alone. Interestingly, the expression of ZIP1, ZIP4 and ZIP5 was repressed by CTV-B2. Both CaLas and the severe strain of CTV, B6, cause chlorosis and symptoms of zinc deficiency on leaves, phloem collapse, and decline of roots, but the mild strain of CTV-B2 does not cause any damage or symptoms on indicator plants.

P-proteins (Phloem-proteins) are components of phloem involved both in the long-distance movement of macromolecules and in signaling [44, 52]. Several phloem proteins PP2-B10, PP2-B11, PP2-B13 and PP2-B15 were over expressed in response to CTV-B6 or CaLas-B232 alone, but PP2-B13 and PP2-B15 were dramatically down-regulated by CTV-B2 [30]. However, no change in expression of PP2-B13 and PP2-B15 was found after co-infection with CTV-B2/CaLas-B232. PP2-B15 has been previously reported to be induced in hosts in response to CaLas [45, 47, 54]. The increased expression of PP2-B13 and PP2-B15 by CaLas appears to be offset by the interaction with CTV-B2. This is also solid evidence that Zn transporters and phloem proteins were connected in a co-expression network as suggested [67, 92]. These results may also support the notion that CTV-B6 acts synergistically with CaLas-B232 to disrupt phloem transport and further stimulate the expression of ZIPs and PP2. Patterns of gene expression activated by CTV-B2 may provide some beneficial effects on infection by CaLas-B232. In addition, induction of ZIP4, ZIP5 and PP2-B15 may serve as biomarkers for HLB disease.

The circadian system

The plant circadian clock regulates expression of as much as one-third of the Arabidopsis genome [71], including carbohydrate metabolism and defense responses [33, 37, 69]. The clock is composed of interlocking transcription-translation feedback loops. These were greatly disrupted by CTV-B2/CaLas-B232 and CTV-B6/CaLas-B232 (Fig. 7), as we observed in our previous study of the pathogens in single infections [30]. Of particular interest is GIGANTEA (GI), a key regulatory protein that is necessary to maintain the appropriate period and amplitude of cca1 and lhy gene expression in the outer feedback loop of the circadian system in Arabidopsis [64]. GI also functions in the sucrose-signaling network of Arabidopsis [15], which probably contributes to the perturbations of sucrose metabolism described above. GI is also involved in the regulatory pathway that determines wall in-growth to provide physical barriers to limit pathogen invasion [13, 23]. Deformation of phloem cell walls is observed in citrus in response to infection by CaLas [29] and stem pitting, caused by in-growths of the phloem cells is a primary symptom of CTV [16]. The expression of GI was increased in response to CTV-B2/CaLas-B232, as well as to CTV-B2, CTV-B6 and CaLas-B232 alone [30], but was repressed by CTV-B6/CaLas-B232 together. The repression of GI would limit the production of intracellular barriers and lead to a weakened defense against the invasion of CaLas-B232 and CTV-B6. Trees inoculated with CaLas-B232 and CTV-B6 together tended to die rapidly rather than linger in diseased states as when inoculated with CTV-B6 or CaLas-B232 alone. Other genes of the circadian system were also differentially regulated in response to CTV-B2/CaLas-B232 and/or CTV-B6/CaLas-B232. These included ELF3 (early flowering), PIF3 (phytochrome interacting factor), and COP1 (constitutive photo morphogenesis protein 1). ELF3 and COP1 control the stability of GI and therefore mediate photoperiodic flowering and circadian function [88]. Whether plants were infected with CaLas and CTV separately or together, the circadian system was heavily disturbed. Alterations of the circadian system may contribute to the generalized reprogramming of the transcriptome observed in response to these pathogens, and has been observed previously for CaLas but not for CTV [30, 47].

Hormones and control of defense responses

The central plant immune responses associated with jasmonic acid, salicylic acid and ethylene [65] interact with plant hormones, including gibberellins [17], abscisic acid [57], auxins [40, 62, 83], cytokinins [40] and brassinosteroids 61]. Pathways leading to production of these hormones were also affected by CaLas-B232 alone [30]. The brassinosteroid, cytokinin and gibberellic acid, ethylene, abscisic acid and salicylic acid associated pathways were affected by CTV-B6 but only the ethylene and SA related pathways were affected by CTV-B2 (Table 1). Lipoxygenase 2 (LOX2) associated with JA synthesis-degradation was highly induced by CaLas-B232 alone but only slightly induced by CTV-B2/CaLas-B232 and repressed by CTV-B6/CaLas-B232 (Additional file 4: Table S2).

The WRKY factors are important regulators of SA-dependent defense responses [82] and some of them have been demonstrated in SA-JA crosstalk. Genes encoding WRKY transcription factors were mostly down-regulated in response to CTV-B2/CaLas-B232 and/or CTV-B6/CaLas-B232 (Additional file 3: Tables S1 and Additional file 4: Table S2), but WRKY 50, WRKY60 and WRKY70 were induced by CTV-B6/CaLas-B232, as well as by CTV-B2, CTV-B6 and CTV-B232 alone [30]. WRKY70, WRKY11 and WRKY17 are involved in SA/JA crosstalk as negative regulators of disease resistance in Arabidopsis [42]. WRKY70 was induced by CTV-B6/CaLas-B232 together, and also by CTV-B2, CTV-B6 and CaLas-B232 alone [30, 54] and in fruit infected with CaLas [55]. These results imply that WRKY70 plays a pivotal role in defense responses to both CaLas and CTV. WRKY33 was induced as a key regulator in modulating hormonal and metabolic responses to Botrytis cinerea infection in Arabidopsis [6], but the expression of WRKY33 was strongly repressed by CTV-B2/CaLas-B232 and CTV-B6/CaLas-B232, while slightly induced by CaLas-B232. This is consistent with different responses to necrotrophic (B. cinerea) and biotrophic (CaLas) pathogens.

Ubiquitin ligases add ubiquitin to proteins to direct them to the proteasome for degradation. Expression of PUB22 (U-box 22) and PUB24 encoding ubiquitin ligases were sharply down-regulated by both CTV-B2/CaLas-B232 and CTV-B6/CaLas-B232 (Additional file 3: Table S1 and Additional file 4: Table S2). This suggests that the subset of proteins whose degradation is regulated by these ubiquitin ligases would accumulate to abnormally high levels in plants infected by CTV and CaLas.

ERF1 (Ethylene response factor 1) is a positive regulator mediating the jasmonate and ethylene signaling pathways [51] and many other ERF members also play important functions in defense responses in Arabidopsis [58]. The expression of ERF1 was repressed in response to both CTV-B2/CaLas-B232 and CTV-B6/CaLas-B232, whereas ERF1 was induced by CTV-B6 and CaLas-B232 and no change was found in response to CTV-B2 when the pathogens were inoculated individually. The gene that encodes jasmonate-ZIM-domain protein1 (JAZ1/TIF10A) was strongly down-regulated by CTV-B2/CaLas-B232 and CTV-B6/CaLas-B232 (Additional file 3: Table S1 and Additional file 4: Table S2) whereas was up-regulated in response to CaLas-B232 alone. JAZ proteins act as repressors of JA-signaling and interact with JIN1/MYC2 [12]. A reduction of JAZ protein releases transcription factor MYC2 and further stimulates the expression of JA-responsive genes [12, 77, 81]. The expression of JAZ1/TIFY10A may be inducible by auxin and crosstalk between auxin and JA signaling [36].

ABA is also an important hormone associated with plant defense responses. NCEDs (nine-cis-epoxycarotenoid dioxygenase 3) are involved in the biosynthesis of ABA and their up- or down-regulation is correlated with levels of ABA. The expression of NCED3, NCED4 and NCED5 was repressed by CTV-B2/CaLas-B232 and CTV-B6/CaLas-B232 together, while NCED3 and NCED4 were over expressed in response to CTV-B6 and CaLas-B232, respectively. ABA plays a central role in plant-pathogen interactions via crosstalk with JA-SA mediated signaling pathways [27].

Auxin related genes were differentially expressed in response to CTV-B2/CaLas-B232, CTV-B6/CaLas-B232 together, and to CaLas-B232 alone, but not by either strain of CTV alone. The auxin-responsive gene GH3.1 (Gretchen Hagen3) was highly repressed by both CTV-B2/CaLas-B232 and CTV-B6/CaLas-B232 (Additional file 3: Table S1 and Additional file 4: Table S2). GH3.5 regulates both SA and auxin signaling against infection by Pseudomonas [90]. Over expression of GH3.8 reinforced resistance of rice against Xanthomonas oryzae pv. oryzae (Xoo) and reduced levels of SA and JA and expression of genes responsive to SA and JA [19], demonstrating a link between the auxin responsive pathway and the SA-JA-ET signaling network in plants. GH3.1 and GH3.6 were strongly repressed by CTV-B2/CaLas-B232 and CTV-B6/CaLas-B232, consistent with down-regulation of auxin responsive genes in young leaves [53], but GH3.6 was slightly induced in response to CaLas-B232 alone [30]. The repression of auxin signaling may be part of the plant immune response and could potentially enhance resistance of plants to pathogens [62, 79].

Brassinosteroids (BRs) are a unique class of plant hormones involved in various developmental processes including cell division, elongation, flowering, senescence, fruit ripening and abiotic/biotic stress responses. The BRs pathways were up regulated by either strain of CTV in conjunction with CaLas, but not by any of the pathogens separately. BR increased rice resistance to Xanthomonas oryzae infection [61]. The titer of CaLas was reduced by treatment with BRs, providing a new insight for the control of HLB [11].

Plants often face multiple pests and pathogens in the natural environment and consequently they apply comprehensive regulatory mechanisms to activate defense responses against various pathogens. Signaling networks and crosstalk among plant hormone mediated defense pathways played more profound and complicated roles in response to co-infection with CTV-B2/CaLas-B232 and CTV-B6/CaLas-B232 than to single infection with CTV-B2, CTV-B6 or CaLas-B232. The expression patterns of some genes related to plant hormones, such as LOX2, WRKY33, ERF1, JAZ1 (Additional file 3: Table S1 and Additional file 4: Table S2), in response to dual pathogen infection were opposite to the response observed with single infections by the same pathogens.

The composition of ribosomal subunits was especially affected by co-infection of CaLas-B232 and CTV-B6, in contrast with infection by the same pathogens in single pathogen infections. The down-regulation of ribosomal biogenesis was also found in both roots and leaves from Fe-deficient Arabidopsis [70]. PSI and PSII are the largest sinks for Fe and majority of Fe is stored in chloroplasts. Decreased photosynthetic performances as well as tetrapyrrole biosynthesis are hallmarks of Fe deficiency in plants [70]. The repression of ribosomal biogenesis we observed could also result from Fe deficiency in the leaves we sampled, particularly when the hosts are infected with both CTV-B6 and CaLas together [70]. A failure to properly recalibrate cellular Fe homeostasis may be responsible for the decrease of ribosomal subunits. This repression is highly organ-specific, rather than general [70, 84].

Conclusion

Carbohydrate, amino acid and lipid metabolism were deeply reconfigured in response to CTV-B2/CaLas-B232 and CTV-B6/CaLas-B232, much more so than in response to infection by any of the pathogens alone. The photosynthetic, glycolytic, oxidative pentose phosphate and tricarboxylic acid pathways were all repressed by CTV-B2/CaLas-B232 and especially by CTV-B6/CaLas-B232. The repression of these energy generating processes limited plant growth and defense responses [7]. The circadian system was also strongly perturbed by both double and single pathogen infection. Defense pathways, including those linked to the hormones jasmonic acid, salicylic acid, gibberellic acid, abscisic acid, auxin, brassinosteroids and cytokinins were all perturbed by CTV-B6/CaLas-B232, and to a lesser extent by CTV-B2/CaLas-B232, CaLas-B232, CTV-B6 and CTV-B2. Transport systems were also heavily perturbed and many transporters were activated. ZIP1, ZIP4, ZIP5 and PP2-B15 showed very different expression patterns in response to double pathogen infection compared to single pathogen infection. CTV-B6 may cooperate with CaLas-B232 in weakening the plant through root decline and phloem blockage. Defense responses activated by the mild strain CTV-B2 may provide some beneficial effects against CaLas-B232. Although many host genes were expressed differently in response to double-infections as compared to single infections with the same pathogens, the symptoms of CaLas and CTV-B6 infection were produced. The interaction of the pathogens within the host may lead to a better or worse result for the host plant.

Methods

Inoculation of experimental trees with CTV and CaLas

CTV mild strain B2 (T30 genotype, Florida), severe strain B6 (Complex genotype, SY568, California) and CaLas strain B232 (Thailand) are maintained in planta as part of the Exotic Pathogens of Citrus Collection (EPCC) at the USDA-ARS Beltsville Agricultural Research Center (BARC) in Beltsville, MD. Experimental seedling trees were ‘Valencia’ sweet orange. CaLas-B232 was inoculated by bud grafting into two groups of 10 trees. Three months later, one group of the trees was inoculated with CTV-B2 and another group was inoculated with CTV-B6. Ten trees were mock-inoculated with healthy buds as controls.

RNA preparation and qPCR

DNA and RNA extractions and RT-PCR and qPCR testing for the presence of pathogens were performed as previously described [30]. Three months after the CTV inoculations, and therefore 6 months after the CaLas inoculations, trees that tested positive for both CTV-B2 and CaLas-B232 (CTV-B2/CaLas-B232) or both CTV-B6 and CaLas-B232 (CTV-B6/CaLas-B232) by PCR were selected for analysis. Three to five young, soft, not fully expanded leaves of uniform size and without symptoms were collected, frozen in liquid nitrogen and kept at −80 °C. RNA was extracted as described [30] and assessed for quantity and quality with a Qubit 2.0 Fluorometer (Invitrogen) and Bioanalyzer 2100 (Agilent Technologies, Santa Clara, CA). Nine RNA samples (RIN ≥ 7.5), comprised of three RNA replicates, each from individual trees for each treatment (CTV-B2/CaLas-B232-infected, CTV-B6/CaLas-B232-infected and healthy citrus) were sent to Otogenetics (Norcross, GA, USA) for paired-end sequencing with the Illumina HiSeq 2500 platform.

Statistical analyses of RNA sequence libraries

Libraries of paired-end reads (2 × 100 nucleotides) for the nine citrus RNA extracts were generated with the Illumina HiSeq2500. After removing low-quality reads and adaptor sequences from the raw reads, high-quality reads were obtained and mapped to the C. sinensis reference genome [85] with Bowtie [46]. Expression levels of transcripts (log2 fold change) were calculated with DESeq2 [53]. Sequence similarity comparisons of the sequences identified in the reference genome of C. sinensis were made with both Arabidopsis thaliana and another C. sinensis genome [32]. Differentially expressed transcripts (DETs) were filtered with cut-off values, Padj ≤0.1, |log2FC| ≥ 1 and e-value ≤ e−5. DETs were enriched by Panther [59] with corresponding Arabidopsis orthologs and further assigned to metabolic pathways by reference to the Kyoto Encyclopedia of Genes and Genomes (KEGG) and functionally classified by Mapman [80].

RT-qPCR analysis

A total of 4 μg of RNA was reverse transcribed for first-strand cDNA syntheses using GoTaq® 2-step RT-qPCR system (Promega, Madison, WI) according to the manufacturer’s instructions. cDNA was diluted three fold with 1 X TE and stored at −20 °C for use. Gene-specific primers (Additional file 5: Table S3) were designed with an online tool (Integrated DNA Technologies, Coralville, IA) with melting temperatures of 60 °C ± 5 °C. qPCR reactions were performed with GoTaq® qPCR Master Mix (Promega) in a Bio-Rad CFX96 system. Twenty microliters of reaction mixture was added to each well with three replicated plant samples for each primer pair for each sample. The reaction program was set at 95 °C for 3 min, and 40 cycles of 95 °C for 10 s, 60 °C for 30 s and 72 °C for 30 s. Melting curves were analyzed to ensure that a single product was amplified. Actin was used as the internal control. The Actin gene has been verified to be relatively stable within the large number of potential references genes [48, 87]; (Additional file 10: Figure S7). The 2-△△Ct method was applied for relative quantification of gene expression [50].